Python common.sanitized函数代码示例

    def rev_list_contains(self, refname, version, fetch=True):
        """
        calls git rev-list with refname and returns True if version
        can be found in rev-list result

        :param refname: a git refname
        :param version: an SHA IDs (if partial, caller is responsible
          for mismatch)
        :returns: True if version is an ancestor commit from refname
        :raises: GitError when call to git fetch fails
        """
        # to avoid listing unnecessarily many rev-ids, we cut off all
        # those we are definitely not interested in
        # $ git rev-list foo bar ^baz ^bez
        # means "list all the commits which are reachable from foo or
        # bar, but not from baz or bez". We use --parents because
        # ^baz also excludes baz itself. We could also use git
        # show --format=%P to get all parents first and use that,
        # not sure what's more performant
        if fetch:
            self._do_fetch()
        if (refname is not None and refname != '' and
                version is not None and version != ''):

            cmd = 'git rev-list %s ^%s --parents' % (sanitized(refname), sanitized(version))
            _, output, _ = run_shell_command(cmd, shell=True, cwd=self._path)
            for line in output.splitlines():
                # can have 1, 2 or 3 elements (commit, parent1, parent2)
                for hashid in line.split(" "):
                    if hashid.startswith(version):
                        return True
        return False

 def checkout(self, url, version='', verbose=False, shallow=False):
     if self.path_exists():
         sys.stderr.write("Error: cannot checkout into existing directory\n")
         return False
     # make sure that the parent directory exists for #3497
     base_path = os.path.split(self.get_path())[0]
     try:
         os.makedirs(base_path)
     except OSError:
         # OSError thrown if directory already exists this is ok
         pass
     cmd = "hg clone %s %s" % (sanitized(url), self._path)
     value, _, _ = run_shell_command(cmd,
                                     shell=True,
                                     no_filter=True)
     if value != 0:
         if self.path_exists():
             sys.stderr.write("Error: cannot checkout into existing directory\n")
         return False
     if version is not None and version.strip() != '':
         cmd = "hg checkout %s" % sanitized(version)
         value, _, _ = run_shell_command(cmd,
                                         cwd=self._path,
                                         shell=True,
                                         no_filter=True)
         if value != 0:
             return False
     return True

 def checkout(self, url, version='', verbose=False, shallow=False):
     if url is None or url.strip() == '':
         raise ValueError('Invalid empty url : "%s"' % url)
     # make sure that the parent directory exists for #3497
     base_path = os.path.split(self.get_path())[0]
     try:
         os.makedirs(base_path)
     except OSError:
         # OSError thrown if directory already exists this is ok
         pass
     cmd = "hg clone %s %s" % (sanitized(url), self._path)
     value, _, msg = run_shell_command(cmd,
                                       shell=True,
                                       no_filter=True)
     if value != 0:
         if msg:
             sys.logger.error('%s' % msg)
         return False
     if version is not None and version.strip() != '':
         cmd = "hg checkout %s" % sanitized(version)
         value, _, msg = run_shell_command(cmd,
                                         cwd=self._path,
                                         shell=True,
                                         no_filter=True)
         if value != 0:
             if msg:
                 sys.stderr.write('%s\n' % msg)
             return False
     return True

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath is None:
            relpath = ""

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = ("--limit %d" % (int(limit))) if limit else ""
            HG_COMMIT_FIELDS = ["id", "author", "email", "date", "message"]
            HG_LOG_FORMAT = (
                "\x1f".join(["{node|short}", "{author|person}", "{autor|email}", "{date|isodate}", "{desc}"]) + "\x1e"
            )

            command = "hg log %s --template '%s' %s" % (sanitized(relpath), HG_LOG_FORMAT, limit_cmd)

            return_code, response_str, stderr = run_shell_command(command, shell=True, cwd=self._path)

            if return_code == 0:
                # Parse response
                response = response_str.strip("\n\x1e").split("\x1e")
                response = [row.strip().split("\x1f") for row in response]
                response = [dict(zip(HG_COMMIT_FIELDS, row)) for row in response]
                # Parse dates
                for entry in response:
                    entry["date"] = dateutil.parser.parse(entry["date"])

        return response

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath is None:
            relpath = ""

        # Compile regexes
        id_regex = re.compile("^revno: ([0-9]+)$", flags=re.MULTILINE)
        committer_regex = re.compile("^committer: (.+)$", flags=re.MULTILINE)
        timestamp_regex = re.compile("^timestamp: (.+)$", flags=re.MULTILINE)
        message_regex = re.compile("^  (.+)$", flags=re.MULTILINE)

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = ("--limit=%d" % (int(limit))) if limit else ""
            command = "bzr log %s %s" % (sanitized(relpath), limit_cmd)
            return_code, text_response, stderr = run_shell_command(command, shell=True, cwd=self._path)
            if return_code == 0:
                revno_match = id_regex.findall(text_response)
                committer_match = committer_regex.findall(text_response)
                timestamp_match = timestamp_regex.findall(text_response)
                message_match = message_regex.findall(text_response)

                # Extract the entries
                for revno, committer, timestamp, message in zip(
                    revno_match, committer_match, timestamp_match, message_match
                ):
                    author, email_address = email.utils.parseaddr(committer)
                    date = dateutil.parser.parse(timestamp)
                    log_data = {"id": revno, "author": author, "email": email_address, "message": message, "date": date}

                    response.append(log_data)

        return response

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath == None:
            relpath = ""

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = ("-n %d" % (int(limit))) if limit else ""

            GIT_COMMIT_FIELDS = ["id", "author", "email", "date", "message"]
            GIT_LOG_FORMAT = "%x1f".join(["%H", "%an", "%ae", "%ad", "%s"]) + "%x1e"

            command = 'git --work-tree=%s log --format="%s" %s %s ' % (
                self._path,
                GIT_LOG_FORMAT,
                limit_cmd,
                sanitized(relpath),
            )
            return_code, response, stderr = run_shell_command(command, shell=True, cwd=self._path)

            if return_code == 0:
                # Parse response
                response = response.strip("\n\x1e").split("\x1e")
                response = [row.strip().split("\x1f") for row in response]
                response = [dict(zip(GIT_COMMIT_FIELDS, row)) for row in response]

            # Parse dates
            for entry in response:
                entry["date"] = dateutil.parser.parse(entry["date"])

        return response

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath is None:
            relpath = ''

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = (("-n %d" % (int(limit))) if limit else "")

            GIT_COMMIT_FIELDS = ['id', 'author', 'email', 'date', 'message']
            GIT_LOG_FORMAT = '%x1f'.join(['%H', '%an', '%ae', '%ad', '%s']) + '%x1e'

            command = "git --work-tree=%s log --format=\"%s\" %s %s " % (self._path, GIT_LOG_FORMAT,
                                                                         limit_cmd, sanitized(relpath))
            return_code, response_str, stderr = run_shell_command(command, shell=True, cwd=self._path)

            if return_code == 0:
                # Parse response
                response = response_str.strip('\n\x1e').split("\x1e")
                response = [row.strip().split("\x1f") for row in response]
                response = [dict(zip(GIT_COMMIT_FIELDS, row)) for row in response]

                # Parse dates
                for entry in response:
                    entry['date'] = dateutil.parser.parse(entry['date'])

        return response

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath == None:
            relpath = ''

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = (("--limit %d" % (int(limit))) if limit else "")
            command = "svn log %s --xml %s" % (limit_cmd, sanitized(relpath) if len(relpath) > 0 else '')
            return_code, xml_response, stderr = run_shell_command(command, shell=True, cwd=self._path)

            # Parse response
            dom = xml.dom.minidom.parseString(xml_response)
            log_entries = dom.getElementsByTagName("logentry")

            # Extract the entries
            for log_entry in log_entries:
                author_tag = log_entry.getElementsByTagName("author")[0]
                date_tag = log_entry.getElementsByTagName("date")[0]
                msg_tags = log_entry.getElementsByTagName("msg")

                log_data = dict()
                log_data['id'] = log_entry.getAttribute("revision")
                log_data['author'] = author_tag.firstChild.nodeValue
                log_data['email'] = None
                log_data['date'] = dateutil.parser.parse(str(date_tag.firstChild.nodeValue))
                if len(msg_tags) > 0:
                    log_data['message'] = msg_tags[0].firstChild.nodeValue

                response.append(log_data)

        return response

 def get_version(self, spec=None):
     """
     :param spec: (optional) token to identify desired version. For
       git, this may be anything accepted by git log, e.g. a tagname,
       branchname, or sha-id.
     :param fetch: When spec is given, can be used to suppress git fetch call
     :returns: current SHA-ID of the repository. Or if spec is
       provided, the SHA-ID of a commit specified by some token if found, else None
     """
     if self.detect_presence():
         command = "git log -1"
         if spec is not None:
             command += " %s" % sanitized(spec)
         command += " --format='%H'"
         output = ''
         #we repeat the call once after fetching if necessary
         for _ in range(2):
             _, output, _ = run_shell_command(command,
                                              shell=True,
                                              cwd=self._path)
             if (output != '' or spec is None):
                 break
             # we try again after fetching if given spec had not been found
             try:
                 self._do_fetch()
             except GitError:
                 return None
         # On Windows the version can have single quotes around it
         output = output.strip("'")
         return output
     return None

 def get_version(self, spec=None):
     """
     :param spec: (optional) revisionspec of desired version.  May
       be any revisionspec as returned by 'bzr help revisionspec',
       e.g. a tagname or 'revno:<number>'
     :returns: the current revision number of the repository. Or if
       spec is provided, the number of a revision specified by some
       token.
     """
     if self.detect_presence():
         if spec is not None:
             command = ['bzr log -r %s .' % sanitized(spec)]
             _, output, _ = run_shell_command(command,
                                              shell=True,
                                              cwd=self._path,
                                              us_env=True)
             if output is None or output.strip() == '' or output.startswith("bzr:"):
                 return None
             else:
                 matches = [l for l in output.split('\n') if l.startswith('revno: ')]
                 if len(matches) == 1:
                     return matches[0].split()[1]
         else:
             _, output, _ = run_shell_command('bzr revno --tree',
                                              shell=True,
                                              cwd=self._path,
                                              us_env=True)
             return output.strip()

    def _get_branch_parent(self, fetch=False, current_branch=None):
        """
        :param fetch: if true, performs git fetch first
        :param current_branch: if not None, this is used as current branch (else extra shell call)
        :returns: (branch, remote) the name of the branch this branch tracks and its remote
        :raises: GitError if fetch fails
        """
        if not self.path_exists():
            return (None, None)
        # get name of configured merge ref.
        branchname = current_branch or self._get_branch()
        if branchname is None:
            return (None, None)

        cmd = 'git config --get %s' % sanitized('branch.%s.merge' % branchname)

        _, output, _ = run_shell_command(cmd,
                                         shell=True,
                                         cwd=self._path)
        if not output:
            return (None, None)
        lines = output.splitlines()
        if len(lines) > 1:
            sys.stderr.write("vcstools unable to handle multiple merge references for branch %s:\n%s\n"
                             % (branchname, output))
            return (None, None)

        # get name of configured remote
        cmd = 'git config --get "branch.%s.remote"' % branchname
        _, output2, _ = run_shell_command(cmd, shell=True, cwd=self._path)
        remote = output2 or self._get_default_remote()

        branch_reference = lines[0]
        # branch_reference is either refname, or /refs/heads/refname, or
        # heads/refname we would like to return refname however,
        # user could also have named any branch
        # "/refs/heads/refname", for some unholy reason check all
        # known branches on remote for refname, then for the odd
        # cases, as git seems to do
        candidate = branch_reference
        if candidate.startswith('refs/'):
            candidate = candidate[len('refs/'):]
        if candidate.startswith('heads/'):
            candidate = candidate[len('heads/'):]
        elif candidate.startswith('tags/'):
            candidate = candidate[len('tags/'):]
        elif candidate.startswith('remotes/'):
            candidate = candidate[len('remotes/'):]

        result = None
        if self._is_remote_branch(candidate, remote_name=remote, fetch=fetch):
            result = candidate
        elif branch_reference != candidate and self._is_remote_branch(branch_reference,
                                                                      remote_name=remote,
                                                                      fetch=False):
            result = branch_reference

        if result is not None:
            return (result, remote)
        return None, None

    def get_log(self, relpath=None, limit=None):
        response = []

        if relpath == None:
            relpath = ''

        if self.path_exists() and os.path.exists(os.path.join(self._path, relpath)):
            # Get the log
            limit_cmd = (("--limit %d" % (int(limit))) if limit else "")
            HG_COMMIT_FIELDS = ['id', 'author', 'email', 'date', 'message']
            HG_LOG_FORMAT = '\x1f'.join(['{node|short}', '{author|person}',
                                         '{autor|email}', '{date|isodate}',
                                         '{desc}']) + '\x1e'

            command = "hg log %s --template '%s' %s" % (sanitized(relpath),
                                                        HG_LOG_FORMAT,
                                                        limit_cmd)

            return_code, response_str, stderr = run_shell_command(command, shell=True, cwd=self._path)

            if return_code == 0:
                # Parse response
                response = response_str.strip('\n\x1e').split("\x1e")
                response = [row.strip().split("\x1f") for row in response]
                response = [dict(zip(HG_COMMIT_FIELDS, row)) for row in response]
                print(response)
                # Parse dates
                for entry in response:
                    entry['date'] = dateutil.parser.parse(entry['date'])

        return response

 def get_diff(self, basepath=None):
     response = ''
     if basepath is None:
         basepath = self._path
     if self.path_exists():
         rel_path = normalized_rel_path(self._path, basepath)
         # git needs special treatment as it only works from inside
         # use HEAD to also show staged changes. Maybe should be option?
         # injection should be impossible using relpath, but to be sure, we check
         cmd = "git diff HEAD --src-prefix=%s/ --dst-prefix=%s/ ." % \
               (sanitized(rel_path), sanitized(rel_path))
         _, response, _ = run_shell_command(cmd, shell=True, cwd=self._path)
         if LooseVersion(self.gitversion) > LooseVersion('1.7'):
             cmd = 'git submodule foreach --recursive git diff HEAD'
             _, output, _ = run_shell_command(cmd, shell=True, cwd=self._path)
             response += _git_diff_path_submodule_change(output, rel_path)
     return response

 def get_diff(self, basepath=None):
     response = None
     if basepath is None:
         basepath = self._path
     if self.path_exists():
         rel_path = normalized_rel_path(self._path, basepath)
         command = "hg diff -g %(path)s --repository %(path)s" % {'path': sanitized(rel_path)}
         _, response, _ = run_shell_command(command, shell=True, cwd=basepath)
         response = _hg_diff_path_change(response, rel_path)
     return response

 def _get_version_from_path(self, spec=None, path=None):
     """
     :param spec: (optional) spec can be what 'svn info --help'
       allows, meaning a revnumber, {date}, HEAD, BASE, PREV, or
       COMMITTED.
     :param path: the url to use, default is for this repo
     :returns: current revision number of the repository. Or if spec
       provided, the number of a revision specified by some
       token.
     """
     if not self.path_exists():
         return None
     command = 'svn info '
     if spec is not None:
         if spec.isdigit():
             # looking up svn with "-r" takes long, and if spec is
             # a number, all we get from svn is the same number,
             # unless we try to look at higher rev numbers (in
             # which case either get the same number, or an error
             # if the rev does not exist). So we first do a very
             # quick svn info, and check revision numbers.
             currentversion = self.get_version(spec=None)
             # currentversion is like '-r12345'
             if currentversion is not None and \
                int(currentversion[2:]) > int(spec):
                 # so if we know revision exist, just return the
                 # number, avoid the long call to svn server
                 return '-r' + spec
         if spec.startswith("-r"):
             command += sanitized(spec)
         else:
             command += sanitized('-r%s' % spec)
     command += " %s" % path
     # #3305: parsing not robust to non-US locales
     _, output, _ = run_shell_command(command, shell=True, us_env=True)
     if output is not None:
         matches = \
             [l for l in output.splitlines() if l.startswith('Last Changed Rev: ')]
         if len(matches) == 1:
             split_str = matches[0].split()
             if len(split_str) == 4:
                 return '-r' + split_str[3]
     return None