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 _
  26 from . import common
  27 from . import metadata
  28 from .exception import BuildException, VCSException
  29
  30 config = None
  31 options = None
  32
  33
  34 def get_gradle_compile_commands(build):
  35     compileCommands = ['compile',          'releaseCompile'
  36                        'provided',         'releaseProvided',
  37                        'apk',              'releaseApk',
  38                        'implementation',   'releaseImplementation',
  39                        'api',              'releaseApi',
  40                        'compileOnly',      'releaseCompileOnly',
  41                        'runtimeOnly',      'releaseRuntimeOnly']
  42     if build.gradle and build.gradle != ['yes']:
  43         compileCommands += [flavor + 'Compile' for flavor in build.gradle]
  44         compileCommands += [flavor + 'ReleaseCompile' for flavor in build.gradle]
  45
  46     return [re.compile(r'\s*' + c, re.IGNORECASE) for c in compileCommands]
  47
  48
  49 def scan_source(build_dir, build=metadata.Build()):
  50     """Scan the source code in the given directory (and all subdirectories)
  51     and return the number of fatal problems encountered
  52     """
  53
  54     count = 0
  55
  56     # Common known non-free blobs (always lower case):
  57     usual_suspects = {
  58         exp: re.compile(r'.*' + exp, re.IGNORECASE) for exp in [
  59             r'flurryagent',
  60             r'paypal.*mpl',
  61             r'google.*analytics',
  62             r'admob.*sdk.*android',
  63             r'google.*ad.*view',
  64             r'google.*admob',
  65             r'google.*play.*services',
  66             r'crittercism',
  67             r'heyzap',
  68             r'jpct.*ae',
  69             r'youtube.*android.*player.*api',
  70             r'bugsense',
  71             r'crashlytics',
  72             r'ouya.*sdk',
  73             r'libspen23',
  74             r'firebase',
  75         ]
  76     }
  77
  78     whitelisted = [
  79         'firebase-jobdispatcher',  # https://github.com/firebase/firebase-jobdispatcher-android/blob/master/LICENSE
  80         'com.firebaseui',          # https://github.com/firebase/FirebaseUI-Android/blob/master/LICENSE
  81         'geofire-android'          # https://github.com/firebase/geofire-java/blob/master/LICENSE
  82     ]
  83
  84     def is_whitelisted(s):
  85         return any(wl in s for wl in whitelisted)
  86
  87     def suspects_found(s):
  88         for n, r in usual_suspects.items():
  89             if r.match(s) and not is_whitelisted(s):
  90                 yield n
  91
  92     gradle_mavenrepo = re.compile(r'maven *{ *(url)? *[\'"]?([^ \'"]*)[\'"]?')
  93
  94     allowed_repos = [re.compile(r'^https?://' + re.escape(repo) + r'/*') for repo in [
  95         'repo1.maven.org/maven2',  # mavenCentral()
  96         'jcenter.bintray.com',     # jcenter()
  97         'jitpack.io',
  98         'repo.maven.apache.org/maven2',
  99         'oss.jfrog.org/artifactory/oss-snapshot-local',
 100         'oss.sonatype.org/content/repositories/snapshots',
 101         'oss.sonatype.org/content/repositories/releases',
 102         'oss.sonatype.org/content/groups/public',
 103         'clojars.org/repo',  # Clojure free software libs
 104         's3.amazonaws.com/repo.commonsware.com',  # CommonsWare
 105         'plugins.gradle.org/m2',  # Gradle plugin repo
 106         'maven.google.com',  # Google Maven Repo, https://developer.android.com/studio/build/dependencies.html#google-maven
 107         ]
 108     ]
 109
 110     scanignore = common.getpaths_map(build_dir, build.scanignore)
 111     scandelete = common.getpaths_map(build_dir, build.scandelete)
 112
 113     scanignore_worked = set()
 114     scandelete_worked = set()
 115
 116     def toignore(path_in_build_dir):
 117         for k, paths in scanignore.items():
 118             for p in paths:
 119                 if path_in_build_dir.startswith(p):
 120                     scanignore_worked.add(k)
 121                     return True
 122         return False
 123
 124     def todelete(path_in_build_dir):
 125         for k, paths in scandelete.items():
 126             for p in paths:
 127                 if path_in_build_dir.startswith(p):
 128                     scandelete_worked.add(k)
 129                     return True
 130         return False
 131
 132     def ignoreproblem(what, path_in_build_dir):
 133         logging.info('Ignoring %s at %s' % (what, path_in_build_dir))
 134         return 0
 135
 136     def removeproblem(what, path_in_build_dir, filepath):
 137         logging.info('Removing %s at %s' % (what, path_in_build_dir))
 138         os.remove(filepath)
 139         return 0
 140
 141     def warnproblem(what, path_in_build_dir):
 142         if toignore(path_in_build_dir):
 143             return
 144         logging.warn('Found %s at %s' % (what, path_in_build_dir))
 145
 146     def handleproblem(what, path_in_build_dir, filepath):
 147         if toignore(path_in_build_dir):
 148             return ignoreproblem(what, path_in_build_dir)
 149         if todelete(path_in_build_dir):
 150             return removeproblem(what, path_in_build_dir, filepath)
 151         logging.error('Found %s at %s' % (what, path_in_build_dir))
 152         return 1
 153
 154     def is_executable(path):
 155         return os.path.exists(path) and os.access(path, os.X_OK)
 156
 157     textchars = bytearray({7, 8, 9, 10, 12, 13, 27} | set(range(0x20, 0x100)) - {0x7f})
 158
 159     def is_binary(path):
 160         d = None
 161         with open(path, 'rb') as f:
 162             d = f.read(1024)
 163         return bool(d.translate(None, textchars))
 164
 165     # False positives patterns for files that are binary and executable.
 166     safe_paths = [re.compile(r) for r in [
 167         r".*/drawable[^/]*/.*\.png$",  # png drawables
 168         r".*/mipmap[^/]*/.*\.png$",    # png mipmaps
 169         ]
 170     ]
 171
 172     def safe_path(path):
 173         for sp in safe_paths:
 174             if sp.match(path):
 175                 return True
 176         return False
 177
 178     gradle_compile_commands = get_gradle_compile_commands(build)
 179
 180     def is_used_by_gradle(line):
 181         return any(command.match(line) for command in gradle_compile_commands)
 182
 183     # Iterate through all files in the source code
 184     for root, dirs, files in os.walk(build_dir, topdown=True):
 185
 186         # It's topdown, so checking the basename is enough
 187         for ignoredir in ('.hg', '.git', '.svn', '.bzr'):
 188             if ignoredir in dirs:
 189                 dirs.remove(ignoredir)
 190
 191         for curfile in files:
 192
 193             if curfile in ['.DS_Store']:
 194                 continue
 195
 196             # Path (relative) to the file
 197             filepath = os.path.join(root, curfile)
 198
 199             if os.path.islink(filepath):
 200                 continue
 201
 202             path_in_build_dir = os.path.relpath(filepath, build_dir)
 203             _ignored, ext = common.get_extension(path_in_build_dir)
 204
 205             if ext == 'so':
 206                 count += handleproblem('shared library', path_in_build_dir, filepath)
 207             elif ext == 'a':
 208                 count += handleproblem('static library', path_in_build_dir, filepath)
 209             elif ext == 'class':
 210                 count += handleproblem('Java compiled class', path_in_build_dir, filepath)
 211             elif ext == 'apk':
 212                 removeproblem('APK file', path_in_build_dir, filepath)
 213
 214             elif ext == 'jar':
 215                 for name in suspects_found(curfile):
 216                     count += handleproblem('usual suspect \'%s\'' % name, path_in_build_dir, filepath)
 217                 if curfile == 'gradle-wrapper.jar':
 218                     removeproblem('gradle-wrapper.jar', path_in_build_dir, filepath)
 219                 else:
 220                     warnproblem('JAR file', path_in_build_dir)
 221
 222             elif ext == 'aar':
 223                 warnproblem('AAR file', path_in_build_dir)
 224
 225             elif ext == 'java':
 226                 if not os.path.isfile(filepath):
 227                     continue
 228                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 229                     for line in f:
 230                         if 'DexClassLoader' in line:
 231                             count += handleproblem('DexClassLoader', path_in_build_dir, filepath)
 232                             break
 233
 234             elif ext == 'gradle':
 235                 if not os.path.isfile(filepath):
 236                     continue
 237                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 238                     lines = f.readlines()
 239                 for i, line in enumerate(lines):
 240                     if is_used_by_gradle(line):
 241                         for name in suspects_found(line):
 242                             count += handleproblem('usual suspect \'%s\' at line %d' % (name, i + 1), path_in_build_dir, filepath)
 243                 noncomment_lines = [l for l in lines if not common.gradle_comment.match(l)]
 244                 joined = re.sub(r'[\n\r\s]+', ' ', ' '.join(noncomment_lines))
 245                 for m in gradle_mavenrepo.finditer(joined):
 246                     url = m.group(2)
 247                     if not any(r.match(url) for r in allowed_repos):
 248                         count += handleproblem('unknown maven repo \'%s\'' % url, path_in_build_dir, filepath)
 249
 250             elif ext in ['', 'bin', 'out', 'exe']:
 251                 if is_binary(filepath):
 252                     count += handleproblem('binary', path_in_build_dir, filepath)
 253
 254             elif is_executable(filepath):
 255                 if is_binary(filepath) and not safe_path(path_in_build_dir):
 256                     warnproblem('possible binary', path_in_build_dir)
 257
 258     for p in scanignore:
 259         if p not in scanignore_worked:
 260             logging.error('Unused scanignore path: %s' % p)
 261             count += 1
 262
 263     for p in scandelete:
 264         if p not in scandelete_worked:
 265             logging.error('Unused scandelete path: %s' % p)
 266             count += 1
 267
 268     return count
 269
 270
 271 def main():
 272
 273     global config, options
 274
 275     # Parse command line...
 276     parser = ArgumentParser(usage="%(prog)s [options] [APPID[:VERCODE] [APPID[:VERCODE] ...]]")
 277     common.setup_global_opts(parser)
 278     parser.add_argument("appid", nargs='*', help=_("applicationId with optional versionCode in the form APPID[:VERCODE]"))
 279     metadata.add_metadata_arguments(parser)
 280     options = parser.parse_args()
 281     metadata.warnings_action = options.W
 282
 283     config = common.read_config(options)
 284
 285     # Read all app and srclib metadata
 286     allapps = metadata.read_metadata()
 287     apps = common.read_app_args(options.appid, allapps, True)
 288
 289     probcount = 0
 290
 291     build_dir = 'build'
 292     if not os.path.isdir(build_dir):
 293         logging.info("Creating build directory")
 294         os.makedirs(build_dir)
 295     srclib_dir = os.path.join(build_dir, 'srclib')
 296     extlib_dir = os.path.join(build_dir, 'extlib')
 297
 298     for appid, app in apps.items():
 299
 300         if app.Disabled:
 301             logging.info(_("Skipping {appid}: disabled").format(appid=appid))
 302             continue
 303
 304         try:
 305             if app.RepoType == 'srclib':
 306                 build_dir = os.path.join('build', 'srclib', app.Repo)
 307             else:
 308                 build_dir = os.path.join('build', appid)
 309
 310             if app.builds:
 311                 logging.info(_("Processing {appid}").format(appid=appid))
 312             else:
 313                 logging.info(_("{appid}: no builds specified, running on current source state")
 314                              .format(appid=appid))
 315                 count = scan_source(build_dir)
 316                 if count > 0:
 317                     logging.warn(_('Scanner found {count} problems in {appid}:')
 318                                  .format(count=count, appid=appid))
 319                     probcount += count
 320                 continue
 321
 322             # Set up vcs interface and make sure we have the latest code...
 323             vcs = common.getvcs(app.RepoType, app.Repo, build_dir)
 324
 325             for build in app.builds:
 326
 327                 if build.disable:
 328                     logging.info("...skipping version %s - %s" % (
 329                         build.versionName, build.get('disable', build.commit[1:])))
 330                     continue
 331
 332                 logging.info("...scanning version " + build.versionName)
 333                 # Prepare the source code...
 334                 common.prepare_source(vcs, app, build,
 335                                       build_dir, srclib_dir,
 336                                       extlib_dir, False)
 337
 338                 count = scan_source(build_dir, build)
 339                 if count > 0:
 340                     logging.warn(_('Scanner found {count} problems in {appid}:{versionCode}:')
 341                                  .format(count=count, appid=appid, versionCode=build.versionCode))
 342                     probcount += count
 343
 344         except BuildException as be:
 345             logging.warn("Could not scan app %s due to BuildException: %s" % (
 346                 appid, be))
 347             probcount += 1
 348         except VCSException as vcse:
 349             logging.warn("VCS error while scanning app %s: %s" % (appid, vcse))
 350             probcount += 1
 351         except Exception:
 352             logging.warn("Could not scan app %s due to unknown error: %s" % (
 353                 appid, traceback.format_exc()))
 354             probcount += 1
 355
 356     logging.info(_("Finished"))
 357     print(_("%d problems found") % probcount)
 358
 359
 360 if __name__ == "__main__":
 361     main()