dulwich/fastexport.py

   1 # __init__.py -- Fast export/import functionality
   2 # Copyright (C) 2010 Jelmer Vernooij <jelmer@samba.org>
   3 #
   4 # This program is free software; you can redistribute it and/or
   5 # modify it under the terms of the GNU General Public License
   6 # as published by the Free Software Foundation; version 2
   7 # of the License or (at your option) any later version of
   8 # the License.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License
  16 # along with this program; if not, write to the Free Software
  17 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
  18 # MA  02110-1301, USA.
  19
  20
  21 """Fast export/import functionality."""
  22
  23 from dulwich.index import (
  24     commit_tree,
  25     )
  26 from dulwich.objects import (
  27     Blob,
  28     Commit,
  29     format_timezone,
  30     parse_timezone,
  31     )
  32
  33 import stat
  34
  35 class FastExporter(object):
  36     """Generate a fast-export output stream for Git objects."""
  37
  38     def __init__(self, outf, store):
  39         self.outf = outf
  40         self.store = store
  41         self.markers = {}
  42         self._marker_idx = 0
  43
  44     def _allocate_marker(self):
  45         self._marker_idx+=1
  46         return self._marker_idx
  47
  48     def _dump_blob(self, blob, marker):
  49         self.outf.write("blob\nmark :%s\n" % marker)
  50         self.outf.write("data %s\n" % blob.raw_length())
  51         for chunk in blob.as_raw_chunks():
  52             self.outf.write(chunk)
  53         self.outf.write("\n")
  54
  55     def export_blob(self, blob):
  56         i = self._allocate_marker()
  57         self.markers[i] = blob.id
  58         self._dump_blob(blob, i)
  59         return i
  60
  61     def _dump_commit(self, commit, marker, ref, file_changes):
  62         self.outf.write("commit %s\n" % ref)
  63         self.outf.write("mark :%s\n" % marker)
  64         self.outf.write("author %s %s %s\n" % (commit.author,
  65             commit.author_time, format_timezone(commit.author_timezone)))
  66         self.outf.write("committer %s %s %s\n" % (commit.committer,
  67             commit.commit_time, format_timezone(commit.commit_timezone)))
  68         self.outf.write("data %s\n" % len(commit.message))
  69         self.outf.write(commit.message)
  70         self.outf.write("\n")
  71         self.outf.write('\n'.join(file_changes))
  72         self.outf.write("\n\n")
  73
  74     def export_commit(self, commit, ref, base_tree=None):
  75         file_changes = []
  76         for (old_path, new_path), (old_mode, new_mode), (old_hexsha, new_hexsha) in \
  77                 self.store.tree_changes(base_tree, commit.tree):
  78             if new_path is None:
  79                 file_changes.append("D %s" % old_path)
  80                 continue
  81             if not stat.S_ISDIR(new_mode):
  82                 marker = self.export_blob(self.store[new_hexsha])
  83             file_changes.append("M %o :%s %s" % (new_mode, marker, new_path))
  84
  85         i = self._allocate_marker()
  86         self._dump_commit(commit, i, ref, file_changes)
  87         return i
  88
  89
  90 class FastImporter(object):
  91     """Class for importing fastimport streams.
  92
  93     Please note that this is mostly a stub implementation at the moment,
  94     doing the bare mimimum.
  95     """
  96
  97     def __init__(self, repo):
  98         self.repo = repo
  99
 100     def _parse_person(self, line):
 101         (name, timestr, timezonestr) = line.rsplit(" ", 2)
 102         return name, int(timestr), parse_timezone(timezonestr)[0]
 103
 104     def _read_blob(self, stream):
 105         line = stream.readline()
 106         if line.startswith("mark :"):
 107             mark = line[len("mark :"):-1]
 108             line = stream.readline()
 109         else:
 110             mark = None
 111         if not line.startswith("data "):
 112             raise ValueError("Blob without valid data line: %s" % line)
 113         size = int(line[len("data "):])
 114         o = Blob()
 115         o.data = stream.read(size)
 116         stream.readline()
 117         self.repo.object_store.add_object(o)
 118         return mark, o.id
 119
 120     def _read_commit(self, stream, contents, marks):
 121         line = stream.readline()
 122         if line.startswith("mark :"):
 123             mark = line[len("mark :"):-1]
 124             line = stream.readline()
 125         else:
 126             mark = None
 127         o = Commit()
 128         o.author = None
 129         o.author_time = None
 130         while line.startswith("author "):
 131             (o.author, o.author_time, o.author_timezone) = \
 132                     self._parse_person(line[len("author "):-1])
 133             line = stream.readline()
 134         while line.startswith("committer "):
 135             (o.committer, o.commit_time, o.commit_timezone) = \
 136                     self._parse_person(line[len("committer "):-1])
 137             line = stream.readline()
 138         if o.author is None:
 139             o.author = o.committer
 140         if o.author_time is None:
 141             o.author_time = o.commit_time
 142             o.author_timezone = o.commit_timezone
 143         if not line.startswith("data "):
 144             raise ValueError("Blob without valid data line: %s" % line)
 145         size = int(line[len("data "):])
 146         o.message = stream.read(size)
 147         stream.readline()
 148         line = stream.readline()[:-1]
 149         while line:
 150             if line.startswith("M "):
 151                 (kind, modestr, val, path) = line.split(" ")
 152                 if val[0] == ":":
 153                     val = marks[int(val[1:])]
 154                 contents[path] = (int(modestr, 8), val)
 155             else:
 156                 raise ValueError(line)
 157             line = stream.readline()[:-1]
 158         try:
 159             o.parents = (self.repo.head(),)
 160         except KeyError:
 161             o.parents = ()
 162         o.tree = commit_tree(self.repo.object_store,
 163             ((path, hexsha, mode) for (path, (mode, hexsha)) in
 164                 contents.iteritems()))
 165         self.repo.object_store.add_object(o)
 166         return mark, o.id
 167
 168     def import_stream(self, stream):
 169         """Import from a file-like object.
 170
 171         :param stream: File-like object to read a fastimport stream from.
 172         :return: Dictionary with marks
 173         """
 174         contents = {}
 175         marks = {}
 176         while True:
 177             line = stream.readline()
 178             if not line:
 179                 break
 180             line = line[:-1]
 181             if line == "" or line[0] == "#":
 182                 continue
 183             if line.startswith("blob"):
 184                 mark, hexsha = self._read_blob(stream)
 185                 if mark is not None:
 186                     marks[int(mark)] = hexsha
 187             elif line.startswith("commit "):
 188                 ref = line[len("commit "):-1]
 189                 mark, hexsha = self._read_commit(stream, contents, marks)
 190                 if mark is not None:
 191                     marks[int(mark)] = hexsha
 192                 self.repo.refs["HEAD"] = self.repo.refs[ref] = hexsha
 193             else:
 194                 raise ValueError("invalid command '%s'" % line)
 195         return marks