5x to 8x speedup in "hg convert"

pygit2.patch

# HG changeset patch
# User Bryan O'Sullivan <[email protected]>
# Date 1331697209 25200
# Branch stable
# Node ID 9c15f20c0418fbad1da202f72dc894372538beba
# Parent  6344043924497cd06d781d9014c66802285072e4
imported patch libgit2.patch

diff -r 634404392449 -r 9c15f20c0418 hgext/convert/git.py
--- a/hgext/convert/git.py	Sun Jan 01 13:37:30 2012 -0600
+++ b/hgext/convert/git.py	Tue Mar 13 20:53:29 2012 -0700
@@ -12,7 +12,22 @@
 
 from common import NoRepo, commit, converter_source, checktool
 
-class convert_git(converter_source):
+class convert_git_base(converter_source):
+    def __init__(self, ui, path, rev):
+        super(convert_git_base, self).__init__(ui, path, rev=rev)
+
+        if os.path.isdir(path + "/.git"):
+            path += "/.git"
+        if not os.path.exists(path + "/objects"):
+            raise NoRepo(_("%s does not look like a Git repository") % path)
+
+        checktool('git', 'git')
+        self.path = path
+
+    # pygit2 0.16 has no support for diffing, so we have to shell out
+    # to git for diffs. the methods below would otherwise belong in
+    # convert_git_plain.
+
     # Windows does not support GIT_DIR= construct while other systems
     # cannot remove environment variable. Just assume none have
     # both issues.
@@ -39,47 +54,6 @@
             else:
                 return util.popen('GIT_DIR=%s %s' % (self.path, s), 'rb')
 
-    def gitread(self, s):
-        fh = self.gitopen(s)
-        data = fh.read()
-        return data, fh.close()
-
-    def __init__(self, ui, path, rev=None):
-        super(convert_git, self).__init__(ui, path, rev=rev)
-
-        if os.path.isdir(path + "/.git"):
-            path += "/.git"
-        if not os.path.exists(path + "/objects"):
-            raise NoRepo(_("%s does not look like a Git repository") % path)
-
-        checktool('git', 'git')
-
-        self.path = path
-
-    def getheads(self):
-        if not self.rev:
-            heads, ret = self.gitread('git rev-parse --branches --remotes')
-            heads = heads.splitlines()
-        else:
-            heads, ret = self.gitread("git rev-parse --verify %s" % self.rev)
-            heads = [heads[:-1]]
-        if ret:
-            raise util.Abort(_('cannot retrieve git heads'))
-        return heads
-
-    def catfile(self, rev, type):
-        if rev == hex(nullid):
-            raise IOError()
-        data, ret = self.gitread("git cat-file %s %s" % (type, rev))
-        if ret:
-            raise util.Abort(_('cannot read %r object at %s') % (type, rev))
-        return data
-
-    def getfile(self, name, rev):
-        data = self.catfile(rev, "blob")
-        mode = self.modecache[(name, rev)]
-        return data, mode
-
     def getchanges(self, version):
         self.modecache = {}
         fh = self.gitopen("git diff-tree -z --root -m -r %s" % version)
@@ -106,6 +80,105 @@
             raise util.Abort(_('cannot read changes in %s') % version)
         return (changes, {})
 
+    def getchangedfiles(self, version, i):
+        changes = []
+        if i is None:
+            fh = self.gitopen("git diff-tree --root -m -r %s" % version)
+            for l in fh:
+                if "\t" not in l:
+                    continue
+                m, f = l[:-1].split("\t")
+                changes.append(f)
+        else:
+            fh = self.gitopen('git diff-tree --name-only --root -r %s "%s^%s" --'
+                             % (version, version, i + 1))
+            changes = [f.rstrip('\n') for f in fh]
+        if fh.close():
+            raise util.Abort(_('cannot read changes in %s') % version)
+
+        return changes
+
+def hexoid(obj):
+    # pygit2's "hex" property is unicode, but "oid" is str
+    return obj.oid.encode('hex')
+
+class convert_git_pygit2(convert_git_base):
+    def __init__(self, ui, path, rev):
+        super(convert_git_pygit2, self).__init__(ui, path, rev=rev)
+        import pygit2
+        self.repo = pygit2.init_repository(path, True)
+
+    def getheads(self):
+        if not self.rev:
+            return [hexoid(self.repo.lookup_reference(r).resolve())
+                    for r in self.repo.listall_references()
+                    if (r.startswith('refs/heads/') or
+                        r.startswith('refs/remotes/'))]
+        else:
+            try:
+                return [hexoid(repo.lookup_reference(self.rev).resolve())]
+            except KeyError:
+                return [hexoid(repo[self.rev])]
+
+    def getcommit(self, rev):
+        def prettyname(sig):
+            if sig.name:
+                return self.recode('%s <%s>' % (sig.name, sig.email))
+            return self.recode(sig.email)
+
+        c = self.repo[rev.decode('hex')]
+        message = c.message
+        author = prettyname(c.author)
+        committer = prettyname(c.committer)
+        if author != committer:
+            message += '\ncommitter: %s\n' % committer
+        return commit(parents=[hexoid(p) for p in c.parents],
+                      date='%s %s' % (c.commit_time, c.commit_time_offset),
+                      author=author, desc=self.recode(message), rev=rev)
+
+    hex_nullid = hex(nullid)
+
+    def getfile(self, name, rev):
+        if rev == self.hex_nullid:
+            raise IOError
+        return self.repo[rev.decode('hex')].data, self.modecache[(name,rev)]
+
+    def gettags(self):
+        return dict((r.split('/',2)[-1],
+                     hexoid(self.repo.lookup_reference(r).resolve()))
+                    for r in self.repo.listall_references()
+                    if r.startswith('refs/tags/'))
+
+class convert_git_plain(convert_git_base):
+    def gitread(self, s):
+        fh = self.gitopen(s)
+        data = fh.read()
+        return data, fh.close()
+
+    def getheads(self):
+        if not self.rev:
+            heads, ret = self.gitread('git rev-parse --branches --remotes')
+            heads = heads.splitlines()
+        else:
+            heads, ret = self.gitread("git rev-parse --verify %s" % self.rev)
+            heads = [heads[:-1]]
+        if ret:
+            raise util.Abort(_('cannot retrieve git heads'))
+        return heads
+
+    def catfile(self, rev, type):
+        if rev == hex(nullid):
+            raise IOError()
+        data, ret = self.gitread("git cat-file %s %s" % (type, rev))
+        if ret:
+            raise util.Abort(_('cannot read %r object at %s') % (type, rev))
+        return data
+
+    def getfile(self, name, rev):
+        data = self.catfile(rev, "blob")
+        mode = self.modecache[(name, rev)]
+        return data, mode
+
     def getcommit(self, version):
         c = self.catfile(version, "commit") # read the commit hash
         end = c.find("\n\n")
@@ -159,24 +232,6 @@
 
         return tags
 
-    def getchangedfiles(self, version, i):
-        changes = []
-        if i is None:
-            fh = self.gitopen("git diff-tree --root -m -r %s" % version)
-            for l in fh:
-                if "\t" not in l:
-                    continue
-                m, f = l[:-1].split("\t")
-                changes.append(f)
-        else:
-            fh = self.gitopen('git diff-tree --name-only --root -r %s "%s^%s" --'
-                             % (version, version, i + 1))
-            changes = [f.rstrip('\n') for f in fh]
-        if fh.close():
-            raise util.Abort(_('cannot read changes in %s') % version)
-
-        return changes
-
     def getbookmarks(self):
         bookmarks = {}
 
@@ -203,3 +258,9 @@
                 pass
 
         return bookmarks
+
+def convert_git(ui, path, rev=None):
+    try:
+        return convert_git_pygit2(ui, path, rev)
+    except ImportError:
+        return convert_git_plain(ui, path, rev)