Add a version-regexp option to customize how to extract version

We get versions like "2.1.28.gd935757" from OpenStack tarballs, with
d935757 being a commit id. This has two issues: this means new versions
can increase or decrease (since there's no guarantee the commit id
always go up), and we already put the commit id elsewhere in the rpm
version.

By specifying a custom regexp ('.*-([^-]+)\.g[a-zA-Z0-9]{7}'), we can
avoid this.

Note that the default ('.*-([^-]+)') matches the old behavior.

git_tarballs

@@ -54,17 +54,27 @@ def get_changelog_from_tarball(tar_name):
     return changelog
 
 
-def get_parent_dir_and_version_from_tarball(tar_name):
+def get_parent_dir_and_version_from_tarball(tar_name, version_regexp):
+    tar = tarfile.open(tar_name)
+    parent_dir = tar.firstmember.name
+    tar.close()
+
     try:
-        tar = tarfile.open(tar_name)
-        parent_dir = tar.firstmember.name
-        version = parent_dir.rsplit('-', 1)[1]
-        return (parent_dir, version)
-    except IndexError:
-        sys.exit("Could not figure out version from directory "
-                 "inside tarball: " + tar.firstmember.name)
-    finally:
-        tar.close()
+        match = re.match(version_regexp, parent_dir)
+    except re.error, e:
+        sys.exit("Could not use '%s' as regular expression to find "
+                 "version: " % (version_regexp, e))
+
+    if match is None:
+        sys.exit("Could not use '%s' as regular expression to find "
+                 "version in '%s': no match" % (version_regexp, parent_dir))
+    elif len(match.groups()) != 1:
+        sys.exit("Could not use '%s' as regular expression to find "
+                 "version in '%s': more than one match" % (version_regexp, parent_dir))
+    else:
+        version = match.group(1)
+
+    return (parent_dir, version)
 
 def get_upstream_commit(changelog):
     try:
@@ -203,6 +213,8 @@ if __name__ == '__main__':
                         help='the OBS package name')
     parser.add_argument('--email', required=True,
                         help='email of the commit author (for the .changes file)')
+    parser.add_argument('--version-regexp', default='.*-([^-]+)',
+                        help='regular expression for extracting version from top-level directory in tarball (default: ".*-([^-]+)")')
     parser.add_argument('--outdir', help='osc service parameter that does nothing')
     args = parser.parse_args()
 
@@ -216,7 +228,7 @@ if __name__ == '__main__':
     changelog = get_changelog_from_tarball(args.filename)
     changes_list = parse_changelog(changelog)
     upstream_commit = get_upstream_commit(changelog)
-    tarball_parent_dir, upstream_version = get_parent_dir_and_version_from_tarball(args.filename)
+    tarball_parent_dir, upstream_version = get_parent_dir_and_version_from_tarball(args.filename, args.version_regexp)
 
     package_commit = get_commit_from_spec(args.package)
     package_version = package_version(upstream_version, upstream_commit)

git_tarballs.service

@@ -15,4 +15,7 @@
     <description>User's email that will be used to sign the rpm's .changes file</description>
     <required/>
   </param>
+  <param name="version-regexp">
+    <description>Regular expression for extracting version from top-level directory in tarball (default: ".*-([^-]+)")</description>
+  </param>
 </service>