Use regular expressions to allow better control for files to export.

1 files changed, 57 insertions, 6 deletions

diff --git a/exporters/svn-fast-export.py b/exporters/svn-fast-export.py
index f2719ae..e44c6cb 100755
--- a/exporters/svn-fast-export.py
+++ b/exporters/svn-fast-export.py
@@ -35,6 +35,56 @@ def dump_file_blob(root, full_path, pool):
     sys.stdout.write("\n")
 
 
+class Matcher(object):
+
+    branch = None
+
+    def __init__(self, trunk_path):
+        self.trunk_path = trunk_path
+
+    def branchname(self):
+        return self.branch
+
+    def __str__(self):
+        return super(Matcher, self).__str__() + ":" + self.trunk_path
+
+    @staticmethod
+    def getMatcher(trunk_path):
+        if trunk_path.startswith("regex:"):
+            return RegexStringMatcher(trunk_path)
+        else:
+            return StaticStringMatcher(trunk_path)
+
+class StaticStringMatcher(Matcher):
+
+    branch = "master"
+
+    def matches(self, path):
+        return path.startswith(trunk_path)
+
+    def replace(self, path):
+        return path.replace(self.trunk_path, '')
+
+class RegexStringMatcher(Matcher):
+
+    def __init__(self, trunk_path):
+        super(RegexStringMatcher, self).__init__(trunk_path)
+        import re
+        self.matcher = re.compile(self.trunk_path[len("regex:"):])
+
+    def matches(self, path):
+        match = self.matcher.match(path)
+        if match:
+            self.branch = match.group(1)
+            return True
+        else:
+            return False
+
+    def replace(self, path):
+        return self.matcher.sub("\g<2>", path)
+
+MATCHER = None
+
 def export_revision(rev, repo, fs, pool):
     sys.stderr.write("Exporting revision %s... " % rev)
 
@@ -55,15 +105,14 @@ def export_revision(rev, repo, fs, pool):
         c_t = ct_short[change_type.change_kind]
         if svn_fs_is_dir(root, path, revpool):
             continue
-
-        if not path.startswith(trunk_path):
+        if not MATCHER.matches(path):
             # We don't handle branches. Or tags. Yet.
             pass
         else:
             if c_t == 'D':
-                file_changes.append("D %s" % path.replace(trunk_path, ''))
+                file_changes.append("D %s" % MATCHER.replace(path))
             else:
-                marks[i] = path.replace(trunk_path, '')
+                marks[i] = MATCHER.replace(path)
                 file_changes.append("M 644 :%s %s" % (i, marks[i]))
                 sys.stdout.write("blob\nmark :%s\n" % i)
                 dump_file_blob(root, path, revpool)
@@ -85,7 +134,7 @@ def export_revision(rev, repo, fs, pool):
 
     svndate = props['svn:date'][0:-8]
     commit_time = mktime(strptime(svndate, '%Y-%m-%dT%H:%M:%S'))
-    sys.stdout.write("commit refs/heads/master\n")
+    sys.stdout.write("commit refs/heads/%s\n" % MATCHER.branchname())
     sys.stdout.write("committer %s %s -0000\n" % (author, int(commit_time)))
     sys.stdout.write("data %s\n" % len(props['svn:log']))
     sys.stdout.write(props['svn:log'])
@@ -129,7 +178,7 @@ if __name__ == '__main__':
     parser.set_usage(usage)
     parser.add_option('-f', '--final-rev', help='Final revision to import', 
                       dest='final_rev', metavar='FINAL_REV', type='int')
-    parser.add_option('-t', '--trunk-path', help='Path in repo to /trunk',
+    parser.add_option('-t', '--trunk-path', help="Path in repo to /trunk, may be `regex:/cvs/(trunk)/proj1/(.*)`\nFirst group is used as branchname, second to match files",
                       dest='trunk_path', metavar='TRUNK_PATH')
     parser.add_option('-b', '--branches-path', help='Path in repo to /branches',
                       dest='branches_path', metavar='BRANCHES_PATH')
@@ -146,6 +195,8 @@ if __name__ == '__main__':
     if options.final_rev != None:
         final_rev = options.final_rev
 
+    MATCHER = Matcher.getMatcher(trunk_path)
+    sys.stderr.write("%s\n" % MATCHER)
     if len(args) != 1:
         parser.print_help()
         sys.exit(2)