hgext/convert/git.py
author Matt Mackall <mpm@selenic.com>
Sat, 23 Jun 2007 13:33:45 -0500
changeset 4687 06a0e0557edc
parent 4534 cc9b79216a76
child 4721 96614af3c679
permissions -rw-r--r--
convert: sometimes git forgets the author
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
4534
cc9b79216a76 Split convert extension into common and repository type modules
Brendan Cully <brendan@kublai.com>
parents: 4532
diff changeset
     1
# git support for the convert extension
3825
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
     2
4534
cc9b79216a76 Split convert extension into common and repository type modules
Brendan Cully <brendan@kublai.com>
parents: 4532
diff changeset
     3
import os
3947
0fab73b3f453 convert-repo: add some smarts
Matt Mackall <mpm@selenic.com>
parents: 3916
diff changeset
     4
4534
cc9b79216a76 Split convert extension into common and repository type modules
Brendan Cully <brendan@kublai.com>
parents: 4532
diff changeset
     5
from common import NoRepo, commit, converter_source
3955
9af4b853ed4d convert-repo: add CVS branch support
Matt Mackall <mpm@selenic.com>
parents: 3954
diff changeset
     6
3825
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
     7
def recode(s):
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
     8
    try:
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
     9
        return s.decode("utf-8").encode("utf-8")
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    10
    except:
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    11
        try:
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    12
            return s.decode("latin-1").encode("utf-8")
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    13
        except:
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    14
            return s.decode("utf-8", "replace").encode("utf-8")
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    15
4447
af013ae3ca10 use documented convert-repo interface
Daniel Holth <dholth@fastmail.fm>
parents: 4446
diff changeset
    16
class convert_git(converter_source):
4513
ac2fe196ac9b Turns convert.py into a real extension
Edouard Gomez <ed.gomez@free.fr>
parents: 4512
diff changeset
    17
    def __init__(self, ui, path):
3947
0fab73b3f453 convert-repo: add some smarts
Matt Mackall <mpm@selenic.com>
parents: 3916
diff changeset
    18
        if os.path.isdir(path + "/.git"):
0fab73b3f453 convert-repo: add some smarts
Matt Mackall <mpm@selenic.com>
parents: 3916
diff changeset
    19
            path += "/.git"
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    20
        self.path = path
4513
ac2fe196ac9b Turns convert.py into a real extension
Edouard Gomez <ed.gomez@free.fr>
parents: 4512
diff changeset
    21
        self.ui = ui
4111
06d65498f73b convert-repo: use .git/objects/ rather than .git/HEAD to detect git repos
Matt Mackall <mpm@selenic.com>
parents: 4082
diff changeset
    22
        if not os.path.exists(path + "/objects"):
3954
fad134931327 convert-repo: add basic CVS import support
Matt Mackall <mpm@selenic.com>
parents: 3948
diff changeset
    23
            raise NoRepo("couldn't open GIT repo %s" % path)
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    24
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    25
    def getheads(self):
2649
e6a7a6a33a62 make convert-repo deal with git symbolic refs.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2093
diff changeset
    26
        fh = os.popen("GIT_DIR=%s git-rev-parse --verify HEAD" % self.path)
e6a7a6a33a62 make convert-repo deal with git symbolic refs.
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 2093
diff changeset
    27
        return [fh.read()[:-1]]
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    28
692
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    29
    def catfile(self, rev, type):
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    30
        if rev == "0" * 40: raise IOError()
4532
c3a78a49d7f0 Some small cleanups for convert extension:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4521
diff changeset
    31
        fh = os.popen("GIT_DIR=%s git-cat-file %s %s 2>/dev/null"
c3a78a49d7f0 Some small cleanups for convert extension:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4521
diff changeset
    32
                      % (self.path, type, rev))
692
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    33
        return fh.read()
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    34
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    35
    def getfile(self, name, rev):
692
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    36
        return self.catfile(rev, "blob")
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    37
3957
558f52943cd2 convert-repo: add CVS exec bit support
Matt Mackall <mpm@selenic.com>
parents: 3955
diff changeset
    38
    def getmode(self, name, rev):
558f52943cd2 convert-repo: add CVS exec bit support
Matt Mackall <mpm@selenic.com>
parents: 3955
diff changeset
    39
        return self.modecache[(name, rev)]
558f52943cd2 convert-repo: add CVS exec bit support
Matt Mackall <mpm@selenic.com>
parents: 3955
diff changeset
    40
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    41
    def getchanges(self, version):
3957
558f52943cd2 convert-repo: add CVS exec bit support
Matt Mackall <mpm@selenic.com>
parents: 3955
diff changeset
    42
        self.modecache = {}
4532
c3a78a49d7f0 Some small cleanups for convert extension:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4521
diff changeset
    43
        fh = os.popen("GIT_DIR=%s git-diff-tree --root -m -r %s"
c3a78a49d7f0 Some small cleanups for convert extension:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4521
diff changeset
    44
                      % (self.path, version))
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    45
        changes = []
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    46
        for l in fh:
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    47
            if "\t" not in l: continue
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    48
            m, f = l[:-1].split("\t")
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    49
            m = m.split()
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    50
            h = m[3]
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    51
            p = (m[1] == "100755")
4082
6b2909e84203 convert-repo converts symlinks from git
Daniel Holth <dholth@fastmail.fm>
parents: 4062
diff changeset
    52
            s = (m[1] == "120000")
6b2909e84203 convert-repo converts symlinks from git
Daniel Holth <dholth@fastmail.fm>
parents: 4062
diff changeset
    53
            self.modecache[(f, h)] = (p and "x") or (s and "l") or ""
3957
558f52943cd2 convert-repo: add CVS exec bit support
Matt Mackall <mpm@selenic.com>
parents: 3955
diff changeset
    54
            changes.append((f, h))
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    55
        return changes
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    56
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    57
    def getcommit(self, version):
692
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    58
        c = self.catfile(version, "commit") # read the commit hash
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    59
        end = c.find("\n\n")
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    60
        message = c[end+2:]
3825
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    61
        message = recode(message)
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    62
        l = c[:end].splitlines()
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    63
        manifest = l[0].split()[1]
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    64
        parents = []
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    65
        for e in l[1:]:
4532
c3a78a49d7f0 Some small cleanups for convert extension:
Thomas Arendsen Hein <thomas@intevation.de>
parents: 4521
diff changeset
    66
            n, v = e.split(" ", 1)
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    67
            if n == "author":
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    68
                p = v.split()
1385
adb3de56635b convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents: 1335
diff changeset
    69
                tm, tz = p[-2:]
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    70
                author = " ".join(p[:-2])
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    71
                if author[0] == "<": author = author[1:-1]
3825
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    72
                author = recode(author)
692
695dd9a491da convert-repo: deal with packed git and other fixes
mpm@selenic.com
parents: 450
diff changeset
    73
            if n == "committer":
431
dfc44f3f587c convert-repo fixups
mpm@selenic.com
parents: 316
diff changeset
    74
                p = v.split()
1385
adb3de56635b convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents: 1335
diff changeset
    75
                tm, tz = p[-2:]
431
dfc44f3f587c convert-repo fixups
mpm@selenic.com
parents: 316
diff changeset
    76
                committer = " ".join(p[:-2])
dfc44f3f587c convert-repo fixups
mpm@selenic.com
parents: 316
diff changeset
    77
                if committer[0] == "<": committer = committer[1:-1]
3825
158fce02dc40 Teach convert-repo to deal with mixed charsets in git
Matt Mackall <mpm@selenic.com>
parents: 2649
diff changeset
    78
                committer = recode(committer)
3910
4bc5a2405b12 convert-repo: fix recoding of committer
Matt Mackall <mpm@selenic.com>
parents: 3825
diff changeset
    79
                message += "\ncommitter: %s\n" % committer
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    80
            if n == "parent": parents.append(v)
1385
adb3de56635b convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents: 1335
diff changeset
    81
adb3de56635b convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents: 1335
diff changeset
    82
        tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:]
2093
5cc414722587 convert-repo: fix reversed time zone offset
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1715
diff changeset
    83
        tz = -int(tzs) * (int(tzh) * 3600 + int(tzm))
1385
adb3de56635b convert-repo: Fix timezone handling
Matt Mackall <mpm@selenic.com>
parents: 1335
diff changeset
    84
        date = tm + " " + str(tz)
4687
06a0e0557edc convert: sometimes git forgets the author
Matt Mackall <mpm@selenic.com>
parents: 4534
diff changeset
    85
        author = author or unknown
3955
9af4b853ed4d convert-repo: add CVS branch support
Matt Mackall <mpm@selenic.com>
parents: 3954
diff changeset
    86
9af4b853ed4d convert-repo: add CVS branch support
Matt Mackall <mpm@selenic.com>
parents: 3954
diff changeset
    87
        c = commit(parents=parents, date=date, author=author, desc=message)
9af4b853ed4d convert-repo: add CVS branch support
Matt Mackall <mpm@selenic.com>
parents: 3954
diff changeset
    88
        return c
316
c48d069163d6 Add new convert-repo script
mpm@selenic.com
parents:
diff changeset
    89
694
51eb248d3348 Teach convert-repo about tags
mpm@selenic.com
parents: 692
diff changeset
    90
    def gettags(self):
51eb248d3348 Teach convert-repo about tags
mpm@selenic.com
parents: 692
diff changeset
    91
        tags = {}
4062
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    92
        fh = os.popen('git-ls-remote --tags "%s" 2>/dev/null' % self.path)
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    93
        prefix = 'refs/tags/'
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    94
        for line in fh:
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    95
            line = line.strip()
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    96
            if not line.endswith("^{}"):
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    97
                continue
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    98
            node, tag = line.split(None, 1)
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
    99
            if not tag.startswith(prefix):
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
   100
                continue
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
   101
            tag = tag[len(prefix):-3]
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
   102
            tags[tag] = node
516f883e3d79 convert-repo: handle packed git tags
Alexis S. L. Carvalho <alexis@cecm.usp.br>
parents: 4047
diff changeset
   103
694
51eb248d3348 Teach convert-repo about tags
mpm@selenic.com
parents: 692
diff changeset
   104
        return tags