fdroidserver/scanner.py

   1 #!/usr/bin/env python3
   2 #
   3 # scanner.py - part of the FDroid server tools
   4 # Copyright (C) 2010-13, Ciaran Gultnieks, ciaran@ciarang.com
   5 #
   6 # This program is free software: you can redistribute it and/or modify
   7 # it under the terms of the GNU Affero General Public License as published by
   8 # the Free Software Foundation, either version 3 of the License, or
   9 # (at your option) any later version.
  10 #
  11 # This program is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 # GNU Affero General Public License for more details.
  15 #
  16 # You should have received a copy of the GNU Affero General Public License
  17 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
  18
  19 import os
  20 import re
  21 import traceback
  22 from argparse import ArgumentParser
  23 import logging
  24
  25 from . import common
  26 from . import metadata
  27 from .exception import BuildException, VCSException
  28
  29 config = None
  30 options = None
  31
  32
  33 def get_gradle_compile_commands(build):
  34     compileCommands = ['compile', 'releaseCompile']
  35     if build.gradle and build.gradle != ['yes']:
  36         compileCommands += [flavor + 'Compile' for flavor in build.gradle]
  37         compileCommands += [flavor + 'ReleaseCompile' for flavor in build.gradle]
  38
  39     return [re.compile(r'\s*' + c, re.IGNORECASE) for c in compileCommands]
  40
  41
  42 def scan_source(build_dir, build):
  43     """Scan the source code in the given directory (and all subdirectories)
  44     and return the number of fatal problems encountered
  45     """
  46
  47     count = 0
  48
  49     # Common known non-free blobs (always lower case):
  50     usual_suspects = {
  51         exp: re.compile(r'.*' + exp, re.IGNORECASE) for exp in [
  52             r'flurryagent',
  53             r'paypal.*mpl',
  54             r'google.*analytics',
  55             r'admob.*sdk.*android',
  56             r'google.*ad.*view',
  57             r'google.*admob',
  58             r'google.*play.*services',
  59             r'crittercism',
  60             r'heyzap',
  61             r'jpct.*ae',
  62             r'youtube.*android.*player.*api',
  63             r'bugsense',
  64             r'crashlytics',
  65             r'ouya.*sdk',
  66             r'libspen23',
  67             r'firebase',
  68         ]
  69     }
  70
  71     def suspects_found(s):
  72         for n, r in usual_suspects.items():
  73             if r.match(s):
  74                 yield n
  75
  76     gradle_mavenrepo = re.compile(r'maven *{ *(url)? *[\'"]?([^ \'"]*)[\'"]?')
  77
  78     allowed_repos = [re.compile(r'^https?://' + re.escape(repo) + r'/*') for repo in [
  79         'repo1.maven.org/maven2',  # mavenCentral()
  80         'jcenter.bintray.com',     # jcenter()
  81         'jitpack.io',
  82         'repo.maven.apache.org/maven2',
  83         'oss.jfrog.org/artifactory/oss-snapshot-local',
  84         'oss.sonatype.org/content/repositories/snapshots',
  85         'oss.sonatype.org/content/repositories/releases',
  86         'oss.sonatype.org/content/groups/public',
  87         'clojars.org/repo',  # Clojure free software libs
  88         's3.amazonaws.com/repo.commonsware.com',  # CommonsWare
  89         'plugins.gradle.org/m2',  # Gradle plugin repo
  90         'maven.google.com',  # Google Maven Repo, https://developer.android.com/studio/build/dependencies.html#google-maven
  91         ]
  92     ]
  93
  94     scanignore = common.getpaths_map(build_dir, build.scanignore)
  95     scandelete = common.getpaths_map(build_dir, build.scandelete)
  96
  97     scanignore_worked = set()
  98     scandelete_worked = set()
  99
 100     def toignore(path_in_build_dir):
 101         for k, paths in scanignore.items():
 102             for p in paths:
 103                 if path_in_build_dir.startswith(p):
 104                     scanignore_worked.add(k)
 105                     return True
 106         return False
 107
 108     def todelete(path_in_build_dir):
 109         for k, paths in scandelete.items():
 110             for p in paths:
 111                 if path_in_build_dir.startswith(p):
 112                     scandelete_worked.add(k)
 113                     return True
 114         return False
 115
 116     def ignoreproblem(what, path_in_build_dir):
 117         logging.info('Ignoring %s at %s' % (what, path_in_build_dir))
 118         return 0
 119
 120     def removeproblem(what, path_in_build_dir, filepath):
 121         logging.info('Removing %s at %s' % (what, path_in_build_dir))
 122         os.remove(filepath)
 123         return 0
 124
 125     def warnproblem(what, path_in_build_dir):
 126         if toignore(path_in_build_dir):
 127             return
 128         logging.warn('Found %s at %s' % (what, path_in_build_dir))
 129
 130     def handleproblem(what, path_in_build_dir, filepath):
 131         if toignore(path_in_build_dir):
 132             return ignoreproblem(what, path_in_build_dir)
 133         if todelete(path_in_build_dir):
 134             return removeproblem(what, path_in_build_dir, filepath)
 135         logging.error('Found %s at %s' % (what, path_in_build_dir))
 136         return 1
 137
 138     def is_executable(path):
 139         return os.path.exists(path) and os.access(path, os.X_OK)
 140
 141     textchars = bytearray({7, 8, 9, 10, 12, 13, 27} | set(range(0x20, 0x100)) - {0x7f})
 142
 143     def is_binary(path):
 144         d = None
 145         with open(path, 'rb') as f:
 146             d = f.read(1024)
 147         return bool(d.translate(None, textchars))
 148
 149     # False positives patterns for files that are binary and executable.
 150     safe_paths = [re.compile(r) for r in [
 151         r".*/drawable[^/]*/.*\.png$",  # png drawables
 152         r".*/mipmap[^/]*/.*\.png$",    # png mipmaps
 153         ]
 154     ]
 155
 156     def safe_path(path):
 157         for sp in safe_paths:
 158             if sp.match(path):
 159                 return True
 160         return False
 161
 162     gradle_compile_commands = get_gradle_compile_commands(build)
 163
 164     def is_used_by_gradle(line):
 165         return any(command.match(line) for command in gradle_compile_commands)
 166
 167     # Iterate through all files in the source code
 168     for dirpath, dirnames, filenames in os.walk(build_dir, topdown=True):
 169
 170         # It's topdown, so checking the basename is enough
 171         for ignoredir in ('.hg', '.git', '.svn', '.bzr'):
 172             if ignoredir in dirnames:
 173                 dirnames.remove(ignoredir)
 174
 175         for curfile in filenames:
 176
 177             if curfile in ['.DS_Store']:
 178                 continue
 179
 180             # Path (relative) to the file
 181             filepath = os.path.join(dirpath, curfile)
 182
 183             if os.path.islink(filepath):
 184                 continue
 185
 186             path_in_build_dir = os.path.relpath(filepath, build_dir)
 187             _, ext = common.get_extension(path_in_build_dir)
 188
 189             if ext == 'so':
 190                 count += handleproblem('shared library', path_in_build_dir, filepath)
 191             elif ext == 'a':
 192                 count += handleproblem('static library', path_in_build_dir, filepath)
 193             elif ext == 'class':
 194                 count += handleproblem('Java compiled class', path_in_build_dir, filepath)
 195             elif ext == 'apk':
 196                 removeproblem('APK file', path_in_build_dir, filepath)
 197
 198             elif ext == 'jar':
 199                 for name in suspects_found(curfile):
 200                     count += handleproblem('usual supect \'%s\'' % name, path_in_build_dir, filepath)
 201                 if curfile == 'gradle-wrapper.jar':
 202                     removeproblem('gradle-wrapper.jar', path_in_build_dir, filepath)
 203                 else:
 204                     count += handleproblem('JAR file', path_in_build_dir, filepath)
 205
 206             elif ext == 'aar':
 207                 count += handleproblem('AAR file', path_in_build_dir, filepath)
 208
 209             elif ext == 'java':
 210                 if not os.path.isfile(filepath):
 211                     continue
 212                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 213                     for line in f:
 214                         if 'DexClassLoader' in line:
 215                             count += handleproblem('DexClassLoader', path_in_build_dir, filepath)
 216                             break
 217
 218             elif ext == 'gradle':
 219                 if not os.path.isfile(filepath):
 220                     continue
 221                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 222                     lines = f.readlines()
 223                 for i, line in enumerate(lines):
 224                     if is_used_by_gradle(line):
 225                         for name in suspects_found(line):
 226                             count += handleproblem('usual supect \'%s\' at line %d' % (name, i + 1), path_in_build_dir, filepath)
 227                 noncomment_lines = [l for l in lines if not common.gradle_comment.match(l)]
 228                 joined = re.sub(r'[\n\r\s]+', ' ', ' '.join(noncomment_lines))
 229                 for m in gradle_mavenrepo.finditer(joined):
 230                     url = m.group(2)
 231                     if not any(r.match(url) for r in allowed_repos):
 232                         count += handleproblem('unknown maven repo \'%s\'' % url, path_in_build_dir, filepath)
 233
 234             elif ext in ['', 'bin', 'out', 'exe']:
 235                 if is_binary(filepath):
 236                     count += handleproblem('binary', path_in_build_dir, filepath)
 237
 238             elif is_executable(filepath):
 239                 if is_binary(filepath) and not safe_path(path_in_build_dir):
 240                     warnproblem('possible binary', path_in_build_dir)
 241
 242     for p in scanignore:
 243         if p not in scanignore_worked:
 244             logging.error('Unused scanignore path: %s' % p)
 245             count += 1
 246
 247     for p in scandelete:
 248         if p not in scandelete_worked:
 249             logging.error('Unused scandelete path: %s' % p)
 250             count += 1
 251
 252     return count
 253
 254
 255 def main():
 256
 257     global config, options
 258
 259     # Parse command line...
 260     parser = ArgumentParser(usage="%(prog)s [options] [APPID[:VERCODE] [APPID[:VERCODE] ...]]")
 261     common.setup_global_opts(parser)
 262     parser.add_argument("appid", nargs='*', help="app-id with optional versionCode in the form APPID[:VERCODE]")
 263     metadata.add_metadata_arguments(parser)
 264     options = parser.parse_args()
 265     metadata.warnings_action = options.W
 266
 267     config = common.read_config(options)
 268
 269     # Read all app and srclib metadata
 270     allapps = metadata.read_metadata()
 271     apps = common.read_app_args(options.appid, allapps, True)
 272
 273     probcount = 0
 274
 275     build_dir = 'build'
 276     if not os.path.isdir(build_dir):
 277         logging.info("Creating build directory")
 278         os.makedirs(build_dir)
 279     srclib_dir = os.path.join(build_dir, 'srclib')
 280     extlib_dir = os.path.join(build_dir, 'extlib')
 281
 282     for appid, app in apps.items():
 283
 284         if app.Disabled:
 285             logging.info("Skipping %s: disabled" % appid)
 286             continue
 287         if not app.builds:
 288             logging.info("Skipping %s: no builds specified" % appid)
 289             continue
 290
 291         logging.info("Processing " + appid)
 292
 293         try:
 294
 295             if app.RepoType == 'srclib':
 296                 build_dir = os.path.join('build', 'srclib', app.Repo)
 297             else:
 298                 build_dir = os.path.join('build', appid)
 299
 300             # Set up vcs interface and make sure we have the latest code...
 301             vcs = common.getvcs(app.RepoType, app.Repo, build_dir)
 302
 303             for build in app.builds:
 304
 305                 if build.disable:
 306                     logging.info("...skipping version %s - %s" % (
 307                         build.versionName, build.get('disable', build.commit[1:])))
 308                 else:
 309                     logging.info("...scanning version " + build.versionName)
 310
 311                     # Prepare the source code...
 312                     common.prepare_source(vcs, app, build,
 313                                           build_dir, srclib_dir,
 314                                           extlib_dir, False)
 315
 316                     # Do the scan...
 317                     count = scan_source(build_dir, build)
 318                     if count > 0:
 319                         logging.warn('Scanner found %d problems in %s (%s)' % (
 320                             count, appid, build.versionCode))
 321                         probcount += count
 322
 323         except BuildException as be:
 324             logging.warn("Could not scan app %s due to BuildException: %s" % (
 325                 appid, be))
 326             probcount += 1
 327         except VCSException as vcse:
 328             logging.warn("VCS error while scanning app %s: %s" % (appid, vcse))
 329             probcount += 1
 330         except Exception:
 331             logging.warn("Could not scan app %s due to unknown error: %s" % (
 332                 appid, traceback.format_exc()))
 333             probcount += 1
 334
 335     logging.info("Finished:")
 336     print("%d problems found" % probcount)
 337
 338
 339 if __name__ == "__main__":
 340     main()