Ejemplo n.º 1
0
    def run(self, parse_now=True):
        """Run git-fast-export."""
        import_marks = self.write_marks()
        export_marks = p4gf_tempfile.new_temp_file(prefix='fe-marks-')

        # Note that we do not ask Git to attempt to detect file renames or
        # copies, as this seems to lead to several bugs, including one that
        # loses data. For now, the safest option is to translate the file
        # operations exactly as they appear in the commit. This also makes the
        # round-trip conversion safer.
        cmd = ['git', 'fast-export', '--no-data']
        if self.ctx.find_copy_rename_enabled:
            cmd.extend(self.ctx.find_copy_rename_args)
        cmd.append("--import-marks={}".format(import_marks.name))
        cmd.append("--export-marks={}".format(export_marks.name))
        if self.last_old_commit:
            cmd.append("{}..{}".format(self.last_old_commit, self.last_new_commit))
        elif isinstance(self.last_new_commit, list):
            cmd.extend(list(set(self.last_new_commit)))
        else:
            cmd.append(self.last_new_commit)
        LOG.debug('cmd={}'.format(cmd))

        try:
            # work around pylint bug where it doesn't know check_output() returns encoded bytes
            result = p4gf_proc.popen_binary(cmd)
            self.script = result['out']
            self.read_marks(export_marks)
            if parse_now:
                self.parse_commands()
        finally:
            import_marks.close()
            export_marks.close()
Ejemplo n.º 2
0
    def write_marks(self):
        """Write a text file with list of every known commit sha1.

        "Known" here means  our Git Fusion knows about it and it has been
        copied to Perforce.".
        """
        log = LOG.getChild('marks')
        marksfile = p4gf_tempfile.new_temp_file(prefix='fastexport-')
        sha1_list = p4gf_object_type.known_commit_sha1_list(self.ctx)
        # If configured to run unpacked, do so. Even to the point of unpacking
        # incoming packfiles. This allows for some time optimizations at the
        # (great!) expense of disk space.
        if not self.ctx.git_autopack:
            p4gf_git.unpack_objects()
        # Ensure hashes are unique and refer to existing objects.
        sha1_list = _prune_missing_objects(sha1_list, self.ctx.repo)
        mark_num = 0
        for sha1 in sha1_list:
            # Don't tell git-fast-export about last_new_commit if we want to
            # force git-fast-export to export it.
            if self.force_export_last_new_commit and sha1 == self.last_new_commit:
                continue

            mark_num += 1
            content = ":{} {}\n".format(mark_num, sha1)
            marksfile.write(content.encode())
            log.debug(content)
        marksfile.flush()
        return marksfile
Ejemplo n.º 3
0
 def _create_database(self):
     """Use sqlite3 maintain separate lists of unique MD5 and tree paths"""
     db_file = p4gf_tempfile.new_temp_file(prefix='repo_size_',
                                           suffix='.db',
                                           delete=False)
     self.sql_db = sqlite3.connect(database=db_file.name,
                                   isolation_level="EXCLUSIVE")
     self.db_file_name = db_file.name
     self.sql_db.execute("PRAGMA synchronous = OFF")
     self.sql_db.execute("CREATE TABLE md5  (key TEXT PRIMARY KEY)")
     self.sql_db.execute("CREATE TABLE tree (key TEXT PRIMARY KEY)")
 def __init__(self, ctx):
     self.ctx = ctx
     self.script = p4gf_tempfile.new_temp_file(prefix='fastimport-')
     self.timezone = ctx.timezone
     self.__tzname = None
     self.project_root_path_length = len(ctx.contentlocalroot)
     self._line_count = 0
     self._byte_count = 0
     self.username_map = dict()
     self.usermap = p4gf_usermap.UserMap(ctx.p4gf,
                                         ctx.email_case_sensitivity)
     self.lfs_files = {}
     self.text_pointers = []
Ejemplo n.º 5
0
 def __enter__(self):
     """Enter the context."""
     if not self.__temp:
         self.__temp = p4gf_tempfile.new_temp_file(mode='w+',
                                                   encoding='UTF-8',
                                                   prefix='http-output-',
                                                   delete=False)
         self.__stdout = sys.stdout
         self.__stderr = sys.stderr
         sys.stdout.flush()
         sys.stderr.flush()
         sys.stdout = self.__temp
         sys.stderr = self.__temp
         LOG.debug("stdout/stderr redirecting to %s...", self.__temp.name)
     return self
    def spec_file_path(self):
        """Lazy-create, then reuse over and over, a single temp file to hold
        our fake changelist spec.
        """
        if self._spec_file_path:
            return self._spec_file_path

            # Don't bother if we're just PASS/FAIL/None
        if not self.needs_spec_file():
            self._spec_file_path = ''
            return self._spec_file_path

        self._spec_file = p4gf_tempfile.new_temp_file(
            prefix='preflight-commit-', delete=False)
        self._spec_file_path = self._spec_file.name
        return self._spec_file_path
Ejemplo n.º 7
0
def read_request_data(environ):
    """Read the incoming request data to a temporary file.

    Handles both WSGI and CGI environments.

    :param dict environ: WSGI request environment.

    :return: name of temporary file containing request data.

    """
    # Read the input from the client.
    incoming = environ['wsgi.input']
    stdin_fobj = p4gf_tempfile.new_temp_file(prefix='http-client-input-', delete=False)
    LOG.debug('read_request_data() writing stdin to %s', stdin_fobj.name)
    if is_cgi():
        # Running in CGI mode as a WSGI application. In a hosted CGI
        # environment, the matter of content-length and transfer-encoding
        # is handled for us by the server. We simply read the input until
        # the EOF is encountered.
        shutil.copyfileobj(incoming, stdin_fobj)
    else:
        # Running within the WSGI simple server.
        # For more information on the idiosyncrasies within WSGI 1.0, see
        # http://blog.dscpl.com.au/2009/10/details-on-wsgi-10-amendmentsclarificat.html
        try:
            content_length = int(environ.get('CONTENT_LENGTH', 0))
        except ValueError:
            content_length = 0
        method = environ['REQUEST_METHOD']
        if TE_HEADER in environ and environ[TE_HEADER] == 'chunked':
            reader = ChunkedTransferReader(stdin_fobj)
            reader.read(incoming)
        elif content_length and (method == "POST" or method == "PUT"):
            # To avoid blocking forever reading input from the client, must
            # read _only_ the number of bytes specified in the request
            # (which happens to permit HTTP/1.1 keep-alive connections).
            while content_length > 0:
                length = min(65536, content_length)
                buf = incoming.read(length)
                if not buf:
                    break
                stdin_fobj.write(buf)
                content_length -= len(buf)
    stdin_fobj.close()
    return stdin_fobj.name
def read_p4changes_from_depot_path(ctx):
    """Read 'p4 changes -l' text and scan for changlists with "push state: complete"."""
    tmp_file = p4gf_tempfile.new_temp_file(mode='w', prefix='annotate-changes', delete=False)
    regex = re.compile(r'^', re.MULTILINE)

    for branch_chunk in ctx.iter_branch_chunks():

        with ctx.switched_to_union(branch_chunk):
            p4_result = ctx.p4run('changes', '-l', '//{}/...'.format(ctx.p4.client),
                                  log_warnings=logging.WARN)
            for rr in p4_result:
                tmp_file.write("Change {} on \n".format(rr['change']))
                desc = re.sub(regex, '    ', rr['desc'])
                tmp_file.write("{}\n".format(desc))
    tmp_file.close()
    complete_change_nums = read_p4changes_from_text_file(tmp_file.name)
    if DEBUG:
        print(_("p4 changes data for this repo written to: '{}'.").format(tmp_file.name))
    else:
        os.remove(tmp_file.name)
    return complete_change_nums
def read_p4files_from_depot_path(p4, repo):
    """Write the depot path for the repo's commits to a temporary file.
    Then call read_p4files_from_file_path."""

    tmp_file = p4gf_tempfile.new_temp_file(mode='w', prefix='annotate-files', delete=False)

    commits_path = '//.git-fusion/objects/repos/{}/commits/...'.format(repo)
    p4_result = p4.run('files', '-e', commits_path)

    for rr in p4_result:
        depot_path = rr.get('depotFile')
        if not depot_path:
            continue
        tmp_file.write("{}\n".format(depot_path))

    tmp_file.close()
    d = read_p4files_from_file_path(tmp_file.name)
    if DEBUG:
        print(_("p4 files data for this repo written to: '{}'.").format(tmp_file.name))
    else:
        os.remove(tmp_file.name)
    return d
def get_gitlogs_from_git(git_dir):
    """Create file with git log output.

    git log --graph --decorate --oneline --all

    """
    tmp_file = p4gf_tempfile.new_temp_file(mode='w', prefix='annotate-gitlog', delete=False)
    if DEBUG:
        print(_("get_gitlogs_from_git: gitdir='{}'.").format(git_dir))
    if not os.path.exists(git_dir):
        LOG.warning("Git repository {} missing stopping.".format(git_dir))
        print(_("Git repository {} missing stopping.").format(git_dir))
        sys.exit(1)
        # it's not the end of the world if the git repo disappears, just recreate it
    else:
        # Ensure the pack config settings are set to desired values.
        git_dir = '--git-dir={}'.format(git_dir)
        cmd = ['git', git_dir, 'log', '--graph',  '--decorate', '--oneline', '--all']
        subprocess.call(cmd, stdout=tmp_file)
    tmp_file.close()
    if DEBUG:
        print(_("git logs data for this repo written to: '{}'.").format(tmp_file.name))
    return tmp_file.name
Ejemplo n.º 11
0
    def run_fast_import(self):
        """Run git-fast-import to create the git commits.

        Returns: a list of commits.  Each line is formatted as
            a change number followed by the SHA1 of the commit.

        The returned list is also written to a file called marks.
        """
        with Timer(OVERALL):
            with Timer(RUN):
                LOG.debug("running git fast-import")
                # tell git-fast-import to export marks to a temp file
                self.script.flush()
                marks_file = p4gf_tempfile.new_temp_file(prefix='marks-')
                try:
                    cmd = [
                        'git', 'fast-import', '--quiet',
                        '--export-marks=' + marks_file.name
                    ]
                    ec = p4gf_proc.wait(cmd, stdin=self.script.name)
                    if ec:
                        _log_crash_report(
                            'git-fast-import failed for {}'.format(
                                self.ctx.config.repo_name))
                        raise CalledProcessError(ec, NTR('git fast-import'))

                    # read the exported marks from file and return result
                    with open(marks_file.name, "r") as marksfile:
                        marks = [l.strip() for l in marksfile.readlines()]
                    if LOG.getChild('marks').isEnabledFor(logging.DEBUG3):
                        LOG.getChild('marks').debug3(
                            'git-fast-import returned marks ct={}\n'.format(
                                len(marks)) + '\n'.join(marks))
                    return marks
                finally:
                    self.script.close()
                    marks_file.close()