git/packwriter: open(..) prohibited in __del__

[bup.git] / lib / bup / git.py

diff --git a/lib/bup/git.py b/lib/bup/git.py
index db0188c5a299784cbef7e3345c715ce0601b0903..4c1a95c52a985e5eaee1bb36ef4ca61651559c78 100644 (file)
--- a/lib/bup/git.py
+++ b/lib/bup/git.py
@@ -4,33 +4,32 @@ interact with the Git data structures.
 """
 
 from __future__ import absolute_import, print_function
 """
 
 from __future__ import absolute_import, print_function

-import errno, os, sys, zlib, time, subprocess, struct, stat, re, tempfile, glob
+import os, sys, zlib, subprocess, struct, stat, re, tempfile, glob

 from array import array
 from binascii import hexlify, unhexlify
 from collections import namedtuple
 from itertools import islice
 from array import array
 from binascii import hexlify, unhexlify
 from collections import namedtuple
 from itertools import islice

-from numbers import Integral

 
 

-from bup import _helpers, compat, hashsplit, path, midx, bloom, xstat
+from bup import _helpers, hashsplit, path, midx, bloom, xstat

 from bup.compat import (buffer,
                         byte_int, bytes_from_byte, bytes_from_uint,
                         environ,
                         items,
 from bup.compat import (buffer,
                         byte_int, bytes_from_byte, bytes_from_uint,
                         environ,
                         items,

+                        pending_raise,

                         range,
                         reraise)
 from bup.io import path_msg
 from bup.helpers import (Sha1, add_error, chunkyreader, debug1, debug2,
                          exo,
                          fdatasync,
                         range,
                         reraise)
 from bup.io import path_msg
 from bup.helpers import (Sha1, add_error, chunkyreader, debug1, debug2,
                          exo,
                          fdatasync,

-                         hostname, localtime, log,
+                         finalized,
+                         log,

                          merge_dict,
                          merge_iter,
                          mmap_read, mmap_readwrite,
                          merge_dict,
                          merge_iter,
                          mmap_read, mmap_readwrite,

-                         parse_num,

                          progress, qprogress, stat_if_exists,
                          unlink,
                          utc_offset_str)
                          progress, qprogress, stat_if_exists,
                          unlink,
                          utc_offset_str)

-from bup.pwdgrp import username, userfullname

 
 
 verbose = 0
 
 
 verbose = 0


@@ -66,14 +65,32 @@ def _git_exo(cmd, **kwargs):
         raise GitError('%r returned %d' % (cmd, proc.returncode))
     return result
 
         raise GitError('%r returned %d' % (cmd, proc.returncode))
     return result
 

-def git_config_get(option, repo_dir=None):
-    cmd = (b'git', b'config', b'--get', option)
-    p = subprocess.Popen(cmd, stdout=subprocess.PIPE,
-                         env=_gitenv(repo_dir=repo_dir),
+def git_config_get(option, repo_dir=None, opttype=None, cfg_file=None):
+    assert not (repo_dir and cfg_file), "repo_dir and cfg_file cannot both be used"
+    cmd = [b'git', b'config', b'--null']
+    if cfg_file:
+        cmd.extend([b'--file', cfg_file])
+    if opttype == 'int':
+        cmd.extend([b'--int'])
+    elif opttype == 'bool':
+        cmd.extend([b'--bool'])
+    else:
+        assert opttype is None
+    cmd.extend([b'--get', option])
+    env=None
+    if repo_dir:
+        env = _gitenv(repo_dir=repo_dir)
+    p = subprocess.Popen(cmd, stdout=subprocess.PIPE, env=env,

                          close_fds=True)
                          close_fds=True)

-    r = p.stdout.read()
+    # with --null, git writes out a trailing \0 after the value
+    r = p.stdout.read()[:-1]

     rc = p.wait()
     if rc == 0:
     rc = p.wait()
     if rc == 0:

+        if opttype == 'int':
+            return int(r)
+        elif opttype == 'bool':
+            # git converts to 'true' or 'false'
+            return r == b'true'

         return r
     if rc != 1:
         raise GitError('%r returned %d' % (cmd, rc))
         return r
     if rc != 1:
         raise GitError('%r returned %d' % (cmd, rc))


@@ -87,9 +104,26 @@ def parse_tz_offset(s):
         return - tz_off
     return tz_off
 
         return - tz_off
     return tz_off
 

+def parse_commit_gpgsig(sig):
+    """Return the original signature bytes.
+
+    i.e. with the "gpgsig " header and the leading space character on
+    each continuation line removed.
+
+    """
+    if not sig:
+        return None
+    assert sig.startswith(b'gpgsig ')
+    sig = sig[7:]
+    return sig.replace(b'\n ', b'\n')

 
 # FIXME: derived from http://git.rsbx.net/Documents/Git_Data_Formats.txt
 # Make sure that's authoritative.
 
 # FIXME: derived from http://git.rsbx.net/Documents/Git_Data_Formats.txt
 # Make sure that's authoritative.

+
+# See also
+# https://github.com/git/git/blob/master/Documentation/technical/signature-format.txt
+# The continuation lines have only one leading space.
+

 _start_end_char = br'[^ .,:;<>"\'\0\n]'
 _content_char = br'[^\0\n<>]'
 _safe_str_rx = br'(?:%s{1,2}|(?:%s%s*%s))' \
 _start_end_char = br'[^ .,:;<>"\'\0\n]'
 _content_char = br'[^\0\n<>]'
 _safe_str_rx = br'(?:%s{1,2}|(?:%s%s*%s))' \


@@ -103,7 +137,7 @@ _mergetag_rx = br'(?:\nmergetag object [abcdefABCDEF0123456789]{40}(?:\n [^\0\n]
 _commit_rx = re.compile(br'''tree (?P<tree>[abcdefABCDEF0123456789]{40})
 (?P<parents>%s*)author (?P<author_name>%s) <(?P<author_mail>%s)> (?P<asec>\d+) (?P<atz>%s)
 committer (?P<committer_name>%s) <(?P<committer_mail>%s)> (?P<csec>\d+) (?P<ctz>%s)(?P<mergetag>%s?)
 _commit_rx = re.compile(br'''tree (?P<tree>[abcdefABCDEF0123456789]{40})
 (?P<parents>%s*)author (?P<author_name>%s) <(?P<author_mail>%s)> (?P<asec>\d+) (?P<atz>%s)
 committer (?P<committer_name>%s) <(?P<committer_mail>%s)> (?P<csec>\d+) (?P<ctz>%s)(?P<mergetag>%s?)

-
+(?P<gpgsig>gpgsig .*\n(?: .*\n)*)?

 (?P<message>(?:.|\n)*)''' % (_parent_rx,
                              _safe_str_rx, _safe_str_rx, _tz_rx,
                              _safe_str_rx, _safe_str_rx, _tz_rx,
 (?P<message>(?:.|\n)*)''' % (_parent_rx,
                              _safe_str_rx, _safe_str_rx, _tz_rx,
                              _safe_str_rx, _safe_str_rx, _tz_rx,


@@ -117,6 +151,7 @@ CommitInfo = namedtuple('CommitInfo', ['tree', 'parents',
                                        'author_sec', 'author_offset',
                                        'committer_name', 'committer_mail',
                                        'committer_sec', 'committer_offset',
                                        'author_sec', 'author_offset',
                                        'committer_name', 'committer_mail',
                                        'committer_sec', 'committer_offset',

+                                       'gpgsig',

                                        'message'])
 
 def parse_commit(content):
                                        'message'])
 
 def parse_commit(content):


@@ -134,6 +169,7 @@ def parse_commit(content):
                       committer_mail=matches['committer_mail'],
                       committer_sec=int(matches['csec']),
                       committer_offset=parse_tz_offset(matches['ctz']),
                       committer_mail=matches['committer_mail'],
                       committer_sec=int(matches['csec']),
                       committer_offset=parse_tz_offset(matches['ctz']),

+                      gpgsig=parse_commit_gpgsig(matches['gpgsig']),

                       message=matches['message'])
 
 
                       message=matches['message'])
 
 


@@ -248,8 +284,7 @@ def demangle_name(name, mode):
     elif name.endswith(b'.bupm'):
         return (name[:-5],
                 BUP_CHUNKED if stat.S_ISDIR(mode) else BUP_NORMAL)
     elif name.endswith(b'.bupm'):
         return (name[:-5],
                 BUP_CHUNKED if stat.S_ISDIR(mode) else BUP_NORMAL)

-    else:
-        return (name, BUP_NORMAL)
+    return (name, BUP_NORMAL)

 
 
 def calc_hash(type, content):
 
 
 def calc_hash(type, content):


@@ -318,27 +353,6 @@ def _encode_packobj(type, content, compression_level=1):
     yield z.flush()
 
 
     yield z.flush()
 
 

-def _encode_looseobj(type, content, compression_level=1):
-    z = zlib.compressobj(compression_level)
-    yield z.compress(b'%s %d\0' % (type, len(content)))
-    yield z.compress(content)
-    yield z.flush()
-
-
-def _decode_looseobj(buf):
-    assert(buf);
-    s = zlib.decompress(buf)
-    i = s.find(b'\0')
-    assert(i > 0)
-    l = s[:i].split(b' ')
-    type = l[0]
-    sz = int(l[1])
-    content = s[i+1:]
-    assert(type in _typemap)
-    assert(sz == len(content))
-    return (type, content)
-
-

 def _decode_packobj(buf):
     assert(buf)
     c = byte_int(buf[0])
 def _decode_packobj(buf):
     assert(buf)
     c = byte_int(buf[0])


@@ -413,7 +427,8 @@ class PackIdxV1(PackIdx):
         return self
 
     def __exit__(self, type, value, traceback):
         return self
 
     def __exit__(self, type, value, traceback):

-        self.close()
+        with pending_raise(value, rethrow=False):
+            self.close()

 
     def __len__(self):
         return int(self.nsha)  # int() from long for python 2
 
     def __len__(self):
         return int(self.nsha)  # int() from long for python 2


@@ -463,7 +478,8 @@ class PackIdxV2(PackIdx):
         return self
 
     def __exit__(self, type, value, traceback):
         return self
 
     def __exit__(self, type, value, traceback):

-        self.close()
+        with pending_raise(value, rethrow=False):
+            self.close()

 
     def __len__(self):
         return int(self.nsha)  # int() from long for python 2
 
     def __len__(self):
         return int(self.nsha)  # int() from long for python 2


@@ -677,6 +693,28 @@ def idxmerge(idxlist, final_progress=True):
     return merge_iter(idxlist, 10024, pfunc, pfinal)
 
 
     return merge_iter(idxlist, 10024, pfunc, pfinal)
 
 

+def create_commit_blob(tree, parent,
+                       author, adate_sec, adate_tz,
+                       committer, cdate_sec, cdate_tz,
+                       msg):
+    if adate_tz is not None:
+        adate_str = _git_date_str(adate_sec, adate_tz)
+    else:
+        adate_str = _local_git_date_str(adate_sec)
+    if cdate_tz is not None:
+        cdate_str = _git_date_str(cdate_sec, cdate_tz)
+    else:
+        cdate_str = _local_git_date_str(cdate_sec)
+    l = []
+    if tree: l.append(b'tree %s' % hexlify(tree))
+    if parent: l.append(b'parent %s' % hexlify(parent))
+    if author: l.append(b'author %s %s' % (author, adate_str))
+    if committer: l.append(b'committer %s %s' % (committer, cdate_str))
+    l.append(b'')
+    l.append(msg)
+    return b'\n'.join(l)
+
+

 def _make_objcache():
     return PackIdxList(repo(b'objects/pack'))
 
 def _make_objcache():
     return PackIdxList(repo(b'objects/pack'))
 


@@ -702,9 +740,8 @@ class PackWriter:
         self.on_pack_finish = on_pack_finish
         if not max_pack_size:
             max_pack_size = git_config_get(b'pack.packSizeLimit',
         self.on_pack_finish = on_pack_finish
         if not max_pack_size:
             max_pack_size = git_config_get(b'pack.packSizeLimit',

-                                           repo_dir=self.repo_dir)
-            if max_pack_size is not None:
-                max_pack_size = parse_num(max_pack_size)
+                                           repo_dir=self.repo_dir,
+                                           opttype='int')

             if not max_pack_size:
                 # larger packs slow down pruning
                 max_pack_size = 1000 * 1000 * 1000
             if not max_pack_size:
                 # larger packs slow down pruning
                 max_pack_size = 1000 * 1000 * 1000


@@ -713,14 +750,12 @@ class PackWriter:
         self.max_pack_objects = max_pack_objects if max_pack_objects \
                                 else max(1, self.max_pack_size // 5000)
 
         self.max_pack_objects = max_pack_objects if max_pack_objects \
                                 else max(1, self.max_pack_size // 5000)
 

-    def __del__(self):
-        self.close()
-

     def __enter__(self):
         return self
 
     def __exit__(self, type, value, traceback):
     def __enter__(self):
         return self
 
     def __exit__(self, type, value, traceback):

-        self.close()
+        with pending_raise(value, rethrow=False):
+            self.close()

 
     def _open(self):
         if not self.file:
 
     def _open(self):
         if not self.file:


@@ -781,12 +816,6 @@ class PackWriter:
             self.breakpoint()
         return sha
 
             self.breakpoint()
         return sha
 

-    def breakpoint(self):
-        """Clear byte and object counts and return the last processed id."""
-        id = self._end(self.run_midx)
-        self.outbytes = self.count = 0
-        return id
-

     def _require_objcache(self):
         if self.objcache is None and self.objcache_maker:
             self.objcache = self.objcache_maker()
     def _require_objcache(self):
         if self.objcache is None and self.objcache_maker:
             self.objcache = self.objcache_maker()


@@ -829,53 +858,32 @@ class PackWriter:
                    msg):
         """Create a commit object in the pack.  The date_sec values must be
         epoch-seconds, and if a tz is None, the local timezone is assumed."""
                    msg):
         """Create a commit object in the pack.  The date_sec values must be
         epoch-seconds, and if a tz is None, the local timezone is assumed."""

-        if adate_tz is not None:
-            adate_str = _git_date_str(adate_sec, adate_tz)
-        else:
-            adate_str = _local_git_date_str(adate_sec)
-        if cdate_tz is not None:
-            cdate_str = _git_date_str(cdate_sec, cdate_tz)
-        else:
-            cdate_str = _local_git_date_str(cdate_sec)
-        l = []
-        if tree: l.append(b'tree %s' % hexlify(tree))
-        if parent: l.append(b'parent %s' % hexlify(parent))
-        if author: l.append(b'author %s %s' % (author, adate_str))
-        if committer: l.append(b'committer %s %s' % (committer, cdate_str))
-        l.append(b'')
-        l.append(msg)
-        return self.maybe_write(b'commit', b'\n'.join(l))
+        content = create_commit_blob(tree, parent,
+                                     author, adate_sec, adate_tz,
+                                     committer, cdate_sec, cdate_tz,
+                                     msg)
+        return self.maybe_write(b'commit', content)
+
+    def _end(self, run_midx=True, abort=False):
+        # Ignores run_midx during abort
+        if not self.file:
+            return None
+        self.file, f = None, self.file
+        self.idx, idx = None, self.idx
+        self.parentfd, pfd, = None, self.parentfd
+        self.objcache = None

 
 

-    def abort(self):
-        """Remove the pack file from disk."""
-        f = self.file
-        if f:
-            pfd = self.parentfd
-            self.file = None
-            self.parentfd = None
-            self.idx = None
-            try:
-                try:
-                    os.unlink(self.filename + b'.pack')
-                finally:
-                    f.close()
-            finally:
-                if pfd is not None:
-                    os.close(pfd)
+        with finalized(pfd, lambda x: x is not None and os.close(x)), \
+             f:

 
 

-    def _end(self, run_midx=True):
-        f = self.file
-        if not f: return None
-        self.file = None
-        try:
-            self.objcache = None
-            idx = self.idx
-            self.idx = None
+            if abort:
+                os.unlink(self.filename + b'.pack')
+                return None

 
             # update object count
             f.seek(8)
             cp = struct.pack('!i', self.count)
 
             # update object count
             f.seek(8)
             cp = struct.pack('!i', self.count)

-            assert(len(cp) == 4)
+            assert len(cp) == 4

             f.write(cp)
 
             # calculate the pack sha1sum
             f.write(cp)
 
             # calculate the pack sha1sum


@@ -885,29 +893,33 @@ class PackWriter:
                 sum.update(b)
             packbin = sum.digest()
             f.write(packbin)
                 sum.update(b)
             packbin = sum.digest()
             f.write(packbin)

+            f.flush()

             fdatasync(f.fileno())
             fdatasync(f.fileno())

-        finally:

             f.close()
 
             f.close()
 

-        obj_list_sha = idx.write(self.filename + b'.idx', packbin)
-        nameprefix = os.path.join(self.repo_dir,
-                                  b'objects/pack/pack-' +  obj_list_sha)
-        if os.path.exists(self.filename + b'.map'):
-            os.unlink(self.filename + b'.map')
-        os.rename(self.filename + b'.pack', nameprefix + b'.pack')
-        os.rename(self.filename + b'.idx', nameprefix + b'.idx')
-        try:
-            os.fsync(self.parentfd)
-        finally:
-            os.close(self.parentfd)
+            idx.write(self.filename + b'.idx', packbin)
+            nameprefix = os.path.join(self.repo_dir,
+                                      b'objects/pack/pack-' +  hexlify(packbin))
+            if os.path.exists(self.filename + b'.map'):
+                os.unlink(self.filename + b'.map')
+            os.rename(self.filename + b'.pack', nameprefix + b'.pack')
+            os.rename(self.filename + b'.idx', nameprefix + b'.idx')
+            os.fsync(pfd)
+            if run_midx:
+                auto_midx(os.path.join(self.repo_dir, b'objects/pack'))
+            if self.on_pack_finish:
+                self.on_pack_finish(nameprefix)
+            return nameprefix

 
 

-        if run_midx:
-            auto_midx(os.path.join(self.repo_dir, b'objects/pack'))
-
-        if self.on_pack_finish:
-            self.on_pack_finish(nameprefix)
+    def abort(self):
+        """Remove the pack file from disk."""
+        self._end(abort=True)

 
 

-        return nameprefix
+    def breakpoint(self):
+        """Clear byte and object counts and return the last processed id."""
+        id = self._end(self.run_midx)
+        self.outbytes = self.count = 0
+        return id

 
     def close(self, run_midx=True):
         """Close the pack file and move it to its definitive path."""
 
     def close(self, run_midx=True):
         """Close the pack file and move it to its definitive path."""


@@ -957,17 +969,13 @@ class PackIdxV2Writer:
             b = idx_f.read(8 + 4*256)
             idx_sum.update(b)
 
             b = idx_f.read(8 + 4*256)
             idx_sum.update(b)
 

-            obj_list_sum = Sha1()

             for b in chunkyreader(idx_f, 20 * self.count):
                 idx_sum.update(b)
             for b in chunkyreader(idx_f, 20 * self.count):
                 idx_sum.update(b)

-                obj_list_sum.update(b)
-            namebase = hexlify(obj_list_sum.digest())

 
             for b in chunkyreader(idx_f):
                 idx_sum.update(b)
             idx_f.write(idx_sum.digest())
             fdatasync(idx_f.fileno())
 
             for b in chunkyreader(idx_f):
                 idx_sum.update(b)
             idx_f.write(idx_sum.digest())
             fdatasync(idx_f.fileno())

-            return namebase

         finally:
             idx_f.close()
 
         finally:
             idx_f.close()
 


@@ -1062,17 +1070,6 @@ def rev_list(ref_or_refs, parse=None, format=None, repo_dir=None):
         raise GitError('git rev-list returned error %d' % rv)
 
 
         raise GitError('git rev-list returned error %d' % rv)
 
 

-def get_commit_dates(refs, repo_dir=None):
-    """Get the dates for the specified commit refs.  For now, every unique
-       string in refs must resolve to a different commit or this
-       function will fail."""
-    result = []
-    for ref in refs:
-        commit = get_commit_items(ref, cp(repo_dir))
-        result.append(commit.author_sec)
-    return result
-
-

 def rev_parse(committish, repo_dir=None):
     """Resolve the full hash for 'committish', if it exists.
 
 def rev_parse(committish, repo_dir=None):
     """Resolve the full hash for 'committish', if it exists.
 


@@ -1280,6 +1277,7 @@ class CatPipe:
         if wait:
             p.wait()
             return p.returncode
         if wait:
             p.wait()
             return p.returncode

+        return None

 
     def restart(self):
         self.close()
 
     def restart(self):
         self.close()


@@ -1310,6 +1308,9 @@ class CatPipe:
         self.p.stdin.write(ref + b'\n')
         self.p.stdin.flush()
         hdr = self.p.stdout.readline()
         self.p.stdin.write(ref + b'\n')
         self.p.stdin.flush()
         hdr = self.p.stdout.readline()

+        if not hdr:
+            raise GitError('unexpected cat-file EOF (last request: %r, exit: %s)'
+                           % (ref, self.p.poll() or 'none'))

         if hdr.endswith(b' missing\n'):
             self.inprogress = None
             yield None, None, None
         if hdr.endswith(b' missing\n'):
             self.inprogress = None
             yield None, None, None