Merge pull request #455 from Ana06/v1-6-0

This release adds the capa explorer rule generator plugin for IDA Pro,
vivisect support for Python 3 and 12 new rules. We appreciate everyone
who opened issues, provided feedback, and contributed code and rules.
Thank you also to the vivisect development team (rakuy0, atlas0
fd00m) for the Python 3 support (v1.0.0) and the fixes for Python 2
(v0.2.1). This is the last capa release which supports Python 2. Next
release will be Python 3 only.

.github/dependabot.yml

@@ -0,0 +1,6 @@
+version: 2
+updates:
+  - package-ecosystem: "pip"
+    directory: "/"
+    schedule:
+      interval: "weekly"

.github/pyinstaller/hooks/hook-smda.py

@@ -0,0 +1,5 @@
+# Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
+import PyInstaller.utils.hooks
+
+# ref: https://groups.google.com/g/pyinstaller/c/amWi0-66uZI/m/miPoKfWjBAAJ
+binaries = PyInstaller.utils.hooks.collect_dynamic_libs("capstone")

.github/pyinstaller/hooks/hook-vivisect.py

@@ -13,3 +13,144 @@ from PyInstaller.utils.hooks import copy_metadata
 #
 # ref: https://github.com/pyinstaller/pyinstaller/issues/1713#issuecomment-162682084
 datas = copy_metadata("vivisect")
+
+excludedimports = [
+    # viv gui requires these heavy libraries,
+    # but viv as a library doesn't.
+    # they shouldn't be installed in our configuration,
+    # but we'll ensure they don't slip in here (such as on developers' systems).
+    "PyQt5",
+    "qt5",
+    "pyqtwebengine",
+    # the above are imported by these viv modules.
+    # so really, we'd want to exclude these submodules of viv.
+    # but i dont think this works.
+    "vqt",
+    "vdb.qt",
+    "envi.qt",
+    # unused by capa
+    "pyasn1",
+]
+
+hiddenimports = [
+    # vivisect does manual/runtime importing of its modules,
+    # so declare the things that could be imported here.
+    "vivisect",
+    "vivisect.analysis",
+    "vivisect.analysis.amd64",
+    "vivisect.analysis.amd64",
+    "vivisect.analysis.amd64.emulation",
+    "vivisect.analysis.amd64.golang",
+    "vivisect.analysis.crypto",
+    "vivisect.analysis.crypto",
+    "vivisect.analysis.crypto.constants",
+    "vivisect.analysis.elf",
+    "vivisect.analysis.elf",
+    "vivisect.analysis.elf.elfplt",
+    "vivisect.analysis.elf.libc_start_main",
+    "vivisect.analysis.generic",
+    "vivisect.analysis.generic",
+    "vivisect.analysis.generic.codeblocks",
+    "vivisect.analysis.generic.emucode",
+    "vivisect.analysis.generic.entrypoints",
+    "vivisect.analysis.generic.funcentries",
+    "vivisect.analysis.generic.impapi",
+    "vivisect.analysis.generic.mkpointers",
+    "vivisect.analysis.generic.pointers",
+    "vivisect.analysis.generic.pointertables",
+    "vivisect.analysis.generic.relocations",
+    "vivisect.analysis.generic.strconst",
+    "vivisect.analysis.generic.switchcase",
+    "vivisect.analysis.generic.thunks",
+    "vivisect.analysis.generic.noret",
+    "vivisect.analysis.i386",
+    "vivisect.analysis.i386",
+    "vivisect.analysis.i386.calling",
+    "vivisect.analysis.i386.golang",
+    "vivisect.analysis.i386.importcalls",
+    "vivisect.analysis.i386.instrhook",
+    "vivisect.analysis.i386.thunk_bx",
+    "vivisect.analysis.ms",
+    "vivisect.analysis.ms",
+    "vivisect.analysis.ms.hotpatch",
+    "vivisect.analysis.ms.localhints",
+    "vivisect.analysis.ms.msvc",
+    "vivisect.analysis.ms.msvcfunc",
+    "vivisect.analysis.ms.vftables",
+    "vivisect.analysis.pe",
+    "vivisect.impapi.posix.amd64",
+    "vivisect.impapi.posix.i386",
+    "vivisect.impapi.windows",
+    "vivisect.impapi.windows.amd64",
+    "vivisect.impapi.windows.i386",
+    "vivisect.impapi.winkern.i386",
+    "vivisect.impapi.winkern.amd64",
+    "vivisect.parsers.blob",
+    "vivisect.parsers.elf",
+    "vivisect.parsers.ihex",
+    "vivisect.parsers.macho",
+    "vivisect.parsers.pe",
+    "vivisect.storage",
+    "vivisect.storage.basicfile",
+    "vstruct.constants",
+    "vstruct.constants.ntstatus",
+    "vstruct.defs",
+    "vstruct.defs.arm7",
+    "vstruct.defs.bmp",
+    "vstruct.defs.dns",
+    "vstruct.defs.elf",
+    "vstruct.defs.gif",
+    "vstruct.defs.ihex",
+    "vstruct.defs.inet",
+    "vstruct.defs.java",
+    "vstruct.defs.kdcom",
+    "vstruct.defs.macho",
+    "vstruct.defs.macho.const",
+    "vstruct.defs.macho.fat",
+    "vstruct.defs.macho.loader",
+    "vstruct.defs.macho.stabs",
+    "vstruct.defs.minidump",
+    "vstruct.defs.pcap",
+    "vstruct.defs.pe",
+    "vstruct.defs.pptp",
+    "vstruct.defs.rar",
+    "vstruct.defs.swf",
+    "vstruct.defs.win32",
+    "vstruct.defs.windows",
+    "vstruct.defs.windows.win_5_1_i386",
+    "vstruct.defs.windows.win_5_1_i386.ntdll",
+    "vstruct.defs.windows.win_5_1_i386.ntoskrnl",
+    "vstruct.defs.windows.win_5_1_i386.win32k",
+    "vstruct.defs.windows.win_5_2_i386",
+    "vstruct.defs.windows.win_5_2_i386.ntdll",
+    "vstruct.defs.windows.win_5_2_i386.ntoskrnl",
+    "vstruct.defs.windows.win_5_2_i386.win32k",
+    "vstruct.defs.windows.win_6_1_amd64",
+    "vstruct.defs.windows.win_6_1_amd64.ntdll",
+    "vstruct.defs.windows.win_6_1_amd64.ntoskrnl",
+    "vstruct.defs.windows.win_6_1_amd64.win32k",
+    "vstruct.defs.windows.win_6_1_i386",
+    "vstruct.defs.windows.win_6_1_i386.ntdll",
+    "vstruct.defs.windows.win_6_1_i386.ntoskrnl",
+    "vstruct.defs.windows.win_6_1_i386.win32k",
+    "vstruct.defs.windows.win_6_1_wow64",
+    "vstruct.defs.windows.win_6_1_wow64.ntdll",
+    "vstruct.defs.windows.win_6_2_amd64",
+    "vstruct.defs.windows.win_6_2_amd64.ntdll",
+    "vstruct.defs.windows.win_6_2_amd64.ntoskrnl",
+    "vstruct.defs.windows.win_6_2_amd64.win32k",
+    "vstruct.defs.windows.win_6_2_i386",
+    "vstruct.defs.windows.win_6_2_i386.ntdll",
+    "vstruct.defs.windows.win_6_2_i386.ntoskrnl",
+    "vstruct.defs.windows.win_6_2_i386.win32k",
+    "vstruct.defs.windows.win_6_2_wow64",
+    "vstruct.defs.windows.win_6_2_wow64.ntdll",
+    "vstruct.defs.windows.win_6_3_amd64",
+    "vstruct.defs.windows.win_6_3_amd64.ntdll",
+    "vstruct.defs.windows.win_6_3_amd64.ntoskrnl",
+    "vstruct.defs.windows.win_6_3_i386",
+    "vstruct.defs.windows.win_6_3_i386.ntdll",
+    "vstruct.defs.windows.win_6_3_i386.ntoskrnl",
+    "vstruct.defs.windows.win_6_3_wow64",
+    "vstruct.defs.windows.win_6_3_wow64.ntdll",
+]

.github/pyinstaller/pyinstaller.spec

@@ -16,9 +16,10 @@ with open('./capa/version.py', 'wb') as f:
     #                 - commits since
     #                   g------- git hash fragment
     version = (subprocess.check_output(["git", "describe", "--always", "--tags", "--long"])
+               .decode("utf-8")
                .strip()
                .replace("tags/", ""))
-    f.write("__version__ = '%s'" % version)
+    f.write(("__version__ = '%s'" % version).encode("utf-8"))
 
 a = Analysis(
     # when invoking pyinstaller from the project root,
@@ -41,128 +42,6 @@ a = Analysis(
         # ref: https://stackoverflow.com/a/62278462/87207
         (os.path.dirname(wcwidth.__file__), 'wcwidth')
     ],
-    hiddenimports=[
-        # vivisect does manual/runtime importing of its modules,
-        # so declare the things that could be imported here.
-        "vivisect",
-        "vivisect.analysis",
-        "vivisect.analysis.amd64",
-        "vivisect.analysis.amd64",
-        "vivisect.analysis.amd64.emulation",
-        "vivisect.analysis.amd64.golang",
-        "vivisect.analysis.crypto",
-        "vivisect.analysis.crypto",
-        "vivisect.analysis.crypto.constants",
-        "vivisect.analysis.elf",
-        "vivisect.analysis.elf",
-        "vivisect.analysis.elf.elfplt",
-        "vivisect.analysis.elf.libc_start_main",
-        "vivisect.analysis.generic",
-        "vivisect.analysis.generic",
-        "vivisect.analysis.generic.codeblocks",
-        "vivisect.analysis.generic.emucode",
-        "vivisect.analysis.generic.entrypoints",
-        "vivisect.analysis.generic.funcentries",
-        "vivisect.analysis.generic.impapi",
-        "vivisect.analysis.generic.mkpointers",
-        "vivisect.analysis.generic.pointers",
-        "vivisect.analysis.generic.pointertables",
-        "vivisect.analysis.generic.relocations",
-        "vivisect.analysis.generic.strconst",
-        "vivisect.analysis.generic.switchcase",
-        "vivisect.analysis.generic.thunks",
-        "vivisect.analysis.i386",
-        "vivisect.analysis.i386",
-        "vivisect.analysis.i386.calling",
-        "vivisect.analysis.i386.golang",
-        "vivisect.analysis.i386.importcalls",
-        "vivisect.analysis.i386.instrhook",
-        "vivisect.analysis.i386.thunk_bx",
-        "vivisect.analysis.ms",
-        "vivisect.analysis.ms",
-        "vivisect.analysis.ms.hotpatch",
-        "vivisect.analysis.ms.localhints",
-        "vivisect.analysis.ms.msvc",
-        "vivisect.analysis.ms.msvcfunc",
-        "vivisect.analysis.ms.vftables",
-        "vivisect.analysis.pe",
-        "vivisect.impapi.posix.amd64",
-        "vivisect.impapi.posix.i386",
-        "vivisect.impapi.windows",
-        "vivisect.impapi.windows.amd64",
-        "vivisect.impapi.windows.i386",
-        "vivisect.impapi.winkern.i386",
-        "vivisect.impapi.winkern.amd64",
-        "vivisect.parsers.blob",
-        "vivisect.parsers.elf",
-        "vivisect.parsers.ihex",
-        "vivisect.parsers.macho",
-        "vivisect.parsers.pe",
-        "vivisect.parsers.utils",
-        "vivisect.storage",
-        "vivisect.storage.basicfile",
-        "vstruct.constants",
-        "vstruct.constants.ntstatus",
-        "vstruct.defs",
-        "vstruct.defs.arm7",
-        "vstruct.defs.bmp",
-        "vstruct.defs.dns",
-        "vstruct.defs.elf",
-        "vstruct.defs.gif",
-        "vstruct.defs.ihex",
-        "vstruct.defs.inet",
-        "vstruct.defs.java",
-        "vstruct.defs.kdcom",
-        "vstruct.defs.macho",
-        "vstruct.defs.macho.const",
-        "vstruct.defs.macho.fat",
-        "vstruct.defs.macho.loader",
-        "vstruct.defs.macho.stabs",
-        "vstruct.defs.minidump",
-        "vstruct.defs.pcap",
-        "vstruct.defs.pe",
-        "vstruct.defs.pptp",
-        "vstruct.defs.rar",
-        "vstruct.defs.swf",
-        "vstruct.defs.win32",
-        "vstruct.defs.windows",
-        "vstruct.defs.windows.win_5_1_i386",
-        "vstruct.defs.windows.win_5_1_i386.ntdll",
-        "vstruct.defs.windows.win_5_1_i386.ntoskrnl",
-        "vstruct.defs.windows.win_5_1_i386.win32k",
-        "vstruct.defs.windows.win_5_2_i386",
-        "vstruct.defs.windows.win_5_2_i386.ntdll",
-        "vstruct.defs.windows.win_5_2_i386.ntoskrnl",
-        "vstruct.defs.windows.win_5_2_i386.win32k",
-        "vstruct.defs.windows.win_6_1_amd64",
-        "vstruct.defs.windows.win_6_1_amd64.ntdll",
-        "vstruct.defs.windows.win_6_1_amd64.ntoskrnl",
-        "vstruct.defs.windows.win_6_1_amd64.win32k",
-        "vstruct.defs.windows.win_6_1_i386",
-        "vstruct.defs.windows.win_6_1_i386.ntdll",
-        "vstruct.defs.windows.win_6_1_i386.ntoskrnl",
-        "vstruct.defs.windows.win_6_1_i386.win32k",
-        "vstruct.defs.windows.win_6_1_wow64",
-        "vstruct.defs.windows.win_6_1_wow64.ntdll",
-        "vstruct.defs.windows.win_6_2_amd64",
-        "vstruct.defs.windows.win_6_2_amd64.ntdll",
-        "vstruct.defs.windows.win_6_2_amd64.ntoskrnl",
-        "vstruct.defs.windows.win_6_2_amd64.win32k",
-        "vstruct.defs.windows.win_6_2_i386",
-        "vstruct.defs.windows.win_6_2_i386.ntdll",
-        "vstruct.defs.windows.win_6_2_i386.ntoskrnl",
-        "vstruct.defs.windows.win_6_2_i386.win32k",
-        "vstruct.defs.windows.win_6_2_wow64",
-        "vstruct.defs.windows.win_6_2_wow64.ntdll",
-        "vstruct.defs.windows.win_6_3_amd64",
-        "vstruct.defs.windows.win_6_3_amd64.ntdll",
-        "vstruct.defs.windows.win_6_3_amd64.ntoskrnl",
-        "vstruct.defs.windows.win_6_3_i386",
-        "vstruct.defs.windows.win_6_3_i386.ntdll",
-        "vstruct.defs.windows.win_6_3_i386.ntoskrnl",
-        "vstruct.defs.windows.win_6_3_wow64",
-        "vstruct.defs.windows.win_6_3_wow64.ntdll",
-    ],
     # when invoking pyinstaller from the project root,
     # this gets run from the project root.
     hookspath=['.github/pyinstaller/hooks'],
@@ -180,6 +59,25 @@ a = Analysis(
         # since we don't spawn a notebook, we can safely remove these.
         "IPython",
         "ipywidgets",
+
+        # these are pulled in by networkx
+        # but we don't need to compute the strongly connected components.
+        "numpy",
+        "scipy",
+        "matplotlib",
+        "pandas",
+        "pytest",
+
+        # deps from viv that we don't use.
+        # this duplicates the entries in `hook-vivisect`,
+        # but works better this way.
+        "vqt",
+        "vdb.qt",
+        "envi.qt",
+        "PyQt5",
+        "qt5",
+        "pyqtwebengine",
+        "pyasn1"
     ])
 
 a.binaries = a.binaries - TOC([

.github/workflows/build.yml

@@ -2,7 +2,7 @@ name: build
 
 on:
   release:
-    types: [created, edited]
+    types: [edited, published]
 
 jobs:
   build:
@@ -26,13 +26,14 @@ jobs:
         uses: actions/checkout@v2
         with:
           submodules: true
-      - name: Set up Python 2.7
+      - name: Set up Python 3.9
         uses: actions/setup-python@v2
         with:
-          python-version: 2.7
+          python-version: 3.9
+      - if: matrix.os == 'ubuntu-latest'
+        run: sudo apt-get install -y libyaml-dev
       - name: Install PyInstaller
-        # pyinstaller 4 doesn't support Python 2.7
-        run: pip install 'pyinstaller==3.*'
+        run: pip install 'pyinstaller==4.2'
       - name: Install capa
         run: pip install -e .
       - name: Build standalone executable
@@ -65,7 +66,7 @@ jobs:
       - name: Set executable flag
         run: chmod +x ${{ matrix.artifact_name }}
       - name: Set zip name
-        run: echo ::set-env name=zip_name::capa-${GITHUB_REF#refs/tags/}-${{ matrix.asset_name }}.zip
+        run: echo "zip_name=capa-${GITHUB_REF#refs/tags/}-${{ matrix.asset_name }}.zip" >> $GITHUB_ENV
       - name: Zip ${{ matrix.artifact_name }} into ${{ env.zip_name }}
         run: zip ${{ env.zip_name }} ${{ matrix.artifact_name }}
       - name: Upload ${{ env.zip_name }} to GH Release
@@ -74,4 +75,3 @@ jobs:
           repo_token: ${{ secrets.GITHUB_TOKEN}}
           file: ${{ env.zip_name }}
           tag: ${{ github.ref }}
-

.github/workflows/tests.yml

@@ -45,13 +45,13 @@ jobs:
     runs-on: ubuntu-latest
     needs: [code_style, rule_linter]
     strategy:
+      fail-fast: false
       matrix:
         include:
           - python: 2.7
-          - python: 3.6
           - python: 3.7
           - python: 3.8
-          - python: '3.9.0-rc.1' # Python latest
+          - python: 3.9.1
     steps:
     - name: Checkout capa with submodules
       uses: actions/checkout@v2
@@ -61,6 +61,8 @@ jobs:
       uses: actions/setup-python@v2
       with:
         python-version: ${{ matrix.python }}
+    - name: Install pyyaml
+      run: sudo apt-get install -y libyaml-dev
     - name: Install capa
       run: pip install -e .[dev]
     - name: Run tests

CHANGELOG.md

@@ -1,5 +1,308 @@
 # Change Log
 
+## v1.6.0 (2021-03-09)
+
+This release adds the capa explorer rule generator plugin for IDA Pro, vivisect support for Python 3 and 12 new rules. We appreciate everyone who opened issues, provided feedback, and contributed code and rules. Thank you also to the vivisect development team (@rakuy0, @atlas0fd00m) for the Python 3 support (`vivisect==1.0.0`) and the fixes for Python 2 (`vivisect==0.2.1`).
+
+### Rule Generator IDA Plugin
+
+The capa explorer IDA plugin now helps you quickly build new capa rules using features extracted directly from your IDA database. Without leaving the plugin interface you can use the features extracted by capa explorer to develop and test new rules and save your work directly to your capa rules directory. To get started select the new `Rule Generator` tab, navigate to a function in the IDA `Disassembly` view, and click `Analyze`. For more information check out the capa explorer [readme](https://github.com/fireeye/capa/blob/master/capa/ida/plugin/README.md).
+
+![](doc/img/rulegen_expanded.png)
+
+### Upcoming changes
+
+**This is the last capa release that supports Python 2.** The next release will be v2.0 and will have breaking changes, including the removal of Python 2 support.
+
+If you have workflows that rely on the Python 2 version and need future maintenance, please reach out. We may be able to supply limited backports of key fixes and features.
+
+### New features
+
+- explorer: Add capa explorer rule generator plugin for IDA Pro. Now capa explorer helps you build new capa rules!  #426, #438, #439 @mike-hunhoff
+- python: Python 3 support in vivisect #421 @Ana06
+- main: Add backend option in Python 3 to select the backend to be used (either SMDA or vivisect) #421 @Ana06
+- python: Python 3 support in IDA #429, #437 @mike-hunhoff
+- ci: test pyinstaller CI #452 @williballenthin
+- scripts: enable multiple backends in `show-features.py` #429 @mike-hunhoff
+- scripts: add `scripts/vivisect-py2-vs-py3.sh`  to compare vivisect Python 2 vs 3 (can easily be modified to test run times and compare different versions) #421 @Ana06
+
+### New Rules (12)
+
+- patch process command line @re-fox @williballenthin (graduated from nursery)
+- compiled with dmd @re-fox
+- compiled with exe4j @johnk3r
+- compiled from Visual Basic @williballenthin
+- capture screenshot in Go @TcM1911
+- compiled with Nim @mike-hunhoff
+- linked against Go process enumeration library @TcM1911
+- linked against Go registry library @TcM1911
+- linked against Go WMI library @TcM1911
+- linked against Go static asset library @TcM1911
+- inspect load icon resource @mike-hunhoff
+- linked against XZip @mr-tz
+
+### Bug Fixes
+
+- ida: check for unmapped addresses when resolving data references #436 @mike-hunhoff
+
+### Changes
+
+- setup: vivisect v1.0.0 is the default backend for Python3 (it was SMDA before) #421 @Ana06
+- setup: bump vivisect to 0.2.1 #454 @mr-tz
+- linter: adding ntoskrnl, ntdll overlap lint #428 @mike-hunhoff
+- ci: use py3.9 and pyinstaller 4.2 to build standalone binaries #452 @williballenthin
+- scripts: remove old migration script #450 @williballenthin
+
+### Development
+
+- main: factor out common cli argument handling #450 @williballenthin
+
+### Raw diffs
+
+  - [capa v1.5.1...v1.6.0](https://github.com/fireeye/capa/compare/v1.5.1...v1.6.0)
+  - [capa-rules v1.5.1...v1.6.0](https://github.com/fireeye/capa-rules/compare/v1.5.1...v1.6.0)
+
+
+## v1.5.1 (2021-02-09)
+
+This release fixes the version number that we forgot to update for v1.5.0 (therefore, v1.5.0 was not published to pypi). It also includes 1 new rule and some rule improvements.
+
+### New Rules (1)
+
+- encrypt data using vest @re-fox
+
+### Raw diffs
+
+  - [capa v1.5.0...v1.5.1](https://github.com/fireeye/capa/compare/v1.5.1...v1.6.0)
+  - [capa-rules v1.5.0...v1.5.1](https://github.com/fireeye/capa-rules/compare/v1.5.1...v1.6.0)
+
+
+## v1.5.0 (2021-02-05)
+
+This release brings support for running capa under Python 3 via [SMDA](https://github.com/danielplohmann/smda), more thorough CI testing and linting, better extraction of strings and byte features, and 50 (!) new rules. We appreciate everyone who opened issues, provided feedback, and contributed code and rules. A special shout out to the following new project contributors:
+
+  - @johnk3r
+  - @doomedraven
+  - @stvemillertime
+  - @itreallynick
+  - @0x534a
+  
+@dzbeck also added [Malware Behavior Catalog](https://github.com/MBCProject/mbc-markdown) (MBC) and ATT&CK mappings for many rules.
+
+Download a standalone binary below and checkout the readme [here on GitHub](https://github.com/fireeye/capa/). Report issues on our [issue tracker](https://github.com/fireeye/capa/issues) and contribute new rules at [capa-rules](https://github.com/fireeye/capa-rules/).
+
+
+### New Features
+
+  - py3 support via SMDA #355 @danielplohmann @jcrussell
+  - scripts: example of using capa as a library #372, #380 @doomedraven
+  - ci: enable dependabot #373 @mr-tz
+  - ci: lint rules @mr-tz
+  - ci: lint rule format #401 @mr-tz
+  - freeze: add base address #391 @mr-tz
+  - json: meta: add base address #412 @mr-tz
+
+### New Rules (50)
+
+  - 64-bit execution via heavens gate @recvfrom
+  - contain anti-disasm techniques @mr-tz
+  - check for microsoft office emulation @re-fox
+  - check for windows sandbox via device @re-fox
+  - check for windows sandbox via dns suffix @re-fox
+  - check for windows sandbox via genuine state @re-fox
+  - check for windows sandbox via process name @re-fox
+  - check for windows sandbox via registry @re-fox
+  - capture microphone audio @re-fox
+  - capture public ip @re-fox
+  - get domain trust relationships @johnk3r
+  - check HTTP status code @mr-tz
+  - compiled with perl2exe @re-fox
+  - compiled with ps2exe @re-fox
+  - compiled with pyarmor @stvemillertime, @itreallynick
+  - validate payment card number using luhn algorithm @re-fox
+  - hash data using fnv @re-fox @mr-tz
+  - generate random numbers via WinAPI @mike-hunhoff @johnk3r
+  - enumerate files recursively @re-fox
+  - get file system object information @mike-hunhoff
+  - read virtual disk @re-fox
+  - register minifilter driver @mike-hunhoff
+  - start minifilter driver @mike-hunhoff
+  - enumerate gui resources @johnk3r
+  - simulate CTRL ALT DEL @mike-hunhoff
+  - hijack thread execution @0x534a
+  - inject dll @0x534a
+  - inject pe @0x534a
+  - create or open registry key @mike-hunhoff
+  - delete registry value @mike-hunhoff
+  - query or enumerate registry key @mike-hunhoff
+  - query or enumerate registry value @mike-hunhoff
+  - resume thread @0x534a
+  - suspend thread @0x534a
+  - allocate memory @0x534a
+  - allocate RW memory @0x534a
+  - contain pusha popa sequence @mr-tz
+  - create or open file @mike-hunhoff
+  - open process @0x534a
+  - open thread @0x534a
+  - get kernel32 base address @mr-tz
+  - get ntdll base address @mr-tz
+  - encrypt or decrypt data via BCrypt @mike-hunhoff
+  - generate random numbers using the Delphi LCG @williballenthin
+  - hash data via BCrypt @mike-hunhoff
+  - migrate process to active window station @williballenthin
+  - patch process command line @williballenthin
+  - resolve function by hash @williballenthin
+  - persist via Winlogon Helper DLL registry key @0x534a
+  - schedule task via command line @0x534a
+
+### Bug Fixes
+
+  - doc: pyinstaller build process @mr-tz
+  - ida: better bytes extraction #409 @mike-hunhoff
+  - viv: better unicode string extraction #364 @mike-hunhoff
+  - viv: better unicode string extraction #378 @mr-tz
+  - viv: more xor instructions #379 @mr-tz
+  - viv: decrease logging verbosity #381 @mr-tz
+  - rules: fix api description syntax #403 @mike-hunhoff
+  - main: disable progress background thread #410 @mike-hunhoff
+  
+### Changes
+
+  - rules: return lib rules for scopes #398 @mr-tz
+  
+### Raw diffs
+
+  - [capa v1.4.1...v1.5.0](https://github.com/fireeye/capa/compare/v1.4.1...v1.5.0)
+  - [capa-rules v1.4.0...v1.5.0](https://github.com/fireeye/capa-rules/compare/v1.4.0...v1.5.0)
+
+## v1.4.1 (2020-10-23)
+
+This release fixes an issue building capa on our CI server, which prevented us from building standalone binaries for v1.4.1.
+
+### Bug Fixes
+
+  - install VC dependencies for Python 2.7 during Windows build
+  
+### Raw diffs
+
+  - [capa v1.4.0...v1.4.1](https://github.com/fireeye/capa/compare/v1.4.0...v1.4.1)
+  - [capa-rules v1.4.0...v1.4.1](https://github.com/fireeye/capa-rules/compare/v1.4.0...v1.4.1)  
+
+## v1.4.0 (2020-10-23)
+
+This capa release includes changes to the rule parsing, enhanced feature extraction, various bug fixes, and improved capa scripts. Everyone should benefit from the improved functionality and performance. The community helped to add 69 new rules. We appreciate everyone who opened issues, provided feedback, and contributed code and rules. A special shout out to the following new project contributors:
+
+  - @mwilliams31
+  - @yt0ng
+
+@dzbeck added [Malware Behavior Catalog](https://github.com/MBCProject/mbc-markdown) (MBC) and ATT&CK mappings for 86 rules.
+
+Download a standalone binary below and checkout the readme [here on GitHub](https://github.com/fireeye/capa/). Report issues on our [issue tracker](https://github.com/fireeye/capa/issues) and contribute new rules at [capa-rules](https://github.com/fireeye/capa-rules/).
+
+### New features
+
+  - script that demonstrates bulk processing @williballenthin #307
+  - main: render MBC table @mr-tz #332
+  - ida backend: improve detection of APIs called via two or more chained thunks @mike-hunhoff #340
+  - viv backend: improve detection of APIs called via two or more chained thunks @mr-tz #341
+  - features: extract APIs called via jmp instruction @mr-tz #337
+
+### New rules
+
+  - clear the Windows event log @mike-hunhoff
+  - crash the Windows event logging service @mike-hunhoff
+  - packed with kkrunchy @re-fox
+  - packed with nspack @re-fox
+  - packed with pebundle @re-fox
+  - packed with pelocknt @re-fox
+  - packed with peshield @re-fox
+  - packed with petite @re-fox
+  - packed with rlpack @re-fox
+  - packed with upack @re-fox
+  - packed with y0da crypter @re-fox
+  - compiled with rust @re-fox
+  - compute adler32 checksum @mwilliams31
+  - encrypt-data-using-hc-128 @recvfrom
+  - manipulate console @williballenthin
+  - references logon banner @re-fox
+  - terminate process via fastfail @re-fox
+  - delete volume shadow copies @mr-tz
+  - authenticate HMAC @mr-tz
+  - compiled from EPL @williballenthin
+  - compiled with Go @williballenthin
+  - create Restart Manager session @mike-hunhoff
+  - decode data using Base64 via WinAPI @mike-hunhoff
+  - empty recycle bin quietly @mwilliams31
+  - enumerate network shares @mike-hunhoff
+  - hook routines via microsoft detours @williballenthin
+  - hooked by API Override @williballenthin
+  - impersonate user @mike-hunhoff
+  - the @williballenthin packer detection package, thanks to Hexacorn for the data, see https://www.hexacorn.com/blog/2016/12/15/pe-section-names-re-visited/
+    - packed with CCG
+    - packed with Crunch
+    - packed with Dragon Armor
+    - packed with enigma
+    - packed with Epack
+    - packed with MaskPE
+    - packed with MEW
+    - packed with Mpress
+    - packed with Neolite
+    - packed with PECompact
+    - packed with Pepack
+    - packed with Perplex
+    - packed with ProCrypt
+    - packed with RPCrypt
+    - packed with SeauSFX
+    - packed with Shrinker
+    - packed with Simple Pack
+    - packed with StarForce
+    - packed with SVKP
+    - packed with Themida
+    - packed with TSULoader
+    - packed with VProtect
+    - packed with WWPACK
+    - rebuilt by ImpRec
+    - packaged as a Pintool
+    - packaged as a CreateInstall installer
+    - packaged as a WinZip self-extracting archive
+  - reference 114DNS DNS server @williballenthin
+  - reference AliDNS DNS server @williballenthin
+  - reference Cloudflare DNS server @williballenthin
+  - reference Comodo Secure DNS server @williballenthin
+  - reference Google Public DNS server @williballenthin
+  - reference Hurricane Electric DNS server @williballenthin
+  - reference kornet DNS server @williballenthin
+  - reference L3 DNS server @williballenthin
+  - reference OpenDNS DNS server @williballenthin
+  - reference Quad9 DNS server @williballenthin
+  - reference Verisign DNS server @williballenthin
+  - run as service @mike-hunhoff
+  - schedule task via ITaskService @mike-hunhoff
+  - references DNS over HTTPS endpoints @yt0ng
+
+### Bug fixes
+
+  - ida plugin: fix tree-view exception @mike-hunhoff #315
+  - ida plugin: fix feature count @mike-hunhoff
+  - main: fix reported total rule count @williballenthin #325
+  - features: fix handling of API names with multiple periods @mike-hunhoff #329
+  - ida backend: find all byte sequences instead of only first @mike-hunhoff #335
+  - features: display 0 value @mr-tz #338
+  - ida backend: extract ordinal and name imports @mr-tz #343
+  - show-features: improvements and support within IDA @mr-tz #342
+  - main: sanity check MBC rendering @williballenthin
+  - main: handle sample path that contains non-ASCII characters @mr-tz #328
+
+### Changes
+
+  - rules: use yaml.CLoader for better performance @williballenthin #306
+  - rules: parse descriptions for statements @mr-tz #312
+
+### Raw diffs
+
+  - [capa v1.3.0...v1.4.0](https://github.com/fireeye/capa/compare/v1.3.0...v1.4.0)
+  - [capa-rules v1.3.0...v1.4.0](https://github.com/fireeye/capa-rules/compare/v1.3.0...v1.4.0)
+
 ## v1.3.0 (2020-09-14)
 
 This release brings newly updated mappings to the [Malware Behavior Catalog version 2.0](https://github.com/MBCProject/mbc-markdown), many enhancements to the IDA Pro plugin, [flare-capa on PyPI](https://pypi.org/project/flare-capa/), a bunch of bug fixes to improve feature extraction, and four new rules. We received contributions from ten reverse engineers, including seven new ones:

README.md

@@ -1,7 +1,7 @@
 ![capa](.github/logo.png)
 
 [![CI status](https://github.com/fireeye/capa/workflows/CI/badge.svg)](https://github.com/fireeye/capa/actions?query=workflow%3ACI+event%3Apush+branch%3Amaster)
-[![Number of rules](https://img.shields.io/badge/rules-345-blue.svg)](https://github.com/fireeye/capa-rules)
+[![Number of rules](https://img.shields.io/badge/rules-469-blue.svg)](https://github.com/fireeye/capa-rules)
 [![License](https://img.shields.io/badge/license-Apache--2.0-green.svg)](LICENSE.txt)
 
 capa detects capabilities in executable files.
@@ -146,11 +146,10 @@ rule:
 The [github.com/fireeye/capa-rules](https://github.com/fireeye/capa-rules) repository contains hundreds of standard library rules that are distributed with capa.
 Please learn to write rules and contribute new entries as you find interesting techniques in malware.
 
-If you use IDA Pro, then you use can use the [capa explorer IDA plugin](capa/ida/plugin/).
-capa explorer lets you quickly identify and navigate to interesting areas of a program and dissect capa rule matches at
-the assembly level.
+If you use IDA Pro, then you can use the [capa explorer plugin](capa/ida/plugin/).
+capa explorer lets you quickly identify and navigate to interesting areas of a program and help you build new capa rules out of the features extracted directly from your IDB.
 
-![capa + IDA Pro integration](doc/img/ida_plugin_intro.gif)
+![capa + IDA Pro integration](doc/img/explorer_expanded.png)
 
 # further information
 ## capa

capa/engine.py

@@ -6,7 +6,6 @@
 #  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 
-import sys
 import copy
 import collections
 

capa/features/__init__.py

@@ -16,6 +16,9 @@ import capa.engine
 logger = logging.getLogger(__name__)
 MAX_BYTES_FEATURE_SIZE = 0x100
 
+# thunks may be chained so we specify a delta to control the depth to which these chains are explored
+THUNK_CHAIN_DEPTH_DELTA = 5
+
 # identifiers for supported architectures names that tweak a feature
 # for example, offset/x32
 ARCH_X32 = "x32"
@@ -74,7 +77,7 @@ class Feature(object):
         return self.value
 
     def __str__(self):
-        if self.value:
+        if self.value is not None:
             if self.description:
                 return "%s(%s = %s)" % (self.name, self.get_value_str(), self.description)
             else:

capa/features/extractors/helpers.py

@@ -42,7 +42,9 @@ def is_ordinal(symbol):
     """
     is the given symbol an ordinal that is prefixed by "#"?
     """
-    return symbol[0] == "#"
+    if symbol:
+        return symbol[0] == "#"
+    return False
 
 
 def generate_symbols(dll, symbol):

capa/features/extractors/ida/file.py

@@ -37,11 +37,11 @@ def check_segment_for_pe(seg):
         )
         for i in range(256)
     ]
-    todo = [
-        (capa.features.extractors.ida.helpers.find_byte_sequence(seg.start_ea, seg.end_ea, mzx), mzx, pex, i)
-        for mzx, pex, i in mz_xor
-    ]
-    todo = [(off, mzx, pex, i) for (off, mzx, pex, i) in todo if off != idaapi.BADADDR]
+
+    todo = []
+    for (mzx, pex, i) in mz_xor:
+        for off in capa.features.extractors.ida.helpers.find_byte_sequence(seg.start_ea, seg.end_ea, mzx):
+            todo.append((off, mzx, pex, i))
 
     while len(todo):
         off, mzx, pex, i = todo.pop()
@@ -61,8 +61,7 @@ def check_segment_for_pe(seg):
         if idc.get_bytes(peoff, 2) == pex:
             yield (off, i)
 
-        nextres = capa.features.extractors.ida.helpers.find_byte_sequence(off + 1, seg.end_ea, mzx)
-        if nextres != -1:
+        for nextres in capa.features.extractors.ida.helpers.find_byte_sequence(off + 1, seg.end_ea, mzx):
             todo.append((nextres, mzx, pex, i))
 
 
@@ -96,7 +95,14 @@ def extract_file_import_names():
      - importname
     """
     for (ea, info) in capa.features.extractors.ida.helpers.get_file_imports().items():
-        if info[1]:
+        if info[1] and info[2]:
+            # e.g. in mimikatz: ('cabinet', 'FCIAddFile', 11L)
+            # extract by name here and by ordinal below
+            for name in capa.features.extractors.helpers.generate_symbols(info[0], info[1]):
+                yield Import(name), ea
+            dll = info[0]
+            symbol = "#%d" % (info[2])
+        elif info[1]:
             dll = info[0]
             symbol = info[1]
         elif info[2]:

capa/features/extractors/ida/helpers.py

@@ -16,17 +16,24 @@ import ida_bytes
 
 
 def find_byte_sequence(start, end, seq):
-    """find byte sequence
+    """yield all ea of a given byte sequence
 
     args:
         start: min virtual address
         end: max virtual address
-        seq: bytes to search e.g. b'\x01\x03'
+        seq: bytes to search e.g. b"\x01\x03"
     """
     if sys.version_info[0] >= 3:
-        return idaapi.find_binary(start, end, " ".join(["%02x" % b for b in seq]), 0, idaapi.SEARCH_DOWN)
+        seq = " ".join(["%02x" % b for b in seq])
     else:
-        return idaapi.find_binary(start, end, " ".join(["%02x" % ord(b) for b in seq]), 0, idaapi.SEARCH_DOWN)
+        seq = " ".join(["%02x" % ord(b) for b in seq])
+
+    while True:
+        ea = idaapi.find_binary(start, end, seq, 0, idaapi.SEARCH_DOWN)
+        if ea == idaapi.BADADDR:
+            break
+        start = ea + 1
+        yield ea
 
 
 def get_functions(start=None, end=None, skip_thunks=False, skip_libs=False):
@@ -159,6 +166,10 @@ def basic_block_size(bb):
 
 def read_bytes_at(ea, count):
     """ """
+    # check if byte has a value, see get_wide_byte doc
+    if not idc.is_loaded(ea):
+        return b""
+
     segm_end = idc.get_segm_end(ea)
     if ea + count > segm_end:
         return idc.get_bytes(ea, segm_end - ea)
@@ -340,6 +351,10 @@ def find_data_reference_from_insn(insn, max_depth=10):
             # break if circular reference
             break
 
+        if not idaapi.is_mapped(data_refs[0]):
+            # break if address is not mapped
+            break
+
         depth += 1
         if depth > max_depth:
             # break if max depth

capa/features/extractors/ida/insn.py

@@ -12,7 +12,15 @@ import idautils
 
 import capa.features.extractors.helpers
 import capa.features.extractors.ida.helpers
-from capa.features import ARCH_X32, ARCH_X64, MAX_BYTES_FEATURE_SIZE, Bytes, String, Characteristic
+from capa.features import (
+    ARCH_X32,
+    ARCH_X64,
+    MAX_BYTES_FEATURE_SIZE,
+    THUNK_CHAIN_DEPTH_DELTA,
+    Bytes,
+    String,
+    Characteristic,
+)
 from capa.features.insn import API, Number, Offset, Mnemonic
 
 # security cookie checks may perform non-zeroing XORs, these are expected within a certain
@@ -46,23 +54,34 @@ def get_imports(ctx):
 
 def check_for_api_call(ctx, insn):
     """ check instruction for API call """
-    if not idaapi.is_call_insn(insn):
+    if not insn.get_canon_mnem() in ("call", "jmp"):
         return
 
-    for ref in idautils.CodeRefsFrom(insn.ea, False):
+    info = ()
+    ref = insn.ea
+
+    # attempt to resolve API calls by following chained thunks to a reasonable depth
+    for _ in range(THUNK_CHAIN_DEPTH_DELTA):
+        # assume only one code/data ref when resolving "call" or "jmp"
+        try:
+            ref = tuple(idautils.CodeRefsFrom(ref, False))[0]
+        except IndexError:
+            try:
+                # thunks may be marked as data refs
+                ref = tuple(idautils.DataRefsFrom(ref))[0]
+            except IndexError:
+                break
+
         info = get_imports(ctx).get(ref, ())
         if info:
-            yield "%s.%s" % (info[0], info[1])
-        else:
-            f = idaapi.get_func(ref)
-            # check if call to thunk
-            # TODO: first instruction might not always be the thunk
-            if f and (f.flags & idaapi.FUNC_THUNK):
-                for thunk_ref in idautils.DataRefsFrom(ref):
-                    # TODO: always data ref for thunk??
-                    info = get_imports(ctx).get(thunk_ref, ())
-                    if info:
-                        yield "%s.%s" % (info[0], info[1])
+            break
+
+        f = idaapi.get_func(ref)
+        if not f or not (f.flags & idaapi.FUNC_THUNK):
+            break
+
+    if info:
+        yield "%s.%s" % (info[0], info[1])
 
 
 def extract_insn_api_features(f, bb, insn):
@@ -129,6 +148,9 @@ def extract_insn_bytes_features(f, bb, insn):
     example:
         push    offset iid_004118d4_IShellLinkA ; riid
     """
+    if idaapi.is_call_insn(insn):
+        return
+
     ref = capa.features.extractors.ida.helpers.find_data_reference_from_insn(insn)
     if ref != insn.ea:
         extracted_bytes = capa.features.extractors.ida.helpers.read_bytes_at(ref, MAX_BYTES_FEATURE_SIZE)
@@ -283,7 +305,7 @@ def extract_insn_nzxor_characteristic_features(f, bb, insn):
         bb (IDA BasicBlock)
         insn (IDA insn_t)
     """
-    if insn.itype != idaapi.NN_xor:
+    if insn.itype not in (idaapi.NN_xor, idaapi.NN_xorpd, idaapi.NN_xorps, idaapi.NN_pxor):
         return
     if capa.features.extractors.ida.helpers.is_operand_equal(insn.Op1, insn.Op2):
         return

capa/features/extractors/smda/__init__.py

@@ -0,0 +1,52 @@
+import sys
+import types
+
+from smda.common.SmdaReport import SmdaReport
+from smda.common.SmdaInstruction import SmdaInstruction
+
+import capa.features.extractors.smda.file
+import capa.features.extractors.smda.insn
+import capa.features.extractors.smda.function
+import capa.features.extractors.smda.basicblock
+from capa.main import UnsupportedRuntimeError
+from capa.features.extractors import FeatureExtractor
+
+
+class SmdaFeatureExtractor(FeatureExtractor):
+    def __init__(self, smda_report: SmdaReport, path):
+        super(SmdaFeatureExtractor, self).__init__()
+        if sys.version_info < (3, 0):
+            raise UnsupportedRuntimeError("SMDA should only be used with Python 3.")
+        self.smda_report = smda_report
+        self.path = path
+
+    def get_base_address(self):
+        return self.smda_report.base_addr
+
+    def extract_file_features(self):
+        for feature, va in capa.features.extractors.smda.file.extract_features(self.smda_report, self.path):
+            yield feature, va
+
+    def get_functions(self):
+        for function in self.smda_report.getFunctions():
+            yield function
+
+    def extract_function_features(self, f):
+        for feature, va in capa.features.extractors.smda.function.extract_features(f):
+            yield feature, va
+
+    def get_basic_blocks(self, f):
+        for bb in f.getBlocks():
+            yield bb
+
+    def extract_basic_block_features(self, f, bb):
+        for feature, va in capa.features.extractors.smda.basicblock.extract_features(f, bb):
+            yield feature, va
+
+    def get_instructions(self, f, bb):
+        for smda_ins in bb.getInstructions():
+            yield smda_ins
+
+    def extract_insn_features(self, f, bb, insn):
+        for feature, va in capa.features.extractors.smda.insn.extract_features(f, bb, insn):
+            yield feature, va

capa/features/extractors/smda/basicblock.py

@@ -0,0 +1,131 @@
+import sys
+import string
+import struct
+
+from capa.features import Characteristic
+from capa.features.basicblock import BasicBlock
+from capa.features.extractors.helpers import MIN_STACKSTRING_LEN
+
+
+def _bb_has_tight_loop(f, bb):
+    """
+    parse tight loops, true if last instruction in basic block branches to bb start
+    """
+    return bb.offset in f.blockrefs[bb.offset] if bb.offset in f.blockrefs else False
+
+
+def extract_bb_tight_loop(f, bb):
+    """ check basic block for tight loop indicators """
+    if _bb_has_tight_loop(f, bb):
+        yield Characteristic("tight loop"), bb.offset
+
+
+def _bb_has_stackstring(f, bb):
+    """
+    extract potential stackstring creation, using the following heuristics:
+      - basic block contains enough moves of constant bytes to the stack
+    """
+    count = 0
+    for instr in bb.getInstructions():
+        if is_mov_imm_to_stack(instr):
+            count += get_printable_len(instr.getDetailed())
+        if count > MIN_STACKSTRING_LEN:
+            return True
+    return False
+
+
+def get_operands(smda_ins):
+    return [o.strip() for o in smda_ins.operands.split(",")]
+
+
+def extract_stackstring(f, bb):
+    """ check basic block for stackstring indicators """
+    if _bb_has_stackstring(f, bb):
+        yield Characteristic("stack string"), bb.offset
+
+
+def is_mov_imm_to_stack(smda_ins):
+    """
+    Return if instruction moves immediate onto stack
+    """
+    if not smda_ins.mnemonic.startswith("mov"):
+        return False
+
+    try:
+        dst, src = get_operands(smda_ins)
+    except ValueError:
+        # not two operands
+        return False
+
+    try:
+        int(src, 16)
+    except ValueError:
+        return False
+
+    if not any(regname in dst for regname in ["ebp", "rbp", "esp", "rsp"]):
+        return False
+
+    return True
+
+
+def is_printable_ascii(chars):
+    return all(c < 127 and chr(c) in string.printable for c in chars)
+
+
+def is_printable_utf16le(chars):
+    if all(c == 0x00 for c in chars[1::2]):
+        return is_printable_ascii(chars[::2])
+
+
+def get_printable_len(instr):
+    """
+    Return string length if all operand bytes are ascii or utf16-le printable
+
+    Works on a capstone instruction
+    """
+    # should have exactly two operands for mov immediate
+    if len(instr.operands) != 2:
+        return 0
+
+    op_value = instr.operands[1].value.imm
+
+    if instr.imm_size == 1:
+        chars = struct.pack("<B", op_value & 0xFF)
+    elif instr.imm_size == 2:
+        chars = struct.pack("<H", op_value & 0xFFFF)
+    elif instr.imm_size == 4:
+        chars = struct.pack("<I", op_value & 0xFFFFFFFF)
+    elif instr.imm_size == 8:
+        chars = struct.pack("<Q", op_value & 0xFFFFFFFFFFFFFFFF)
+    else:
+        raise ValueError("Unhandled operand data type 0x%x." % instr.imm_size)
+
+    if is_printable_ascii(chars):
+        return instr.imm_size
+    if is_printable_utf16le(chars):
+        return instr.imm_size // 2
+
+    return 0
+
+
+def extract_features(f, bb):
+    """
+    extract features from the given basic block.
+
+    args:
+      f (smda.common.SmdaFunction): the function from which to extract features
+      bb (smda.common.SmdaBasicBlock): the basic block to process.
+
+    yields:
+      Feature, set[VA]: the features and their location found in this basic block.
+    """
+    yield BasicBlock(), bb.offset
+    for bb_handler in BASIC_BLOCK_HANDLERS:
+        for feature, va in bb_handler(f, bb):
+            yield feature, va
+
+
+BASIC_BLOCK_HANDLERS = (
+    extract_bb_tight_loop,
+    extract_stackstring,
+)

capa/features/extractors/smda/file.py

@@ -0,0 +1,139 @@
+import struct
+
+# if we have SMDA we definitely have lief
+import lief
+
+import capa.features.extractors.helpers
+import capa.features.extractors.strings
+from capa.features import String, Characteristic
+from capa.features.file import Export, Import, Section
+
+
+def carve(pbytes, offset=0):
+    """
+    Return a list of (offset, size, xor) tuples of embedded PEs
+
+    Based on the version from vivisect:
+    https://github.com/vivisect/vivisect/blob/7be4037b1cecc4551b397f840405a1fc606f9b53/PE/carve.py#L19
+    And its IDA adaptation:
+    capa/features/extractors/ida/file.py
+    """
+    mz_xor = [
+        (
+            capa.features.extractors.helpers.xor_static(b"MZ", i),
+            capa.features.extractors.helpers.xor_static(b"PE", i),
+            i,
+        )
+        for i in range(256)
+    ]
+
+    pblen = len(pbytes)
+    todo = [(pbytes.find(mzx, offset), mzx, pex, i) for mzx, pex, i in mz_xor]
+    todo = [(off, mzx, pex, i) for (off, mzx, pex, i) in todo if off != -1]
+
+    while len(todo):
+
+        off, mzx, pex, i = todo.pop()
+
+        # The MZ header has one field we will check
+        # e_lfanew is at 0x3c
+        e_lfanew = off + 0x3C
+        if pblen < (e_lfanew + 4):
+            continue
+
+        newoff = struct.unpack("<I", capa.features.extractors.helpers.xor_static(pbytes[e_lfanew : e_lfanew + 4], i))[0]
+
+        nextres = pbytes.find(mzx, off + 1)
+        if nextres != -1:
+            todo.append((nextres, mzx, pex, i))
+
+        peoff = off + newoff
+        if pblen < (peoff + 2):
+            continue
+
+        if pbytes[peoff : peoff + 2] == pex:
+            yield (off, i)
+
+
+def extract_file_embedded_pe(smda_report, file_path):
+    with open(file_path, "rb") as f:
+        fbytes = f.read()
+
+    for offset, i in carve(fbytes, 1):
+        yield Characteristic("embedded pe"), offset
+
+
+def extract_file_export_names(smda_report, file_path):
+    lief_binary = lief.parse(file_path)
+    if lief_binary is not None:
+        for function in lief_binary.exported_functions:
+            yield Export(function.name), function.address
+
+
+def extract_file_import_names(smda_report, file_path):
+    # extract import table info via LIEF
+    lief_binary = lief.parse(file_path)
+    if not isinstance(lief_binary, lief.PE.Binary):
+        return
+    for imported_library in lief_binary.imports:
+        library_name = imported_library.name.lower()
+        library_name = library_name[:-4] if library_name.endswith(".dll") else library_name
+        for func in imported_library.entries:
+            if func.name:
+                va = func.iat_address + smda_report.base_addr
+                for name in capa.features.extractors.helpers.generate_symbols(library_name, func.name):
+                    yield Import(name), va
+            elif func.is_ordinal:
+                for name in capa.features.extractors.helpers.generate_symbols(library_name, "#%s" % func.ordinal):
+                    yield Import(name), va
+
+
+def extract_file_section_names(smda_report, file_path):
+    lief_binary = lief.parse(file_path)
+    if not isinstance(lief_binary, lief.PE.Binary):
+        return
+    if lief_binary and lief_binary.sections:
+        base_address = lief_binary.optional_header.imagebase
+        for section in lief_binary.sections:
+            yield Section(section.name), base_address + section.virtual_address
+
+
+def extract_file_strings(smda_report, file_path):
+    """
+    extract ASCII and UTF-16 LE strings from file
+    """
+    with open(file_path, "rb") as f:
+        b = f.read()
+
+    for s in capa.features.extractors.strings.extract_ascii_strings(b):
+        yield String(s.s), s.offset
+
+    for s in capa.features.extractors.strings.extract_unicode_strings(b):
+        yield String(s.s), s.offset
+
+
+def extract_features(smda_report, file_path):
+    """
+    extract file features from given workspace
+
+    args:
+      smda_report (smda.common.SmdaReport): a SmdaReport
+      file_path: path to the input file
+
+    yields:
+      Tuple[Feature, VA]: a feature and its location.
+    """
+
+    for file_handler in FILE_HANDLERS:
+        result = file_handler(smda_report, file_path)
+        for feature, va in file_handler(smda_report, file_path):
+            yield feature, va
+
+
+FILE_HANDLERS = (
+    extract_file_embedded_pe,
+    extract_file_export_names,
+    extract_file_import_names,
+    extract_file_section_names,
+    extract_file_strings,
+)

capa/features/extractors/smda/function.py

@@ -0,0 +1,38 @@
+from capa.features import Characteristic
+from capa.features.extractors import loops
+
+
+def extract_function_calls_to(f):
+    for inref in f.inrefs:
+        yield Characteristic("calls to"), inref
+
+
+def extract_function_loop(f):
+    """
+    parse if a function has a loop
+    """
+    edges = []
+    for bb_from, bb_tos in f.blockrefs.items():
+        for bb_to in bb_tos:
+            edges.append((bb_from, bb_to))
+
+    if edges and loops.has_loop(edges):
+        yield Characteristic("loop"), f.offset
+
+
+def extract_features(f):
+    """
+    extract features from the given function.
+
+    args:
+      f (smda.common.SmdaFunction): the function from which to extract features
+
+    yields:
+      Feature, set[VA]: the features and their location found in this function.
+    """
+    for func_handler in FUNCTION_HANDLERS:
+        for feature, va in func_handler(f):
+            yield feature, va
+
+
+FUNCTION_HANDLERS = (extract_function_calls_to, extract_function_loop)

capa/features/extractors/smda/insn.py

@@ -0,0 +1,393 @@
+import re
+import string
+import struct
+
+from smda.common.SmdaReport import SmdaReport
+
+import capa.features.extractors.helpers
+from capa.features import (
+    ARCH_X32,
+    ARCH_X64,
+    MAX_BYTES_FEATURE_SIZE,
+    THUNK_CHAIN_DEPTH_DELTA,
+    Bytes,
+    String,
+    Characteristic,
+)
+from capa.features.insn import API, Number, Offset, Mnemonic
+
+# security cookie checks may perform non-zeroing XORs, these are expected within a certain
+# byte range within the first and returning basic blocks, this helps to reduce FP features
+SECURITY_COOKIE_BYTES_DELTA = 0x40
+PATTERN_HEXNUM = re.compile(r"[+\-] (?P<num>0x[a-fA-F0-9]+)")
+PATTERN_SINGLENUM = re.compile(r"[+\-] (?P<num>[0-9])")
+
+
+def get_arch(smda_report):
+    if smda_report.architecture == "intel":
+        if smda_report.bitness == 32:
+            return ARCH_X32
+        elif smda_report.bitness == 64:
+            return ARCH_X64
+    else:
+        raise NotImplementedError
+
+
+def extract_insn_api_features(f, bb, insn):
+    """parse API features from the given instruction."""
+    if insn.offset in f.apirefs:
+        api_entry = f.apirefs[insn.offset]
+        # reformat
+        dll_name, api_name = api_entry.split("!")
+        dll_name = dll_name.split(".")[0]
+        dll_name = dll_name.lower()
+        for name in capa.features.extractors.helpers.generate_symbols(dll_name, api_name):
+            yield API(name), insn.offset
+    elif insn.offset in f.outrefs:
+        current_function = f
+        current_instruction = insn
+        for index in range(THUNK_CHAIN_DEPTH_DELTA):
+            if current_function and len(current_function.outrefs[current_instruction.offset]) == 1:
+                target = current_function.outrefs[current_instruction.offset][0]
+                referenced_function = current_function.smda_report.getFunction(target)
+                if referenced_function:
+                    # TODO SMDA: implement this function for both jmp and call, checking if function has 1 instruction which refs an API
+                    if referenced_function.isApiThunk():
+                        api_entry = (
+                            referenced_function.apirefs[target] if target in referenced_function.apirefs else None
+                        )
+                        if api_entry:
+                            # reformat
+                            dll_name, api_name = api_entry.split("!")
+                            dll_name = dll_name.split(".")[0]
+                            dll_name = dll_name.lower()
+                            for name in capa.features.extractors.helpers.generate_symbols(dll_name, api_name):
+                                yield API(name), insn.offset
+                    elif referenced_function.num_instructions == 1 and referenced_function.num_outrefs == 1:
+                        current_function = referenced_function
+                        current_instruction = [i for i in referenced_function.getInstructions()][0]
+                else:
+                    return
+
+
+def extract_insn_number_features(f, bb, insn):
+    """parse number features from the given instruction."""
+    # example:
+    #
+    #     push    3136B0h         ; dwControlCode
+    operands = [o.strip() for o in insn.operands.split(",")]
+    if insn.mnemonic == "add" and operands[0] in ["esp", "rsp"]:
+        # skip things like:
+        #
+        #    .text:00401140                 call    sub_407E2B
+        #    .text:00401145                 add     esp, 0Ch
+        return
+    for operand in operands:
+        try:
+            yield Number(int(operand, 16)), insn.offset
+            yield Number(int(operand, 16), arch=get_arch(f.smda_report)), insn.offset
+        except:
+            continue
+
+
+def read_bytes(smda_report, va, num_bytes=None):
+    """
+    read up to MAX_BYTES_FEATURE_SIZE from the given address.
+    """
+
+    rva = va - smda_report.base_addr
+    if smda_report.buffer is None:
+        return
+    buffer_end = len(smda_report.buffer)
+    max_bytes = num_bytes if num_bytes is not None else MAX_BYTES_FEATURE_SIZE
+    if rva + max_bytes > buffer_end:
+        return smda_report.buffer[rva:]
+    else:
+        return smda_report.buffer[rva : rva + max_bytes]
+
+
+def derefs(smda_report, p):
+    """
+    recursively follow the given pointer, yielding the valid memory addresses along the way.
+    useful when you may have a pointer to string, or pointer to pointer to string, etc.
+
+    this is a "do what i mean" type of helper function.
+
+    based on the implementation in viv/insn.py
+    """
+    depth = 0
+    while True:
+        if not smda_report.isAddrWithinMemoryImage(p):
+            return
+        yield p
+
+        bytes_ = read_bytes(smda_report, p, num_bytes=4)
+        val = struct.unpack("I", bytes_)[0]
+
+        # sanity: pointer points to self
+        if val == p:
+            return
+
+        # sanity: avoid chains of pointers that are unreasonably deep
+        depth += 1
+        if depth > 10:
+            return
+
+        p = val
+
+
+def extract_insn_bytes_features(f, bb, insn):
+    """
+    parse byte sequence features from the given instruction.
+    example:
+        #     push    offset iid_004118d4_IShellLinkA ; riid
+    """
+    for data_ref in insn.getDataRefs():
+        for v in derefs(f.smda_report, data_ref):
+            bytes_read = read_bytes(f.smda_report, v)
+            if bytes_read is None:
+                continue
+            if capa.features.extractors.helpers.all_zeros(bytes_read):
+                continue
+
+            yield Bytes(bytes_read), insn.offset
+
+
+def detect_ascii_len(smda_report, offset):
+    if smda_report.buffer is None:
+        return 0
+    ascii_len = 0
+    rva = offset - smda_report.base_addr
+    char = smda_report.buffer[rva]
+    while char < 127 and chr(char) in string.printable:
+        ascii_len += 1
+        rva += 1
+        char = smda_report.buffer[rva]
+    if char == 0:
+        return ascii_len
+    return 0
+
+
+def detect_unicode_len(smda_report, offset):
+    if smda_report.buffer is None:
+        return 0
+    unicode_len = 0
+    rva = offset - smda_report.base_addr
+    char = smda_report.buffer[rva]
+    second_char = smda_report.buffer[rva + 1]
+    while char < 127 and chr(char) in string.printable and second_char == 0:
+        unicode_len += 2
+        rva += 2
+        char = smda_report.buffer[rva]
+        second_char = smda_report.buffer[rva + 1]
+    if char == 0 and second_char == 0:
+        return unicode_len
+    return 0
+
+
+def read_string(smda_report, offset):
+    alen = detect_ascii_len(smda_report, offset)
+    if alen > 1:
+        return read_bytes(smda_report, offset, alen).decode("utf-8")
+    ulen = detect_unicode_len(smda_report, offset)
+    if ulen > 2:
+        return read_bytes(smda_report, offset, ulen).decode("utf-16")
+
+
+def extract_insn_string_features(f, bb, insn):
+    """parse string features from the given instruction."""
+    # example:
+    #
+    #     push    offset aAcr     ; "ACR  > "
+    for data_ref in insn.getDataRefs():
+        for v in derefs(f.smda_report, data_ref):
+            string_read = read_string(f.smda_report, v)
+            if string_read:
+                yield String(string_read.rstrip("\x00")), insn.offset
+
+
+def extract_insn_offset_features(f, bb, insn):
+    """parse structure offset features from the given instruction."""
+    # examples:
+    #
+    #     mov eax, [esi + 4]
+    #     mov eax, [esi + ecx + 16384]
+    operands = [o.strip() for o in insn.operands.split(",")]
+    for operand in operands:
+        if not "ptr" in operand:
+            continue
+        if "esp" in operand or "ebp" in operand or "rbp" in operand:
+            continue
+        number = 0
+        number_hex = re.search(PATTERN_HEXNUM, operand)
+        number_int = re.search(PATTERN_SINGLENUM, operand)
+        if number_hex:
+            number = int(number_hex.group("num"), 16)
+            number = -1 * number if number_hex.group().startswith("-") else number
+        elif number_int:
+            number = int(number_int.group("num"))
+            number = -1 * number if number_int.group().startswith("-") else number
+        yield Offset(number), insn.offset
+        yield Offset(number, arch=get_arch(f.smda_report)), insn.offset
+
+
+def is_security_cookie(f, bb, insn):
+    """
+    check if an instruction is related to security cookie checks
+    """
+    # security cookie check should use SP or BP
+    operands = [o.strip() for o in insn.operands.split(",")]
+    if operands[1] not in ["esp", "ebp", "rsp", "rbp"]:
+        return False
+    for index, block in enumerate(f.getBlocks()):
+        # expect security cookie init in first basic block within first bytes (instructions)
+        block_instructions = [i for i in block.getInstructions()]
+        if index == 0 and insn.offset < (block_instructions[0].offset + SECURITY_COOKIE_BYTES_DELTA):
+            return True
+        # ... or within last bytes (instructions) before a return
+        if block_instructions[-1].mnemonic.startswith("ret") and insn.offset > (
+            block_instructions[-1].offset - SECURITY_COOKIE_BYTES_DELTA
+        ):
+            return True
+    return False
+
+
+def extract_insn_nzxor_characteristic_features(f, bb, insn):
+    """
+    parse non-zeroing XOR instruction from the given instruction.
+    ignore expected non-zeroing XORs, e.g. security cookies.
+    """
+
+    if insn.mnemonic not in ("xor", "xorpd", "xorps", "pxor"):
+        return
+
+    operands = [o.strip() for o in insn.operands.split(",")]
+    if operands[0] == operands[1]:
+        return
+
+    if is_security_cookie(f, bb, insn):
+        return
+
+    yield Characteristic("nzxor"), insn.offset
+
+
+def extract_insn_mnemonic_features(f, bb, insn):
+    """parse mnemonic features from the given instruction."""
+    yield Mnemonic(insn.mnemonic), insn.offset
+
+
+def extract_insn_peb_access_characteristic_features(f, bb, insn):
+    """
+    parse peb access from the given function. fs:[0x30] on x86, gs:[0x60] on x64
+    """
+
+    if insn.mnemonic not in ["push", "mov"]:
+        return
+
+    operands = [o.strip() for o in insn.operands.split(",")]
+    for operand in operands:
+        if "fs:" in operand and "0x30" in operand:
+            yield Characteristic("peb access"), insn.offset
+        elif "gs:" in operand and "0x60" in operand:
+            yield Characteristic("peb access"), insn.offset
+
+
+def extract_insn_segment_access_features(f, bb, insn):
+    """ parse the instruction for access to fs or gs """
+    operands = [o.strip() for o in insn.operands.split(",")]
+    for operand in operands:
+        if "fs:" in operand:
+            yield Characteristic("fs access"), insn.offset
+        elif "gs:" in operand:
+            yield Characteristic("gs access"), insn.offset
+
+
+def extract_insn_cross_section_cflow(f, bb, insn):
+    """
+    inspect the instruction for a CALL or JMP that crosses section boundaries.
+    """
+    if insn.mnemonic in ["call", "jmp"]:
+        if insn.offset in f.apirefs:
+            return
+
+        smda_report = insn.smda_function.smda_report
+        if insn.offset in f.outrefs:
+            for target in f.outrefs[insn.offset]:
+                if smda_report.getSection(insn.offset) != smda_report.getSection(target):
+                    yield Characteristic("cross section flow"), insn.offset
+        elif insn.operands.startswith("0x"):
+            target = int(insn.operands, 16)
+            if smda_report.getSection(insn.offset) != smda_report.getSection(target):
+                yield Characteristic("cross section flow"), insn.offset
+
+
+# this is a feature that's most relevant at the function scope,
+# however, its most efficient to extract at the instruction scope.
+def extract_function_calls_from(f, bb, insn):
+    if insn.mnemonic != "call":
+        return
+
+    if insn.offset in f.outrefs:
+        for outref in f.outrefs[insn.offset]:
+            yield Characteristic("calls from"), outref
+
+            if outref == f.offset:
+                # if we found a jump target and it's the function address
+                # mark as recursive
+                yield Characteristic("recursive call"), outref
+    if insn.offset in f.apirefs:
+        yield Characteristic("calls from"), insn.offset
+
+
+# this is a feature that's most relevant at the function or basic block scope,
+# however, its most efficient to extract at the instruction scope.
+def extract_function_indirect_call_characteristic_features(f, bb, insn):
+    """
+    extract indirect function call characteristic (e.g., call eax or call dword ptr [edx+4])
+    does not include calls like => call ds:dword_ABD4974
+    """
+    if insn.mnemonic != "call":
+        return
+    if insn.operands.startswith("0x"):
+        return False
+    if "qword ptr" in insn.operands and "rip" in insn.operands:
+        return False
+    if insn.operands.startswith("dword ptr [0x"):
+        return False
+    # call edx
+    # call dword ptr [eax+50h]
+    # call qword ptr [rsp+78h]
+    yield Characteristic("indirect call"), insn.offset
+
+
+def extract_features(f, bb, insn):
+    """
+    extract features from the given insn.
+
+    args:
+      f (smda.common.SmdaFunction): the function to process.
+      bb (smda.common.SmdaBasicBlock): the basic block to process.
+      insn (smda.common.SmdaInstruction): the instruction to process.
+
+    yields:
+      Feature, set[VA]: the features and their location found in this insn.
+    """
+    for insn_handler in INSTRUCTION_HANDLERS:
+        for feature, va in insn_handler(f, bb, insn):
+            yield feature, va
+
+
+INSTRUCTION_HANDLERS = (
+    extract_insn_api_features,
+    extract_insn_number_features,
+    extract_insn_string_features,
+    extract_insn_bytes_features,
+    extract_insn_offset_features,
+    extract_insn_nzxor_characteristic_features,
+    extract_insn_mnemonic_features,
+    extract_insn_peb_access_characteristic_features,
+    extract_insn_cross_section_cflow,
+    extract_insn_segment_access_features,
+    extract_function_calls_from,
+    extract_function_indirect_call_characteristic_features,
+)

capa/features/extractors/viv/__init__.py

@@ -8,11 +8,7 @@
 
 import types
 
-import file
-import insn
-import function
 import viv_utils
-import basicblock
 
 import capa.features.extractors
 import capa.features.extractors.viv.file
@@ -42,7 +38,7 @@ def add_va_int_cast(o):
     this bit of skullduggery lets use cast viv-utils objects as ints.
     the correct way of doing this is to update viv-utils (or subclass the objects here).
     """
-    setattr(o, "__int__", types.MethodType(get_va, o, type(o)))
+    setattr(o, "__int__", types.MethodType(get_va, o))
     return o
 
 

capa/features/extractors/viv/basicblock.py

@@ -125,11 +125,16 @@ def get_printable_len(oper):
 
 
 def is_printable_ascii(chars):
-    return all(ord(c) < 127 and c in string.printable for c in chars)
+    try:
+        chars_str = chars.decode("ascii")
+    except UnicodeDecodeError:
+        return False
+    else:
+        return all(c in string.printable for c in chars_str)
 
 
 def is_printable_utf16le(chars):
-    if all(c == "\x00" for c in chars[1::2]):
+    if all(c == b"\x00" for c in chars[1::2]):
         return is_printable_ascii(chars[::2])
 
 

capa/features/extractors/viv/helpers.py

@@ -0,0 +1,20 @@
+# Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at: [package root]/LICENSE.txt
+# Unless required by applicable law or agreed to in writing, software distributed under the License
+#  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and limitations under the License.
+from vivisect.const import XR_TO, REF_CODE
+
+
+def get_coderef_from(vw, va):
+    """
+    return first code `tova` whose origin is the specified va
+    return None if no code reference is found
+    """
+    xrefs = vw.getXrefsFrom(va, REF_CODE)
+    if len(xrefs) > 0:
+        return xrefs[0][XR_TO]
+    else:
+        return None

capa/features/extractors/viv/indirect_calls.py

@@ -132,7 +132,7 @@ def is_indirect_call(vw, va, insn=None):
     if insn is None:
         insn = vw.parseOpcode(va)
 
-    return insn.mnem == "call" and isinstance(insn.opers[0], envi.archs.i386.disasm.i386RegOper)
+    return insn.mnem in ("call", "jmp") and isinstance(insn.opers[0], envi.archs.i386.disasm.i386RegOper)
 
 
 def resolve_indirect_call(vw, va, insn=None):

capa/features/extractors/viv/insn.py

@@ -7,11 +7,19 @@
 # See the License for the specific language governing permissions and limitations under the License.
 
 import envi.memory
-import vivisect.const
 import envi.archs.i386.disasm
 
 import capa.features.extractors.helpers
-from capa.features import ARCH_X32, ARCH_X64, MAX_BYTES_FEATURE_SIZE, Bytes, String, Characteristic
+import capa.features.extractors.viv.helpers
+from capa.features import (
+    ARCH_X32,
+    ARCH_X64,
+    MAX_BYTES_FEATURE_SIZE,
+    THUNK_CHAIN_DEPTH_DELTA,
+    Bytes,
+    String,
+    Characteristic,
+)
 from capa.features.insn import API, Number, Offset, Mnemonic
 from capa.features.extractors.viv.indirect_calls import NotFoundError, resolve_indirect_call
 
@@ -67,9 +75,13 @@ def extract_insn_api_features(f, bb, insn):
     #
     #    call dword [0x00473038]
 
-    if insn.mnem != "call":
+    if insn.mnem not in ("call", "jmp"):
         return
 
+    if insn.mnem == "jmp":
+        if f.vw.getFunctionMeta(f.va, "Thunk"):
+            return
+
     # traditional call via IAT
     if isinstance(insn.opers[0], envi.archs.i386.disasm.i386ImmMemOper):
         oper = insn.opers[0]
@@ -86,21 +98,29 @@ def extract_insn_api_features(f, bb, insn):
     #
     # this is also how calls to internal functions may be decoded on x64.
     # see Lab21-01.exe_:0x140001178
-    elif isinstance(insn.opers[0], envi.archs.i386.disasm.i386PcRelOper):
-        target = insn.opers[0].getOperValue(insn)
+    #
+    # follow chained thunks, e.g. in 82bf6347acf15e5d883715dc289d8a2b at 0x14005E0FF in
+    # 0x140059342 (viv) / 0x14005E0C0 (IDA)
+    # 14005E0FF call    j_ElfClearEventLogFileW (14005AAF8)
+    #   14005AAF8 jmp     ElfClearEventLogFileW (14005E196)
+    #     14005E196 jmp     cs:__imp_ElfClearEventLogFileW
 
-        try:
-            thunk = f.vw.getFunctionMeta(target, "Thunk")
-        except vivisect.exc.InvalidFunction:
+    elif isinstance(insn.opers[0], envi.archs.i386.disasm.i386PcRelOper):
+        imports = get_imports(f.vw)
+        target = capa.features.extractors.viv.helpers.get_coderef_from(f.vw, insn.va)
+        if not target:
             return
-        else:
-            if thunk:
-                dll, _, symbol = thunk.rpartition(".")
-                if symbol.startswith("ord"):
-                    symbol = "#" + symbol[len("ord") :]
+
+        for _ in range(THUNK_CHAIN_DEPTH_DELTA):
+            if target in imports:
+                dll, symbol = imports[target]
                 for name in capa.features.extractors.helpers.generate_symbols(dll, symbol):
                     yield API(name), insn.va
 
+            target = capa.features.extractors.viv.helpers.get_coderef_from(f.vw, target)
+            if not target:
+                return
+
     # call via import on x64
     # see Lab21-01.exe_:0x14000118C
     elif isinstance(insn.opers[0], envi.archs.amd64.disasm.Amd64RipRelOper):
@@ -219,7 +239,7 @@ def read_bytes(vw, va):
     """
     segm = vw.getSegment(va)
     if not segm:
-        raise envi.SegmentationViolation()
+        raise envi.SegmentationViolation(va)
 
     segm_end = segm[0] + segm[1]
     try:
@@ -238,10 +258,10 @@ def extract_insn_bytes_features(f, bb, insn):
     example:
         #     push    offset iid_004118d4_IShellLinkA ; riid
     """
-    for oper in insn.opers:
-        if insn.mnem == "call":
-            continue
+    if insn.mnem == "call":
+        return
 
+    for oper in insn.opers:
         if isinstance(oper, envi.archs.i386.disasm.i386ImmOper):
             v = oper.getOperValue(oper)
         elif isinstance(oper, envi.archs.i386.disasm.i386RegMemOper):
@@ -291,6 +311,10 @@ def read_string(vw, offset):
                 # vivisect seems to mis-detect the end unicode strings
                 # off by one, too short
                 ulen += 1
+            else:
+                # vivisect seems to mis-detect the end unicode strings
+                # off by two, too short
+                ulen += 2
             return read_memory(vw, offset, ulen).decode("utf-16")
 
     raise ValueError("not a string", offset)
@@ -305,6 +329,9 @@ def extract_insn_string_features(f, bb, insn):
     for oper in insn.opers:
         if isinstance(oper, envi.archs.i386.disasm.i386ImmOper):
             v = oper.getOperValue(oper)
+        elif isinstance(oper, envi.archs.i386.disasm.i386ImmMemOper):
+            # like 0x10056CB4 in `lea eax, dword [0x10056CB4]`
+            v = oper.imm
         elif isinstance(oper, envi.archs.i386.disasm.i386SibOper):
             # like 0x401000 in `mov eax, 0x401000[2 * ebx]`
             v = oper.imm
@@ -395,7 +422,7 @@ def extract_insn_nzxor_characteristic_features(f, bb, insn):
     parse non-zeroing XOR instruction from the given instruction.
     ignore expected non-zeroing XORs, e.g. security cookies.
     """
-    if insn.mnem != "xor":
+    if insn.mnem not in ("xor", "xorpd", "xorps", "pxor"):
         return
 
     if insn.opers[0] == insn.opers[1]:
@@ -472,6 +499,10 @@ def extract_insn_cross_section_cflow(f, bb, insn):
     inspect the instruction for a CALL or JMP that crosses section boundaries.
     """
     for va, flags in insn.getBranches():
+        if va is None:
+            # va may be none for dynamic branches that haven't been resolved, such as `jmp eax`.
+            continue
+
         if flags & envi.BR_FALL:
             continue
 

capa/features/freeze.py

@@ -5,6 +5,7 @@ json format:
 
     {
       'version': 1,
+      'base address': int(base address),
       'functions': {
         int(function va): {
           'basic blocks': {
@@ -86,6 +87,7 @@ def dumps(extractor):
     """
     ret = {
         "version": 1,
+        "base address": extractor.get_base_address(),
         "functions": {},
         "scopes": {
             "file": [],
@@ -147,6 +149,7 @@ def loads(s):
         raise ValueError("unsupported freeze format version: %d" % (doc.get("version")))
 
     features = {
+        "base address": doc.get("base address"),
         "file features": [],
         "functions": {},
     }
@@ -261,6 +264,15 @@ def main(argv=None):
     parser.add_argument(
         "-f", "--format", choices=[f[0] for f in formats], default="auto", help="Select sample format, %s" % format_help
     )
+    if sys.version_info >= (3, 0):
+        parser.add_argument(
+            "-b",
+            "--backend",
+            type=str,
+            help="select the backend to use",
+            choices=(capa.main.BACKEND_VIV, capa.main.BACKEND_SMDA),
+            default=capa.main.BACKEND_VIV,
+        )
     args = parser.parse_args(args=argv)
 
     if args.quiet:
@@ -273,7 +285,8 @@ def main(argv=None):
         logging.basicConfig(level=logging.INFO)
         logging.getLogger().setLevel(logging.INFO)
 
-    extractor = capa.main.get_extractor(args.sample, args.format)
+    backend = args.backend if sys.version_info > (3, 0) else capa.main.BACKEND_VIV
+    extractor = capa.main.get_extractor(args.sample, args.format, backend)
     with open(args.output, "wb") as f:
         f.write(dump(extractor))
 

capa/features/insn.py

@@ -13,10 +13,10 @@ class API(Feature):
     def __init__(self, name, description=None):
         # Downcase library name if given
         if "." in name:
-            modname, impname = name.split(".")
+            modname, _, impname = name.rpartition(".")
             name = modname.lower() + "." + impname
 
-        super(API, self).__init__(name, description)
+        super(API, self).__init__(name, description=description)
 
 
 class Number(Feature):

capa/ida/helpers/__init__.py

@@ -82,14 +82,26 @@ def get_func_start_ea(ea):
     return f if f is None else f.start_ea
 
 
-def collect_metadata():
+def get_file_md5():
+    """ """
     md5 = idautils.GetInputFileMD5()
     if not isinstance(md5, six.string_types):
         md5 = capa.features.bytes_to_str(md5)
+    return md5
 
+
+def get_file_sha256():
+    """ """
     sha256 = idaapi.retrieve_input_file_sha256()
     if not isinstance(sha256, six.string_types):
         sha256 = capa.features.bytes_to_str(sha256)
+    return sha256
+
+
+def collect_metadata():
+    """ """
+    md5 = get_file_md5()
+    sha256 = get_file_sha256()
 
     return {
         "timestamp": datetime.datetime.now().isoformat(),
@@ -103,6 +115,7 @@ def collect_metadata():
         "analysis": {
             "format": idaapi.get_file_type_name(),
             "extractor": "ida",
+            "base_address": idaapi.get_imagebase(),
         },
         "version": capa.version.__version__,
     }

capa/ida/plugin/README.md

@@ -1,49 +1,37 @@
-# capa explorer
+![capa explorer](../../../.github/capa-explorer-logo.png)
 
-capa explorer is an IDA Pro plugin that integrates the FLARE team's open-source framework, capa, with IDA. capa is a framework that uses a well-defined collection of rules to 
+capa explorer is an IDA Pro plugin written in Python that integrates the FLARE team's open-source framework, capa, with IDA. capa is a framework that uses a well-defined collection of rules to 
 identify capabilities in a program. You can run capa against a PE file or shellcode and it tells you what it thinks the program can do. For example, it might suggest that 
-the program is a backdoor, can install services, or relies on HTTP to communicate. 
+the program is a backdoor, can install services, or relies on HTTP to communicate. You can use capa explorer to run capa directly on an IDA database without requiring access
+to or execution of the source binary. Once a database has been analyzed, capa explorer can be used to quickly identify and navigate to interesting areas of a program and manually build new capa rules out
+of the features extracted directly from your IDB.
 
-The capa explorer IDA plugin brings capa's detection capabilities to IDA. You can use capa explorer to run capa directly on an IDA database without needing access
-to the source binary. Once a database has been analyzed, capa explorer can be used to quickly identify and navigate to interesting areas of a program 
-and dissect capa rule matches at the assembly level.
-
-To illustrate, we use capa explorer to analyze Lab 14-02 from [Practical Malware Analysis](https://nostarch.com/malware) (PMA) available [here](https://practicalmalwareanalysis.com/labs/). Our
-goal is to understand the program's functionality.
+We love using capa explorer during malware analysis because it teaches us what parts of a program suggest a behavior. As we click on rows, capa explorer jumps directly 
+to important addresses in the IDA Pro database and highlights key features in the Disassembly view so they stand out visually. To illustrate, we use capa explorer to 
+analyze Lab 14-02 from [Practical Malware Analysis](https://nostarch.com/malware) (PMA) available [here](https://practicalmalwareanalysis.com/labs/). Our goal is to understand 
+the program's functionality.
 
 After loading Lab 14-02 into IDA and analyzing the database with capa explorer, we see that capa detected a rule match for `self delete via COMSPEC environment variable`:
 
-![](../../../doc/img/ida_plugin_example_1.png)
+![](../../../doc/img/explorer_condensed.png)
 
 We can use capa explorer to navigate the IDA Disassembly view directly to the suspect function and get an assembly-level breakdown of why capa matched `self delete via COMSPEC environment variable` 
 for this particular function.
 
-![](../../../doc/img/ida_plugin_example_2.png)
+![](../../../doc/img/explorer_expanded.png)
 
 Using the `Rule Information` and `Details` columns capa explorer shows us that the suspect function matched `self delete via COMSPEC environment variable` because it contains capa rule matches for `create process`, `get COMSPEC environment variable`,
 and `query environment variable`, references to the strings `COMSPEC`, ` > nul`, and `/c del`, and calls to the Windows API functions `GetEnvironmentVariableA` and `ShellExecuteEx`.
 
+capa explorer also helps you build new capa rules. To start select the `Rule Generator` tab, navigate to a function in the IDA `Disassembly` view,
+and click `Analyze`. capa explorer will extract features from this function and display them in the `Function Features` pane. You can add features listed in this pane to the `Editor` pane
+by either double-clicking a feature or using multi-select + right-click to add multiple features at once. The `Preview` and `Editor` panes help edit your rule. Use the `Preview` pane
+to modify the rule text directly and the `Editor` pane to construct and rearrange your hierarchy of statements and features. When you finish a rule you can save it directly to a file by clicking `Save`.
+
+![](../../../doc/img/rulegen_expanded.png)
+
 For more information on the FLARE team's open-source framework, capa, check out the overview in our first [blog](https://www.fireeye.com/blog/threat-research/2020/07/capa-automatically-identify-malware-capabilities.html).
 
-## Features
-
-![](../../../doc/img/ida_plugin_intro.gif)
-
-* Display capa results in an interactive tree view of rule matches and their locations in the current database
-* Search for keywords or phrases found in the `Rule Information`, `Address`, or `Details` columns
-* Display rule source content when a user hovers their cursor over a rule match
-* Double-click `Address` column to view associated feature in the IDA Disassembly view
-* Limit tree view results to the function currently displayed in the IDA Disassembly view; update results as a user navigates to different functions
-* Export results as formatted JSON by navigating to `File > Export results...`
-* Remember a user's capa rules directory for future runs; change capa rules directory by navigating to `Rules > Change rules directory...`
-* Automatically re-analyze database when user performs a program rebase
-* Automatically update results when IDA is used to rename a function
-* Select one or more checkboxes to highlight the associated addresses in the IDA Disassembly view
-* Right-click a function match to rename it; the new function name is propagated to the current IDA database
-* Right-click to copy a result by column or by row
-* Sort results by column
-* Reset tree view and IDA Disassembly view highlighting by clicking `Reset`
-
 ## Getting Started
 
 ### Requirements
@@ -65,7 +53,7 @@ capa explorer is limited to the file types supported by capa, which includes:
 
 You can install capa explorer using the following steps:
 
-1. Install capa for the Python interpreter used by your IDA installation:
+1. Install capa and its dependencies from PyPI for the Python interpreter used by your IDA installation:
     ```
     $ pip install flare-capa
     ```
@@ -74,30 +62,42 @@ You can install capa explorer using the following steps:
 
 ### Usage
 
-1. Run IDA and analyze a supported file type (select `Manual Load` and `Load Resources` for best results)
+1. Run IDA and analyze a supported file type (select the `Manual Load` and `Load Resources` options in IDA for best results)
 2. Open capa explorer in IDA by navigating to `Edit > Plugins > FLARE capa explorer` or using the keyboard shortcut `Alt+F5`
-3. Click `Analyze`
+3. Select the `Program Analysis` tab
+4. Click the `Analyze` button
 
 When running capa explorer for the first time you are prompted to select a file directory containing capa rules. The plugin conveniently
-remembers your selection for future runs; you can change this selection by navigating to `Rules > Change rules directory...`. We recommend 
+remembers your selection for future runs; you can change this selection by navigating to `Settings > Change default rules directory...`. We recommend 
 downloading and using the [standard collection of capa rules](https://github.com/fireeye/capa-rules) when getting started with the plugin.
 
-#### Tips
+#### Tips for Program Analysis
 
-* Start analysis by clicking `Analyze`
-* Reset the plugin user interface and remove highlighting from IDA disassembly view by clicking `Reset`
-* Change your capa rules directory by navigating to `Rules > Change rules directory...`
+* Start analysis by clicking the `Analyze` button
+* Reset the plugin user interface and remove highlighting from IDA disassembly view by clicking the `Reset` button
+* Change your capa rules directory by navigating to `Settings > Change default rules directory...` from the plugin menu
 * Hover your cursor over a rule match to view the source content of the rule
 * Double-click the `Address` column to navigate the IDA Disassembly view to the associated feature
 * Double-click a result in the `Rule Information` column to expand its children
 * Select a checkbox in the `Rule Information` column to highlight the address of the associated feature in the IDA Dissasembly view
 
+#### Tips for Rule Generator
+
+* Navigate to a function in the `Disassembly` view and click`Analyze` to get started
+* Double-click or multi-select + right-click in the `Function Features` pane to add features to the `Editor` pane
+* Right-click features in the `Editor` pane to make modifications
+* Drag-and-drop (single click + multi-select support) features in the `Editor` pane to quickly build a hierarchy of statements and features
+* Right-click anywhere in the `Editor` pane not on a feature to quickly remove all features
+* Add descriptions/comments by placing editing the appropriate column in the `Editor` pane
+* Directly edit rule text, including rule metadata fields using the `Preview` pane
+* Change the default rule author and default scope displayed in the `Preview` pane by navigating to `Settings`
+
 ## Development
 
 Because capa explorer is packaged with capa you will need to install capa locally for development.
 
 You can install capa locally by following the steps outlined in `Method 3: Inspecting the capa source code` of the [capa 
-installation guide](https://github.com/fireeye/capa/blob/ida_plugin_documentation/doc/installation.md#method-3-inspecting-the-capa-source-code). Once installed, copy [capa_explorer.py](https://raw.githubusercontent.com/fireeye/capa/master/capa/ida/plugin/capa_explorer.py) 
+installation guide](https://github.com/fireeye/capa/blob/master/doc/installation.md#method-3-inspecting-the-capa-source-code). Once installed, copy [capa_explorer.py](https://raw.githubusercontent.com/fireeye/capa/master/capa/ida/plugin/capa_explorer.py) 
 to your IDA plugins directory to run the plugin in IDA.
 
 ### Components
@@ -107,5 +107,5 @@ capa explorer consists of two main components:
 * An IDA [feature extractor](https://github.com/fireeye/capa/tree/master/capa/features/extractors/ida) built on top of IDA's binary analysis engine
   * This component uses IDAPython to extract [capa features](https://github.com/fireeye/capa-rules/blob/master/doc/format.md#extracted-features) from the IDA database such as strings, 
 disassembly, and control flow; these extracted features are used by capa to find feature combinations that result in a rule match
-* An [interactive plugin](https://github.com/fireeye/capa/tree/master/capa/ida/plugin) for displaying and exploring capa rule matches
+* An [interactive user interface](https://github.com/fireeye/capa/tree/master/capa/ida/plugin) for displaying and exploring capa rule matches
   * This component integrates the IDA feature extractor and capa, providing an interactive user interface to dissect rule matches found by capa using features extracted by the IDA feature extractor

capa/ida/plugin/item.py

@@ -35,20 +35,19 @@ def location_to_hex(location):
 class CapaExplorerDataItem(object):
     """store data for CapaExplorerDataModel"""
 
-    def __init__(self, parent, data):
+    def __init__(self, parent, data, can_check=True):
         """initialize item"""
         self.pred = parent
         self._data = data
         self.children = []
         self._checked = False
+        self._can_check = can_check
 
         # default state for item
-        self.flags = (
-            QtCore.Qt.ItemIsEnabled
-            | QtCore.Qt.ItemIsSelectable
-            | QtCore.Qt.ItemIsTristate
-            | QtCore.Qt.ItemIsUserCheckable
-        )
+        self.flags = QtCore.Qt.ItemIsEnabled | QtCore.Qt.ItemIsSelectable
+
+        if self._can_check:
+            self.flags = self.flags | QtCore.Qt.ItemIsUserCheckable | QtCore.Qt.ItemIsTristate
 
         if self.pred:
             self.pred.appendChild(self)
@@ -70,6 +69,10 @@ class CapaExplorerDataItem(object):
         """
         self._checked = checked
 
+    def canCheck(self):
+        """ """
+        return self._can_check
+
     def isChecked(self):
         """get item is checked"""
         return self._checked
@@ -165,7 +168,7 @@ class CapaExplorerRuleItem(CapaExplorerDataItem):
 
     fmt = "%s (%d matches)"
 
-    def __init__(self, parent, name, namespace, count, source):
+    def __init__(self, parent, name, namespace, count, source, can_check=True):
         """initialize item
 
         @param parent: parent node
@@ -175,7 +178,7 @@ class CapaExplorerRuleItem(CapaExplorerDataItem):
         @param source: rule source (tooltip)
         """
         display = self.fmt % (name, count) if count > 1 else name
-        super(CapaExplorerRuleItem, self).__init__(parent, [display, "", namespace])
+        super(CapaExplorerRuleItem, self).__init__(parent, [display, "", namespace], can_check)
         self._source = source
 
     @property
@@ -208,14 +211,14 @@ class CapaExplorerFunctionItem(CapaExplorerDataItem):
 
     fmt = "function(%s)"
 
-    def __init__(self, parent, location):
+    def __init__(self, parent, location, can_check=True):
         """initialize item
 
         @param parent: parent node
         @param location: virtual address of function as seen by IDA
         """
         super(CapaExplorerFunctionItem, self).__init__(
-            parent, [self.fmt % idaapi.get_name(location), location_to_hex(location), ""]
+            parent, [self.fmt % idaapi.get_name(location), location_to_hex(location), ""], can_check
         )
 
     @property

capa/ida/plugin/model.py

@@ -6,7 +6,7 @@
 #  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 
-from collections import deque
+from collections import deque, defaultdict
 
 import idc
 import idaapi
@@ -110,6 +110,8 @@ class CapaExplorerDataModel(QtCore.QAbstractItemModel):
 
         if role == QtCore.Qt.CheckStateRole and column == CapaExplorerDataModel.COLUMN_INDEX_RULE_INFORMATION:
             # inform view how to display content of checkbox - un/checked
+            if not item.canCheck():
+                return None
             return QtCore.Qt.Checked if item.isChecked() else QtCore.Qt.Unchecked
 
         if role == QtCore.Qt.FontRole and column in (
@@ -424,14 +426,28 @@ class CapaExplorerDataModel(QtCore.QAbstractItemModel):
         for child in match.get("children", []):
             self.render_capa_doc_match(parent2, child, doc)
 
-    def render_capa_doc(self, doc):
-        """render capa features specified in doc
-
-        @param doc: capa result doc
-        """
-        # inform model that changes are about to occur
-        self.beginResetModel()
+    def render_capa_doc_by_function(self, doc):
+        """ """
+        matches_by_function = {}
+        for rule in rutils.capability_rules(doc):
+            for ea in rule["matches"].keys():
+                ea = capa.ida.helpers.get_func_start_ea(ea)
+                if ea is None:
+                    # file scope, skip for rendering in this mode
+                    continue
+                if None is matches_by_function.get(ea, None):
+                    matches_by_function[ea] = CapaExplorerFunctionItem(self.root_node, ea, can_check=False)
+                CapaExplorerRuleItem(
+                    matches_by_function[ea],
+                    rule["meta"]["name"],
+                    rule["meta"].get("namespace"),
+                    len(rule["matches"]),
+                    rule["source"],
+                    can_check=False,
+                )
 
+    def render_capa_doc_by_program(self, doc):
+        """ """
         for rule in rutils.capability_rules(doc):
             rule_name = rule["meta"]["name"]
             rule_namespace = rule["meta"].get("namespace")
@@ -451,6 +467,19 @@ class CapaExplorerDataModel(QtCore.QAbstractItemModel):
 
                 self.render_capa_doc_match(parent2, match, doc)
 
+    def render_capa_doc(self, doc, by_function):
+        """render capa features specified in doc
+
+        @param doc: capa result doc
+        """
+        # inform model that changes are about to occur
+        self.beginResetModel()
+
+        if by_function:
+            self.render_capa_doc_by_function(doc)
+        else:
+            self.render_capa_doc_by_program(doc)
+
         # inform model changes have ended
         self.endResetModel()
 

capa/ida/plugin/view.py

@@ -5,15 +5,812 @@
 # Unless required by applicable law or agreed to in writing, software distributed under the License
 #  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
+import re
+from collections import Counter
 
 import idc
-from PyQt5 import QtCore, QtWidgets
+from PyQt5 import QtGui, QtCore, QtWidgets
 
+import capa.rules
+import capa.engine
+import capa.ida.helpers
+import capa.features.basicblock
 from capa.ida.plugin.item import CapaExplorerFunctionItem
 from capa.ida.plugin.model import CapaExplorerDataModel
 
 MAX_SECTION_SIZE = 750
 
+# default colors used in views
+COLOR_GREEN_RGB = (79, 121, 66)
+COLOR_BLUE_RGB = (37, 147, 215)
+
+
+def calc_level_by_indent(line, prev_level=0):
+    """ """
+    if not len(line.strip()):
+        # blank line, which may occur for comments so we simply use the last level
+        return prev_level
+    stripped = line.lstrip()
+    if stripped.startswith("description"):
+        # need to adjust two spaces when encountering string description
+        line = line[2:]
+    # calc line level based on preceding whitespace
+    return len(line) - len(stripped)
+
+
+def parse_feature_for_node(feature):
+    """ """
+    description = ""
+    comment = ""
+
+    if feature.startswith("- count"):
+        # count is weird, we need to handle special
+        # first, we need to grab the comment, if exists
+        # next, we need to check for an embedded description
+        feature, _, comment = feature.partition("#")
+        m = re.search(r"- count\(([a-zA-Z]+)\((.+)\s+=\s+(.+)\)\):\s*(.+)", feature)
+        if m:
+            # reconstruct count without description
+            feature, value, description, count = m.groups()
+            feature = "- count(%s(%s)): %s" % (feature, value, count)
+    elif not feature.startswith("#"):
+        feature, _, comment = feature.partition("#")
+        feature, _, description = feature.partition("=")
+
+    return map(lambda o: o.strip(), (feature, description, comment))
+
+
+def parse_node_for_feature(feature, description, comment, depth):
+    """ """
+    depth = (depth * 2) + 4
+    display = ""
+
+    if feature.startswith("#"):
+        display += "%s%s\n" % (" " * depth, feature)
+    elif description:
+        if feature.startswith(("- and", "- or", "- optional", "- basic block", "- not")):
+            display += "%s%s" % (" " * depth, feature)
+            if comment:
+                display += " # %s" % comment
+            display += "\n%s- description: %s\n" % (" " * (depth + 2), description)
+        elif feature.startswith("- string"):
+            display += "%s%s" % (" " * depth, feature)
+            if comment:
+                display += " # %s" % comment
+            display += "\n%sdescription: %s\n" % (" " * (depth + 2), description)
+        elif feature.startswith("- count"):
+            # count is weird, we need to format description based on feature type, so we parse with regex
+            # assume format - count(<feature_name>(<feature_value>)): <count>
+            m = re.search(r"- count\(([a-zA-Z]+)\((.+)\)\): (.+)", feature)
+            if m:
+                name, value, count = m.groups()
+                if name in ("string",):
+                    display += "%s%s" % (" " * depth, feature)
+                    if comment:
+                        display += " # %s" % comment
+                    display += "\n%sdescription: %s\n" % (" " * (depth + 2), description)
+                else:
+                    display += "%s- count(%s(%s = %s)): %s" % (
+                        " " * depth,
+                        name,
+                        value,
+                        description,
+                        count,
+                    )
+                    if comment:
+                        display += " # %s\n" % comment
+        else:
+            display += "%s%s = %s" % (" " * depth, feature, description)
+            if comment:
+                display += " # %s\n" % comment
+    else:
+        display += "%s%s" % (" " * depth, feature)
+        if comment:
+            display += " # %s\n" % comment
+
+    return display if display.endswith("\n") else display + "\n"
+
+
+def yaml_to_nodes(s):
+    level = 0
+    for line in s.splitlines():
+        feature, description, comment = parse_feature_for_node(line.strip())
+
+        o = QtWidgets.QTreeWidgetItem(None)
+
+        # set node attributes
+        setattr(o, "capa_level", calc_level_by_indent(line, level))
+
+        if feature.startswith(("- and:", "- or:", "- not:", "- basic block:", "- optional:")):
+            setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_expression())
+        elif feature.startswith("#"):
+            setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_comment())
+        else:
+            setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_feature())
+
+        # set node text
+        for (i, v) in enumerate((feature, description, comment)):
+            o.setText(i, v)
+
+        yield o
+
+
+def iterate_tree(o):
+    """ """
+    itr = QtWidgets.QTreeWidgetItemIterator(o)
+    while itr.value():
+        yield itr.value()
+        itr += 1
+
+
+def calc_item_depth(o):
+    """ """
+    depth = 0
+    while True:
+        if not o.parent():
+            break
+        depth += 1
+        o = o.parent()
+    return depth
+
+
+def build_action(o, display, data, slot):
+    """ """
+    action = QtWidgets.QAction(display, o)
+
+    action.setData(data)
+    action.triggered.connect(lambda checked: slot(action))
+
+    return action
+
+
+def build_context_menu(o, actions):
+    """ """
+    menu = QtWidgets.QMenu()
+
+    for action in actions:
+        if isinstance(action, QtWidgets.QMenu):
+            menu.addMenu(action)
+        else:
+            menu.addAction(build_action(o, *action))
+
+    return menu
+
+
+class CapaExplorerRulgenPreview(QtWidgets.QTextEdit):
+    def __init__(self, parent=None):
+        """ """
+        super(CapaExplorerRulgenPreview, self).__init__(parent)
+
+        self.setFont(QtGui.QFont("Courier", weight=QtGui.QFont.Bold))
+        self.setLineWrapMode(QtWidgets.QTextEdit.NoWrap)
+        self.setHorizontalScrollBarPolicy(QtCore.Qt.ScrollBarAsNeeded)
+
+    def reset_view(self):
+        """ """
+        self.clear()
+
+    def load_preview_meta(self, ea, author, scope):
+        """ """
+        metadata_default = [
+            "# generated using capa explorer for IDA Pro",
+            "rule:",
+            "  meta:",
+            "    name: <insert_name>",
+            "    namespace: <insert_namespace>",
+            "    author: %s" % author,
+            "    scope: %s" % scope,
+            "    references: <insert_references>",
+            "    examples:",
+            "      - %s:0x%X" % (capa.ida.helpers.get_file_md5().upper(), ea)
+            if ea
+            else "      - %s" % (capa.ida.helpers.get_file_md5().upper()),
+            "  features:",
+        ]
+        self.setText("\n".join(metadata_default))
+
+    def keyPressEvent(self, e):
+        """ """
+        if e.key() == QtCore.Qt.Key_Tab:
+            self.insertPlainText(" " * 2)
+        else:
+            super(CapaExplorerRulgenPreview, self).keyPressEvent(e)
+
+
+class CapaExplorerRulgenEditor(QtWidgets.QTreeWidget):
+
+    updated = QtCore.pyqtSignal()
+
+    def __init__(self, preview, parent=None):
+        """ """
+        super(CapaExplorerRulgenEditor, self).__init__(parent)
+
+        self.preview = preview
+
+        self.setHeaderLabels(["Feature", "Description", "Comment"])
+        self.header().setSectionResizeMode(QtWidgets.QHeaderView.ResizeToContents)
+        self.header().setStretchLastSection(False)
+        self.setExpandsOnDoubleClick(False)
+        self.setEditTriggers(QtWidgets.QAbstractItemView.NoEditTriggers)
+        self.setContextMenuPolicy(QtCore.Qt.CustomContextMenu)
+        self.setSelectionMode(QtWidgets.QAbstractItemView.ExtendedSelection)
+        self.setStyleSheet("QTreeView::item {padding-right: 15 px;padding-bottom: 2 px;}")
+
+        # enable drag and drop
+        self.setDragEnabled(True)
+        self.setAcceptDrops(True)
+        self.setDragDropMode(QtWidgets.QAbstractItemView.InternalMove)
+
+        # connect slots
+        self.itemChanged.connect(self.slot_item_changed)
+        self.customContextMenuRequested.connect(self.slot_custom_context_menu_requested)
+        self.itemDoubleClicked.connect(self.slot_item_double_clicked)
+
+        self.root = None
+        self.reset_view()
+
+        self.is_editing = False
+
+    @staticmethod
+    def get_column_feature_index():
+        """ """
+        return 0
+
+    @staticmethod
+    def get_column_description_index():
+        """ """
+        return 1
+
+    @staticmethod
+    def get_column_comment_index():
+        """ """
+        return 2
+
+    @staticmethod
+    def get_node_type_expression():
+        """ """
+        return 0
+
+    @staticmethod
+    def get_node_type_feature():
+        """ """
+        return 1
+
+    @staticmethod
+    def get_node_type_comment():
+        """ """
+        return 2
+
+    def dragMoveEvent(self, e):
+        """ """
+        super(CapaExplorerRulgenEditor, self).dragMoveEvent(e)
+
+    def dragEventEnter(self, e):
+        """ """
+        super(CapaExplorerRulgenEditor, self).dragEventEnter(e)
+
+    def dropEvent(self, e):
+        """ """
+        if not self.indexAt(e.pos()).isValid():
+            return
+
+        super(CapaExplorerRulgenEditor, self).dropEvent(e)
+
+        # self.prune_expressions()
+        self.update_preview()
+        self.expandAll()
+
+    def reset_view(self):
+        """ """
+        self.root = None
+        self.clear()
+
+    def slot_item_changed(self, item, column):
+        """ """
+        if self.is_editing:
+            self.update_preview()
+            self.is_editing = False
+
+    def slot_remove_selected(self, action):
+        """ """
+        for o in self.selectedItems():
+            if o == self.root:
+                self.takeTopLevelItem(self.indexOfTopLevelItem(o))
+                self.root = None
+                continue
+            o.parent().removeChild(o)
+
+    def slot_nest_features(self, action):
+        """ """
+        # create a new parent under root node, by default; new node added last position in tree
+        new_parent = self.new_expression_node(self.root, (action.data()[0], ""))
+
+        for o in self.get_features(selected=True):
+            # take child from its parent by index, add to new parent
+            new_parent.addChild(o.parent().takeChild(o.parent().indexOfChild(o)))
+
+        # ensure new parent expanded
+        new_parent.setExpanded(True)
+
+    def slot_edit_expression(self, action):
+        """ """
+        expression, o = action.data()
+        o.setText(CapaExplorerRulgenEditor.get_column_feature_index(), expression)
+
+    def slot_clear_all(self, action):
+        """ """
+        self.reset_view()
+
+    def slot_custom_context_menu_requested(self, pos):
+        """ """
+        if not self.indexAt(pos).isValid():
+            # user selected invalid index
+            self.load_custom_context_menu_invalid_index(pos)
+        elif self.itemAt(pos).capa_type == CapaExplorerRulgenEditor.get_node_type_expression():
+            # user selected expression node
+            self.load_custom_context_menu_expression(pos)
+        else:
+            # user selected feature node
+            self.load_custom_context_menu_feature(pos)
+
+        self.update_preview()
+
+    def slot_item_double_clicked(self, o, column):
+        """ """
+        if column in (
+            CapaExplorerRulgenEditor.get_column_comment_index(),
+            CapaExplorerRulgenEditor.get_column_description_index(),
+        ):
+            o.setFlags(o.flags() | QtCore.Qt.ItemIsEditable)
+            self.editItem(o, column)
+            o.setFlags(o.flags() & ~QtCore.Qt.ItemIsEditable)
+            self.is_editing = True
+
+    def update_preview(self):
+        """ """
+        rule_text = self.preview.toPlainText()
+
+        if -1 != rule_text.find("features:"):
+            rule_text = rule_text[: rule_text.find("features:") + len("features:")]
+            rule_text += "\n"
+        else:
+            rule_text = rule_text.rstrip()
+            rule_text += "\n  features:\n"
+
+        for o in iterate_tree(self):
+            feature, description, comment = map(lambda o: o.strip(), tuple(o.text(i) for i in range(3)))
+            rule_text += parse_node_for_feature(feature, description, comment, calc_item_depth(o))
+
+        # FIXME we avoid circular update by disabling signals when updating
+        # the preview. Preferably we would refactor the code to avoid this
+        # in the first place
+        self.preview.blockSignals(True)
+        self.preview.setPlainText(rule_text)
+        self.preview.blockSignals(False)
+
+        # emit signal so views can update
+        self.updated.emit()
+
+    def load_custom_context_menu_invalid_index(self, pos):
+        """ """
+        actions = (("Remove all", (), self.slot_clear_all),)
+
+        menu = build_context_menu(self.parent(), actions)
+        menu.exec_(self.viewport().mapToGlobal(pos))
+
+    def load_custom_context_menu_feature(self, pos):
+        """ """
+        actions = (("Remove selection", (), self.slot_remove_selected),)
+
+        sub_actions = (
+            ("and", ("- and:",), self.slot_nest_features),
+            ("or", ("- or:",), self.slot_nest_features),
+            ("not", ("- not:",), self.slot_nest_features),
+            ("optional", ("- optional:",), self.slot_nest_features),
+            ("basic block", ("- basic block:",), self.slot_nest_features),
+        )
+
+        # build submenu with modify actions
+        sub_menu = build_context_menu(self.parent(), sub_actions)
+        sub_menu.setTitle("Nest feature%s" % ("" if len(tuple(self.get_features(selected=True))) == 1 else "s"))
+
+        # build main menu with submenu + main actions
+        menu = build_context_menu(self.parent(), (sub_menu,) + actions)
+
+        menu.exec_(self.viewport().mapToGlobal(pos))
+
+    def load_custom_context_menu_expression(self, pos):
+        """ """
+        actions = (("Remove expression", (), self.slot_remove_selected),)
+
+        sub_actions = (
+            ("and", ("- and:", self.itemAt(pos)), self.slot_edit_expression),
+            ("or", ("- or:", self.itemAt(pos)), self.slot_edit_expression),
+            ("not", ("- not:", self.itemAt(pos)), self.slot_edit_expression),
+            ("optional", ("- optional:", self.itemAt(pos)), self.slot_edit_expression),
+            ("basic block", ("- basic block:", self.itemAt(pos)), self.slot_edit_expression),
+        )
+
+        # build submenu with modify actions
+        sub_menu = build_context_menu(self.parent(), sub_actions)
+        sub_menu.setTitle("Modify")
+
+        # build main menu with submenu + main actions
+        menu = build_context_menu(self.parent(), (sub_menu,) + actions)
+
+        menu.exec_(self.viewport().mapToGlobal(pos))
+
+    def style_expression_node(self, o):
+        """ """
+        font = QtGui.QFont()
+        font.setBold(True)
+
+        o.setFont(CapaExplorerRulgenEditor.get_column_feature_index(), font)
+
+    def style_feature_node(self, o):
+        """ """
+        font = QtGui.QFont()
+        brush = QtGui.QBrush()
+
+        font.setFamily("Courier")
+        font.setWeight(QtGui.QFont.Medium)
+        brush.setColor(QtGui.QColor(*COLOR_GREEN_RGB))
+
+        o.setFont(CapaExplorerRulgenEditor.get_column_feature_index(), font)
+        o.setForeground(CapaExplorerRulgenEditor.get_column_feature_index(), brush)
+
+    def style_comment_node(self, o):
+        """ """
+        font = QtGui.QFont()
+        font.setBold(True)
+        font.setFamily("Courier")
+
+        o.setFont(CapaExplorerRulgenEditor.get_column_feature_index(), font)
+
+    def set_expression_node(self, o):
+        """ """
+        setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_expression())
+        self.style_expression_node(o)
+
+    def set_feature_node(self, o):
+        """ """
+        setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_feature())
+        o.setFlags(o.flags() & ~QtCore.Qt.ItemIsDropEnabled)
+        self.style_feature_node(o)
+
+    def set_comment_node(self, o):
+        """ """
+        setattr(o, "capa_type", CapaExplorerRulgenEditor.get_node_type_comment())
+        o.setFlags(o.flags() & ~QtCore.Qt.ItemIsDropEnabled)
+
+        self.style_comment_node(o)
+
+    def new_expression_node(self, parent, values=()):
+        """ """
+        o = QtWidgets.QTreeWidgetItem(parent)
+        self.set_expression_node(o)
+        for (i, v) in enumerate(values):
+            o.setText(i, v)
+        return o
+
+    def new_feature_node(self, parent, values=()):
+        """ """
+        o = QtWidgets.QTreeWidgetItem(parent)
+        self.set_feature_node(o)
+        for (i, v) in enumerate(values):
+            o.setText(i, v)
+        return o
+
+    def new_comment_node(self, parent, values=()):
+        """ """
+        o = QtWidgets.QTreeWidgetItem(parent)
+        self.set_comment_node(o)
+        for (i, v) in enumerate(values):
+            o.setText(i, v)
+        return o
+
+    def update_features(self, features):
+        """ """
+        if not self.root:
+            # root node does not exist, create default node, set expanded
+            self.root = self.new_expression_node(self, ("- or:", ""))
+
+        # build feature counts
+        counted = list(zip(Counter(features).keys(), Counter(features).values()))
+
+        # single features
+        for (k, v) in filter(lambda t: t[1] == 1, counted):
+            self.new_feature_node(self.root, ("- %s: %s" % (k.name.lower(), k.get_value_str()), ""))
+
+        # n > 1 features
+        for (k, v) in filter(lambda t: t[1] > 1, counted):
+            self.new_feature_node(self.root, ("- count(%s): %d" % (str(k), v), ""))
+
+        self.expandAll()
+        self.update_preview()
+
+    def load_features_from_yaml(self, rule_text, update_preview=False):
+        """ """
+
+        def add_node(parent, node):
+            if node.text(0).startswith("description:"):
+                if parent.childCount():
+                    parent.child(parent.childCount() - 1).setText(1, node.text(0).lstrip("description:").lstrip())
+                else:
+                    parent.setText(1, node.text(0).lstrip("description:").lstrip())
+            elif node.text(0).startswith("- description:"):
+                parent.setText(1, node.text(0).lstrip("- description:").lstrip())
+            else:
+                parent.addChild(node)
+
+        def build(parent, nodes):
+            if nodes:
+                child_lvl = nodes[0].capa_level
+                while nodes:
+                    node = nodes.pop(0)
+                    if node.capa_level == child_lvl:
+                        add_node(parent, node)
+                    elif node.capa_level > child_lvl:
+                        nodes.insert(0, node)
+                        build(parent.child(parent.childCount() - 1), nodes)
+                    else:
+                        parent = parent.parent() if parent.parent() else parent
+                        add_node(parent, node)
+
+        self.reset_view()
+
+        # check for lack of features block
+        if -1 == rule_text.find("features:"):
+            return
+
+        rule_features = rule_text[rule_text.find("features:") + len("features:") :].strip()
+        rule_nodes = list(yaml_to_nodes(rule_features))
+
+        # check for lack of nodes
+        if not rule_nodes:
+            return
+
+        for o in rule_nodes:
+            (self.set_expression_node, self.set_feature_node, self.set_comment_node)[o.capa_type](o)
+
+        self.root = rule_nodes.pop(0)
+        self.addTopLevelItem(self.root)
+
+        if update_preview:
+            self.preview.blockSignals(True)
+            self.preview.setPlainText(rule_text)
+            self.preview.blockSignals(False)
+
+        build(self.root, rule_nodes)
+
+        self.expandAll()
+
+    def get_features(self, selected=False, ignore=()):
+        """ """
+        for feature in filter(
+            lambda o: o.capa_type
+            in (CapaExplorerRulgenEditor.get_node_type_feature(), CapaExplorerRulgenEditor.get_node_type_comment()),
+            tuple(iterate_tree(self)),
+        ):
+            if feature in ignore:
+                continue
+            if selected and not feature.isSelected():
+                continue
+            yield feature
+
+    def get_expressions(self, selected=False, ignore=()):
+        """ """
+        for expression in filter(
+            lambda o: o.capa_type == CapaExplorerRulgenEditor.get_node_type_expression(), tuple(iterate_tree(self))
+        ):
+            if expression in ignore:
+                continue
+            if selected and not expression.isSelected():
+                continue
+            yield expression
+
+
+class CapaExplorerRulegenFeatures(QtWidgets.QTreeWidget):
+    def __init__(self, editor, parent=None):
+        """ """
+        super(CapaExplorerRulegenFeatures, self).__init__(parent)
+
+        self.parent_items = {}
+        self.editor = editor
+
+        self.setHeaderLabels(["Feature", "Virtual Address"])
+        self.header().setSectionResizeMode(QtWidgets.QHeaderView.ResizeToContents)
+        self.setStyleSheet("QTreeView::item {padding-right: 15 px;padding-bottom: 2 px;}")
+
+        self.setExpandsOnDoubleClick(False)
+        self.setContextMenuPolicy(QtCore.Qt.CustomContextMenu)
+        self.setSelectionMode(QtWidgets.QAbstractItemView.ExtendedSelection)
+
+        # connect slots
+        self.itemDoubleClicked.connect(self.slot_item_double_clicked)
+        self.customContextMenuRequested.connect(self.slot_custom_context_menu_requested)
+
+        self.reset_view()
+
+    @staticmethod
+    def get_column_feature_index():
+        """ """
+        return 0
+
+    @staticmethod
+    def get_column_address_index():
+        """ """
+        return 1
+
+    @staticmethod
+    def get_node_type_parent():
+        """ """
+        return 0
+
+    @staticmethod
+    def get_node_type_leaf():
+        """ """
+        return 1
+
+    def reset_view(self):
+        """ """
+        self.clear()
+
+    def slot_add_selected_features(self, action):
+        """ """
+        selected = [item.data(0, 0x100) for item in self.selectedItems()]
+        if selected:
+            self.editor.update_features(selected)
+
+    def slot_custom_context_menu_requested(self, pos):
+        """ """
+        actions = []
+        action_add_features_fmt = ""
+
+        selected_items_count = len(self.selectedItems())
+        if selected_items_count == 0:
+            return
+
+        if selected_items_count == 1:
+            action_add_features_fmt = "Add feature"
+        else:
+            action_add_features_fmt = "Add %d features" % selected_items_count
+
+        actions.append((action_add_features_fmt, (), self.slot_add_selected_features))
+
+        menu = build_context_menu(self.parent(), actions)
+        menu.exec_(self.viewport().mapToGlobal(pos))
+
+    def slot_item_double_clicked(self, o, column):
+        """ """
+        if column == CapaExplorerRulegenFeatures.get_column_address_index() and o.text(column):
+            idc.jumpto(int(o.text(column), 0x10))
+        elif o.capa_type == CapaExplorerRulegenFeatures.get_node_type_leaf():
+            self.editor.update_features([o.data(0, 0x100)])
+
+    def show_all_items(self):
+        """ """
+        for o in iterate_tree(self):
+            o.setHidden(False)
+            o.setExpanded(False)
+
+    def filter_items_by_text(self, text):
+        """ """
+        if text:
+            for o in iterate_tree(self):
+                data = o.data(0, 0x100)
+                if data and text.lower() not in data.get_value_str().lower():
+                    o.setHidden(True)
+                    continue
+                o.setHidden(False)
+                o.setExpanded(True)
+        else:
+            self.show_all_items()
+
+    def style_parent_node(self, o):
+        """ """
+        font = QtGui.QFont()
+        font.setBold(True)
+
+        o.setFont(CapaExplorerRulegenFeatures.get_column_feature_index(), font)
+
+    def style_leaf_node(self, o):
+        """ """
+        font = QtGui.QFont("Courier", weight=QtGui.QFont.Bold)
+        brush = QtGui.QBrush()
+
+        o.setFont(CapaExplorerRulegenFeatures.get_column_feature_index(), font)
+        o.setFont(CapaExplorerRulegenFeatures.get_column_address_index(), font)
+
+        brush.setColor(QtGui.QColor(*COLOR_GREEN_RGB))
+        o.setForeground(CapaExplorerRulegenFeatures.get_column_feature_index(), brush)
+
+        brush.setColor(QtGui.QColor(*COLOR_BLUE_RGB))
+        o.setForeground(CapaExplorerRulegenFeatures.get_column_address_index(), brush)
+
+    def set_parent_node(self, o):
+        """ """
+        o.setFlags(o.flags() & ~QtCore.Qt.ItemIsSelectable)
+        setattr(o, "capa_type", CapaExplorerRulegenFeatures.get_node_type_parent())
+        self.style_parent_node(o)
+
+    def set_leaf_node(self, o):
+        """ """
+        setattr(o, "capa_type", CapaExplorerRulegenFeatures.get_node_type_leaf())
+        self.style_leaf_node(o)
+
+    def new_parent_node(self, parent, data, feature=None):
+        """ """
+        o = QtWidgets.QTreeWidgetItem(parent)
+
+        self.set_parent_node(o)
+        for (i, v) in enumerate(data):
+            o.setText(i, v)
+        if feature:
+            o.setData(0, 0x100, feature)
+
+        return o
+
+    def new_leaf_node(self, parent, data, feature=None):
+        """ """
+        o = QtWidgets.QTreeWidgetItem(parent)
+
+        self.set_leaf_node(o)
+        for (i, v) in enumerate(data):
+            o.setText(i, v)
+        if feature:
+            o.setData(0, 0x100, feature)
+
+        return o
+
+    def load_features(self, file_features, func_features={}):
+        """ """
+        self.parse_features_for_tree(self.new_parent_node(self, ("File Scope",)), file_features)
+        if func_features:
+            self.parse_features_for_tree(self.new_parent_node(self, ("Function/Basic Block Scope",)), func_features)
+
+    def parse_features_for_tree(self, parent, features):
+        """ """
+        self.parent_items = {}
+
+        def format_address(e):
+            return "%X" % e if e else ""
+
+        for (feature, eas) in sorted(features.items(), key=lambda k: sorted(k[1])):
+            if isinstance(feature, capa.features.basicblock.BasicBlock):
+                # filter basic blocks for now, we may want to add these back in some time
+                # in the future
+                continue
+
+            if isinstance(feature, capa.features.String):
+                # strip string for display
+                feature.value = feature.value.strip()
+
+            # level 0
+            if type(feature) not in self.parent_items:
+                self.parent_items[type(feature)] = self.new_parent_node(parent, (feature.name.lower(),))
+
+            # level 1
+            if feature not in self.parent_items:
+                if len(eas) > 1:
+                    self.parent_items[feature] = self.new_parent_node(
+                        self.parent_items[type(feature)], (str(feature),), feature=feature
+                    )
+                else:
+                    self.parent_items[feature] = self.new_leaf_node(
+                        self.parent_items[type(feature)], (str(feature),), feature=feature
+                    )
+
+            # level n > 1
+            if len(eas) > 1:
+                for ea in sorted(eas):
+                    self.new_leaf_node(self.parent_items[feature], (str(feature), format_address(ea)), feature=feature)
+            else:
+                ea = eas.pop()
+                for (i, v) in enumerate((str(feature), format_address(ea))):
+                    self.parent_items[feature].setText(i, v)
+                self.parent_items[feature].setData(0, 0x100, feature)
+
 
 class CapaExplorerQtreeView(QtWidgets.QTreeView):
     """tree view used to display hierarchical capa results
@@ -108,6 +905,7 @@ class CapaExplorerQtreeView(QtWidgets.QTreeView):
 
         if not model_index.isValid():
             raise ValueError("invalid index")
+
         return model_index.internalPointer()
 
     def send_data_to_clipboard(self, data):
@@ -254,6 +1052,10 @@ class CapaExplorerQtreeView(QtWidgets.QTreeView):
         @param pos: cursor position
         """
         model_index = self.indexAt(pos)
+
+        if not model_index.isValid():
+            return
+
         item = self.map_index_to_source_item(model_index)
 
         column = model_index.column()

capa/main.py

@@ -32,7 +32,9 @@ import capa.features.extractors
 from capa.helpers import oint, get_file_taste
 
 RULES_PATH_DEFAULT_STRING = "(embedded rules)"
-SUPPORTED_FILE_MAGIC = set(["MZ"])
+SUPPORTED_FILE_MAGIC = set([b"MZ"])
+BACKEND_VIV = "vivisect"
+BACKEND_SMDA = "smda"
 
 
 logger = logging.getLogger("capa")
@@ -40,8 +42,11 @@ logger = logging.getLogger("capa")
 
 def set_vivisect_log_level(level):
     logging.getLogger("vivisect").setLevel(level)
+    logging.getLogger("vivisect.base").setLevel(level)
+    logging.getLogger("vivisect.impemu").setLevel(level)
     logging.getLogger("vtrace").setLevel(level)
     logging.getLogger("envi").setLevel(level)
+    logging.getLogger("envi.codeflow").setLevel(level)
 
 
 def find_function_capabilities(ruleset, extractor, f):
@@ -112,7 +117,13 @@ def find_capabilities(ruleset, extractor, disable_progress=None):
         }
     }
 
-    for f in tqdm.tqdm(list(extractor.get_functions()), disable=disable_progress, desc="matching", unit=" functions"):
+    pbar = tqdm.tqdm
+    if disable_progress:
+        # do not use tqdm to avoid unnecessary side effects when caller intends
+        # to disable progress completely
+        pbar = lambda s, *args, **kwargs: s
+
+    for f in pbar(list(extractor.get_functions()), desc="matching", unit=" functions"):
         function_matches, bb_matches, feature_count = find_function_capabilities(ruleset, extractor, f)
         meta["feature_counts"]["functions"][f.__int__()] = feature_count
         logger.debug("analyzed function 0x%x and extracted %d features", f.__int__(), feature_count)
@@ -271,6 +282,8 @@ def get_workspace(path, format, should_save=True):
         vw = get_shellcode_vw(path, arch="i386", should_save=should_save)
     elif format == "sc64":
         vw = get_shellcode_vw(path, arch="amd64", should_save=should_save)
+    else:
+        raise ValueError("unexpected format: " + format)
     logger.debug("%s", get_meta_str(vw))
     return vw
 
@@ -294,17 +307,43 @@ class UnsupportedRuntimeError(RuntimeError):
     pass
 
 
-def get_extractor_py3(path, format, disable_progress=False):
-    raise UnsupportedRuntimeError()
+def get_extractor_py3(path, format, backend, disable_progress=False):
+    if backend == "smda":
+        from smda.SmdaConfig import SmdaConfig
+        from smda.Disassembler import Disassembler
+
+        import capa.features.extractors.smda
+
+        smda_report = None
+        with halo.Halo(text="analyzing program", spinner="simpleDots", stream=sys.stderr, enabled=not disable_progress):
+            config = SmdaConfig()
+            config.STORE_BUFFER = True
+            smda_disasm = Disassembler(config)
+            smda_report = smda_disasm.disassembleFile(path)
+
+        return capa.features.extractors.smda.SmdaFeatureExtractor(smda_report, path)
+    else:
+        import capa.features.extractors.viv
+
+        with halo.Halo(text="analyzing program", spinner="simpleDots", stream=sys.stderr, enabled=not disable_progress):
+            vw = get_workspace(path, format, should_save=False)
+
+            try:
+                vw.saveWorkspace()
+            except IOError:
+                # see #168 for discussion around how to handle non-writable directories
+                logger.info("source directory is not writable, won't save intermediate workspace")
+
+        return capa.features.extractors.viv.VivisectFeatureExtractor(vw, path)
 
 
-def get_extractor(path, format, disable_progress=False):
+def get_extractor(path, format, backend, disable_progress=False):
     """
     raises:
       UnsupportedFormatError:
     """
     if sys.version_info >= (3, 0):
-        return get_extractor_py3(path, format, disable_progress=disable_progress)
+        return get_extractor_py3(path, format, backend, disable_progress=disable_progress)
     else:
         return get_extractor_py2(path, format, disable_progress=disable_progress)
 
@@ -351,7 +390,13 @@ def get_rules(rule_path, disable_progress=False):
 
     rules = []
 
-    for rule_path in tqdm.tqdm(list(rule_paths), disable=disable_progress, desc="loading ", unit="     rules"):
+    pbar = tqdm.tqdm
+    if disable_progress:
+        # do not use tqdm to avoid unnecessary side effects when caller intends
+        # to disable progress completely
+        pbar = lambda s, *args, **kwargs: s
+
+    for rule_path in pbar(list(rule_paths), desc="loading ", unit="     rules"):
         try:
             rule = capa.rules.Rule.from_yaml_file(rule_path)
         except capa.rules.InvalidRule:
@@ -401,19 +446,162 @@ def collect_metadata(argv, sample_path, rules_path, format, extractor):
     }
 
 
+def install_common_args(parser, wanted=None):
+    """
+    register a common set of command line arguments for re-use by main & scripts.
+    these are things like logging/coloring/etc.
+    also enable callers to opt-in to common arguments, like specifying the input sample.
+
+    this routine lets many script use the same language for cli arguments.
+    see `handle_common_args` to do common configuration.
+
+    args:
+      parser (argparse.ArgumentParser): a parser to update in place, adding common arguments.
+      wanted (Set[str]): collection of arguments to opt-into, including:
+        - "sample": required positional argument to input file.
+        - "format": flag to override file format.
+        - "backend": flag to override analysis backend under py3.
+        - "rules": flag to override path to capa rules.
+        - "tag": flag to override/specify which rules to match.
+    """
+    if wanted is None:
+        wanted = set()
+
+    #
+    # common arguments that all scripts will have
+    #
+
+    parser.add_argument("--version", action="version", version="%(prog)s {:s}".format(capa.version.__version__))
+    parser.add_argument(
+        "-v", "--verbose", action="store_true", help="enable verbose result document (no effect with --json)"
+    )
+    parser.add_argument(
+        "-vv", "--vverbose", action="store_true", help="enable very verbose result document (no effect with --json)"
+    )
+    parser.add_argument("-d", "--debug", action="store_true", help="enable debugging output on STDERR")
+    parser.add_argument("-q", "--quiet", action="store_true", help="disable all output but errors")
+    parser.add_argument(
+        "--color",
+        type=str,
+        choices=("auto", "always", "never"),
+        default="auto",
+        help="enable ANSI color codes in results, default: only during interactive session",
+    )
+
+    #
+    # arguments that may be opted into:
+    #
+    #   - sample
+    #   - format
+    #   - rules
+    #   - tag
+    #
+
+    if "sample" in wanted:
+        if sys.version_info >= (3, 0):
+            parser.add_argument(
+                # Python 3 str handles non-ASCII arguments correctly
+                "sample",
+                type=str,
+                help="path to sample to analyze",
+            )
+        else:
+            parser.add_argument(
+                # in #328 we noticed that the sample path is not handled correctly if it contains non-ASCII characters
+                # https://stackoverflow.com/a/22947334/ offers a solution and decoding using getfilesystemencoding works
+                # in our testing, however other sources suggest `sys.stdin.encoding` (https://stackoverflow.com/q/4012571/)
+                "sample",
+                type=lambda s: s.decode(sys.getfilesystemencoding()),
+                help="path to sample to analyze",
+            )
+
+    if "format" in wanted:
+        formats = [
+            ("auto", "(default) detect file type automatically"),
+            ("pe", "Windows PE file"),
+            ("sc32", "32-bit shellcode"),
+            ("sc64", "64-bit shellcode"),
+            ("freeze", "features previously frozen by capa"),
+        ]
+        format_help = ", ".join(["%s: %s" % (f[0], f[1]) for f in formats])
+        parser.add_argument(
+            "-f",
+            "--format",
+            choices=[f[0] for f in formats],
+            default="auto",
+            help="select sample format, %s" % format_help,
+        )
+
+    if "backend" in wanted and sys.version_info >= (3, 0):
+        parser.add_argument(
+            "-b",
+            "--backend",
+            type=str,
+            help="select the backend to use",
+            choices=(BACKEND_VIV, BACKEND_SMDA),
+            default=BACKEND_VIV,
+        )
+
+    if "rules" in wanted:
+        parser.add_argument(
+            "-r",
+            "--rules",
+            type=str,
+            default=RULES_PATH_DEFAULT_STRING,
+            help="path to rule file or directory, use embedded rules by default",
+        )
+
+    if "tag" in wanted:
+        parser.add_argument("-t", "--tag", type=str, help="filter on rule meta field values")
+
+
+def handle_common_args(args):
+    """
+    handle the global config specified by `install_common_args`,
+    such as configuring logging/coloring/etc.
+
+    args:
+      args (argparse.Namespace): parsed arguments that included at least `install_common_args` args.
+    """
+    if args.quiet:
+        logging.basicConfig(level=logging.WARNING)
+        logging.getLogger().setLevel(logging.WARNING)
+    elif args.debug:
+        logging.basicConfig(level=logging.DEBUG)
+        logging.getLogger().setLevel(logging.DEBUG)
+    else:
+        logging.basicConfig(level=logging.INFO)
+        logging.getLogger().setLevel(logging.INFO)
+
+    # disable vivisect-related logging, it's verbose and not relevant for capa users
+    set_vivisect_log_level(logging.CRITICAL)
+
+    # py2 doesn't know about cp65001, which is a variant of utf-8 on windows
+    # tqdm bails when trying to render the progress bar in this setup.
+    # because cp65001 is utf-8, we just map that codepage to the utf-8 codec.
+    # see #380 and: https://stackoverflow.com/a/3259271/87207
+    import codecs
+
+    codecs.register(lambda name: codecs.lookup("utf-8") if name == "cp65001" else None)
+
+    if args.color == "always":
+        colorama.init(strip=False)
+    elif args.color == "auto":
+        # colorama will detect:
+        #  - when on Windows console, and fixup coloring, and
+        #  - when not an interactive session, and disable coloring
+        # renderers should use coloring and assume it will be stripped out if necessary.
+        colorama.init()
+    elif args.color == "never":
+        colorama.init(strip=True)
+    else:
+        raise RuntimeError("unexpected --color value: " + args.color)
+
+
 def main(argv=None):
     if argv is None:
         argv = sys.argv[1:]
 
-    formats = [
-        ("auto", "(default) detect file type automatically"),
-        ("pe", "Windows PE file"),
-        ("sc32", "32-bit shellcode"),
-        ("sc64", "64-bit shellcode"),
-        ("freeze", "features previously frozen by capa"),
-    ]
-    format_help = ", ".join(["%s: %s" % (f[0], f[1]) for f in formats])
-
     desc = "The FLARE team's open-source tool to identify capabilities in executable files."
     epilog = textwrap.dedent(
         """
@@ -446,64 +634,19 @@ def main(argv=None):
     parser = argparse.ArgumentParser(
         description=desc, epilog=epilog, formatter_class=argparse.RawDescriptionHelpFormatter
     )
-    parser.add_argument("sample", type=str, help="path to sample to analyze")
-    parser.add_argument("--version", action="version", version="%(prog)s {:s}".format(capa.version.__version__))
-    parser.add_argument(
-        "-r",
-        "--rules",
-        type=str,
-        default=RULES_PATH_DEFAULT_STRING,
-        help="path to rule file or directory, use embedded rules by default",
-    )
-    parser.add_argument(
-        "-f", "--format", choices=[f[0] for f in formats], default="auto", help="select sample format, %s" % format_help
-    )
-    parser.add_argument("-t", "--tag", type=str, help="filter on rule meta field values")
+    install_common_args(parser, {"sample", "format", "backend", "rules", "tag"})
     parser.add_argument("-j", "--json", action="store_true", help="emit JSON instead of text")
-    parser.add_argument(
-        "-v", "--verbose", action="store_true", help="enable verbose result document (no effect with --json)"
-    )
-    parser.add_argument(
-        "-vv", "--vverbose", action="store_true", help="enable very verbose result document (no effect with --json)"
-    )
-    parser.add_argument("-d", "--debug", action="store_true", help="enable debugging output on STDERR")
-    parser.add_argument("-q", "--quiet", action="store_true", help="disable all output but errors")
-    parser.add_argument(
-        "--color",
-        type=str,
-        choices=("auto", "always", "never"),
-        default="auto",
-        help="enable ANSI color codes in results, default: only during interactive session",
-    )
     args = parser.parse_args(args=argv)
-
-    if args.quiet:
-        logging.basicConfig(level=logging.WARNING)
-        logging.getLogger().setLevel(logging.WARNING)
-    elif args.debug:
-        logging.basicConfig(level=logging.DEBUG)
-        logging.getLogger().setLevel(logging.DEBUG)
-    else:
-        logging.basicConfig(level=logging.INFO)
-        logging.getLogger().setLevel(logging.INFO)
-
-    # disable vivisect-related logging, it's verbose and not relevant for capa users
-    set_vivisect_log_level(logging.CRITICAL)
+    handle_common_args(args)
 
     try:
         taste = get_file_taste(args.sample)
     except IOError as e:
-        logger.error("%s", str(e))
+        # per our research there's not a programmatic way to render the IOError with non-ASCII filename unless we
+        # handle the IOError separately and reach into the args
+        logger.error("%s", e.args[0])
         return -1
 
-    # py2 doesn't know about cp65001, which is a variant of utf-8 on windows
-    # tqdm bails when trying to render the progress bar in this setup.
-    # because cp65001 is utf-8, we just map that codepage to the utf-8 codec.
-    # see #380 and: https://stackoverflow.com/a/3259271/87207
-    import codecs
-
-    codecs.register(lambda name: codecs.lookup("utf-8") if name == "cp65001" else None)
-
     if args.rules == RULES_PATH_DEFAULT_STRING:
         logger.debug("-" * 80)
         logger.debug(" Using default embedded rules.")
@@ -536,7 +679,13 @@ def main(argv=None):
     try:
         rules = get_rules(rules_path, disable_progress=args.quiet)
         rules = capa.rules.RuleSet(rules)
-        logger.debug("successfully loaded %s rules", len(rules))
+        logger.debug(
+            "successfully loaded %s rules",
+            # during the load of the RuleSet, we extract subscope statements into their own rules
+            # that are subsequently `match`ed upon. this inflates the total rule count.
+            # so, filter out the subscope rules when reporting total number of loaded rules.
+            len([i for i in filter(lambda r: "capa/subscope-rule" not in r.meta, rules.rules.values())]),
+        )
         if args.tag:
             rules = rules.filter_rules_by_meta(args.tag)
             logger.debug("selected %s rules", len(rules))
@@ -554,7 +703,8 @@ def main(argv=None):
     else:
         format = args.format
         try:
-            extractor = get_extractor(args.sample, args.format, disable_progress=args.quiet)
+            backend = args.backend if sys.version_info > (3, 0) else BACKEND_VIV
+            extractor = get_extractor(args.sample, args.format, backend, disable_progress=args.quiet)
         except UnsupportedFormatError:
             logger.error("-" * 80)
             logger.error(" Input file does not appear to be a PE file.")
@@ -587,19 +737,6 @@ def main(argv=None):
         if not (args.verbose or args.vverbose or args.json):
             return -1
 
-    if args.color == "always":
-        colorama.init(strip=False)
-    elif args.color == "auto":
-        # colorama will detect:
-        #  - when on Windows console, and fixup coloring, and
-        #  - when not an interactive session, and disable coloring
-        # renderers should use coloring and assume it will be stripped out if necessary.
-        colorama.init()
-    elif args.color == "never":
-        colorama.init(strip=True)
-    else:
-        raise RuntimeError("unexpected --color value: " + args.color)
-
     if args.json:
         print(capa.render.render_json(meta, rules, capabilities))
     elif args.vverbose:

capa/render/default.py

@@ -161,6 +161,65 @@ def render_attack(doc, ostream):
         ostream.write("\n")
 
 
+def render_mbc(doc, ostream):
+    """
+    example::
+
+        +--------------------------+------------------------------------------------------------+
+        | MBC Objective            | MBC Behavior                                               |
+        |--------------------------+------------------------------------------------------------|
+        | ANTI-BEHAVIORAL ANALYSIS | Virtual Machine Detection::Instruction Testing [B0009.029] |
+        | COLLECTION               | Keylogging::Polling [F0002.002]                            |
+        | COMMUNICATION            | Interprocess Communication::Create Pipe [C0003.001]        |
+        |                          | Interprocess Communication::Write Pipe [C0003.004]         |
+        | IMPACT                   | Remote Access::Reverse Shell [B0022.001]                   |
+        +--------------------------+------------------------------------------------------------+
+    """
+    objectives = collections.defaultdict(set)
+    for rule in rutils.capability_rules(doc):
+        if not rule["meta"].get("mbc"):
+            continue
+
+        mbcs = rule["meta"]["mbc"]
+        if not isinstance(mbcs, list):
+            raise ValueError("invalid rule: MBC mapping is not a list")
+
+        for mbc in mbcs:
+            objective, _, rest = mbc.partition("::")
+            if "::" in rest:
+                behavior, _, rest = rest.partition("::")
+                method, _, id = rest.rpartition(" ")
+                objectives[objective].add((behavior, method, id))
+            else:
+                behavior, _, id = rest.rpartition(" ")
+                objectives[objective].add((behavior, id))
+
+    rows = []
+    for objective, behaviors in sorted(objectives.items()):
+        inner_rows = []
+        for spec in sorted(behaviors):
+            if len(spec) == 2:
+                behavior, id = spec
+                inner_rows.append("%s %s" % (rutils.bold(behavior), id))
+            elif len(spec) == 3:
+                behavior, method, id = spec
+                inner_rows.append("%s::%s %s" % (rutils.bold(behavior), method, id))
+            else:
+                raise RuntimeError("unexpected MBC spec format")
+        rows.append(
+            (
+                rutils.bold(objective.upper()),
+                "\n".join(inner_rows),
+            )
+        )
+
+    if rows:
+        ostream.write(
+            tabulate.tabulate(rows, headers=[width("MBC Objective", 25), width("MBC Behavior", 75)], tablefmt="psql")
+        )
+        ostream.write("\n")
+
+
 def render_default(doc):
     ostream = rutils.StringIO()
 
@@ -168,6 +227,8 @@ def render_default(doc):
     ostream.write("\n")
     render_attack(doc, ostream)
     ostream.write("\n")
+    render_mbc(doc, ostream)
+    ostream.write("\n")
     render_capabilities(doc, ostream)
 
     return ostream.getvalue()

capa/rules.py

@@ -6,13 +6,20 @@
 #  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 
+import re
 import uuid
 import codecs
 import logging
 import binascii
 import functools
 
+try:
+    from functools import lru_cache
+except ImportError:
+    from backports.functools_lru_cache import lru_cache
+
 import six
+import yaml
 import ruamel.yaml
 
 import capa.engine
@@ -25,7 +32,6 @@ from capa.features import MAX_BYTES_FEATURE_SIZE
 
 logger = logging.getLogger(__name__)
 
-
 # these are the standard metadata fields, in the preferred order.
 # when reformatted, any custom keys will come after these.
 META_KEYS = (
@@ -271,27 +277,63 @@ def parse_description(s, value_type, description=None):
     return value, description
 
 
+def pop_statement_description_entry(d):
+    """
+    extracts the description for statements and removes the description entry from the document
+    a statement can only have one description
+
+    example:
+    the features definition
+      - or:
+        - description: statement description
+        - number: 1
+          description: feature description
+
+    becomes
+      <statement>: [
+        { "description": "statement description" },  <-- extracted here
+        { "number": 1, "description": "feature description" }
+      ]
+    """
+    if not isinstance(d, list):
+        return None
+
+    # identify child of form '{ "description": <description> }'
+    descriptions = list(filter(lambda c: isinstance(c, dict) and len(c) == 1 and "description" in c, d))
+    if len(descriptions) > 1:
+        raise InvalidRule("statements can only have one description")
+
+    if not descriptions:
+        return None
+
+    description = descriptions[0]
+    d.remove(description)
+
+    return description["description"]
+
+
 def build_statements(d, scope):
     if len(d.keys()) > 2:
         raise InvalidRule("too many statements")
 
     key = list(d.keys())[0]
+    description = pop_statement_description_entry(d[key])
     if key == "and":
-        return And([build_statements(dd, scope) for dd in d[key]], description=d.get("description"))
+        return And([build_statements(dd, scope) for dd in d[key]], description=description)
     elif key == "or":
-        return Or([build_statements(dd, scope) for dd in d[key]], description=d.get("description"))
+        return Or([build_statements(dd, scope) for dd in d[key]], description=description)
     elif key == "not":
         if len(d[key]) != 1:
             raise InvalidRule("not statement must have exactly one child statement")
-        return Not(build_statements(d[key][0], scope), description=d.get("description"))
+        return Not(build_statements(d[key][0], scope), description=description)
     elif key.endswith(" or more"):
         count = int(key[: -len("or more")])
-        return Some(count, [build_statements(dd, scope) for dd in d[key]], description=d.get("description"))
+        return Some(count, [build_statements(dd, scope) for dd in d[key]], description=description)
     elif key == "optional":
         # `optional` is an alias for `0 or more`
         # which is useful for documenting behaviors,
         # like with `write file`, we might say that `WriteFile` is optionally found alongside `CreateFileA`.
-        return Some(0, [build_statements(dd, scope) for dd in d[key]], description=d.get("description"))
+        return Some(0, [build_statements(dd, scope) for dd in d[key]], description=description)
 
     elif key == "function":
         if scope != FILE_SCOPE:
@@ -350,18 +392,18 @@ def build_statements(d, scope):
 
         count = d[key]
         if isinstance(count, int):
-            return Range(feature, min=count, max=count, description=d.get("description"))
+            return Range(feature, min=count, max=count, description=description)
         elif count.endswith(" or more"):
             min = parse_int(count[: -len(" or more")])
             max = None
-            return Range(feature, min=min, max=max, description=d.get("description"))
+            return Range(feature, min=min, max=max, description=description)
         elif count.endswith(" or fewer"):
             min = None
             max = parse_int(count[: -len(" or fewer")])
-            return Range(feature, min=min, max=max, description=d.get("description"))
+            return Range(feature, min=min, max=max, description=description)
         elif count.startswith("("):
             min, max = parse_range(count)
-            return Range(feature, min=min, max=max, description=d.get("description"))
+            return Range(feature, min=min, max=max, description=description)
         else:
             raise InvalidRule("unexpected range: %s" % (count))
     elif key == "string" and not isinstance(d[key], six.string_types):
@@ -385,26 +427,6 @@ def second(s):
     return s[1]
 
 
-# we use the ruamel.yaml parser because it supports roundtripping of documents with comments.
-yaml = ruamel.yaml.YAML(typ="rt")
-
-
-# use block mode, not inline json-like mode
-yaml.default_flow_style = False
-
-
-# indent lists by two spaces below their parent
-#
-#     features:
-#       - or:
-#         - mnemonic: aesdec
-#         - mnemonic: vaesdec
-yaml.indent(sequence=2, offset=2)
-
-# avoid word wrapping
-yaml.width = 4096
-
-
 class Rule(object):
     def __init__(self, name, scope, statement, meta, definition=""):
         super(Rule, self).__init__()
@@ -533,7 +555,7 @@ class Rule(object):
         return self.statement.evaluate(features)
 
     @classmethod
-    def from_dict(cls, d, s):
+    def from_dict(cls, d, definition):
         name = d["rule"]["meta"]["name"]
         # if scope is not specified, default to function scope.
         # this is probably the mode that rule authors will start with.
@@ -551,17 +573,65 @@ class Rule(object):
         if scope not in SUPPORTED_FEATURES.keys():
             raise InvalidRule("{:s} is not a supported scope".format(scope))
 
-        return cls(name, scope, build_statements(statements[0], scope), d["rule"]["meta"], s)
+        return cls(name, scope, build_statements(statements[0], scope), d["rule"]["meta"], definition)
+
+    @staticmethod
+    @lru_cache()
+    def _get_yaml_loader():
+        try:
+            # prefer to use CLoader to be fast, see #306
+            # on Linux, make sure you install libyaml-dev or similar
+            # on Windows, get WHLs from pyyaml.org/pypi
+            loader = yaml.CLoader
+            logger.debug("using libyaml CLoader.")
+        except:
+            loader = yaml.Loader
+            logger.debug("unable to import libyaml CLoader, falling back to Python yaml parser.")
+            logger.debug("this will be slower to load rules.")
+
+        return loader
+
+    @staticmethod
+    def _get_ruamel_yaml_parser():
+        # use ruamel to enable nice formatting
+
+        # we use the ruamel.yaml parser because it supports roundtripping of documents with comments.
+        y = ruamel.yaml.YAML(typ="rt")
+
+        # use block mode, not inline json-like mode
+        y.default_flow_style = False
+
+        # leave quotes unchanged
+        y.preserve_quotes = True
+
+        # indent lists by two spaces below their parent
+        #
+        #     features:
+        #       - or:
+        #         - mnemonic: aesdec
+        #         - mnemonic: vaesdec
+        y.indent(sequence=2, offset=2)
+
+        # avoid word wrapping
+        y.width = 4096
+
+        return y
 
     @classmethod
-    def from_yaml(cls, s):
-        return cls.from_dict(yaml.load(s), s)
+    def from_yaml(cls, s, use_ruamel=False):
+        if use_ruamel:
+            # ruamel enables nice formatting and doc roundtripping with comments
+            doc = cls._get_ruamel_yaml_parser().load(s)
+        else:
+            # use pyyaml because it can be much faster than ruamel (pure python)
+            doc = yaml.load(s, Loader=cls._get_yaml_loader())
+        return cls.from_dict(doc, s)
 
     @classmethod
-    def from_yaml_file(cls, path):
+    def from_yaml_file(cls, path, use_ruamel=False):
         with open(path, "rb") as f:
             try:
-                return cls.from_yaml(f.read().decode("utf-8"))
+                return cls.from_yaml(f.read().decode("utf-8"), use_ruamel=use_ruamel)
             except InvalidRule as e:
                 raise InvalidRuleWithPath(path, str(e))
 
@@ -575,12 +645,25 @@ class Rule(object):
         # but not for rule logic.
         # programmatic generation of rules is not yet supported.
 
-        definition = yaml.load(self.definition)
-        # definition retains a reference to `meta`,
-        # so we're updating that in place.
-        definition["rule"]["meta"] = self.meta
-        meta = self.meta
+        # use ruamel because it supports round tripping.
+        # pyyaml will lose the existing ordering of rule statements.
+        definition = self._get_ruamel_yaml_parser().load(self.definition)
 
+        # we want to apply any updates that have been made to `meta`.
+        # so we would like to assigned it like this:
+        #
+        #     definition["rule"]["meta"] = self.meta
+        #
+        # however, `self.meta` is not ordered, its just a dict, so subsequent formatting doesn't work.
+        # so, we'll manually copy the keys over, re-using the existing ordereddict/CommentedMap
+        meta = definition["rule"]["meta"]
+        for k in meta.keys():
+            if k not in self.meta:
+                del meta[k]
+        for k, v in self.meta.items():
+            meta[k] = v
+
+        # the name and scope of the rule instance overrides anything in meta.
         meta["name"] = self.name
         meta["scope"] = self.scope
 
@@ -617,7 +700,7 @@ class Rule(object):
             del meta[key]
 
         ostream = six.BytesIO()
-        yaml.dump(definition, ostream)
+        self._get_ruamel_yaml_parser().dump(definition, ostream)
 
         for key, value in hidden_meta.items():
             if value is None:
@@ -641,7 +724,18 @@ class Rule(object):
         # tweaking `ruamel.indent()` doesn't quite give us the control we want.
         # so, add the two extra spaces that we've determined we need through experimentation.
         # see #263
-        doc = doc.replace("  description:", "    description:")
+        # only do this for the features section, so the meta description doesn't get reformatted
+        # assumes features section always exists
+        features_offset = doc.find("features")
+        doc = doc[:features_offset] + doc[features_offset:].replace("  description:", "    description:")
+
+        # for negative hex numbers, yaml dump outputs:
+        # - offset: !!int '0x-30'
+        # we prefer:
+        # - offset: -0x30
+        # the below regex makes these adjustments and while ugly, we don't have to explore the ruamel.yaml insides
+        doc = re.sub(r"!!int '0x-([0-9a-fA-F]+)'", r"-0x\1", doc)
+
         return doc
 
 
@@ -791,7 +885,8 @@ class RuleSet(object):
         given a collection of rules, collect the rules that are needed at the given scope.
         these rules are ordered topologically.
 
-        don't include "lib" rules, unless they are dependencies of other rules.
+        don't include auto-generated "subscope" rules.
+        we want to include general "lib" rules here - even if they are not dependencies of other rules, see #398
         """
         scope_rules = set([])
 
@@ -800,7 +895,7 @@ class RuleSet(object):
         #  at lower scope, e.g. function scope.
         # so, we find all dependencies of all rules, and later will filter them down.
         for rule in rules:
-            if rule.meta.get("lib", False):
+            if rule.meta.get("capa/subscope-rule", False):
                 continue
 
             scope_rules.update(get_rules_and_dependencies(rules, rule.name))

capa/version.py

@@ -1 +1 @@
-__version__ = "1.3.0"
+__version__ = "1.6.0"

doc/installation.md

@@ -22,15 +22,15 @@ By default, on MacOS Catalina or greater, Gatekeeper will block execution of the
 ![approve dialog](img/approve.png)
 
 ## Method 2: Using capa as a Python library
-To install capa as a Python library, you'll need to install a few dependencies, and then use `pip` to fetch the capa module.
+To install capa as a Python library use `pip` to fetch the `flare-capa` module.
 
 #### *Note*:
-This method is appropriate for integrating capa in an existing project. It is not the right choice for local tool usage, such as within IDA Pro - see Method 3, instead.
-That's because this technique doesn't pull the default rule set, so you should check it out separately from [capa-rules](https://github.com/fireeye/capa-rules/) and pass the directory to the entrypoint using `-r`.
+This method is appropriate for integrating capa in an existing project.
+This technique doesn't pull the default rule set, so you should check it out separately from [capa-rules](https://github.com/fireeye/capa-rules/) and pass the directory to the entrypoint using `-r` or set the rules path in the IDA Pro plugin.
+Alternatively, see Method 3 below.
 
 ### 1. Install capa module
-Second, use `pip` to install the capa module to your local Python environment. This fetches the library code to your computer but does not keep editable source files around for you to hack on. If you'd like to edit the source files, see below.
-`$ pip install https://github.com/fireeye/capa/archive/master.zip`
+Use `pip` to install the capa module to your local Python environment. This fetches the library code to your computer but does not keep editable source files around for you to hack on. If you'd like to edit the source files, see below. `$ pip install flare-capa`
 
 ### 2. Use capa
 You can now import the `capa` module from a Python script or use the IDA Pro plugins from the `capa/ida` directory. For more information please see the [usage](usage.md) documentation.
@@ -74,8 +74,20 @@ Note that some development dependencies (including the black code formatter) req
 To check the code style, formatting and run the tests you can run the script `scripts/ci.sh`.
 You can run it with the argument `no_tests` to skip the tests and only run the code style and formatting: `scripts/ci.sh no_tests`
 
-### 3. Setup hooks [optional]
+### 3. Compile binary using PyInstaller
+We compile capa standalone binaries using PyInstaller. To reproduce the build process check out the source code as described above and follow these steps.
 
+#### Install PyInstaller:
+For Python 2.7: `$ pip install 'pyinstaller==3.*'` (PyInstaller 4 doesn't support Python 2.7)
+
+For Python 3: `$ pip install 'pyinstaller`
+
+#### Run Pyinstaller
+`$ pyinstaller .github/pyinstaller/pyinstaller.spec`
+
+You can find the compiled binary in the created directory `dist/`.
+
+### 4. Setup hooks [optional]
 If you plan to contribute to capa, you may want to setup the hooks.
 Run `scripts/setup-hooks.sh` to set the following hooks up:
 - The `pre-commit` hook runs checks before every `git commit`.
@@ -84,4 +96,3 @@ Run `scripts/setup-hooks.sh` to set the following hooks up:
 - The `pre-push` hook runs checks before every `git push`.
   It runs `scripts/ci.sh` aborting the push if there are code style or rule linter offenses or if the tests fail.
   This way you can ensure everything is alright before sending a pull request.
-

doc/usage.md

@@ -4,11 +4,10 @@ See `capa -h` for all supported arguments and usage examples.
 
 ## tips and tricks
 
-  - [match only rules by given author or namespace](#only-run-selected-rules)
-  - [IDA Pro capa explorer](#capa-explorer)
-  - [IDA Pro rule generator](#rule-generator)
-
 ### only run selected rules
 Use the `-t` option to run rules with the given metadata value (see the rule fields `rule.meta.*`).
 For example, `capa -t william.ballenthin@mandiant.com` runs rules that reference Willi's email address (probably as the author), or
 `capa -t communication` runs rules with the namespace `communication`.
+
+### IDA Pro plugin: capa explorer
+Please check out the [capa explorer documentation](/capa/ida/plugin/README.md).

scripts/bulk-process.py

@@ -0,0 +1,220 @@
+#!/usr/bin/env python
+"""
+bulk-process
+
+Invoke capa recursively against a directory of samples
+and emit a JSON document mapping the file paths to their results.
+
+By default, this will use subprocesses for parallelism.
+Use `-n/--parallelism` to change the subprocess count from
+ the default of current CPU count.
+Use `--no-mp` to use threads instead of processes,
+ which is probably not useful unless you set `--parallelism=1`.
+
+example:
+
+    $ python scripts/bulk-process /tmp/suspicious
+    {
+      "/tmp/suspicious/suspicious.dll_": {
+        "rules": {
+          "encode data using XOR": {
+            "matches": {
+              "268440358": {
+              [...]
+      "/tmp/suspicious/1.dll_": { ... }
+      "/tmp/suspicious/2.dll_": { ... }
+    }
+
+
+usage:
+
+    usage: bulk-process.py [-h] [-r RULES] [-d] [-q] [-n PARALLELISM] [--no-mp]
+                           input
+
+    detect capabilities in programs.
+
+    positional arguments:
+      input                 Path to directory of files to recursively analyze
+
+    optional arguments:
+      -h, --help            show this help message and exit
+      -r RULES, --rules RULES
+                            Path to rule file or directory, use embedded rules by
+                            default
+      -d, --debug           Enable debugging output on STDERR
+      -q, --quiet           Disable all output but errors
+      -n PARALLELISM, --parallelism PARALLELISM
+                            parallelism factor
+      --no-mp               disable subprocesses
+
+Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
+Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+You may obtain a copy of the License at: [package root]/LICENSE.txt
+Unless required by applicable law or agreed to in writing, software distributed under the License
+ is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and limitations under the License.
+"""
+import sys
+import json
+import logging
+import os.path
+import argparse
+import multiprocessing
+import multiprocessing.pool
+
+import capa
+import capa.main
+import capa.rules
+import capa.render
+
+logger = logging.getLogger("capa")
+
+
+def get_capa_results(args):
+    """
+    run capa against the file at the given path, using the given rules.
+
+    args is a tuple, containing:
+      rules (capa.rules.RuleSet): the rules to match
+      format (str): the name of the sample file format
+      path (str): the file system path to the sample to process
+
+    args is a tuple because i'm not quite sure how to unpack multiple arguments using `map`.
+
+    returns an dict with two required keys:
+      path (str): the file system path of the sample to process
+      status (str): either "error" or "ok"
+
+    when status == "error", then a human readable message is found in property "error".
+    when status == "ok", then the capa results are found in the property "ok".
+
+    the capa results are a dictionary with the following keys:
+      meta (dict): the meta analysis results
+      capabilities (dict): the matched capabilities and their result objects
+    """
+    rules, format, path = args
+    logger.info("computing capa results for: %s", path)
+    try:
+        extractor = capa.main.get_extractor(path, format, capa.main.BACKEND_VIV, disable_progress=True)
+    except capa.main.UnsupportedFormatError:
+        # i'm 100% sure if multiprocessing will reliably raise exceptions across process boundaries.
+        # so instead, return an object with explicit success/failure status.
+        #
+        # if success, then status=ok, and results found in property "ok"
+        # if error, then status=error, and human readable message in property "error"
+        return {
+            "path": path,
+            "status": "error",
+            "error": "input file does not appear to be a PE file: %s" % path,
+        }
+    except capa.main.UnsupportedRuntimeError:
+        return {
+            "path": path,
+            "status": "error",
+            "error": "unsupported runtime or Python interpreter",
+        }
+    except Exception as e:
+        return {
+            "path": path,
+            "status": "error",
+            "error": "unexpected error: %s" % (e),
+        }
+
+    meta = capa.main.collect_metadata("", path, "", format, extractor)
+    capabilities, counts = capa.main.find_capabilities(rules, extractor, disable_progress=True)
+    meta["analysis"].update(counts)
+
+    return {
+        "path": path,
+        "status": "ok",
+        "ok": {
+            "meta": meta,
+            "capabilities": capabilities,
+        },
+    }
+
+
+def main(argv=None):
+    if argv is None:
+        argv = sys.argv[1:]
+
+        parser = argparse.ArgumentParser(description="detect capabilities in programs.")
+        capa.main.install_common_args(parser, wanted={"rules"})
+        parser.add_argument("input", type=str, help="Path to directory of files to recursively analyze")
+        parser.add_argument(
+            "-n", "--parallelism", type=int, default=multiprocessing.cpu_count(), help="parallelism factor"
+        )
+        parser.add_argument("--no-mp", action="store_true", help="disable subprocesses")
+        args = parser.parse_args(args=argv)
+        capa.main.handle_common_args(args)
+
+        if args.rules == "(embedded rules)":
+            logger.info("using default embedded rules")
+            logger.debug("detected running from source")
+            args.rules = os.path.join(os.path.dirname(__file__), "..", "rules")
+            logger.debug("default rule path (source method): %s", args.rules)
+        else:
+            logger.info("using rules path: %s", args.rules)
+
+        try:
+            rules = capa.main.get_rules(args.rules)
+            rules = capa.rules.RuleSet(rules)
+            logger.info("successfully loaded %s rules", len(rules))
+        except (IOError, capa.rules.InvalidRule, capa.rules.InvalidRuleSet) as e:
+            logger.error("%s", str(e))
+            return -1
+
+        samples = []
+        for (base, directories, files) in os.walk(args.input):
+            for file in files:
+                samples.append(os.path.join(base, file))
+
+        def pmap(f, args, parallelism=multiprocessing.cpu_count()):
+            """apply the given function f to the given args using subprocesses"""
+            return multiprocessing.Pool(parallelism).imap(f, args)
+
+        def tmap(f, args, parallelism=multiprocessing.cpu_count()):
+            """apply the given function f to the given args using threads"""
+            return multiprocessing.pool.ThreadPool(parallelism).imap(f, args)
+
+        def map(f, args, parallelism=None):
+            """apply the given function f to the given args in the current thread"""
+            for arg in args:
+                yield f(arg)
+
+        if args.no_mp:
+            if args.parallelism == 1:
+                logger.debug("using current thread mapper")
+                mapper = map
+            else:
+                logger.debug("using threading mapper")
+                mapper = tmap
+        else:
+            logger.debug("using process mapper")
+            mapper = pmap
+
+        results = {}
+        for result in mapper(
+            get_capa_results, [(rules, "pe", sample) for sample in samples], parallelism=args.parallelism
+        ):
+            if result["status"] == "error":
+                logger.warning(result["error"])
+            elif result["status"] == "ok":
+                meta = result["ok"]["meta"]
+                capabilities = result["ok"]["capabilities"]
+                # our renderer expects to emit a json document for a single sample
+                # so we deserialize the json document, store it in a larger dict, and we'll subsequently re-encode.
+                results[result["path"]] = json.loads(capa.render.render_json(meta, rules, capabilities))
+            else:
+                raise ValueError("unexpected status: %s" % (result["status"]))
+
+        print(json.dumps(results))
+
+        logger.info("done.")
+
+        return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

scripts/capa_as_library.py

@@ -0,0 +1,215 @@
+#!/usr/bin/env python3
+
+import json
+import collections
+
+import capa.main
+import capa.rules
+import capa.engine
+import capa.render
+import capa.features
+import capa.render.utils as rutils
+from capa.engine import *
+from capa.render import convert_capabilities_to_result_document
+
+# edit this to set the path for file to analyze and rule directory
+RULES_PATH = "/tmp/capa/rules/"
+
+# load rules from disk
+rules = capa.main.get_rules(RULES_PATH, disable_progress=True)
+rules = capa.rules.RuleSet(rules)
+
+# == Render ddictionary helpers
+def render_meta(doc, ostream):
+    ostream["md5"] = doc["meta"]["sample"]["md5"]
+    ostream["sha1"] = doc["meta"]["sample"]["sha1"]
+    ostream["sha256"] = doc["meta"]["sample"]["sha256"]
+    ostream["path"] = doc["meta"]["sample"]["path"]
+
+
+def find_subrule_matches(doc):
+    """
+    collect the rule names that have been matched as a subrule match.
+    this way we can avoid displaying entries for things that are too specific.
+    """
+    matches = set([])
+
+    def rec(node):
+        if not node["success"]:
+            # there's probably a bug here for rules that do `not: match: ...`
+            # but we don't have any examples of this yet
+            return
+
+        elif node["node"]["type"] == "statement":
+            for child in node["children"]:
+                rec(child)
+
+        elif node["node"]["type"] == "feature":
+            if node["node"]["feature"]["type"] == "match":
+                matches.add(node["node"]["feature"]["match"])
+
+    for rule in rutils.capability_rules(doc):
+        for node in rule["matches"].values():
+            rec(node)
+
+    return matches
+
+
+def render_capabilities(doc, ostream):
+    """
+    example::
+        {'CAPABILITY': {'accept command line arguments': 'host-interaction/cli',
+                'allocate thread local storage (2 matches)': 'host-interaction/process',
+                'check for time delay via GetTickCount': 'anti-analysis/anti-debugging/debugger-detection',
+                'check if process is running under wine': 'anti-analysis/anti-emulation/wine',
+                'contain a resource (.rsrc) section': 'executable/pe/section/rsrc',
+                'write file (3 matches)': 'host-interaction/file-system/write'}
+        }
+    """
+    subrule_matches = find_subrule_matches(doc)
+
+    ostream["CAPABILITY"] = dict()
+    for rule in rutils.capability_rules(doc):
+        if rule["meta"]["name"] in subrule_matches:
+            # rules that are also matched by other rules should not get rendered by default.
+            # this cuts down on the amount of output while giving approx the same detail.
+            # see #224
+            continue
+
+        count = len(rule["matches"])
+        if count == 1:
+            capability = rule["meta"]["name"]
+        else:
+            capability = "%s (%d matches)" % (rule["meta"]["name"], count)
+
+        ostream["CAPABILITY"].setdefault(rule["meta"]["namespace"], list())
+        ostream["CAPABILITY"][rule["meta"]["namespace"]].append(capability)
+
+
+def render_attack(doc, ostream):
+    """
+    example::
+        {'ATT&CK': {'COLLECTION': ['Input Capture::Keylogging [T1056.001]'],
+            'DEFENSE EVASION': ['Obfuscated Files or Information [T1027]',
+                                'Virtualization/Sandbox Evasion::System Checks '
+                                '[T1497.001]'],
+            'DISCOVERY': ['File and Directory Discovery [T1083]',
+                          'Query Registry [T1012]',
+                          'System Information Discovery [T1082]'],
+            'EXECUTION': ['Shared Modules [T1129]']}
+        }
+    """
+    ostream["ATTCK"] = dict()
+    tactics = collections.defaultdict(set)
+    for rule in rutils.capability_rules(doc):
+        if not rule["meta"].get("att&ck"):
+            continue
+
+        for attack in rule["meta"]["att&ck"]:
+            tactic, _, rest = attack.partition("::")
+            if "::" in rest:
+                technique, _, rest = rest.partition("::")
+                subtechnique, _, id = rest.rpartition(" ")
+                tactics[tactic].add((technique, subtechnique, id))
+            else:
+                technique, _, id = rest.rpartition(" ")
+                tactics[tactic].add((technique, id))
+
+    for tactic, techniques in sorted(tactics.items()):
+        inner_rows = []
+        for spec in sorted(techniques):
+            if len(spec) == 2:
+                technique, id = spec
+                inner_rows.append("%s %s" % (technique, id))
+            elif len(spec) == 3:
+                technique, subtechnique, id = spec
+                inner_rows.append("%s::%s %s" % (technique, subtechnique, id))
+            else:
+                raise RuntimeError("unexpected ATT&CK spec format")
+        ostream["ATTCK"].setdefault(tactic.upper(), inner_rows)
+
+
+def render_mbc(doc, ostream):
+    """
+    example::
+        {'MBC': {'ANTI-BEHAVIORAL ANALYSIS': ['Debugger Detection::Timing/Delay Check '
+                                      'GetTickCount [B0001.032]',
+                                      'Emulator Detection [B0004]',
+                                      'Virtual Machine Detection::Instruction '
+                                      'Testing [B0009.029]',
+                                      'Virtual Machine Detection [B0009]'],
+         'COLLECTION': ['Keylogging::Polling [F0002.002]'],
+         'CRYPTOGRAPHY': ['Encrypt Data::RC4 [C0027.009]',
+                          'Generate Pseudo-random Sequence::RC4 PRGA '
+                          '[C0021.004]']}
+        }
+    """
+    ostream["MBC"] = dict()
+    objectives = collections.defaultdict(set)
+    for rule in rutils.capability_rules(doc):
+        if not rule["meta"].get("mbc"):
+            continue
+
+        mbcs = rule["meta"]["mbc"]
+        if not isinstance(mbcs, list):
+            raise ValueError("invalid rule: MBC mapping is not a list")
+
+        for mbc in mbcs:
+            objective, _, rest = mbc.partition("::")
+            if "::" in rest:
+                behavior, _, rest = rest.partition("::")
+                method, _, id = rest.rpartition(" ")
+                objectives[objective].add((behavior, method, id))
+            else:
+                behavior, _, id = rest.rpartition(" ")
+                objectives[objective].add((behavior, id))
+
+    for objective, behaviors in sorted(objectives.items()):
+        inner_rows = []
+        for spec in sorted(behaviors):
+            if len(spec) == 2:
+                behavior, id = spec
+                inner_rows.append("%s %s" % (behavior, id))
+            elif len(spec) == 3:
+                behavior, method, id = spec
+                inner_rows.append("%s::%s %s" % (behavior, method, id))
+            else:
+                raise RuntimeError("unexpected MBC spec format")
+        ostream["MBC"].setdefault(objective.upper(), inner_rows)
+
+
+def render_dictionary(doc):
+    ostream = dict()
+    render_meta(doc, ostream)
+    render_attack(doc, ostream)
+    render_mbc(doc, ostream)
+    render_capabilities(doc, ostream)
+
+    return ostream
+
+
+# ==== render dictionary helpers
+def capa_details(file_path, output_format="dictionary"):
+
+    # extract features and find capabilities
+    extractor = capa.main.get_extractor(file_path, "auto", capa.main.BACKEND_VIV, disable_progress=True)
+    capabilities, counts = capa.main.find_capabilities(rules, extractor, disable_progress=True)
+
+    # collect metadata (used only to make rendering more complete)
+    meta = capa.main.collect_metadata("", file_path, RULES_PATH, "auto", extractor)
+    meta["analysis"].update(counts)
+
+    capa_output = False
+    if output_format == "dictionary":
+        # ...as python dictionary, simplified as textable but in dictionary
+        doc = convert_capabilities_to_result_document(meta, rules, capabilities)
+        capa_output = render_dictionary(doc)
+    elif output_format == "json":
+        # render results
+        # ...as json
+        capa_output = json.loads(capa.render.render_json(meta, rules, capabilities))
+    elif output_format == "texttable":
+        # ...as human readable text table
+        capa_output = capa.render.render_default(meta, rules, capabilities)
+
+    return capa_output

scripts/capafmt.py

@@ -38,6 +38,12 @@ def main(argv=None):
     )
     parser.add_argument("-v", "--verbose", action="store_true", help="Enable debug logging")
     parser.add_argument("-q", "--quiet", action="store_true", help="Disable all output but errors")
+    parser.add_argument(
+        "-c",
+        "--check",
+        action="store_true",
+        help="Don't output (reformatted) rule, only return status. 0 = no changes, 1 = would reformat",
+    )
     args = parser.parse_args(args=argv)
 
     if args.verbose:
@@ -50,12 +56,22 @@ def main(argv=None):
     logging.basicConfig(level=level)
     logging.getLogger("capafmt").setLevel(level)
 
-    rule = capa.rules.Rule.from_yaml_file(args.path)
+    rule = capa.rules.Rule.from_yaml_file(args.path, use_ruamel=True)
+    reformatted_rule = rule.to_yaml()
+
+    if args.check:
+        if rule.definition == reformatted_rule:
+            logger.info("rule is formatted correctly, nice! (%s)", rule.name)
+            return 0
+        else:
+            logger.info("rule requires reformatting (%s)", rule.name)
+            return 1
+
     if args.in_place:
         with open(args.path, "wb") as f:
-            f.write(rule.to_yaml().encode("utf-8"))
+            f.write(reformatted_rule.encode("utf-8"))
     else:
-        print(rule.to_yaml().rstrip("\n"))
+        print(reformatted_rule)
 
     return 0
 

scripts/import-to-ida.py

@@ -31,10 +31,8 @@ See the License for the specific language governing permissions and limitations
 import json
 import logging
 
-import idc
 import idautils
 import ida_funcs
-import ida_idaapi
 import ida_kernwin
 
 logger = logging.getLogger("capa")

scripts/lint.py

@@ -15,7 +15,9 @@ See the License for the specific language governing permissions and limitations
 """
 import os
 import sys
+import time
 import string
+import difflib
 import hashlib
 import logging
 import os.path
@@ -24,6 +26,7 @@ import itertools
 import posixpath
 
 import capa.main
+import capa.rules
 import capa.engine
 import capa.features
 import capa.features.insn
@@ -32,7 +35,11 @@ logger = logging.getLogger("capa.lint")
 
 
 class Lint(object):
+    WARN = "WARN"
+    FAIL = "FAIL"
+
     name = "lint"
+    level = FAIL
     recommendation = ""
 
     def check_rule(self, ctx, rule):
@@ -194,7 +201,7 @@ class DoesntMatchExample(Lint):
                 continue
 
             try:
-                extractor = capa.main.get_extractor(path, "auto")
+                extractor = capa.main.get_extractor(path, "auto", capa.main.BACKEND_VIV, disable_progress=True)
                 capabilities, meta = capa.main.find_capabilities(ctx["rules"], extractor, disable_progress=True)
             except Exception as e:
                 logger.error("failed to extract capabilities: %s %s %s", rule.name, path, e)
@@ -232,7 +239,7 @@ class LibRuleNotInLibDirectory(Lint):
         if "lib" not in rule.meta:
             return False
 
-        return "/lib/" not in get_normpath(rule.meta["capa/path"])
+        return "lib/" not in get_normpath(rule.meta["capa/path"])
 
 
 class LibRuleHasNamespace(Lint):
@@ -276,6 +283,50 @@ class FeatureNegativeNumber(Lint):
         return False
 
 
+class FeatureNtdllNtoskrnlApi(Lint):
+    name = "feature api may overlap with ntdll and ntoskrnl"
+    level = Lint.WARN
+    recommendation = (
+        "check if {:s} is exported by both ntdll and ntoskrnl; if true, consider removing {:s} "
+        "module requirement to improve detection"
+    )
+
+    def check_features(self, ctx, features):
+        for feature in features:
+            if isinstance(feature, capa.features.insn.API):
+                modname, _, impname = feature.value.rpartition(".")
+                if modname in ("ntdll", "ntoskrnl"):
+                    self.recommendation = self.recommendation.format(impname, modname)
+                    return True
+        return False
+
+
+class FormatSingleEmptyLineEOF(Lint):
+    name = "EOF format"
+    recommendation = "end file with a single empty line"
+
+    def check_rule(self, ctx, rule):
+        if rule.definition.endswith("\n") and not rule.definition.endswith("\n\n"):
+            return False
+        return True
+
+
+class FormatIncorrect(Lint):
+    name = "rule format incorrect"
+    recommendation_template = "use scripts/capafmt.py or adjust as follows\n{:s}"
+
+    def check_rule(self, ctx, rule):
+        actual = rule.definition
+        expected = capa.rules.Rule.from_yaml(rule.definition, use_ruamel=True).to_yaml()
+
+        if actual != expected:
+            diff = difflib.ndiff(actual.splitlines(1), expected.splitlines(True))
+            self.recommendation = self.recommendation_template.format("".join(diff))
+            return True
+
+        return False
+
+
 def run_lints(lints, ctx, rule):
     for lint in lints:
         if lint.check_rule(ctx, rule):
@@ -325,14 +376,7 @@ def lint_meta(ctx, rule):
     return run_lints(META_LINTS, ctx, rule)
 
 
-FEATURE_LINTS = (
-    FeatureStringTooShort(),
-    FeatureNegativeNumber(),
-)
-
-
-def get_normpath(path):
-    return posixpath.normpath(path).replace(os.sep, "/")
+FEATURE_LINTS = (FeatureStringTooShort(), FeatureNegativeNumber(), FeatureNtdllNtoskrnlApi())
 
 
 def lint_features(ctx, rule):
@@ -340,6 +384,20 @@ def lint_features(ctx, rule):
     return run_feature_lints(FEATURE_LINTS, ctx, features)
 
 
+FORMAT_LINTS = (
+    FormatSingleEmptyLineEOF(),
+    FormatIncorrect(),
+)
+
+
+def lint_format(ctx, rule):
+    return run_lints(FORMAT_LINTS, ctx, rule)
+
+
+def get_normpath(path):
+    return posixpath.normpath(path).replace(os.sep, "/")
+
+
 def get_features(ctx, rule):
     # get features from rule and all dependencies including subscopes and matched rules
     features = []
@@ -390,6 +448,7 @@ def lint_rule(ctx, rule):
             lint_meta(ctx, rule),
             lint_logic(ctx, rule),
             lint_features(ctx, rule),
+            lint_format(ctx, rule),
         )
     )
 
@@ -406,25 +465,28 @@ def lint_rule(ctx, rule):
             )
         )
 
-        level = "WARN" if is_nursery_rule(rule) else "FAIL"
-
         for violation in violations:
             print(
                 "%s  %s: %s: %s"
                 % (
                     "    " if is_nursery_rule(rule) else "",
-                    level,
+                    Lint.WARN if is_nursery_rule(rule) else violation.level,
                     violation.name,
                     violation.recommendation,
                 )
             )
 
-    elif len(violations) == 0 and is_nursery_rule(rule):
+        print("")
+
+    lints_failed = any(map(lambda v: v.level == Lint.FAIL, violations))
+
+    if not lints_failed and is_nursery_rule(rule):
         print("")
         print("%s%s" % ("    (nursery) ", rule.name))
-        print("%s  %s: %s: %s" % ("    ", "WARN", "no violations", "Graduate the rule"))
+        print("%s  %s: %s: %s" % ("    ", Lint.WARN, "no lint failures", "Graduate the rule"))
+        print("")
 
-    return len(violations) > 0 and not is_nursery_rule(rule)
+    return lints_failed and not is_nursery_rule(rule)
 
 
 def lint(ctx, rules):
@@ -493,6 +555,7 @@ def main(argv=None):
     samples_path = os.path.join(os.path.dirname(__file__), "..", "tests", "data")
 
     parser = argparse.ArgumentParser(description="A program.")
+    capa.main.install_common_args(parser, wanted={"tag"})
     parser.add_argument("rules", type=str, help="Path to rules")
     parser.add_argument("--samples", type=str, default=samples_path, help="Path to samples")
     parser.add_argument(
@@ -500,31 +563,24 @@ def main(argv=None):
         action="store_true",
         help="Enable thorough linting - takes more time, but does a better job",
     )
-    parser.add_argument("-v", "--verbose", action="store_true", help="Enable debug logging")
-    parser.add_argument("-q", "--quiet", action="store_true", help="Disable all output but errors")
     args = parser.parse_args(args=argv)
+    capa.main.handle_common_args(args)
 
-    if args.verbose:
-        level = logging.DEBUG
-    elif args.quiet:
-        level = logging.ERROR
-    else:
-        level = logging.INFO
-
-    logging.basicConfig(level=level)
-    logging.getLogger("capa.lint").setLevel(level)
-
-    capa.main.set_vivisect_log_level(logging.CRITICAL)
     logging.getLogger("capa").setLevel(logging.CRITICAL)
+    logging.getLogger("viv_utils").setLevel(logging.CRITICAL)
+
+    time0 = time.time()
 
     try:
-        rules = capa.main.get_rules(args.rules)
+        rules = capa.main.get_rules(args.rules, disable_progress=True)
         rules = capa.rules.RuleSet(rules)
         logger.info("successfully loaded %s rules", len(rules))
-    except IOError as e:
-        logger.error("%s", str(e))
-        return -1
-    except capa.rules.InvalidRule as e:
+        if args.tag:
+            rules = rules.filter_rules_by_meta(args.tag)
+            logger.debug("selected %s rules", len(rules))
+            for i, r in enumerate(rules.rules, 1):
+                logger.debug(" %d. %s", i, r)
+    except (IOError, capa.rules.InvalidRule, capa.rules.InvalidRuleSet) as e:
         logger.error("%s", str(e))
         return -1
 
@@ -542,8 +598,12 @@ def main(argv=None):
     }
 
     did_violate = lint(ctx, rules)
+
+    min, sec = divmod(time.time() - time0, 60)
+    logger.debug("lints ran for ~ %02d:%02dm", min, sec)
+
     if not did_violate:
-        logger.info("no suggestions, nice!")
+        logger.info("no lints failed, nice!")
         return 0
     else:
         return 1

scripts/migrate-rules.py

@@ -1,167 +0,0 @@
-#!/usr/bin/env python
-"""
-migrate rules and their namespaces.
-
-example:
-
-    $ python scripts/migrate-rules.py migration.csv ./rules ./new-rules
-
-Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
-Licensed under the Apache License, Version 2.0 (the "License");
- you may not use this file except in compliance with the License.
-You may obtain a copy of the License at: [package root]/LICENSE.txt
-Unless required by applicable law or agreed to in writing, software distributed under the License
- is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and limitations under the License.
-"""
-import os
-import csv
-import sys
-import logging
-import os.path
-import argparse
-import collections
-
-import capa.rules
-
-logger = logging.getLogger("migrate-rules")
-
-
-def read_plan(plan_path):
-    with open(plan_path, "rb") as f:
-        return list(
-            csv.DictReader(
-                f,
-                restkey="other",
-                fieldnames=(
-                    "existing path",
-                    "existing name",
-                    "existing rule-category",
-                    "proposed name",
-                    "proposed namespace",
-                    "ATT&CK",
-                    "MBC",
-                    "comment1",
-                ),
-            )
-        )
-
-
-def read_rules(rule_directory):
-    rules = {}
-    for root, dirs, files in os.walk(rule_directory):
-        for file in files:
-            path = os.path.join(root, file)
-            if not path.endswith(".yml"):
-                logger.info("skipping file: %s", path)
-                continue
-
-            rule = capa.rules.Rule.from_yaml_file(path)
-            rules[rule.name] = rule
-
-            if "nursery" in path:
-                rule.meta["capa/nursery"] = True
-    return rules
-
-
-def main(argv=None):
-    if argv is None:
-        argv = sys.argv[1:]
-
-    parser = argparse.ArgumentParser(description="migrate rules.")
-    parser.add_argument("plan", type=str, help="Path to CSV describing migration")
-    parser.add_argument("source", type=str, help="Source directory of rules")
-    parser.add_argument("destination", type=str, help="Destination directory of rules")
-    args = parser.parse_args(args=argv)
-
-    logging.basicConfig(level=logging.INFO)
-    logging.getLogger().setLevel(logging.INFO)
-
-    plan = read_plan(args.plan)
-    logger.info("read %d plan entries", len(plan))
-
-    rules = read_rules(args.source)
-    logger.info("read %d rules", len(rules))
-
-    planned_rules = set([row["existing name"] for row in plan])
-    unplanned_rules = [rule for (name, rule) in rules.items() if name not in planned_rules]
-
-    if unplanned_rules:
-        logger.error("plan does not account for %d rules:" % (len(unplanned_rules)))
-        for rule in unplanned_rules:
-            logger.error("  " + rule.name)
-        return -1
-
-    # pairs of strings (needle, replacement)
-    match_translations = []
-
-    for row in plan:
-        if not row["existing name"]:
-            continue
-
-        rule = rules[row["existing name"]]
-
-        if rule.meta["name"] != row["proposed name"]:
-            logger.info("renaming rule '%s' -> '%s'", rule.meta["name"], row["proposed name"])
-
-            # assume the yaml is formatted like `- match: $rule-name`.
-            # but since its been linted, this should be ok.
-            match_translations.append(("- match: " + rule.meta["name"], "- match: " + row["proposed name"]))
-
-            rule.meta["name"] = row["proposed name"]
-            rule.name = row["proposed name"]
-
-        if "rule-category" in rule.meta:
-            logger.info("deleting rule category '%s'", rule.meta["rule-category"])
-            del rule.meta["rule-category"]
-
-        rule.meta["namespace"] = row["proposed namespace"]
-
-        if row["ATT&CK"] != "n/a" and row["ATT&CK"] != "":
-            tag = row["ATT&CK"]
-            name, _, id = tag.rpartition(" ")
-            tag = "%s [%s]" % (name, id)
-            rule.meta["att&ck"] = [tag]
-
-        if row["MBC"] != "n/a" and row["MBC"] != "":
-            tag = row["MBC"]
-            rule.meta["mbc"] = [tag]
-
-    for rule in rules.values():
-        filename = rule.name
-        filename = filename.lower()
-        filename = filename.replace(" ", "-")
-        filename = filename.replace("(", "")
-        filename = filename.replace(")", "")
-        filename = filename.replace("+", "")
-        filename = filename.replace("/", "")
-        filename = filename + ".yml"
-
-        try:
-            if rule.meta.get("capa/nursery"):
-                directory = os.path.join(args.destination, "nursery")
-            elif rule.meta.get("lib"):
-                directory = os.path.join(args.destination, "lib")
-            else:
-                directory = os.path.join(args.destination, rule.meta.get("namespace"))
-            os.makedirs(directory)
-        except OSError:
-            pass
-        else:
-            logger.info("created namespace: %s", directory)
-
-        path = os.path.join(directory, filename)
-        logger.info("writing rule %s", path)
-
-        doc = rule.to_yaml().decode("utf-8")
-        for (needle, replacement) in match_translations:
-            doc = doc.replace(needle, replacement)
-
-        with open(path, "wb") as f:
-            f.write(doc.encode("utf-8"))
-
-    return 0
-
-
-if __name__ == "__main__":
-    sys.exit(main())

scripts/show-capabilities-by-function.py

@@ -63,7 +63,6 @@ import capa.render
 import capa.features
 import capa.render.utils as rutils
 import capa.features.freeze
-import capa.features.extractors.viv
 from capa.helpers import get_file_taste
 
 logger = logging.getLogger("capa.show-capabilities-by-function")
@@ -111,143 +110,93 @@ def main(argv=None):
     if argv is None:
         argv = sys.argv[1:]
 
-        formats = [
-            ("auto", "(default) detect file type automatically"),
-            ("pe", "Windows PE file"),
-            ("sc32", "32-bit shellcode"),
-            ("sc64", "64-bit shellcode"),
-            ("freeze", "features previously frozen by capa"),
-        ]
-        format_help = ", ".join(["%s: %s" % (f[0], f[1]) for f in formats])
+    parser = argparse.ArgumentParser(description="detect capabilities in programs.")
+    capa.main.install_common_args(parser, wanted={"format", "sample", "rules", "tag"})
+    args = parser.parse_args(args=argv)
+    capa.main.handle_common_args(args)
 
-        parser = argparse.ArgumentParser(description="detect capabilities in programs.")
-        parser.add_argument("sample", type=str, help="Path to sample to analyze")
-        parser.add_argument(
-            "-r",
-            "--rules",
-            type=str,
-            default="(embedded rules)",
-            help="Path to rule file or directory, use embedded rules by default",
-        )
-        parser.add_argument("-t", "--tag", type=str, help="Filter on rule meta field values")
-        parser.add_argument("-d", "--debug", action="store_true", help="Enable debugging output on STDERR")
-        parser.add_argument("-q", "--quiet", action="store_true", help="Disable all output but errors")
-        parser.add_argument(
-            "-f",
-            "--format",
-            choices=[f[0] for f in formats],
-            default="auto",
-            help="Select sample format, %s" % format_help,
-        )
-        args = parser.parse_args(args=argv)
+    try:
+        taste = get_file_taste(args.sample)
+    except IOError as e:
+        logger.error("%s", str(e))
+        return -1
 
-        if args.quiet:
-            logging.basicConfig(level=logging.ERROR)
-            logging.getLogger().setLevel(logging.ERROR)
-        elif args.debug:
-            logging.basicConfig(level=logging.DEBUG)
-            logging.getLogger().setLevel(logging.DEBUG)
-        else:
-            logging.basicConfig(level=logging.INFO)
-            logging.getLogger().setLevel(logging.INFO)
+    if args.rules == "(embedded rules)":
+        logger.info("-" * 80)
+        logger.info(" Using default embedded rules.")
+        logger.info(" To provide your own rules, use the form `capa.exe -r ./path/to/rules/  /path/to/mal.exe`.")
+        logger.info(" You can see the current default rule set here:")
+        logger.info("     https://github.com/fireeye/capa-rules")
+        logger.info("-" * 80)
 
-        # disable vivisect-related logging, it's verbose and not relevant for capa users
-        capa.main.set_vivisect_log_level(logging.CRITICAL)
+        logger.debug("detected running from source")
+        args.rules = os.path.join(os.path.dirname(__file__), "..", "rules")
+        logger.debug("default rule path (source method): %s", args.rules)
+    else:
+        logger.info("using rules path: %s", args.rules)
 
+    try:
+        rules = capa.main.get_rules(args.rules)
+        rules = capa.rules.RuleSet(rules)
+        logger.info("successfully loaded %s rules", len(rules))
+        if args.tag:
+            rules = rules.filter_rules_by_meta(args.tag)
+            logger.info("selected %s rules", len(rules))
+    except (IOError, capa.rules.InvalidRule, capa.rules.InvalidRuleSet) as e:
+        logger.error("%s", str(e))
+        return -1
+
+    if (args.format == "freeze") or (args.format == "auto" and capa.features.freeze.is_freeze(taste)):
+        format = "freeze"
+        with open(args.sample, "rb") as f:
+            extractor = capa.features.freeze.load(f.read())
+    else:
+        format = args.format
         try:
-            taste = get_file_taste(args.sample)
-        except IOError as e:
-            logger.error("%s", str(e))
+            extractor = capa.main.get_extractor(args.sample, args.format)
+        except capa.main.UnsupportedFormatError:
+            logger.error("-" * 80)
+            logger.error(" Input file does not appear to be a PE file.")
+            logger.error(" ")
+            logger.error(
+                " capa currently only supports analyzing PE files (or shellcode, when using --format sc32|sc64)."
+            )
+            logger.error(" If you don't know the input file type, you can try using the `file` utility to guess it.")
+            logger.error("-" * 80)
+            return -1
+        except capa.main.UnsupportedRuntimeError:
+            logger.error("-" * 80)
+            logger.error(" Unsupported runtime or Python interpreter.")
+            logger.error(" ")
+            logger.error(" capa supports running under Python 2.7 using Vivisect for binary analysis.")
+            logger.error(" It can also run within IDA Pro, using either Python 2.7 or 3.5+.")
+            logger.error(" ")
+            logger.error(" If you're seeing this message on the command line, please ensure you're running Python 2.7.")
+            logger.error("-" * 80)
             return -1
 
-        # py2 doesn't know about cp65001, which is a variant of utf-8 on windows
-        # tqdm bails when trying to render the progress bar in this setup.
-        # because cp65001 is utf-8, we just map that codepage to the utf-8 codec.
-        # see #380 and: https://stackoverflow.com/a/3259271/87207
-        import codecs
+    meta = capa.main.collect_metadata(argv, args.sample, args.rules, format, extractor)
+    capabilities, counts = capa.main.find_capabilities(rules, extractor)
+    meta["analysis"].update(counts)
 
-        codecs.register(lambda name: codecs.lookup("utf-8") if name == "cp65001" else None)
-
-        if args.rules == "(embedded rules)":
-            logger.info("-" * 80)
-            logger.info(" Using default embedded rules.")
-            logger.info(" To provide your own rules, use the form `capa.exe -r ./path/to/rules/  /path/to/mal.exe`.")
-            logger.info(" You can see the current default rule set here:")
-            logger.info("     https://github.com/fireeye/capa-rules")
-            logger.info("-" * 80)
-
-            logger.debug("detected running from source")
-            args.rules = os.path.join(os.path.dirname(__file__), "..", "rules")
-            logger.debug("default rule path (source method): %s", args.rules)
-        else:
-            logger.info("using rules path: %s", args.rules)
-
-        try:
-            rules = capa.main.get_rules(args.rules)
-            rules = capa.rules.RuleSet(rules)
-            logger.info("successfully loaded %s rules", len(rules))
-            if args.tag:
-                rules = rules.filter_rules_by_meta(args.tag)
-                logger.info("selected %s rules", len(rules))
-        except (IOError, capa.rules.InvalidRule, capa.rules.InvalidRuleSet) as e:
-            logger.error("%s", str(e))
+    if capa.main.has_file_limitation(rules, capabilities):
+        # bail if capa encountered file limitation e.g. a packed binary
+        # do show the output in verbose mode, though.
+        if not (args.verbose or args.vverbose or args.json):
             return -1
 
-        if (args.format == "freeze") or (args.format == "auto" and capa.features.freeze.is_freeze(taste)):
-            format = "freeze"
-            with open(args.sample, "rb") as f:
-                extractor = capa.features.freeze.load(f.read())
-        else:
-            format = args.format
-            try:
-                extractor = capa.main.get_extractor(args.sample, args.format)
-            except capa.main.UnsupportedFormatError:
-                logger.error("-" * 80)
-                logger.error(" Input file does not appear to be a PE file.")
-                logger.error(" ")
-                logger.error(
-                    " capa currently only supports analyzing PE files (or shellcode, when using --format sc32|sc64)."
-                )
-                logger.error(
-                    " If you don't know the input file type, you can try using the `file` utility to guess it."
-                )
-                logger.error("-" * 80)
-                return -1
-            except capa.main.UnsupportedRuntimeError:
-                logger.error("-" * 80)
-                logger.error(" Unsupported runtime or Python interpreter.")
-                logger.error(" ")
-                logger.error(" capa supports running under Python 2.7 using Vivisect for binary analysis.")
-                logger.error(" It can also run within IDA Pro, using either Python 2.7 or 3.5+.")
-                logger.error(" ")
-                logger.error(
-                    " If you're seeing this message on the command line, please ensure you're running Python 2.7."
-                )
-                logger.error("-" * 80)
-                return -1
+    # colorama will detect:
+    #  - when on Windows console, and fixup coloring, and
+    #  - when not an interactive session, and disable coloring
+    # renderers should use coloring and assume it will be stripped out if necessary.
+    colorama.init()
+    doc = capa.render.convert_capabilities_to_result_document(meta, rules, capabilities)
+    print(render_matches_by_function(doc))
+    colorama.deinit()
 
-        meta = capa.main.collect_metadata(argv, args.sample, args.rules, format, extractor)
-        capabilities, counts = capa.main.find_capabilities(rules, extractor)
-        meta["analysis"].update(counts)
+    logger.info("done.")
 
-        if capa.main.has_file_limitation(rules, capabilities):
-            # bail if capa encountered file limitation e.g. a packed binary
-            # do show the output in verbose mode, though.
-            if not (args.verbose or args.vverbose or args.json):
-                return -1
-
-        # colorama will detect:
-        #  - when on Windows console, and fixup coloring, and
-        #  - when not an interactive session, and disable coloring
-        # renderers should use coloring and assume it will be stripped out if necessary.
-        colorama.init()
-        doc = capa.render.convert_capabilities_to_result_document(meta, rules, capabilities)
-        print(render_matches_by_function(doc))
-        colorama.deinit()
-
-        logger.info("done.")
-
-        return 0
+    return 0
 
 
 if __name__ == "__main__":

scripts/show-features.py

@@ -1,5 +1,13 @@
 #!/usr/bin/env python2
 """
+Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
+Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+You may obtain a copy of the License at: [package root]/LICENSE.txt
+Unless required by applicable law or agreed to in writing, software distributed under the License
+ is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and limitations under the License.
+
 show-features
 
 Show the features that capa extracts from the given sample,
@@ -55,14 +63,6 @@ Example::
     insn: 0x10001027: number(0x1)
     insn: 0x10001027: mnemonic(shl)
     ...
-
-Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
-Licensed under the Apache License, Version 2.0 (the "License");
- you may not use this file except in compliance with the License.
-You may obtain a copy of the License at: [package root]/LICENSE.txt
-Unless required by applicable law or agreed to in writing, software distributed under the License
- is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and limitations under the License.
 """
 import sys
 import logging
@@ -71,41 +71,56 @@ import argparse
 import capa.main
 import capa.rules
 import capa.engine
+import capa.helpers
 import capa.features
 import capa.features.freeze
-import capa.features.extractors.viv
+
+logger = logging.getLogger("capa.show-features")
 
 
 def main(argv=None):
     if argv is None:
         argv = sys.argv[1:]
 
-    formats = [
-        ("auto", "(default) detect file type automatically"),
-        ("pe", "Windows PE file"),
-        ("sc32", "32-bit shellcode"),
-        ("sc64", "64-bit shellcode"),
-        ("freeze", "features previously frozen by capa"),
-    ]
-    format_help = ", ".join(["%s: %s" % (f[0], f[1]) for f in formats])
+    parser = argparse.ArgumentParser(description="Show the features that capa extracts from the given sample")
+    capa.main.install_common_args(parser, wanted={"format", "sample"})
 
-    parser = argparse.ArgumentParser(description="detect capabilities in programs.")
-    parser.add_argument("sample", type=str, help="Path to sample to analyze")
-    parser.add_argument(
-        "-f", "--format", choices=[f[0] for f in formats], default="auto", help="Select sample format, %s" % format_help
-    )
-    parser.add_argument("-F", "--function", type=lambda x: int(x, 0), help="Show features for specific function")
+    parser.add_argument("-F", "--function", type=lambda x: int(x, 0x10), help="Show features for specific function")
     args = parser.parse_args(args=argv)
+    capa.main.handle_common_args(args)
 
-    logging.basicConfig(level=logging.INFO)
-    logging.getLogger().setLevel(logging.INFO)
+    try:
+        taste = capa.helpers.get_file_taste(args.sample)
+    except IOError as e:
+        logger.error("%s", str(e))
+        return -1
 
-    if args.format == "freeze":
+    if (args.format == "freeze") or (args.format == "auto" and capa.features.freeze.is_freeze(taste)):
         with open(args.sample, "rb") as f:
             extractor = capa.features.freeze.load(f.read())
     else:
-        vw = capa.main.get_workspace(args.sample, args.format)
-        extractor = capa.features.extractors.viv.VivisectFeatureExtractor(vw, args.sample)
+        try:
+            extractor = capa.main.get_extractor(args.sample, args.format, capa.main.BACKEND_VIV)
+        except capa.main.UnsupportedFormatError:
+            logger.error("-" * 80)
+            logger.error(" Input file does not appear to be a PE file.")
+            logger.error(" ")
+            logger.error(
+                " capa currently only supports analyzing PE files (or shellcode, when using --format sc32|sc64)."
+            )
+            logger.error(" If you don't know the input file type, you can try using the `file` utility to guess it.")
+            logger.error("-" * 80)
+            return -1
+        except capa.main.UnsupportedRuntimeError:
+            logger.error("-" * 80)
+            logger.error(" Unsupported runtime or Python interpreter.")
+            logger.error(" ")
+            logger.error(" capa supports running under Python 2.7 using Vivisect for binary analysis.")
+            logger.error(" It can also run within IDA Pro, using either Python 2.7 or 3.5+.")
+            logger.error(" ")
+            logger.error(" If you're seeing this message on the command line, please ensure you're running Python 2.7.")
+            logger.error("-" * 80)
+            return -1
 
     if not args.function:
         for feature, va in extractor.extract_file_features():
@@ -118,10 +133,52 @@ def main(argv=None):
 
     if args.function:
         if args.format == "freeze":
-            functions = filter(lambda f: f == args.function, functions)
+            functions = tuple(filter(lambda f: f == args.function, functions))
         else:
-            functions = filter(lambda f: f.va == args.function, functions)
+            functions = tuple(filter(lambda f: capa.helpers.oint(f) == args.function, functions))
 
+            if args.function not in [capa.helpers.oint(f) for f in functions]:
+                print("0x%X not a function" % args.function)
+                return -1
+
+        if len(functions) == 0:
+            print("0x%X not a function")
+            return -1
+
+    print_features(functions, extractor)
+
+    return 0
+
+
+def ida_main():
+    function = idc.get_func_attr(idc.here(), idc.FUNCATTR_START)
+    print("getting features for current function 0x%X" % function)
+
+    extractor = capa.features.extractors.ida.IdaFeatureExtractor()
+
+    if not function:
+        for feature, va in extractor.extract_file_features():
+            if va:
+                print("file: 0x%08x: %s" % (va, feature))
+            else:
+                print("file: 0x00000000: %s" % (feature))
+        return
+
+    functions = extractor.get_functions()
+
+    if function:
+        functions = tuple(filter(lambda f: f.start_ea == function, functions))
+
+        if len(functions) == 0:
+            print("0x%X not a function" % function)
+            return -1
+
+    print_features(functions, extractor)
+
+    return 0
+
+
+def print_features(functions, extractor):
     for f in functions:
         for feature, va in extractor.extract_function_features(f):
             print("func: 0x%08x: %s" % (va, feature))
@@ -138,8 +195,9 @@ def main(argv=None):
                         # may be an issue while piping to less and encountering non-ascii characters
                         continue
 
-    return 0
-
 
 if __name__ == "__main__":
-    sys.exit(main())
+    if capa.main.is_runtime_ida():
+        ida_main()
+    else:
+        sys.exit(main())

scripts/vivisect-py2-vs-py3.sh

@@ -0,0 +1,69 @@
+#!/usr/bin/env bash
+
+int() {
+  int=$(bc <<< "scale=0; ($1 + 0.5)/1")
+}
+
+export TIMEFORMAT='%3R'
+threshold_time=90
+threshold_py3_time=60 # Do not warn if it doesn't take at least 1 minute to run
+rm tests/data/*.viv 2>/dev/null
+mkdir results
+for file in tests/data/*
+do
+  file=$(printf %q "$file") # Handle names with white spaces
+  file_name=$(basename $file)
+  echo $file_name
+
+  rm "$file.viv" 2>/dev/null
+  py3_time=$(sh -c "time python3 scripts/show-features.py $file >> results/p3-$file_name.out 2>/dev/null" 2>&1)
+  rm "$file.viv" 2>/dev/null
+  py2_time=$(sh -c "time python2 scripts/show-features.py $file >> results/p2-$file_name.out 2>/dev/null" 2>&1)
+
+  int $py3_time
+  if (($int > $threshold_py3_time))
+  then
+    percentage=$(bc <<< "scale=3; $py2_time/$py3_time*100 + 0.5")
+    int $percentage
+    if (($int < $threshold_py3_time))
+    then
+      echo -n "  SLOWER ($percentage): "
+    fi
+  fi
+  echo "  PY2($py2_time) PY3($py3_time)"
+done
+
+threshold_features=98
+counter=0
+average=0
+results_for() {
+  py3=$(cat "results/p3-$file_name.out" | grep "$1" | wc -l)
+  py2=$(cat "results/p2-$file_name.out" | grep "$1" | wc -l)
+  if (($py2 > 0))
+  then
+    percentage=$(bc <<< "scale=2; 100*$py3/$py2")
+    average=$(bc <<< "scale=2; $percentage + $average")
+    count=$(($count + 1))
+    int $percentage
+    if (($int < $threshold_features))
+    then
+      echo -e "$1: py2($py2) py3($py3) $percentage% - $file_name"
+    fi
+  fi
+}
+
+rm tests/data/*.viv 2>/dev/null
+echo -e '\nRESULTS:'
+for file in tests/data/*
+do
+  file_name=$(basename $file)
+  if test -f "results/p2-$file_name.out"; then
+    results_for 'insn'
+    results_for 'file'
+    results_for 'func'
+    results_for 'bb'
+  fi
+done
+
+average=$(bc <<< "scale=2; $average/$count")
+echo "TOTAL: $average"

setup.py

@@ -11,7 +11,6 @@ import sys
 
 import setuptools
 
-# halo==0.0.30 is the last version to support py2.7
 requirements = [
     "six",
     "tqdm",
@@ -21,36 +20,52 @@ requirements = [
     "termcolor",
     "ruamel.yaml",
     "wcwidth",
-    "halo==0.0.30",
     "ida-settings==2.1.0",
 ]
 
 if sys.version_info >= (3, 0):
     # py3
+    requirements.append("halo")
     requirements.append("networkx")
+    requirements.append("vivisect==1.0.0")
+    requirements.append("viv-utils==0.3.19")
+    requirements.append("smda==1.5.13")
 else:
     # py2
     requirements.append("enum34==1.1.6")  # v1.1.6 is needed by halo 0.0.30 / spinners 0.0.24
-    requirements.append("vivisect==0.1.0")
-    requirements.append("viv-utils")
+    requirements.append("halo==0.0.30")  # halo==0.0.30 is the last version to support py2.7
+    requirements.append("vivisect==0.2.1")
+    requirements.append("viv-utils==0.3.19")
     requirements.append("networkx==2.2")  # v2.2 is last version supported by Python 2.7
     requirements.append("backports.functools-lru-cache")
 
 # this sets __version__
 # via: http://stackoverflow.com/a/7071358/87207
 # and: http://stackoverflow.com/a/2073599/87207
-with open(os.path.join("capa", "version.py"), "rb") as f:
+with open(os.path.join("capa", "version.py"), "r") as f:
     exec(f.read())
 
 
+# via: https://packaging.python.org/guides/making-a-pypi-friendly-readme/
+this_directory = os.path.abspath(os.path.dirname(__file__))
+with open(os.path.join(this_directory, "README.md"), "r") as f:
+    long_description = f.read()
+
+
 setuptools.setup(
     name="flare-capa",
     version=__version__,
     description="The FLARE team's open-source tool to identify capabilities in executable files.",
-    long_description="",
+    long_description=long_description,
+    long_description_content_type="text/markdown",
     author="Willi Ballenthin, Moritz Raabe",
     author_email="william.ballenthin@mandiant.com, moritz.raabe@mandiant.com",
     url="https://www.github.com/fireeye/capa",
+    project_urls={
+        "Documentation": "https://github.com/fireeye/capa/tree/master/doc",
+        "Rules": "https://github.com/fireeye/capa-rules",
+        "Rules Documentation": "https://github.com/fireeye/capa-rules/tree/master/doc",
+    },
     packages=setuptools.find_packages(exclude=["tests"]),
     package_dir={"capa": "capa"},
     entry_points={
@@ -72,12 +87,15 @@ setuptools.setup(
         ]
     },
     zip_safe=False,
-    keywords="capa",
+    keywords="capa malware analysis capability detection FLARE",
     classifiers=[
-        "Development Status :: 3 - Alpha",
+        "Development Status :: 5 - Production/Stable",
         "Intended Audience :: Developers",
+        "Intended Audience :: Information Technology",
+        "License :: OSI Approved :: Apache Software License",
         "Natural Language :: English",
-        "Programming Language :: Python :: 2",
+        "Programming Language :: Python :: 2.7",
         "Programming Language :: Python :: 3",
+        "Topic :: Security",
     ],
 )

tests/fixtures.py

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 # Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
 # Licensed under the Apache License, Version 2.0 (the "License");
 #  you may not use this file except in compliance with the License.
@@ -9,6 +10,7 @@
 import os
 import sys
 import os.path
+import binascii
 import contextlib
 import collections
 
@@ -77,7 +79,33 @@ def get_viv_extractor(path):
         vw = capa.main.get_workspace(path, "sc64", should_save=False)
     else:
         vw = capa.main.get_workspace(path, "auto", should_save=True)
-    return capa.features.extractors.viv.VivisectFeatureExtractor(vw, path)
+    extractor = capa.features.extractors.viv.VivisectFeatureExtractor(vw, path)
+    fixup_viv(path, extractor)
+    return extractor
+
+
+def fixup_viv(path, extractor):
+    """
+    vivisect fixups to overcome differences between backends
+    """
+    if "3b13b" in path:
+        # vivisect only recognizes calling thunk function at 0x10001573
+        extractor.vw.makeFunction(0x10006860)
+
+
+@lru_cache()
+def get_smda_extractor(path):
+    from smda.SmdaConfig import SmdaConfig
+    from smda.Disassembler import Disassembler
+
+    import capa.features.extractors.smda
+
+    config = SmdaConfig()
+    config.STORE_BUFFER = True
+    disasm = Disassembler(config)
+    report = disasm.disassembleFile(path)
+
+    return capa.features.extractors.smda.SmdaFeatureExtractor(report, path)
 
 
 @lru_cache()
@@ -128,6 +156,8 @@ def get_data_path_by_name(name):
         return os.path.join(CD, "data", "Practical Malware Analysis Lab 21-01.exe_")
     elif name == "al-khaser x86":
         return os.path.join(CD, "data", "al-khaser_x86.exe_")
+    elif name == "al-khaser x64":
+        return os.path.join(CD, "data", "al-khaser_x64.exe_")
     elif name.startswith("39c05"):
         return os.path.join(CD, "data", "39c05b15e9834ac93f206bc114d0a00c357c888db567ba8f5345da0529cbed41.dll_")
     elif name.startswith("499c2"):
@@ -144,8 +174,16 @@ def get_data_path_by_name(name):
         return os.path.join(CD, "data", "c91887d861d9bd4a5872249b641bc9f9.exe_")
     elif name.startswith("64d9f"):
         return os.path.join(CD, "data", "64d9f7d96b99467f36e22fada623c3bb.dll_")
+    elif name.startswith("82bf6"):
+        return os.path.join(CD, "data", "82BF6347ACF15E5D883715DC289D8A2B.exe_")
+    elif name.startswith("pingtaest"):
+        return os.path.join(CD, "data", "ping_täst.exe_")
+    elif name.startswith("77329"):
+        return os.path.join(CD, "data", "773290480d5445f11d3dc1b800728966.exe_")
+    elif name.startswith("3b13b"):
+        return os.path.join(CD, "data", "3b13b6f1d7cd14dc4a097a12e2e505c0a4cff495262261e2bfc991df238b9b04.dll_")
     else:
-        raise ValueError("unexpected sample fixture")
+        raise ValueError("unexpected sample fixture: %s" % name)
 
 
 def get_sample_md5_by_name(name):
@@ -164,6 +202,8 @@ def get_sample_md5_by_name(name):
         return "c8403fb05244e23a7931c766409b5e22"
     elif name == "al-khaser x86":
         return "db648cd247281954344f1d810c6fd590"
+    elif name == "al-khaser x64":
+        return "3cb21ae76ff3da4b7e02d77ff76e82be"
     elif name.startswith("39c05"):
         return "b7841b9d5dc1f511a93cc7576672ec0c"
     elif name.startswith("499c2"):
@@ -180,8 +220,15 @@ def get_sample_md5_by_name(name):
         return "c91887d861d9bd4a5872249b641bc9f9"
     elif name.startswith("64d9f"):
         return "64d9f7d96b99467f36e22fada623c3bb"
+    elif name.startswith("82bf6"):
+        return "82bf6347acf15e5d883715dc289d8a2b"
+    elif name.startswith("77329"):
+        return "773290480d5445f11d3dc1b800728966"
+    elif name.startswith("3b13b"):
+        # file name is SHA256 hash
+        return "56a6ffe6a02941028cc8235204eef31d"
     else:
-        raise ValueError("unexpected sample fixture")
+        raise ValueError("unexpected sample fixture: %s" % name)
 
 
 def resolve_sample(sample):
@@ -369,6 +416,12 @@ FEATURE_PRESENCE_TESTS = [
         True,
     ),
     ("kernel32-64", "function=0x1800202B0", capa.features.insn.API("RtlCaptureContext"), True),
+    # insn/api: x64 nested thunk
+    ("al-khaser x64", "function=0x14004B4F0", capa.features.insn.API("__vcrt_GetModuleHandle"), True),
+    # insn/api: call via jmp
+    ("mimikatz", "function=0x40B3C6", capa.features.insn.API("LocalFree"), True),
+    ("c91887...", "function=0x40156F", capa.features.insn.API("CloseClipboard"), True),
+    # TODO ignore thunk functions that call via jmp?
     # insn/api: resolve indirect calls
     ("c91887...", "function=0x401A77", capa.features.insn.API("kernel32.CreatePipe"), True),
     ("c91887...", "function=0x401A77", capa.features.insn.API("kernel32.SetHandleInformation"), True),
@@ -379,16 +432,21 @@ FEATURE_PRESENCE_TESTS = [
     ("mimikatz", "function=0x40105D", capa.features.String("SCardTransmit"), True),
     ("mimikatz", "function=0x40105D", capa.features.String("ACR  > "), True),
     ("mimikatz", "function=0x40105D", capa.features.String("nope"), False),
+    ("773290...", "function=0x140001140", capa.features.String(r"%s:\\OfficePackagesForWDAG"), True),
     # insn/regex, issue #262
     ("pma16-01", "function=0x4021B0", capa.features.Regex("HTTP/1.0"), True),
     ("pma16-01", "function=0x4021B0", capa.features.Regex("www.practicalmalwareanalysis.com"), False),
     # insn/string, pointer to string
     ("mimikatz", "function=0x44EDEF", capa.features.String("INPUTEVENT"), True),
+    # insn/string, direct memory reference
+    ("mimikatz", "function=0x46D6CE", capa.features.String("(null)"), True),
     # insn/bytes
     ("mimikatz", "function=0x40105D", capa.features.Bytes("SCardControl".encode("utf-16le")), True),
     ("mimikatz", "function=0x40105D", capa.features.Bytes("SCardTransmit".encode("utf-16le")), True),
     ("mimikatz", "function=0x40105D", capa.features.Bytes("ACR  > ".encode("utf-16le")), True),
     ("mimikatz", "function=0x40105D", capa.features.Bytes("nope".encode("ascii")), False),
+    # IDA features included byte sequences read from invalid memory, fixed in #409
+    ("mimikatz", "function=0x44570F", capa.features.Bytes(binascii.unhexlify("FF" * 256)), False),
     # insn/bytes, pointer to bytes
     ("mimikatz", "function=0x44EDEF", capa.features.Bytes("INPUTEVENT".encode("utf-16le")), True),
     # insn/characteristic(nzxor)
@@ -396,6 +454,9 @@ FEATURE_PRESENCE_TESTS = [
     ("mimikatz", "function=0x40105D", capa.features.Characteristic("nzxor"), False),
     # insn/characteristic(nzxor): no security cookies
     ("mimikatz", "function=0x46D534", capa.features.Characteristic("nzxor"), False),
+    # insn/characteristic(nzxor): xorps
+    # viv needs fixup to recognize function, see above
+    ("3b13b...", "function=0x10006860", capa.features.Characteristic("nzxor"), True),
     # insn/characteristic(peb access)
     ("kernel32-64", "function=0x1800017D0", capa.features.Characteristic("peb access"), True),
     ("mimikatz", "function=0x4556E5", capa.features.Characteristic("peb access"), False),
@@ -421,6 +482,12 @@ FEATURE_PRESENCE_TESTS = [
     ("mimikatz", "function=0x4556E5", capa.features.Characteristic("calls to"), False),
 ]
 
+FEATURE_PRESENCE_TESTS_IDA = [
+    # file/imports
+    # IDA can recover more names of APIs imported by ordinal
+    ("mimikatz", "file", capa.features.file.Import("cabinet.FCIAddFile"), True),
+]
+
 FEATURE_COUNT_TESTS = [
     ("mimikatz", "function=0x40E5C2", capa.features.basicblock.BasicBlock(), 7),
     ("mimikatz", "function=0x4702FD", capa.features.Characteristic("calls from"), 0),
@@ -453,11 +520,7 @@ def do_test_feature_count(get_extractor, sample, scope, feature, expected):
 
 
 def get_extractor(path):
-    if sys.version_info >= (3, 0):
-        raise RuntimeError("no supported py3 backends yet")
-    else:
-        extractor = get_viv_extractor(path)
-
+    extractor = get_viv_extractor(path)
     # overload the extractor so that the fixture exposes `extractor.path`
     setattr(extractor, "path", path)
     return extractor
@@ -526,3 +589,8 @@ def z499c2_extractor():
 @pytest.fixture
 def al_khaser_x86_extractor():
     return get_extractor(get_data_path_by_name("al-khaser x86"))
+
+
+@pytest.fixture
+def pingtaest_extractor():
+    return get_extractor(get_data_path_by_name("pingtaest"))

tests/test_ida_features.py

@@ -44,7 +44,7 @@ def get_ida_extractor(_path):
 
 @pytest.mark.skip(reason="IDA Pro tests must be run within IDA")
 def test_ida_features():
-    for (sample, scope, feature, expected) in FEATURE_PRESENCE_TESTS:
+    for (sample, scope, feature, expected) in FEATURE_PRESENCE_TESTS + FEATURE_PRESENCE_TESTS_IDA:
         id = make_test_id((sample, scope, feature, expected))
 
         try:

tests/test_main.py

@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 # Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
 # Licensed under the Apache License, Version 2.0 (the "License");
 #  you may not use this file except in compliance with the License.
@@ -6,6 +7,7 @@
 #  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and limitations under the License.
 import sys
+import json
 import textwrap
 
 import pytest
@@ -18,7 +20,6 @@ import capa.features
 from capa.engine import *
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_main(z9324d_extractor):
     # tests rules can be loaded successfully and all output modes
     path = z9324d_extractor.path
@@ -28,7 +29,6 @@ def test_main(z9324d_extractor):
     assert capa.main.main([path]) == 0
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_main_single_rule(z9324d_extractor, tmpdir):
     # tests a single rule can be loaded successfully
     RULE_CONTENT = textwrap.dedent(
@@ -57,7 +57,34 @@ def test_main_single_rule(z9324d_extractor, tmpdir):
     )
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
+def test_main_non_ascii_filename(pingtaest_extractor, tmpdir, capsys):
+    # on py2.7, need to be careful about str (which can hold bytes)
+    #  vs unicode (which is only unicode characters).
+    # on py3, this should not be needed.
+    #
+    # here we print a string with unicode characters in it
+    # (specifically, a byte string with utf-8 bytes in it, see file encoding)
+    assert capa.main.main(["-q", pingtaest_extractor.path]) == 0
+
+    std = capsys.readouterr()
+    # but here, we have to use a unicode instance,
+    # because capsys has decoded the output for us.
+    if sys.version_info >= (3, 0):
+        assert pingtaest_extractor.path in std.out
+    else:
+        assert pingtaest_extractor.path.decode("utf-8") in std.out
+
+
+def test_main_non_ascii_filename_nonexistent(tmpdir, caplog):
+    NON_ASCII_FILENAME = "täst_not_there.exe"
+    assert capa.main.main(["-q", NON_ASCII_FILENAME]) == -1
+
+    if sys.version_info >= (3, 0):
+        assert NON_ASCII_FILENAME in caplog.text
+    else:
+        assert NON_ASCII_FILENAME.decode("utf-8") in caplog.text
+
+
 def test_main_shellcode(z499c2_extractor):
     path = z499c2_extractor.path
     assert capa.main.main([path, "-vv", "-f", "sc32"]) == 0
@@ -112,7 +139,6 @@ def test_ruleset():
     assert len(rules.basic_block_rules) == 1
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_match_across_scopes_file_function(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [
@@ -176,7 +202,6 @@ def test_match_across_scopes_file_function(z9324d_extractor):
     assert ".text section and install service" in capabilities
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_match_across_scopes(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [
@@ -239,7 +264,6 @@ def test_match_across_scopes(z9324d_extractor):
     assert "kill thread program" in capabilities
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_subscope_bb_rules(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [
@@ -264,7 +288,6 @@ def test_subscope_bb_rules(z9324d_extractor):
     assert "test rule" in capabilities
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_byte_matching(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [
@@ -287,7 +310,6 @@ def test_byte_matching(z9324d_extractor):
     assert "byte match test" in capabilities
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_count_bb(z9324d_extractor):
     rules = capa.rules.RuleSet(
         [
@@ -311,7 +333,6 @@ def test_count_bb(z9324d_extractor):
     assert "count bb" in capabilities
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_fix262(pma16_01_extractor, capsys):
     # tests rules can be loaded successfully and all output modes
     path = pma16_01_extractor.path
@@ -322,7 +343,6 @@ def test_fix262(pma16_01_extractor, capsys):
     assert "www.practicalmalwareanalysis.com" not in std.out
 
 
-@pytest.mark.xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2")
 def test_not_render_rules_also_matched(z9324d_extractor, capsys):
     # rules that are also matched by other rules should not get rendered by default.
     # this cuts down on the amount of output while giving approx the same detail.
@@ -346,3 +366,20 @@ def test_not_render_rules_also_matched(z9324d_extractor, capsys):
     assert "act as TCP client" in std.out
     assert "connect TCP socket" in std.out
     assert "create TCP socket" in std.out
+
+
+# It tests main works with different backends
+def test_backend_option(capsys):
+    if sys.version_info > (3, 0):
+        path = get_data_path_by_name("pma16-01")
+        assert capa.main.main([path, "-j", "-b", capa.main.BACKEND_VIV]) == 0
+        std = capsys.readouterr()
+        std_json = json.loads(std.out)
+        assert std_json["meta"]["analysis"]["extractor"] == "VivisectFeatureExtractor"
+        assert len(std_json["rules"]) > 0
+
+        assert capa.main.main([path, "-j", "-b", capa.main.BACKEND_SMDA]) == 0
+        std = capsys.readouterr()
+        std_json = json.loads(std.out)
+        assert std_json["meta"]["analysis"]["extractor"] == "SmdaFeatureExtractor"
+        assert len(std_json["rules"]) > 0

tests/test_rules.py

@@ -69,46 +69,63 @@ def test_rule_yaml_complex():
     assert r.evaluate({Number(6): {1}, Number(7): {1}, Number(8): {1}}) == False
 
 
-def test_rule_yaml_descriptions():
+def test_rule_descriptions():
     rule = textwrap.dedent(
         """
         rule:
-            meta:
-                name: test rule
-            features:
+          meta:
+            name: test rule
+          features:
+            - and:
+              - description: and description
+              - number: 1 = number description
+              - string: mystring
+                description: string description
+              - string: '/myregex/'
+                description: regex description
+              # TODO - count(number(2 = number description)): 2
+              - or:
+                - description: or description
                 - and:
-                    - number: 1 = This is the number 1
-                    - string: This program cannot be run in DOS mode.
-                      description: MS-DOS stub message
-                    - string: '/SELECT.*FROM.*WHERE/i'
-                      description: SQL WHERE Clause
-                    - count(number(2 = AF_INET/SOCK_DGRAM)): 2
-                    - or:
-                        - and:
-                            - offset: 0x50 = IMAGE_NT_HEADERS.OptionalHeader.SizeOfImage
-                            - offset: 0x34 = IMAGE_NT_HEADERS.OptionalHeader.ImageBase
-                          description: 32-bits
-                        - and:
-                            - offset: 0x50 = IMAGE_NT_HEADERS64.OptionalHeader.SizeOfImage
-                            - offset: 0x30 = IMAGE_NT_HEADERS64.OptionalHeader.ImageBase
-                          description: 64-bits
-                      description: PE headers offsets
+                  - offset: 0x50 = offset description
+                  - offset: 0x34 = offset description
+                  - description: and description
+                - and:
+                  - description: and description
+                  - offset/x64: 0x50 = offset/x64 description
+                  - offset/x64: 0x30 = offset/x64 description
         """
     )
     r = capa.rules.Rule.from_yaml(rule)
-    assert (
-        r.evaluate(
-            {
-                Number(1): {1},
-                Number(2): {2, 3},
-                String("This program cannot be run in DOS mode."): {4},
-                String("SELECT password FROM hidden_table WHERE user == admin"): {5},
-                Offset(0x50): {6},
-                Offset(0x30): {7},
-            }
+
+    def rec(statement):
+        if isinstance(statement, capa.engine.Statement):
+            assert statement.description == statement.name.lower() + " description"
+            for child in statement.get_children():
+                rec(child)
+        else:
+            assert statement.description == statement.name + " description"
+
+    rec(r.statement)
+
+
+def test_invalid_rule_statement_descriptions():
+    # statements can only have one description
+    with pytest.raises(capa.rules.InvalidRule):
+        capa.rules.Rule.from_yaml(
+            textwrap.dedent(
+                """
+                rule:
+                  meta:
+                    name: test rule
+                  features:
+                    - or:
+                      - number: 1 = This is the number 1
+                      - description: description
+                      - description: another description (invalid)
+                """
+            )
         )
-        == True
-    )
 
 
 def test_rule_yaml_not():
@@ -265,7 +282,8 @@ def test_lib_rules():
             ),
         ]
     )
-    assert len(rules.function_rules) == 1
+    # lib rules are added to the rule set
+    assert len(rules.function_rules) == 2
 
 
 def test_subscope_rules():

tests/test_smda_features.py

@@ -0,0 +1,30 @@
+# Copyright (C) 2020 FireEye, Inc. All Rights Reserved.
+# Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at: [package root]/LICENSE.txt
+# Unless required by applicable law or agreed to in writing, software distributed under the License
+#  is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and limitations under the License.
+import sys
+
+from fixtures import *
+
+
+@parametrize(
+    "sample,scope,feature,expected",
+    FEATURE_PRESENCE_TESTS,
+    indirect=["sample", "scope"],
+)
+def test_smda_features(sample, scope, feature, expected):
+    with xfail(sys.version_info < (3, 0), reason="SMDA only works on py3"):
+        do_test_feature_presence(get_smda_extractor, sample, scope, feature, expected)
+
+
+@parametrize(
+    "sample,scope,feature,expected",
+    FEATURE_COUNT_TESTS,
+    indirect=["sample", "scope"],
+)
+def test_smda_feature_counts(sample, scope, feature, expected):
+    with xfail(sys.version_info < (3, 0), reason="SMDA only works on py3"):
+        do_test_feature_count(get_smda_extractor, sample, scope, feature, expected)

tests/test_viv_features.py

@@ -16,8 +16,7 @@ from fixtures import *
     indirect=["sample", "scope"],
 )
 def test_viv_features(sample, scope, feature, expected):
-    with xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2"):
-        do_test_feature_presence(get_viv_extractor, sample, scope, feature, expected)
+    do_test_feature_presence(get_viv_extractor, sample, scope, feature, expected)
 
 
 @parametrize(
@@ -26,5 +25,4 @@ def test_viv_features(sample, scope, feature, expected):
     indirect=["sample", "scope"],
 )
 def test_viv_feature_counts(sample, scope, feature, expected):
-    with xfail(sys.version_info >= (3, 0), reason="vivsect only works on py2"):
-        do_test_feature_count(get_viv_extractor, sample, scope, feature, expected)
+    do_test_feature_count(get_viv_extractor, sample, scope, feature, expected)