Improve version guessing.

Change-Id: I3a462f170521c4296e82ea1d2567a7ca7ed476c9

1 files changed, 72 insertions, 1 deletions

diff --git a/morphlib/plugins/cve_check_plugin.py b/morphlib/plugins/cve_check_plugin.py
index 26bb987c..3179d797 100644
--- a/morphlib/plugins/cve_check_plugin.py
+++ b/morphlib/plugins/cve_check_plugin.py
@@ -19,6 +19,7 @@
 
 import warnings
 
+import re
 import cliapp
 import morphlib
 
@@ -55,6 +56,7 @@ class CVECheckPlugin(cliapp.Plugin):
         self.lrc, self.rrc = morphlib.util.new_repo_caches(self.app)
         self.resolver = morphlib.artifactresolver.ArtifactResolver()
         self.cve_db = CVEDataBase()
+        self.version_guesser = VersionGuesser()
 
         for system_filename in system_filenames:
             self.certify_system(repo, ref, system_filename)
@@ -100,7 +102,7 @@ class CVECheckPlugin(cliapp.Plugin):
                 self.lrc.cache_repo(source.repo_name)
 
             cached = self.lrc.get_repo(source.repo_name)
-            version = cached.version_guess(ref)
+            version = self.version_guesser.guess_version(cached, ref)
 
             self.cve_db.check_vulnerability(name, version)
 
@@ -180,3 +182,72 @@ class CVEDataBase:
 
             s.check_vulnerability(version)
             break
+
+
+class ProjectVersionGuesser(object):
+
+    def __init__(self, interesting_files):
+        self.interesting_files = interesting_files
+
+    def file_contents(self, cached, ref, tree):
+        filenames = [x for x in self.interesting_files if x in tree]
+        if filenames:
+            for filename in filenames:
+                yield filename, cached.read_file(filename, ref)
+
+class AutotoolsVersionGuesser(ProjectVersionGuesser):
+
+    def __init__(self):
+        ProjectVersionGuesser.__init__(self, [
+            'configure.ac',
+            'configure.in',
+            'configure.ac.in',
+            'configure.in.in',
+        ])
+
+    def guess_version(self, cached, ref, tree):
+        version = None
+        for filename, data in self.file_contents(cached, ref, tree):
+            # First, try to grep for AC_INIT()
+            version = self._check_ac_init(data)
+            if version:
+                break
+
+        return version
+
+    def _check_ac_init(self, data):
+        data = data.replace('\n', ' ')
+        for macro in ['AC_INIT', 'AM_INIT_AUTOMAKE']:
+            pattern = r'.*%s\((.*?)\).*' % macro
+            if not re.match(pattern, data):
+                continue
+            acinit = re.sub(pattern, r'\1', data)
+            if acinit:
+                version = acinit.split(',')
+                if macro == 'AM_INIT_AUTOMAKE' and len(version) == 1:
+                    continue
+                version = version[0] if len(version) == 1 else version[1]
+                version = re.sub('[\[\]]', '', version).strip()
+                version = version.split()[0]
+                if version:
+                    if version and version[0].isdigit():
+                        return version
+        return None
+
+class VersionGuesser(object):
+
+    def __init__(self):
+        self.guessers = [
+            AutotoolsVersionGuesser()
+        ]
+
+    def guess_version(self, cached, ref):
+        version = None
+        tree = cached.list_files(ref, recurse=False)
+        for guesser in self.guessers:
+            version = guesser.guess_version(cached, ref, tree)
+            if version:
+                return version
+
+        # Fall back to `git describe` which always gives something
+        return cached.version_guess(ref)