Mercurial > hg-git
annotate hggit/hg2git.py @ 707:d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Upcoming patches will start incrementally exporting from a particular commit
instead of from null. This function will be used for that..
author | Siddharth Agarwal <sid0@fb.com> |
---|---|
date | Fri, 14 Mar 2014 19:17:09 -0700 |
parents | fbfa6353d96c |
children | 5c7943ca051f |
rev | line source |
---|---|
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
1 # This file contains code dealing specifically with converting Mercurial |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
2 # repositories to Git repositories. Code in this file is meant to be a generic |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
3 # library and should be usable outside the context of hg-git or an hg command. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
4 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
5 import os |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
6 import stat |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
7 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
8 import dulwich.objects as dulobjs |
707
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
9 from dulwich import diff_tree |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
10 import mercurial.node |
637
23d7caeed05a
hg2git: store ctx instead of rev
Siddharth Agarwal <sid0@fb.com>
parents:
636
diff
changeset
|
11 import mercurial.context |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
12 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
13 import util |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
14 |
671
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
15 def parse_subrepos(ctx): |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
16 sub = util.OrderedDict() |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
17 if '.hgsub' in ctx: |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
18 sub = util.parse_hgsub(ctx['.hgsub'].data().splitlines()) |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
19 substate = util.OrderedDict() |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
20 if '.hgsubstate' in ctx: |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
21 substate = util.parse_hgsubstate( |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
22 ctx['.hgsubstate'].data().splitlines()) |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
23 return sub, substate |
71fb5dd678bc
hg2git: move parse_subrepos to top level
Siddharth Agarwal <sid0@fb.com>
parents:
649
diff
changeset
|
24 |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
25 class IncrementalChangesetExporter(object): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
26 """Incrementally export Mercurial changesets to Git trees. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
27 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
28 The purpose of this class is to facilitate Git tree export that is more |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
29 optimal than brute force. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
30 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
31 A "dumb" implementations of Mercurial to Git export would iterate over |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
32 every file present in a Mercurial changeset and would convert each to |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
33 a Git blob and then conditionally add it to a Git repository if it didn't |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
34 yet exist. This is suboptimal because the overhead associated with |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
35 obtaining every file's raw content and converting it to a Git blob is |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
36 not trivial! |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
37 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
38 This class works around the suboptimality of brute force export by |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
39 leveraging the information stored in Mercurial - the knowledge of what |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
40 changed between changesets - to only export Git objects corresponding to |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
41 changes in Mercurial. In the context of converting Mercurial repositories |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
42 to Git repositories, we only export objects Git (possibly) hasn't seen yet. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
43 This prevents a lot of redundant work and is thus faster. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
44 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
45 Callers instantiate an instance of this class against a mercurial.localrepo |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
46 instance. They then associate it with a specific changesets by calling |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
47 update_changeset(). On each call to update_changeset(), the instance |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
48 computes the difference between the current and new changesets and emits |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
49 Git objects that haven't yet been encountered during the lifetime of the |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
50 class instance. In other words, it expresses Mercurial changeset deltas in |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
51 terms of Git objects. Callers then (usually) take this set of Git objects |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
52 and add them to the Git repository. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
53 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
54 This class only emits Git blobs and trees, not commits. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
55 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
56 The tree calculation part of this class is essentially a reimplementation |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
57 of dulwich.index.commit_tree. However, since our implementation reuses |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
58 Tree instances and only recalculates SHA-1 when things change, we are |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
59 more efficient. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
60 """ |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
61 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
62 def __init__(self, hg_repo): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
63 """Create an instance against a mercurial.localrepo.""" |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
64 self._hg = hg_repo |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
65 |
637
23d7caeed05a
hg2git: store ctx instead of rev
Siddharth Agarwal <sid0@fb.com>
parents:
636
diff
changeset
|
66 # Our current revision's context. |
23d7caeed05a
hg2git: store ctx instead of rev
Siddharth Agarwal <sid0@fb.com>
parents:
636
diff
changeset
|
67 self._ctx = mercurial.context.changectx(hg_repo, 'null') |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
68 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
69 # Path to dulwich.objects.Tree. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
70 self._dirs = {} |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
71 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
72 # Mercurial file nodeid to Git blob SHA-1. Used to prevent redundant |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
73 # blob calculation. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
74 self._blob_cache = {} |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
75 |
707
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
76 def _init_dirs(self, store, commit): |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
77 """Initialize self._dirs for a Git object store and commit.""" |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
78 self._dirs = {} |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
79 if commit is None: |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
80 return |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
81 dirkind = stat.S_IFDIR |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
82 # depth-first order, chosen arbitrarily |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
83 todo = [('', store[commit.tree])] |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
84 while todo: |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
85 path, tree = todo.pop() |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
86 self._dirs[path] = tree |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
87 for entry in tree.iteritems(): |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
88 if entry.mode == dirkind: |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
89 todo.append((path + '/' + entry.path, store[entry.sha])) |
d5facc1be5f8
hg2git: implement a method to initialize _dirs from a Git commit
Siddharth Agarwal <sid0@fb.com>
parents:
672
diff
changeset
|
90 |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
91 @property |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
92 def root_tree_sha(self): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
93 """The SHA-1 of the root Git tree. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
94 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
95 This is needed to construct a Git commit object. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
96 """ |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
97 return self._dirs[''].id |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
98 |
636
0ab89bd32c8e
hg2git: rename ctx to newctx in update_changeset
Siddharth Agarwal <sid0@fb.com>
parents:
598
diff
changeset
|
99 def update_changeset(self, newctx): |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
100 """Set the tree to track a new Mercurial changeset. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
101 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
102 This is a generator of 2-tuples. The first item in each tuple is a |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
103 dulwich object, either a Blob or a Tree. The second item is the |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
104 corresponding Mercurial nodeid for the item, if any. Only blobs will |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
105 have nodeids. Trees do not correspond to a specific nodeid, so it does |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
106 not make sense to emit a nodeid for them. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
107 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
108 When exporting trees from Mercurial, callers typically write the |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
109 returned dulwich object to the Git repo via the store's add_object(). |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
110 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
111 Some emitted objects may already exist in the Git repository. This |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
112 class does not know about the Git repository, so it's up to the caller |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
113 to conditionally add the object, etc. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
114 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
115 Emitted objects are those that have changed since the last call to |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
116 update_changeset. If this is the first call to update_chanageset, all |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
117 objects in the tree are emitted. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
118 """ |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
119 # Our general strategy is to accumulate dulwich.objects.Blob and |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
120 # dulwich.objects.Tree instances for the current Mercurial changeset. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
121 # We do this incremental by iterating over the Mercurial-reported |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
122 # changeset delta. We rely on the behavior of Mercurial to lazy |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
123 # calculate a Tree's SHA-1 when we modify it. This is critical to |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
124 # performance. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
125 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
126 # In theory we should be able to look at changectx.files(). This is |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
127 # *much* faster. However, it may not be accurate, especially with older |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
128 # repositories, which may not record things like deleted files |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
129 # explicitly in the manifest (which is where files() gets its data). |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
130 # The only reliable way to get the full set of changes is by looking at |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
131 # the full manifest. And, the easy way to compare two manifests is |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
132 # localrepo.status(). |
638
f828d82c35dc
hg2git: call status on newctx, not newctx.rev()
Siddharth Agarwal <sid0@fb.com>
parents:
637
diff
changeset
|
133 modified, added, removed = self._hg.status(self._ctx, newctx)[0:3] |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
134 |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
135 # We track which directories/trees have modified in this update and we |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
136 # only export those. |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
137 dirty_trees = set() |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
138 |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
139 subadded, subremoved = [], [] |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
140 |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
141 for s in modified, added, removed: |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
142 if '.hgsub' in s or '.hgsubstate' in s: |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
143 subadded, subremoved = self._handle_subrepos(newctx) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
144 break |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
145 |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
146 # We first process subrepo and file removals so we can prune dead trees. |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
147 for path in subremoved: |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
148 self._remove_path(path, dirty_trees) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
149 |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
150 for path in removed: |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
151 if path == '.hgsubstate' or path == '.hgsub': |
649
53423381c540
hg2git: call _handle_subrepos when .hgsubstate is removed
Siddharth Agarwal <sid0@fb.com>
parents:
648
diff
changeset
|
152 continue |
53423381c540
hg2git: call _handle_subrepos when .hgsubstate is removed
Siddharth Agarwal <sid0@fb.com>
parents:
648
diff
changeset
|
153 |
645
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
154 self._remove_path(path, dirty_trees) |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
155 |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
156 for path, sha in subadded: |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
157 d = os.path.dirname(path) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
158 tree = self._dirs.setdefault(d, dulobjs.Tree()) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
159 dirty_trees.add(d) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
160 tree.add(os.path.basename(path), dulobjs.S_IFGITLINK, sha) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
161 |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
162 # For every file that changed or was added, we need to calculate the |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
163 # corresponding Git blob and its tree entry. We emit the blob |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
164 # immediately and update trees to be aware of its presence. |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
165 for path in set(modified) | set(added): |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
166 if path == '.hgsubstate' or path == '.hgsub': |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
167 continue |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
168 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
169 d = os.path.dirname(path) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
170 tree = self._dirs.setdefault(d, dulobjs.Tree()) |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
171 dirty_trees.add(d) |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
172 |
636
0ab89bd32c8e
hg2git: rename ctx to newctx in update_changeset
Siddharth Agarwal <sid0@fb.com>
parents:
598
diff
changeset
|
173 fctx = newctx[path] |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
174 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
175 entry, blob = IncrementalChangesetExporter.tree_entry(fctx, |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
176 self._blob_cache) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
177 if blob is not None: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
178 yield (blob, fctx.filenode()) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
179 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
180 tree.add(*entry) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
181 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
182 # Now that all the trees represent the current changeset, recalculate |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
183 # the tree IDs and emit them. Note that we wait until now to calculate |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
184 # tree SHA-1s. This is an important difference between us and |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
185 # dulwich.index.commit_tree(), which builds new Tree instances for each |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
186 # series of blobs. |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
187 for obj in self._populate_tree_entries(dirty_trees): |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
188 yield (obj, None) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
189 |
637
23d7caeed05a
hg2git: store ctx instead of rev
Siddharth Agarwal <sid0@fb.com>
parents:
636
diff
changeset
|
190 self._ctx = newctx |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
191 |
645
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
192 def _remove_path(self, path, dirty_trees): |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
193 """Remove a path (file or git link) from the current changeset. |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
194 |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
195 If the tree containing this path is empty, it might be removed.""" |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
196 d = os.path.dirname(path) |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
197 tree = self._dirs.get(d, dulobjs.Tree()) |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
198 |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
199 del tree[os.path.basename(path)] |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
200 dirty_trees.add(d) |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
201 |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
202 # If removing this file made the tree empty, we should delete this |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
203 # tree. This could result in parent trees losing their only child |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
204 # and so on. |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
205 if not len(tree): |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
206 self._remove_tree(d) |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
207 else: |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
208 self._dirs[d] = tree |
104f536be5c7
hg2git: factor out remove path logic into a separate function
Siddharth Agarwal <sid0@fb.com>
parents:
638
diff
changeset
|
209 |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
210 def _remove_tree(self, path): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
211 """Remove a (presumably empty) tree from the current changeset. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
212 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
213 A now-empty tree may be the only child of its parent. So, we traverse |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
214 up the chain to the root tree, deleting any empty trees along the way. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
215 """ |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
216 try: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
217 del self._dirs[path] |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
218 except KeyError: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
219 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
220 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
221 # Now we traverse up to the parent and delete any references. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
222 if path == '': |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
223 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
224 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
225 basename = os.path.basename(path) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
226 parent = os.path.dirname(path) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
227 while True: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
228 tree = self._dirs.get(parent, None) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
229 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
230 # No parent entry. Nothing to remove or update. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
231 if tree is None: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
232 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
233 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
234 try: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
235 del tree[basename] |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
236 except KeyError: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
237 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
238 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
239 if len(tree): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
240 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
241 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
242 # The parent tree is empty. Se, we can delete it. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
243 del self._dirs[parent] |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
244 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
245 if parent == '': |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
246 return |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
247 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
248 basename = os.path.basename(parent) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
249 parent = os.path.dirname(parent) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
250 |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
251 def _populate_tree_entries(self, dirty_trees): |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
252 self._dirs.setdefault('', dulobjs.Tree()) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
253 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
254 # Fill in missing directories. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
255 for path in self._dirs.keys(): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
256 parent = os.path.dirname(path) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
257 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
258 while parent != '': |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
259 parent_tree = self._dirs.get(parent, None) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
260 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
261 if parent_tree is not None: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
262 break |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
263 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
264 self._dirs[parent] = dulobjs.Tree() |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
265 parent = os.path.dirname(parent) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
266 |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
267 for dirty in list(dirty_trees): |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
268 parent = os.path.dirname(dirty) |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
269 |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
270 while parent != '': |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
271 if parent in dirty_trees: |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
272 break |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
273 |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
274 dirty_trees.add(parent) |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
275 parent = os.path.dirname(parent) |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
276 |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
277 # The root tree is always dirty but doesn't always get updated. |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
278 dirty_trees.add('') |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
279 |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
280 # We only need to recalculate and export dirty trees. |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
281 for d in sorted(dirty_trees, key=len, reverse=True): |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
282 # Only happens for deleted directories. |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
283 try: |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
284 tree = self._dirs[d] |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
285 except KeyError: |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
286 continue |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
287 |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
288 yield tree |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
289 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
290 if d == '': |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
291 continue |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
292 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
293 parent_tree = self._dirs[os.path.dirname(d)] |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
294 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
295 # Accessing the tree's ID is what triggers SHA-1 calculation and is |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
296 # the expensive part (at least if the tree has been modified since |
598
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
297 # the last time we retrieved its ID). Also, assigning an entry to a |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
298 # tree (even if it already exists) invalidates the existing tree |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
299 # and incurs SHA-1 recalculation. So, it's in our interest to avoid |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
300 # invalidating trees. Since we only update the entries of dirty |
792955be68dd
Only export modified Git trees
Gregory Szorc <gregory.szorc@gmail.com>
parents:
596
diff
changeset
|
301 # trees, this should hold true. |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
302 parent_tree[os.path.basename(d)] = (stat.S_IFDIR, tree.id) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
303 |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
304 def _handle_subrepos(self, newctx): |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
305 sub, substate = parse_subrepos(self._ctx) |
647
3ceacdd23abe
hg2git: add 'new' prefix to _handle_subrepos variables
Siddharth Agarwal <sid0@fb.com>
parents:
646
diff
changeset
|
306 newsub, newsubstate = parse_subrepos(newctx) |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
307 |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
308 # For each path, the logic is described by the following table. 'no' |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
309 # stands for 'the subrepo doesn't exist', 'git' stands for 'git |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
310 # subrepo', and 'hg' stands for 'hg or other subrepo'. |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
311 # |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
312 # old new | action |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
313 # * git | link (1) |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
314 # git hg | delete (2) |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
315 # git no | delete (3) |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
316 # |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
317 # All other combinations are 'do nothing'. |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
318 # |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
319 # git links without corresponding submodule paths are stored as subrepos |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
320 # with a substate but without an entry in .hgsub. |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
321 |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
322 # 'added' is both modified and added |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
323 added, removed = [], [] |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
324 |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
325 def isgit(sub, path): |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
326 return path not in sub or sub[path].startswith('[git]') |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
327 |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
328 for path, sha in substate.iteritems(): |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
329 if not isgit(sub, path): |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
330 # old = hg -- will be handled in next loop |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
331 continue |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
332 # old = git |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
333 if path not in newsubstate or not isgit(newsub, path): |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
334 # new = hg or no, case (2) or (3) |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
335 removed.append(path) |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
336 |
647
3ceacdd23abe
hg2git: add 'new' prefix to _handle_subrepos variables
Siddharth Agarwal <sid0@fb.com>
parents:
646
diff
changeset
|
337 for path, sha in newsubstate.iteritems(): |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
338 if not isgit(newsub, path): |
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
339 # new = hg or no; the only cases we care about are handled above |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
340 continue |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
341 |
648
bd63cdfbc1de
hg2git: make _handle_subrepos worked in the removed case
Siddharth Agarwal <sid0@fb.com>
parents:
647
diff
changeset
|
342 # case (1) |
672
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
343 added.append((path, sha)) |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
344 |
fbfa6353d96c
hg2git: fix subrepo handling to be deterministic
Siddharth Agarwal <sid0@fb.com>
parents:
671
diff
changeset
|
345 return added, removed |
596
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
346 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
347 @staticmethod |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
348 def tree_entry(fctx, blob_cache): |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
349 """Compute a dulwich TreeEntry from a filectx. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
350 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
351 A side effect is the TreeEntry is stored in the passed cache. |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
352 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
353 Returns a 2-tuple of (dulwich.objects.TreeEntry, dulwich.objects.Blob). |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
354 """ |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
355 blob_id = blob_cache.get(fctx.filenode(), None) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
356 blob = None |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
357 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
358 if blob_id is None: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
359 blob = dulobjs.Blob.from_string(fctx.data()) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
360 blob_id = blob.id |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
361 blob_cache[fctx.filenode()] = blob_id |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
362 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
363 flags = fctx.flags() |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
364 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
365 if 'l' in flags: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
366 mode = 0120000 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
367 elif 'x' in flags: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
368 mode = 0100755 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
369 else: |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
370 mode = 0100644 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
371 |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
372 return (dulobjs.TreeEntry(os.path.basename(fctx.path()), mode, blob_id), |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
373 blob) |
d6b9c30a3e0f
Export Git objects from incremental Mercurial changes
Gregory Szorc <gregory.szorc@gmail.com>
parents:
diff
changeset
|
374 |