chiark / gitweb /
PEP8 fix E225 missing whitespace around operator
[fdroidserver.git] / fdroidserver / scanner.py
index 7f7c6aa9b226a496e80acb436fc5c1efdc76dc08..e41e557057a8da4d9d5a71f5095504fedeed6819 100644 (file)
@@ -31,49 +31,81 @@ config = None
 options = None
 
 
+def get_gradle_compile_commands(build):
+    compileCommands = ['compile', 'releaseCompile']
+    if build.gradle and build.gradle != ['yes']:
+        compileCommands += [flavor + 'Compile' for flavor in build.gradle]
+        compileCommands += [flavor + 'ReleaseCompile' for flavor in build.gradle]
+
+    return [re.compile(r'\s*' + c, re.IGNORECASE) for c in compileCommands]
+
+
 # Scan the source code in the given directory (and all subdirectories)
 # and return the number of fatal problems encountered
-def scan_source(build_dir, root_dir, thisbuild):
+def scan_source(build_dir, root_dir, build):
 
     count = 0
 
     # Common known non-free blobs (always lower case):
-    usual_suspects = [
-        re.compile(r'.*flurryagent', re.IGNORECASE),
-        re.compile(r'.*paypal.*mpl', re.IGNORECASE),
-        re.compile(r'.*google.*analytics', re.IGNORECASE),
-        re.compile(r'.*admob.*sdk.*android', re.IGNORECASE),
-        re.compile(r'.*google.*ad.*view', re.IGNORECASE),
-        re.compile(r'.*google.*admob', re.IGNORECASE),
-        re.compile(r'.*google.*play.*services', re.IGNORECASE),
-        re.compile(r'.*crittercism', re.IGNORECASE),
-        re.compile(r'.*heyzap', re.IGNORECASE),
-        re.compile(r'.*jpct.*ae', re.IGNORECASE),
-        re.compile(r'.*youtube.*android.*player.*api', re.IGNORECASE),
-        re.compile(r'.*bugsense', re.IGNORECASE),
-        re.compile(r'.*crashlytics', re.IGNORECASE),
-        re.compile(r'.*ouya.*sdk', re.IGNORECASE),
-        re.compile(r'.*libspen23', re.IGNORECASE),
+    usual_suspects = {
+        exp: re.compile(r'.*' + exp, re.IGNORECASE) for exp in [
+            r'flurryagent',
+            r'paypal.*mpl',
+            r'google.*analytics',
+            r'admob.*sdk.*android',
+            r'google.*ad.*view',
+            r'google.*admob',
+            r'google.*play.*services',
+            r'crittercism',
+            r'heyzap',
+            r'jpct.*ae',
+            r'youtube.*android.*player.*api',
+            r'bugsense',
+            r'crashlytics',
+            r'ouya.*sdk',
+            r'libspen23',
+        ]
+    }
+
+    def suspects_found(s):
+        for n, r in usual_suspects.iteritems():
+            if r.match(s):
+                yield n
+
+    gradle_mavenrepo = re.compile(r'maven *{ *(url)? *[\'"]?([^ \'"]*)[\'"]?')
+
+    allowed_repos = [re.compile(r'^https?://' + re.escape(repo) + r'/*') for repo in [
+        'repo1.maven.org/maven2',  # mavenCentral()
+        'jcenter.bintray.com',     # jcenter()
+        'jitpack.io',
+        'repo.maven.apache.org/maven2',
+        'oss.sonatype.org/content/repositories/snapshots',
+        'oss.sonatype.org/content/repositories/releases',
+        'oss.sonatype.org/content/groups/public',
+        'clojars.org/repo',  # Clojure free software libs
+        ]
     ]
 
-    scanignore = common.getpaths(build_dir, thisbuild, 'scanignore')
-    scandelete = common.getpaths(build_dir, thisbuild, 'scandelete')
+    scanignore = common.getpaths_map(build_dir, build.scanignore)
+    scandelete = common.getpaths_map(build_dir, build.scandelete)
 
     scanignore_worked = set()
     scandelete_worked = set()
 
     def toignore(fd):
-        for p in scanignore:
-            if fd.startswith(p):
-                scanignore_worked.add(p)
-                return True
+        for k, paths in scanignore.iteritems():
+            for p in paths:
+                if fd.startswith(p):
+                    scanignore_worked.add(k)
+                    return True
         return False
 
     def todelete(fd):
-        for p in scandelete:
-            if fd.startswith(p):
-                scandelete_worked.add(p)
-                return True
+        for k, paths in scandelete.iteritems():
+            for p in paths:
+                if fd.startswith(p):
+                    scandelete_worked.add(k)
+                    return True
         return False
 
     def ignoreproblem(what, fd, fp):
@@ -86,6 +118,8 @@ def scan_source(build_dir, root_dir, thisbuild):
         return 0
 
     def warnproblem(what, fd):
+        if toignore(fd):
+            return
         logging.warn('Found %s at %s' % (what, fd))
 
     def handleproblem(what, fd, fp):
@@ -107,6 +141,24 @@ def scan_source(build_dir, root_dir, thisbuild):
             d = f.read(1024)
         return bool(d.translate(None, textchars))
 
+    # False positives patterns for files that are binary and executable.
+    safe_paths = [re.compile(r) for r in [
+        r".*/drawable[^/]*/.*\.png$",  # png drawables
+        r".*/mipmap[^/]*/.*\.png$",    # png mipmaps
+        ]
+    ]
+
+    def safe_path(path):
+        for sp in safe_paths:
+            if sp.match(path):
+                return True
+        return False
+
+    gradle_compile_commands = get_gradle_compile_commands(build)
+
+    def is_used_by_gradle(line):
+        return any(command.match(line) for command in gradle_compile_commands)
+
     # Iterate through all files in the source code
     for r, d, f in os.walk(build_dir, topdown=True):
 
@@ -117,11 +169,17 @@ def scan_source(build_dir, root_dir, thisbuild):
 
         for curfile in f:
 
+            if curfile in ['.DS_Store']:
+                continue
+
             # Path (relative) to the file
             fp = os.path.join(r, curfile)
-            fd = fp[len(build_dir) + 1:]
 
-            ext = common.get_extension(fd)
+            if os.path.islink(fp):
+                continue
+
+            fd = fp[len(build_dir) + 1:]
+            _, ext = common.get_extension(fd)
 
             if ext == 'so':
                 count += handleproblem('shared library', fd, fp)
@@ -133,10 +191,9 @@ def scan_source(build_dir, root_dir, thisbuild):
                 removeproblem('APK file', fd, fp)
 
             elif ext == 'jar':
-                if any(suspect.match(curfile) for suspect in usual_suspects):
-                    count += handleproblem('usual supect', fd, fp)
-                else:
-                    warnproblem('JAR file', fd)
+                for name in suspects_found(curfile):
+                    count += handleproblem('usual supect \'%s\'' % name, fd, fp)
+                warnproblem('JAR file', fd)
 
             elif ext == 'java':
                 if not os.path.isfile(fp):
@@ -149,17 +206,26 @@ def scan_source(build_dir, root_dir, thisbuild):
             elif ext == 'gradle':
                 if not os.path.isfile(fp):
                     continue
-                for i, line in enumerate(file(fp)):
-                    i = i + 1
-                    if any(suspect.match(line) for suspect in usual_suspects):
-                        count += handleproblem('usual suspect at line %d' % i, fd, fp)
-                        break
-
-            elif is_binary(fp):
-                if is_executable(fp):
-                    count += handleproblem('executable binary', fd, fp)
-                elif ext == '':
-                    count += handleproblem('unknown binary', fd, fp)
+                with open(fp, 'r') as f:
+                    lines = f.readlines()
+                for i, line in enumerate(lines):
+                    if is_used_by_gradle(line):
+                        for name in suspects_found(line):
+                            count += handleproblem('usual supect \'%s\' at line %d' % (name, i + 1), fd, fp)
+                noncomment_lines = [l for l in lines if not common.gradle_comment.match(l)]
+                joined = re.sub(r'[\n\r\s]+', ' ', ' '.join(noncomment_lines))
+                for m in gradle_mavenrepo.finditer(joined):
+                    url = m.group(2)
+                    if not any(r.match(url) for r in allowed_repos):
+                        count += handleproblem('unknown maven repo \'%s\'' % url, fd, fp)
+
+            elif ext in ['', 'bin', 'out', 'exe']:
+                if is_binary(fp):
+                    count += handleproblem('binary', fd, fp)
+
+            elif is_executable(fp):
+                if is_binary(fp) and not safe_path(fd):
+                    warnproblem('possible binary', fd)
 
     for p in scanignore:
         if p not in scanignore_worked:
@@ -171,14 +237,6 @@ def scan_source(build_dir, root_dir, thisbuild):
             logging.error('Unused scandelete path: %s' % p)
             count += 1
 
-    # Presence of a jni directory without buildjni=yes might
-    # indicate a problem (if it's not a problem, explicitly use
-    # buildjni=no to bypass this check)
-    if (os.path.exists(os.path.join(root_dir, 'jni')) and
-            not thisbuild['buildjni']):
-        logging.error('Found jni directory, but buildjni is not enabled. Set it to \'no\' to ignore.')
-        count += 1
-
     return count
 
 
@@ -209,10 +267,10 @@ def main():
 
     for appid, app in apps.iteritems():
 
-        if app['Disabled']:
+        if app.Disabled:
             logging.info("Skipping %s: disabled" % appid)
             continue
-        if not app['builds']:
+        if not app.builds:
             logging.info("Skipping %s: no builds specified" % appid)
             continue
 
@@ -220,32 +278,32 @@ def main():
 
         try:
 
-            if app['Repo Type'] == 'srclib':
-                build_dir = os.path.join('build', 'srclib', app['Repo'])
+            if app.RepoType == 'srclib':
+                build_dir = os.path.join('build', 'srclib', app.Repo)
             else:
                 build_dir = os.path.join('build', appid)
 
             # Set up vcs interface and make sure we have the latest code...
-            vcs = common.getvcs(app['Repo Type'], app['Repo'], build_dir)
+            vcs = common.getvcs(app.RepoType, app.Repo, build_dir)
 
-            for thisbuild in app['builds']:
+            for build in app.builds:
 
-                if thisbuild['disable']:
+                if build.disable:
                     logging.info("...skipping version %s - %s" % (
-                        thisbuild['version'], thisbuild.get('disable', thisbuild['commit'][1:])))
+                        build.version, build.get('disable', build.commit[1:])))
                 else:
-                    logging.info("...scanning version " + thisbuild['version'])
+                    logging.info("...scanning version " + build.version)
 
                     # Prepare the source code...
-                    root_dir, _ = common.prepare_source(vcs, app, thisbuild,
+                    root_dir, _ = common.prepare_source(vcs, app, build,
                                                         build_dir, srclib_dir,
                                                         extlib_dir, False)
 
                     # Do the scan...
-                    count = scan_source(build_dir, root_dir, thisbuild)
+                    count = scan_source(build_dir, root_dir, build)
                     if count > 0:
                         logging.warn('Scanner found %d problems in %s (%s)' % (
-                            count, appid, thisbuild['vercode']))
+                            count, appid, build.vercode))
                         probcount += count
 
         except BuildException as be:
@@ -261,7 +319,7 @@ def main():
             probcount += 1
 
     logging.info("Finished:")
-    print "%d problems found" % probcount
+    print("%d problems found" % probcount)
 
 if __name__ == "__main__":
     main()