From fd440f0e6b812eb97b7b43688cda6ef3c4747bc6 Mon Sep 17 00:00:00 2001
From: Benoit Sigoure <tsuna@lrde.epita.fr>
Date: Sun, 20 Jan 2008 09:55:12 +0100
Subject: [PATCH] Find the merge parent faster.

	* svn-merge2git.sh (find_merge_parent): Generate an ERE that
	intelligently matches the possible merge parents so that the first
	commit found by rev-list is the merge parent (instead of finding all
	the commits in the merged branch and piping this all through a Perl
	script that finds the good parent).
	(rewrite_history): Make sure that filter-branch does not fail.
	(top-level aka `main'): Count the total number of commits processed.

Signed-off-by: Benoit Sigoure <tsuna@lrde.epita.fr>
---
 svn-merge2git.sh | 51 ++++++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 40 insertions(+), 11 deletions(-)

diff --git a/svn-merge2git.sh b/svn-merge2git.sh
index bab7ee4..cdea480 100755
--- a/svn-merge2git.sh
+++ b/svn-merge2git.sh
@@ -139,16 +139,39 @@ find_merge_parent()
     merge_parent='unknown'
     return 0
   fi
-  svn_merge_parent=`git rev-list --all --header \
-             --grep="^ *git-svn-id: .*/$svn_branch_name@[0-9]* [-0-9a-f]*\\$" |
-             perl -we 'my $rev = 0; my $t = 0;
-             while(<>) {
-               m{^ *git-svn-id: .*/\Q'"$svn_branch_name"'\E@([0-9]+) [-0-9a-f]+$}
-                 or next;
-               $t = int($1);
-               $rev = $t if $rev < $t and $t <= '"$svn_merge_to"';
-             }
-             print $rev'`
+  # Create a range to intelligently limit the match of rev-list.  This will
+  # produce a RE that rules out all the impossible revision numbers (that is,
+  # the revisions >TO).  e.g:
+  # 7    -> ([0-7])
+  # 42   -> (4[0-2]|[0-3][0-9]|[1-9])
+  # 123  -> (12[0-3]|1[0-1][0-9]|0[0-9][0-9]|[1-9][0-9]{0,1})
+  # 6951 -> (695[0-1]|69[0-4][0-9]|6[0-8][0-9][0-9]|[0-5][0-9][0-9][0-9]|[1-9][0-9]{0,2})
+  perl_tmp='$_ = "'"$svn_merge_to"'";
+            my $l = length($_);
+            my @r;
+            foreach my $i (0 .. $l - 1) {
+              /^(\d*)(\d)(\d{$i})$/;
+              my ($a, $b, $c) = ($1, int($2), $3);
+              if ($i != 0) {
+                # Avoid pitfalls e.g. 10[0-9] or 0[0-9][0-9] for 101
+                next if $b == 0 or ($b == 1 and $a eq "");
+                --$b;
+              }
+              $b = "[0-$b]" if $b;
+              $c =~ s/./[0-9]/g;
+              push(@r, "$a$b$c");
+            }
+            push(@r, "[1-9]" . ($l - 2 ? "[0-9]{0," . ($l - 2) . "}" : ""))
+              if $l > 1;
+            print "(" . join("|", @r) . ")";'
+  rev_range=`perl -we "$perl_tmp"`
+  sed_tmp='s/^ *git-svn-id: .*@\([0-9]*\) [-0-9a-f]*$/\1/p'
+  svn_merge_parent=`git rev-list --all -1 --header -E \
+       --grep="^ *git-svn-id: .*/$svn_branch_name@$rev_range [-0-9a-f]*\\$" \
+       | sed -n "$sed_tmp"`
+  case $svn_merge_parent in #(
+    '' | *[^0-9]*) fatal "invalid svn_merge_parent: '$svn_merge_parent'";;
+  esac
   rv=$?
   test $rv -eq 0 || fatal "perl returned $rv"
   if $opt_verbose; then
@@ -206,7 +229,8 @@ create_graft()
 rewrite_history()
 {
   $opt_dryrun && return 0
-  git filter-branch --parent-filter cat -- --all
+  git filter-branch --parent-filter cat -- --all \
+    || fatal "git filter-branch returned $?"
 }
 
 # doit <REF>
@@ -361,14 +385,19 @@ else
   has_log_exclude=:
 fi
 
+totalmerge=0
+totalconverted=0
 for refspec
 do
   nconverted=0
   nmerge=0
   doit "$refspec"
   echo ">>> processed $nconverted/$nmerge merges in $refspec"
+  totalmerge=$(($totalmerge + $nmerge))
+  totalconverted=$(($totalconverted + $nconverted))
 done
 
 rewrite_history
 
+echo "Done.  Processed $totalconverted/$totalmerge merges"
 test $warnings -eq 0 || warn "job completed with $warnings warnings"
-- 
2.11.4.GIT