From 27bf2140b31da95e6e12d9e46b68362effec134a Mon Sep 17 00:00:00 2001
From: Siddharth Agarwal <sid0@fb.com>
Date: Tue, 14 Oct 2014 15:54:58 -0700
Subject: [PATCH] git_handler: move get_git_incoming to a separate module

git_handler's getting far too big, and we're going to add some more git-to-hg
specific logic soon, so now seems like a good time to add a new module.
---
 hggit/git2hg.py      | 66 ++++++++++++++++++++++++++++++++++++++++++++
 hggit/git_handler.py | 59 ++-------------------------------------
 2 files changed, 68 insertions(+), 57 deletions(-)
 create mode 100644 hggit/git2hg.py

diff --git a/hggit/git2hg.py b/hggit/git2hg.py
new file mode 100644
index 0000000000..edba5d6276
--- /dev/null
+++ b/hggit/git2hg.py
@@ -0,0 +1,66 @@
+# git2hg.py - convert Git repositories and commits to Mercurial ones
+
+from dulwich.objects import Commit, Tag
+
+def find_incoming(git_object_store, git_map, refs):
+    '''find what commits need to be imported
+
+    git_object_store is a dulwich object store.
+    git_map is a map with keys being Git commits that have already been imported
+    refs is a map of refs to SHAs that we're interested in.'''
+
+    # import heads and fetched tags as remote references
+    todo = []
+    done = set()
+    commit_cache = {}
+
+    # get a list of all the head shas
+    seenheads = set()
+    if refs:
+        for sha in refs.itervalues():
+            # refs contains all the refs in the server, not just the ones
+            # we are pulling
+            if sha in git_object_store:
+                obj = git_object_store[sha]
+                while isinstance(obj, Tag):
+                    obj_type, sha = obj.object
+                    obj = git_object_store[sha]
+                if isinstance (obj, Commit) and sha not in seenheads:
+                    seenheads.add(sha)
+                    todo.append(sha)
+
+    # sort by commit date
+    def commitdate(sha):
+        obj = git_object_store[sha]
+        return obj.commit_time-obj.commit_timezone
+
+    todo.sort(key=commitdate, reverse=True)
+
+    # traverse the heads getting a list of all the unique commits in
+    # topological order
+    commits = []
+    seen = set(todo)
+    while todo:
+        sha = todo[-1]
+        if sha in done or sha in git_map:
+            todo.pop()
+            continue
+        assert isinstance(sha, str)
+        if sha in commit_cache:
+            obj = commit_cache[sha]
+        else:
+            obj = git_object_store[sha]
+            commit_cache[sha] = obj
+        assert isinstance(obj, Commit)
+        for p in obj.parents:
+            if p not in done and p not in git_map:
+                todo.append(p)
+                # process parents of a commit before processing the
+                # commit itself, and come back to this commit later
+                break
+        else:
+            commits.append(sha)
+            done.add(sha)
+            todo.pop()
+
+    return commit_cache, commits
diff --git a/hggit/git_handler.py b/hggit/git_handler.py
index 55c4962497..3fc2105af3 100644
--- a/hggit/git_handler.py
+++ b/hggit/git_handler.py
@@ -26,6 +26,7 @@ from mercurial import error
 from mercurial import url
 
 import _ssh
+import git2hg
 import hg2git
 import util
 from overlay import overlayrepo
@@ -663,63 +664,7 @@ class GitHandler(object):
     def get_git_incoming(self, refs):
         if refs is None:
             refs = self.git.refs.as_dict()
-        git_object_store = self.git.object_store
-
-        # import heads and fetched tags as remote references
-        todo = []
-        done = set()
-        commit_cache = {}
-
-        # get a list of all the head shas
-        seenheads = set()
-        if refs:
-            for sha in refs.itervalues():
-                # refs contains all the refs in the server, not just the ones
-                # we are pulling
-                if sha in git_object_store:
-                    obj = git_object_store[sha]
-                    while isinstance(obj, Tag):
-                        obj_type, sha = obj.object
-                        obj = git_object_store[sha]
-                    if isinstance (obj, Commit) and sha not in seenheads:
-                        seenheads.add(sha)
-                        todo.append(sha)
-
-        # sort by commit date
-        def commitdate(sha):
-            obj = git_object_store[sha]
-            return obj.commit_time-obj.commit_timezone
-
-        todo.sort(key=commitdate, reverse=True)
-
-        # traverse the heads getting a list of all the unique commits in
-        # topological order
-        commits = []
-        seen = set(todo)
-        while todo:
-            sha = todo[-1]
-            if sha in done or sha in self._map_git:
-                todo.pop()
-                continue
-            assert isinstance(sha, str)
-            if sha in commit_cache:
-                obj = commit_cache[sha]
-            else:
-                obj = git_object_store[sha]
-                commit_cache[sha] = obj
-            assert isinstance(obj, Commit)
-            for p in obj.parents:
-                if p not in done and p not in self._map_git:
-                    todo.append(p)
-                    # process parents of a commit before processing the
-                    # commit itself, and come back to this commit later
-                    break
-            else:
-                commits.append(sha)
-                done.add(sha)
-                todo.pop()
-
-        return commit_cache, commits
+        return git2hg.find_incoming(self.git.object_store, self._map_git, refs)
 
     def import_git_objects(self, remote_name=None, refs=None):
         commit_cache, commits = self.get_git_incoming(refs)