From: Germán Póo-Caamaño <gpoo@gnome.org>
Date: Thu, 23 Jun 2011 02:15:59 +0000 (-0700)
Subject: Added workaround for svn tags imported wrongly
X-Git-Url: http://xenbits.xensource.com/gitweb?a=commitdiff_plain;h=1a85acef6b98f27621e5f55e395c82691ed89693;p=people%2Flarsk%2Fxenproject-org-gitdm.git

Added workaround for svn tags imported wrongly

When some projects have migrated from Subversion to Git, there
were several tags that were treated as new commits, which shows
a change in the whole project (code added/removed) when nothing
really happened.  For instance, in GNOME a lot svn tags were
catched during the migration, but not all of them.

svn tags in git repositories brings bad stats because double count
commits, and in project with a lot history it may may involve several thousands of source of lines of code.

Signed-off-by: Germán Póo-Caamaño <gpoo@gnome.org>
---

diff --git a/gitdm b/gitdm
index fe5473c..28df314 100755
--- a/gitdm
+++ b/gitdm
@@ -341,6 +341,22 @@ def ApplyFileFilter (line, ignore):
             return 0
     return ignore
 
+def is_svntag(logpatch):
+    """
+        This is a workaround for a bug on the migration to Git
+        from Subversion found in GNOME.  It may happen in other
+        repositories as well.
+    """
+
+    for Line in logpatch:
+        m = patterns['svn-tag'].match(Line.strip())
+        if m:
+            sys.stderr.write ('(W) detected a commit on a svn tag: %s\n' %
+                              (m.group (0),))
+            return True
+
+    return False
+
 #
 # If this patch is signed off by both Andrew Morton and Linus Torvalds,
 # remove the (redundant) Linus signoff.
@@ -384,6 +400,15 @@ for logpatch in patches:
     if (printcount % 50) == 0:
         print >> sys.stderr, 'Grabbing changesets...%d\r' % printcount,
     printcount += 1
+
+    # We want to ignore commits on svn tags since in Subversion
+    # thats mean a copy of the whole repository, which leads to
+    # wrong results.  Some migrations from Subversion to Git does
+    # not catch all this tags/copy and import them just as a new
+    # big changeset.
+    if is_svntag(logpatch):
+        continue
+
     p = grabpatch(logpatch)
     if not p:
         break
diff --git a/patterns.py b/patterns.py
index b679049..803e532 100644
--- a/patterns.py
+++ b/patterns.py
@@ -46,4 +46,6 @@ patterns = {
     # It implies --numstat
     'numstat': re.compile('^(\d+|-)\s+(\d+|-)\s+(.*)$'),
     'rename' : re.compile('(.*)\{(.*) => (.*)\}(.*)'),
+    # Detect errors on svn conversions
+    'svn-tag': re.compile("^svn path=/tags/(.*)/?; revision=([0-9]+)$"),
 }