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                 warnproblem('JAR file', path_in_build_dir)
 202
 203             elif ext == 'aar':
 204                 warnproblem('AAR file', path_in_build_dir)
 205
 206             elif ext == 'java':
 207                 if not os.path.isfile(filepath):
 208                     continue
 209                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 210                     for line in f:
 211                         if 'DexClassLoader' in line:
 212                             count += handleproblem('DexClassLoader', path_in_build_dir, filepath)
 213                             break
 214
 215             elif ext == 'gradle':
 216                 if not os.path.isfile(filepath):
 217                     continue
 218                 with open(filepath, 'r', encoding='utf8', errors='replace') as f:
 219                     lines = f.readlines()
 220                 for i, line in enumerate(lines):
 221                     if is_used_by_gradle(line):
 222                         for name in suspects_found(line):
 223                             count += handleproblem('usual supect \'%s\' at line %d' % (name, i + 1), path_in_build_dir, filepath)
 224                 noncomment_lines = [l for l in lines if not common.gradle_comment.match(l)]
 225                 joined = re.sub(r'[\n\r\s]+', ' ', ' '.join(noncomment_lines))
 226                 for m in gradle_mavenrepo.finditer(joined):
 227                     url = m.group(2)
 228                     if not any(r.match(url) for r in allowed_repos):
 229                         count += handleproblem('unknown maven repo \'%s\'' % url, path_in_build_dir, filepath)
 230
 231             elif ext in ['', 'bin', 'out', 'exe']:
 232                 if is_binary(filepath):
 233                     count += handleproblem('binary', path_in_build_dir, filepath)
 234
 235             elif is_executable(filepath):
 236                 if is_binary(filepath) and not safe_path(path_in_build_dir):
 237                     warnproblem('possible binary', path_in_build_dir)
 238
 239     for p in scanignore:
 240         if p not in scanignore_worked:
 241             logging.error('Unused scanignore path: %s' % p)
 242             count += 1
 243
 244     for p in scandelete:
 245         if p not in scandelete_worked:
 246             logging.error('Unused scandelete path: %s' % p)
 247             count += 1
 248
 249     return count
 250
 251
 252 def main():
 253
 254     global config, options
 255
 256     # Parse command line...
 257     parser = ArgumentParser(usage="%(prog)s [options] [APPID[:VERCODE] [APPID[:VERCODE] ...]]")
 258     common.setup_global_opts(parser)
 259     parser.add_argument("appid", nargs='*', help="app-id with optional versionCode in the form APPID[:VERCODE]")
 260     metadata.add_metadata_arguments(parser)
 261     options = parser.parse_args()
 262     metadata.warnings_action = options.W
 263
 264     config = common.read_config(options)
 265
 266     # Read all app and srclib metadata
 267     allapps = metadata.read_metadata()
 268     apps = common.read_app_args(options.appid, allapps, True)
 269
 270     probcount = 0
 271
 272     build_dir = 'build'
 273     if not os.path.isdir(build_dir):
 274         logging.info("Creating build directory")
 275         os.makedirs(build_dir)
 276     srclib_dir = os.path.join(build_dir, 'srclib')
 277     extlib_dir = os.path.join(build_dir, 'extlib')
 278
 279     for appid, app in apps.items():
 280
 281         if app.Disabled:
 282             logging.info("Skipping %s: disabled" % appid)
 283             continue
 284         if not app.builds:
 285             logging.info("Skipping %s: no builds specified" % appid)
 286             continue
 287
 288         logging.info("Processing " + appid)
 289
 290         try:
 291
 292             if app.RepoType == 'srclib':
 293                 build_dir = os.path.join('build', 'srclib', app.Repo)
 294             else:
 295                 build_dir = os.path.join('build', appid)
 296
 297             # Set up vcs interface and make sure we have the latest code...
 298             vcs = common.getvcs(app.RepoType, app.Repo, build_dir)
 299
 300             for build in app.builds:
 301
 302                 if build.disable:
 303                     logging.info("...skipping version %s - %s" % (
 304                         build.versionName, build.get('disable', build.commit[1:])))
 305                 else:
 306                     logging.info("...scanning version " + build.versionName)
 307
 308                     # Prepare the source code...
 309                     common.prepare_source(vcs, app, build,
 310                                           build_dir, srclib_dir,
 311                                           extlib_dir, False)
 312
 313                     # Do the scan...
 314                     count = scan_source(build_dir, build)
 315                     if count > 0:
 316                         logging.warn('Scanner found %d problems in %s (%s)' % (
 317                             count, appid, build.versionCode))
 318                         probcount += count
 319
 320         except BuildException as be:
 321             logging.warn("Could not scan app %s due to BuildException: %s" % (
 322                 appid, be))
 323             probcount += 1
 324         except VCSException as vcse:
 325             logging.warn("VCS error while scanning app %s: %s" % (appid, vcse))
 326             probcount += 1
 327         except Exception:
 328             logging.warn("Could not scan app %s due to unknown error: %s" % (
 329                 appid, traceback.format_exc()))
 330             probcount += 1
 331
 332     logging.info("Finished:")
 333     print("%d problems found" % probcount)
 334
 335
 336 if __name__ == "__main__":
 337     main()