[mirror_edk2.git] / BaseTools / Scripts / PatchCheck.py

## @file\r
#  Check a patch for various format issues\r
#\r
#  Copyright (c) 2015 - 2019, Intel Corporation. All rights reserved.<BR>\r
#  Copyright (C) 2020, Red Hat, Inc.<BR>\r
#\r
#  SPDX-License-Identifier: BSD-2-Clause-Patent\r
#\r
\r
from __future__ import print_function\r
\r
VersionNumber = '0.1'\r
__copyright__ = "Copyright (c) 2015 - 2016, Intel Corporation  All rights reserved."\r
\r
import email\r
import argparse\r
import os\r
import re\r
import subprocess\r
import sys\r
\r
class Verbose:\r
    SILENT, ONELINE, NORMAL = range(3)\r
    level = NORMAL\r
\r
class EmailAddressCheck:\r
    """Checks an email address."""\r
\r
    def __init__(self, email):\r
        self.ok = True\r
\r
        if email is None:\r
            self.error('Email address is missing!')\r
            return\r
\r
        self.check_email_address(email)\r
\r
    def error(self, *err):\r
        if self.ok and Verbose.level > Verbose.ONELINE:\r
            print('The email address is not valid:')\r
        self.ok = False\r
        if Verbose.level < Verbose.NORMAL:\r
            return\r
        count = 0\r
        for line in err:\r
            prefix = (' *', '  ')[count > 0]\r
            print(prefix, line)\r
            count += 1\r
\r
    email_re1 = re.compile(r'(?:\s*)(.*?)(\s*)<(.+)>\s*$',\r
                           re.MULTILINE|re.IGNORECASE)\r
\r
    def check_email_address(self, email):\r
        email = email.strip()\r
        mo = self.email_re1.match(email)\r
        if mo is None:\r
            self.error("Email format is invalid: " + email.strip())\r
            return\r
\r
        name = mo.group(1).strip()\r
        if name == '':\r
            self.error("Name is not provided with email address: " +\r
                       email)\r
        else:\r
            quoted = len(name) > 2 and name[0] == '"' and name[-1] == '"'\r
            if name.find(',') >= 0 and not quoted:\r
                self.error('Add quotes (") around name with a comma: ' +\r
                           name)\r
\r
        if mo.group(2) == '':\r
            self.error("There should be a space between the name and " +\r
                       "email address: " + email)\r
\r
        if mo.group(3).find(' ') >= 0:\r
            self.error("The email address cannot contain a space: " +\r
                       mo.group(3))\r
\r
class CommitMessageCheck:\r
    """Checks the contents of a git commit message."""\r
\r
    def __init__(self, subject, message):\r
        self.ok = True\r
\r
        if subject is None and  message is None:\r
            self.error('Commit message is missing!')\r
            return\r
\r
        self.subject = subject\r
        self.msg = message\r
\r
        self.check_contributed_under()\r
        self.check_signed_off_by()\r
        self.check_misc_signatures()\r
        self.check_overall_format()\r
        self.report_message_result()\r
\r
    url = 'https://github.com/tianocore/tianocore.github.io/wiki/Commit-Message-Format'\r
\r
    def report_message_result(self):\r
        if Verbose.level < Verbose.NORMAL:\r
            return\r
        if self.ok:\r
            # All checks passed\r
            return_code = 0\r
            print('The commit message format passed all checks.')\r
        else:\r
            return_code = 1\r
        if not self.ok:\r
            print(self.url)\r
\r
    def error(self, *err):\r
        if self.ok and Verbose.level > Verbose.ONELINE:\r
            print('The commit message format is not valid:')\r
        self.ok = False\r
        if Verbose.level < Verbose.NORMAL:\r
            return\r
        count = 0\r
        for line in err:\r
            prefix = (' *', '  ')[count > 0]\r
            print(prefix, line)\r
            count += 1\r
\r
    # Find 'contributed-under:' at the start of a line ignoring case and\r
    # requires ':' to be present.  Matches if there is white space before\r
    # the tag or between the tag and the ':'.\r
    contributed_under_re = \\r
        re.compile(r'^\s*contributed-under\s*:', re.MULTILINE|re.IGNORECASE)\r
\r
    def check_contributed_under(self):\r
        match = self.contributed_under_re.search(self.msg)\r
        if match is not None:\r
            self.error('Contributed-under! (Note: this must be ' +\r
                       'removed by the code contributor!)')\r
\r
    @staticmethod\r
    def make_signature_re(sig, re_input=False):\r
        if re_input:\r
            sub_re = sig\r
        else:\r
            sub_re = sig.replace('-', r'[-\s]+')\r
        re_str = (r'^(?P<tag>' + sub_re +\r
                  r')(\s*):(\s*)(?P<value>\S.*?)(?:\s*)$')\r
        try:\r
            return re.compile(re_str, re.MULTILINE|re.IGNORECASE)\r
        except Exception:\r
            print("Tried to compile re:", re_str)\r
            raise\r
\r
    sig_block_re = \\r
        re.compile(r'''^\r
                        (?: (?P<tag>[^:]+) \s* : \s*\r
                            (?P<value>\S.*?) )\r
                            |\r
                        (?: \[ (?P<updater>[^:]+) \s* : \s*\r
                               (?P<note>.+?) \s* \] )\r
                    \s* $''',\r
                   re.VERBOSE | re.MULTILINE)\r
\r
    def find_signatures(self, sig):\r
        if not sig.endswith('-by') and sig != 'Cc':\r
            sig += '-by'\r
        regex = self.make_signature_re(sig)\r
\r
        sigs = regex.findall(self.msg)\r
\r
        bad_case_sigs = filter(lambda m: m[0] != sig, sigs)\r
        for s in bad_case_sigs:\r
            self.error("'" +s[0] + "' should be '" + sig + "'")\r
\r
        for s in sigs:\r
            if s[1] != '':\r
                self.error('There should be no spaces between ' + sig +\r
                           " and the ':'")\r
            if s[2] != ' ':\r
                self.error("There should be a space after '" + sig + ":'")\r
\r
            EmailAddressCheck(s[3])\r
\r
        return sigs\r
\r
    def check_signed_off_by(self):\r
        sob='Signed-off-by'\r
        if self.msg.find(sob) < 0:\r
            self.error('Missing Signed-off-by! (Note: this must be ' +\r
                       'added by the code contributor!)')\r
            return\r
\r
        sobs = self.find_signatures('Signed-off')\r
\r
        if len(sobs) == 0:\r
            self.error('Invalid Signed-off-by format!')\r
            return\r
\r
    sig_types = (\r
        'Reviewed',\r
        'Reported',\r
        'Tested',\r
        'Suggested',\r
        'Acked',\r
        'Cc'\r
        )\r
\r
    def check_misc_signatures(self):\r
        for sig in self.sig_types:\r
            self.find_signatures(sig)\r
\r
    def check_overall_format(self):\r
        lines = self.msg.splitlines()\r
\r
        if len(lines) >= 1 and lines[0].endswith('\r\n'):\r
            empty_line = '\r\n'\r
        else:\r
            empty_line = '\n'\r
\r
        lines.insert(0, empty_line)\r
        lines.insert(0, self.subject + empty_line)\r
\r
        count = len(lines)\r
\r
        if count <= 0:\r
            self.error('Empty commit message!')\r
            return\r
\r
        if count >= 1 and len(lines[0].rstrip()) >= 72:\r
            self.error('First line of commit message (subject line) ' +\r
                       'is too long.')\r
\r
        if count >= 1 and len(lines[0].strip()) == 0:\r
            self.error('First line of commit message (subject line) ' +\r
                       'is empty.')\r
\r
        if count >= 2 and lines[1].strip() != '':\r
            self.error('Second line of commit message should be ' +\r
                       'empty.')\r
\r
        for i in range(2, count):\r
            if (len(lines[i]) >= 76 and\r
                len(lines[i].split()) > 1 and\r
                not lines[i].startswith('git-svn-id:')):\r
                self.error('Line %d of commit message is too long.' % (i + 1))\r
\r
        last_sig_line = None\r
        for i in range(count - 1, 0, -1):\r
            line = lines[i]\r
            mo = self.sig_block_re.match(line)\r
            if mo is None:\r
                if line.strip() == '':\r
                    break\r
                elif last_sig_line is not None:\r
                    err2 = 'Add empty line before "%s"?' % last_sig_line\r
                    self.error('The line before the signature block ' +\r
                               'should be empty', err2)\r
                else:\r
                    self.error('The signature block was not found')\r
                break\r
            last_sig_line = line.strip()\r
\r
(START, PRE_PATCH, PATCH) = range(3)\r
\r
class GitDiffCheck:\r
    """Checks the contents of a git diff."""\r
\r
    def __init__(self, diff):\r
        self.ok = True\r
        self.format_ok = True\r
        self.lines = diff.splitlines(True)\r
        self.count = len(self.lines)\r
        self.line_num = 0\r
        self.state = START\r
        self.new_bin = []\r
        while self.line_num < self.count and self.format_ok:\r
            line_num = self.line_num\r
            self.run()\r
            assert(self.line_num > line_num)\r
        self.report_message_result()\r
\r
    def report_message_result(self):\r
        if Verbose.level < Verbose.NORMAL:\r
            return\r
        if self.ok:\r
            print('The code passed all checks.')\r
        if self.new_bin:\r
            print('\nWARNING - The following binary files will be added ' +\r
                  'into the repository:')\r
            for binary in self.new_bin:\r
                print('  ' + binary)\r
\r
    def run(self):\r
        line = self.lines[self.line_num]\r
\r
        if self.state in (PRE_PATCH, PATCH):\r
            if line.startswith('diff --git'):\r
                self.state = START\r
        if self.state == PATCH:\r
            if line.startswith('@@ '):\r
                self.state = PRE_PATCH\r
            elif len(line) >= 1 and line[0] not in ' -+' and \\r
                 not line.startswith('\r\n') and  \\r
                 not line.startswith(r'\ No newline ') and not self.binary:\r
                for line in self.lines[self.line_num + 1:]:\r
                    if line.startswith('diff --git'):\r
                        self.format_error('diff found after end of patch')\r
                        break\r
                self.line_num = self.count\r
                return\r
\r
        if self.state == START:\r
            if line.startswith('diff --git'):\r
                self.state = PRE_PATCH\r
                self.filename = line[13:].split(' ', 1)[0]\r
                self.is_newfile = False\r
                self.force_crlf = not self.filename.endswith('.sh')\r
            elif len(line.rstrip()) != 0:\r
                self.format_error("didn't find diff command")\r
            self.line_num += 1\r
        elif self.state == PRE_PATCH:\r
            if line.startswith('@@ '):\r
                self.state = PATCH\r
                self.binary = False\r
            elif line.startswith('GIT binary patch') or \\r
                 line.startswith('Binary files'):\r
                self.state = PATCH\r
                self.binary = True\r
                if self.is_newfile:\r
                    self.new_bin.append(self.filename)\r
            else:\r
                ok = False\r
                self.is_newfile = self.newfile_prefix_re.match(line)\r
                for pfx in self.pre_patch_prefixes:\r
                    if line.startswith(pfx):\r
                        ok = True\r
                if not ok:\r
                    self.format_error("didn't find diff hunk marker (@@)")\r
            self.line_num += 1\r
        elif self.state == PATCH:\r
            if self.binary:\r
                pass\r
            elif line.startswith('-'):\r
                pass\r
            elif line.startswith('+'):\r
                self.check_added_line(line[1:])\r
            elif line.startswith('\r\n'):\r
                pass\r
            elif line.startswith(r'\ No newline '):\r
                pass\r
            elif not line.startswith(' '):\r
                self.format_error("unexpected patch line")\r
            self.line_num += 1\r
\r
    pre_patch_prefixes = (\r
        '--- ',\r
        '+++ ',\r
        'index ',\r
        'new file ',\r
        'deleted file ',\r
        'old mode ',\r
        'new mode ',\r
        'similarity index ',\r
        'copy from ',\r
        'copy to ',\r
        'rename ',\r
        )\r
\r
    line_endings = ('\r\n', '\n\r', '\n', '\r')\r
\r
    newfile_prefix_re = \\r
        re.compile(r'''^\r
                       index\ 0+\.\.\r
                   ''',\r
                   re.VERBOSE)\r
\r
    def added_line_error(self, msg, line):\r
        lines = [ msg ]\r
        if self.filename is not None:\r
            lines.append('File: ' + self.filename)\r
        lines.append('Line: ' + line)\r
\r
        self.error(*lines)\r
\r
    old_debug_re = \\r
        re.compile(r'''\r
                        DEBUG \s* \( \s* \( \s*\r
                        (?: DEBUG_[A-Z_]+ \s* \| \s*)*\r
                        EFI_D_ ([A-Z_]+)\r
                   ''',\r
                   re.VERBOSE)\r
\r
    def check_added_line(self, line):\r
        eol = ''\r
        for an_eol in self.line_endings:\r
            if line.endswith(an_eol):\r
                eol = an_eol\r
                line = line[:-len(eol)]\r
\r
        stripped = line.rstrip()\r
\r
        if self.force_crlf and eol != '\r\n':\r
            self.added_line_error('Line ending (%s) is not CRLF' % repr(eol),\r
                                  line)\r
        if '\t' in line:\r
            self.added_line_error('Tab character used', line)\r
        if len(stripped) < len(line):\r
            self.added_line_error('Trailing whitespace found', line)\r
\r
        mo = self.old_debug_re.search(line)\r
        if mo is not None:\r
            self.added_line_error('EFI_D_' + mo.group(1) + ' was used, '\r
                                  'but DEBUG_' + mo.group(1) +\r
                                  ' is now recommended', line)\r
\r
    split_diff_re = re.compile(r'''\r
                                   (?P<cmd>\r
                                       ^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
                                   )\r
                                   (?P<index>\r
                                       ^ index \s+ .+ $\r
                                   )\r
                               ''',\r
                               re.IGNORECASE | re.VERBOSE | re.MULTILINE)\r
\r
    def format_error(self, err):\r
        self.format_ok = False\r
        err = 'Patch format error: ' + err\r
        err2 = 'Line: ' + self.lines[self.line_num].rstrip()\r
        self.error(err, err2)\r
\r
    def error(self, *err):\r
        if self.ok and Verbose.level > Verbose.ONELINE:\r
            print('Code format is not valid:')\r
        self.ok = False\r
        if Verbose.level < Verbose.NORMAL:\r
            return\r
        count = 0\r
        for line in err:\r
            prefix = (' *', '  ')[count > 0]\r
            print(prefix, line)\r
            count += 1\r
\r
class CheckOnePatch:\r
    """Checks the contents of a git email formatted patch.\r
\r
    Various checks are performed on both the commit message and the\r
    patch content.\r
    """\r
\r
    def __init__(self, name, patch):\r
        self.patch = patch\r
        self.find_patch_pieces()\r
\r
        msg_check = CommitMessageCheck(self.commit_subject, self.commit_msg)\r
        msg_ok = msg_check.ok\r
\r
        diff_ok = True\r
        if self.diff is not None:\r
            diff_check = GitDiffCheck(self.diff)\r
            diff_ok = diff_check.ok\r
\r
        self.ok = msg_ok and diff_ok\r
\r
        if Verbose.level == Verbose.ONELINE:\r
            if self.ok:\r
                result = 'ok'\r
            else:\r
                result = list()\r
                if not msg_ok:\r
                    result.append('commit message')\r
                if not diff_ok:\r
                    result.append('diff content')\r
                result = 'bad ' + ' and '.join(result)\r
            print(name, result)\r
\r
\r
    git_diff_re = re.compile(r'''\r
                                 ^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
                             ''',\r
                             re.IGNORECASE | re.VERBOSE | re.MULTILINE)\r
\r
    stat_re = \\r
        re.compile(r'''\r
                       (?P<commit_message> [\s\S\r\n]* )\r
                       (?P<stat>\r
                           ^ --- $ [\r\n]+\r
                           (?: ^ \s+ .+ \s+ \| \s+ \d+ \s+ \+* \-*\r
                               $ [\r\n]+ )+\r
                           [\s\S\r\n]+\r
                       )\r
                   ''',\r
                   re.IGNORECASE | re.VERBOSE | re.MULTILINE)\r
\r
    subject_prefix_re = \\r
        re.compile(r'''^\r
                       \s* (\[\r
                        [^\[\]]* # Allow all non-brackets\r
                       \])* \s*\r
                   ''',\r
                   re.VERBOSE)\r
\r
    def find_patch_pieces(self):\r
        if sys.version_info < (3, 0):\r
            patch = self.patch.encode('ascii', 'ignore')\r
        else:\r
            patch = self.patch\r
\r
        self.commit_msg = None\r
        self.stat = None\r
        self.commit_subject = None\r
        self.commit_prefix = None\r
        self.diff = None\r
\r
        if patch.startswith('diff --git'):\r
            self.diff = patch\r
            return\r
\r
        pmail = email.message_from_string(patch)\r
        parts = list(pmail.walk())\r
        assert(len(parts) == 1)\r
        assert(parts[0].get_content_type() == 'text/plain')\r
        content = parts[0].get_payload(decode=True).decode('utf-8', 'ignore')\r
\r
        mo = self.git_diff_re.search(content)\r
        if mo is not None:\r
            self.diff = content[mo.start():]\r
            content = content[:mo.start()]\r
\r
        mo = self.stat_re.search(content)\r
        if mo is None:\r
            self.commit_msg = content\r
        else:\r
            self.stat = mo.group('stat')\r
            self.commit_msg = mo.group('commit_message')\r
\r
        self.commit_subject = pmail['subject'].replace('\r\n', '')\r
        self.commit_subject = self.commit_subject.replace('\n', '')\r
        self.commit_subject = self.subject_prefix_re.sub('', self.commit_subject, 1)\r
\r
class CheckGitCommits:\r
    """Reads patches from git based on the specified git revision range.\r
\r
    The patches are read from git, and then checked.\r
    """\r
\r
    def __init__(self, rev_spec, max_count):\r
        commits = self.read_commit_list_from_git(rev_spec, max_count)\r
        if len(commits) == 1 and Verbose.level > Verbose.ONELINE:\r
            commits = [ rev_spec ]\r
        self.ok = True\r
        blank_line = False\r
        for commit in commits:\r
            if Verbose.level > Verbose.ONELINE:\r
                if blank_line:\r
                    print()\r
                else:\r
                    blank_line = True\r
                print('Checking git commit:', commit)\r
            patch = self.read_patch_from_git(commit)\r
            self.ok &= CheckOnePatch(commit, patch).ok\r
        if not commits:\r
            print("Couldn't find commit matching: '{}'".format(rev_spec))\r
\r
    def read_commit_list_from_git(self, rev_spec, max_count):\r
        # Run git to get the commit patch\r
        cmd = [ 'rev-list', '--abbrev-commit', '--no-walk' ]\r
        if max_count is not None:\r
            cmd.append('--max-count=' + str(max_count))\r
        cmd.append(rev_spec)\r
        out = self.run_git(*cmd)\r
        return out.split() if out else []\r
\r
    def read_patch_from_git(self, commit):\r
        # Run git to get the commit patch\r
        return self.run_git('show', '--pretty=email', '--no-textconv', commit)\r
\r
    def run_git(self, *args):\r
        cmd = [ 'git' ]\r
        cmd += args\r
        p = subprocess.Popen(cmd,\r
                     stdout=subprocess.PIPE,\r
                     stderr=subprocess.STDOUT)\r
        Result = p.communicate()\r
        return Result[0].decode('utf-8', 'ignore') if Result[0] and Result[0].find(b"fatal")!=0 else None\r
\r
class CheckOnePatchFile:\r
    """Performs a patch check for a single file.\r
\r
    stdin is used when the filename is '-'.\r
    """\r
\r
    def __init__(self, patch_filename):\r
        if patch_filename == '-':\r
            patch = sys.stdin.read()\r
            patch_filename = 'stdin'\r
        else:\r
            f = open(patch_filename, 'rb')\r
            patch = f.read().decode('utf-8', 'ignore')\r
            f.close()\r
        if Verbose.level > Verbose.ONELINE:\r
            print('Checking patch file:', patch_filename)\r
        self.ok = CheckOnePatch(patch_filename, patch).ok\r
\r
class CheckOneArg:\r
    """Performs a patch check for a single command line argument.\r
\r
    The argument will be handed off to a file or git-commit based\r
    checker.\r
    """\r
\r
    def __init__(self, param, max_count=None):\r
        self.ok = True\r
        if param == '-' or os.path.exists(param):\r
            checker = CheckOnePatchFile(param)\r
        else:\r
            checker = CheckGitCommits(param, max_count)\r
        self.ok = checker.ok\r
\r
class PatchCheckApp:\r
    """Checks patches based on the command line arguments."""\r
\r
    def __init__(self):\r
        self.parse_options()\r
        patches = self.args.patches\r
\r
        if len(patches) == 0:\r
            patches = [ 'HEAD' ]\r
\r
        self.ok = True\r
        self.count = None\r
        for patch in patches:\r
            self.process_one_arg(patch)\r
\r
        if self.count is not None:\r
            self.process_one_arg('HEAD')\r
\r
        if self.ok:\r
            self.retval = 0\r
        else:\r
            self.retval = -1\r
\r
    def process_one_arg(self, arg):\r
        if len(arg) >= 2 and arg[0] == '-':\r
            try:\r
                self.count = int(arg[1:])\r
                return\r
            except ValueError:\r
                pass\r
        self.ok &= CheckOneArg(arg, self.count).ok\r
        self.count = None\r
\r
    def parse_options(self):\r
        parser = argparse.ArgumentParser(description=__copyright__)\r
        parser.add_argument('--version', action='version',\r
                            version='%(prog)s ' + VersionNumber)\r
        parser.add_argument('patches', nargs='*',\r
                            help='[patch file | git rev list]')\r
        group = parser.add_mutually_exclusive_group()\r
        group.add_argument("--oneline",\r
                           action="store_true",\r
                           help="Print one result per line")\r
        group.add_argument("--silent",\r
                           action="store_true",\r
                           help="Print nothing")\r
        self.args = parser.parse_args()\r
        if self.args.oneline:\r
            Verbose.level = Verbose.ONELINE\r
        if self.args.silent:\r
            Verbose.level = Verbose.SILENT\r
\r
if __name__ == "__main__":\r
    sys.exit(PatchCheckApp().retval)\r
Commit	Line	Data
a7e173b0 JJ	1	## @file\r
	2	# Check a patch for various format issues\r
	3	#\r
a2813610	4	# Copyright (c) 2015 - 2019, Intel Corporation. All rights reserved.<BR>\r
8ffa47fb	5	# Copyright (C) 2020, Red Hat, Inc.<BR>\r
a7e173b0	6	#\r
2e351cbe	7	# SPDX-License-Identifier: BSD-2-Clause-Patent\r
a7e173b0 JJ	8	#\r
	9	\r
	10	from __future__ import print_function\r
	11	\r
	12	VersionNumber = '0.1'\r
e6140670	13	__copyright__ = "Copyright (c) 2015 - 2016, Intel Corporation All rights reserved."\r
a7e173b0 JJ	14	\r
	15	import email\r
	16	import argparse\r
	17	import os\r
	18	import re\r
	19	import subprocess\r
	20	import sys\r
	21	\r
	22	class Verbose:\r
	23	SILENT, ONELINE, NORMAL = range(3)\r
	24	level = NORMAL\r
	25	\r
8ffa47fb PMD	26	class EmailAddressCheck:\r
	27	"""Checks an email address."""\r
	28	\r
	29	def __init__(self, email):\r
	30	self.ok = True\r
	31	\r
	32	if email is None:\r
	33	self.error('Email address is missing!')\r
	34	return\r
	35	\r
	36	self.check_email_address(email)\r
	37	\r
	38	def error(self, *err):\r
	39	if self.ok and Verbose.level > Verbose.ONELINE:\r
	40	print('The email address is not valid:')\r
	41	self.ok = False\r
	42	if Verbose.level < Verbose.NORMAL:\r
	43	return\r
	44	count = 0\r
	45	for line in err:\r
	46	prefix = (' *', ' ')[count > 0]\r
	47	print(prefix, line)\r
	48	count += 1\r
	49	\r
	50	email_re1 = re.compile(r'(?:\s)(.?)(\s)<(.+)>\s$',\r
	51	re.MULTILINE\|re.IGNORECASE)\r
	52	\r
	53	def check_email_address(self, email):\r
	54	email = email.strip()\r
	55	mo = self.email_re1.match(email)\r
	56	if mo is None:\r
	57	self.error("Email format is invalid: " + email.strip())\r
	58	return\r
	59	\r
	60	name = mo.group(1).strip()\r
	61	if name == '':\r
	62	self.error("Name is not provided with email address: " +\r
	63	email)\r
	64	else:\r
	65	quoted = len(name) > 2 and name[0] == '"' and name[-1] == '"'\r
	66	if name.find(',') >= 0 and not quoted:\r
	67	self.error('Add quotes (") around name with a comma: ' +\r
	68	name)\r
	69	\r
	70	if mo.group(2) == '':\r
	71	self.error("There should be a space between the name and " +\r
	72	"email address: " + email)\r
	73	\r
	74	if mo.group(3).find(' ') >= 0:\r
	75	self.error("The email address cannot contain a space: " +\r
	76	mo.group(3))\r
	77	\r
a7e173b0 JJ	78	class CommitMessageCheck:\r
	79	"""Checks the contents of a git commit message."""\r
	80	\r
	81	def __init__(self, subject, message):\r
	82	self.ok = True\r
	83	\r
	84	if subject is None and message is None:\r
	85	self.error('Commit message is missing!')\r
	86	return\r
	87	\r
	88	self.subject = subject\r
	89	self.msg = message\r
	90	\r
	91	self.check_contributed_under()\r
	92	self.check_signed_off_by()\r
	93	self.check_misc_signatures()\r
	94	self.check_overall_format()\r
	95	self.report_message_result()\r
	96	\r
	97	url = 'https://github.com/tianocore/tianocore.github.io/wiki/Commit-Message-Format'\r
	98	\r
	99	def report_message_result(self):\r
	100	if Verbose.level < Verbose.NORMAL:\r
	101	return\r
	102	if self.ok:\r
	103	# All checks passed\r
	104	return_code = 0\r
	105	print('The commit message format passed all checks.')\r
	106	else:\r
	107	return_code = 1\r
	108	if not self.ok:\r
	109	print(self.url)\r
	110	\r
	111	def error(self, *err):\r
	112	if self.ok and Verbose.level > Verbose.ONELINE:\r
	113	print('The commit message format is not valid:')\r
	114	self.ok = False\r
	115	if Verbose.level < Verbose.NORMAL:\r
	116	return\r
	117	count = 0\r
	118	for line in err:\r
	119	prefix = (' *', ' ')[count > 0]\r
	120	print(prefix, line)\r
	121	count += 1\r
	122	\r
a2813610 MK	123	# Find 'contributed-under:' at the start of a line ignoring case and\r
	124	# requires ':' to be present. Matches if there is white space before\r
	125	# the tag or between the tag and the ':'.\r
	126	contributed_under_re = \\r
	127	re.compile(r'^\scontributed-under\s:', re.MULTILINE\|re.IGNORECASE)\r
	128	\r
a7e173b0	129	def check_contributed_under(self):\r
a2813610 MK	130	match = self.contributed_under_re.search(self.msg)\r
	131	if match is not None:\r
	132	self.error('Contributed-under! (Note: this must be ' +\r
	133	'removed by the code contributor!)')\r
a7e173b0 JJ	134	\r
	135	@staticmethod\r
	136	def make_signature_re(sig, re_input=False):\r
	137	if re_input:\r
	138	sub_re = sig\r
	139	else:\r
	140	sub_re = sig.replace('-', r'[-\s]+')\r
	141	re_str = (r'^(?P<tag>' + sub_re +\r
	142	r')(\s):(\s)(?P<value>\S.?)(?:\s)$')\r
	143	try:\r
	144	return re.compile(re_str, re.MULTILINE\|re.IGNORECASE)\r
	145	except Exception:\r
	146	print("Tried to compile re:", re_str)\r
	147	raise\r
	148	\r
	149	sig_block_re = \\r
	150	re.compile(r'''^\r
	151	(?: (?P<tag>[^:]+) \s* : \s*\r
	152	(?P<value>\S.*?) )\r
	153	\|\r
	154	(?: \[ (?P<updater>[^:]+) \s* : \s*\r
	155	(?P<note>.+?) \s* \] )\r
	156	\s* $''',\r
	157	re.VERBOSE \| re.MULTILINE)\r
	158	\r
	159	def find_signatures(self, sig):\r
	160	if not sig.endswith('-by') and sig != 'Cc':\r
	161	sig += '-by'\r
	162	regex = self.make_signature_re(sig)\r
	163	\r
	164	sigs = regex.findall(self.msg)\r
	165	\r
	166	bad_case_sigs = filter(lambda m: m[0] != sig, sigs)\r
	167	for s in bad_case_sigs:\r
	168	self.error("'" +s[0] + "' should be '" + sig + "'")\r
	169	\r
	170	for s in sigs:\r
	171	if s[1] != '':\r
	172	self.error('There should be no spaces between ' + sig +\r
	173	" and the ':'")\r
	174	if s[2] != ' ':\r
	175	self.error("There should be a space after '" + sig + ":'")\r
	176	\r
8ffa47fb	177	EmailAddressCheck(s[3])\r
a7e173b0 JJ	178	\r
	179	return sigs\r
	180	\r
a7e173b0 JJ	181	def check_signed_off_by(self):\r
	182	sob='Signed-off-by'\r
	183	if self.msg.find(sob) < 0:\r
	184	self.error('Missing Signed-off-by! (Note: this must be ' +\r
	185	'added by the code contributor!)')\r
	186	return\r
	187	\r
	188	sobs = self.find_signatures('Signed-off')\r
	189	\r
	190	if len(sobs) == 0:\r
	191	self.error('Invalid Signed-off-by format!')\r
	192	return\r
	193	\r
	194	sig_types = (\r
	195	'Reviewed',\r
	196	'Reported',\r
	197	'Tested',\r
	198	'Suggested',\r
	199	'Acked',\r
	200	'Cc'\r
	201	)\r
	202	\r
	203	def check_misc_signatures(self):\r
	204	for sig in self.sig_types:\r
	205	self.find_signatures(sig)\r
	206	\r
	207	def check_overall_format(self):\r
	208	lines = self.msg.splitlines()\r
	209	\r
	210	if len(lines) >= 1 and lines[0].endswith('\r\n'):\r
	211	empty_line = '\r\n'\r
	212	else:\r
	213	empty_line = '\n'\r
	214	\r
	215	lines.insert(0, empty_line)\r
	216	lines.insert(0, self.subject + empty_line)\r
	217	\r
	218	count = len(lines)\r
	219	\r
	220	if count <= 0:\r
	221	self.error('Empty commit message!')\r
	222	return\r
	223	\r
2649a735	224	if count >= 1 and len(lines[0].rstrip()) >= 72:\r
a7e173b0 JJ	225	self.error('First line of commit message (subject line) ' +\r
	226	'is too long.')\r
	227	\r
	228	if count >= 1 and len(lines[0].strip()) == 0:\r
	229	self.error('First line of commit message (subject line) ' +\r
	230	'is empty.')\r
	231	\r
	232	if count >= 2 and lines[1].strip() != '':\r
	233	self.error('Second line of commit message should be ' +\r
	234	'empty.')\r
	235	\r
	236	for i in range(2, count):\r
e6140670	237	if (len(lines[i]) >= 76 and\r
a7e173b0 JJ	238	len(lines[i].split()) > 1 and\r
	239	not lines[i].startswith('git-svn-id:')):\r
	240	self.error('Line %d of commit message is too long.' % (i + 1))\r
	241	\r
	242	last_sig_line = None\r
	243	for i in range(count - 1, 0, -1):\r
	244	line = lines[i]\r
	245	mo = self.sig_block_re.match(line)\r
	246	if mo is None:\r
	247	if line.strip() == '':\r
	248	break\r
	249	elif last_sig_line is not None:\r
	250	err2 = 'Add empty line before "%s"?' % last_sig_line\r
	251	self.error('The line before the signature block ' +\r
	252	'should be empty', err2)\r
	253	else:\r
	254	self.error('The signature block was not found')\r
	255	break\r
	256	last_sig_line = line.strip()\r
	257	\r
	258	(START, PRE_PATCH, PATCH) = range(3)\r
	259	\r
	260	class GitDiffCheck:\r
	261	"""Checks the contents of a git diff."""\r
	262	\r
	263	def __init__(self, diff):\r
	264	self.ok = True\r
	265	self.format_ok = True\r
	266	self.lines = diff.splitlines(True)\r
	267	self.count = len(self.lines)\r
	268	self.line_num = 0\r
	269	self.state = START\r
6a69dd49	270	self.new_bin = []\r
a7e173b0 JJ	271	while self.line_num < self.count and self.format_ok:\r
	272	line_num = self.line_num\r
	273	self.run()\r
	274	assert(self.line_num > line_num)\r
	275	self.report_message_result()\r
	276	\r
	277	def report_message_result(self):\r
	278	if Verbose.level < Verbose.NORMAL:\r
	279	return\r
	280	if self.ok:\r
	281	print('The code passed all checks.')\r
6a69dd49 HW	282	if self.new_bin:\r
	283	print('\nWARNING - The following binary files will be added ' +\r
	284	'into the repository:')\r
	285	for binary in self.new_bin:\r
	286	print(' ' + binary)\r
a7e173b0 JJ	287	\r
	288	def run(self):\r
	289	line = self.lines[self.line_num]\r
	290	\r
	291	if self.state in (PRE_PATCH, PATCH):\r
	292	if line.startswith('diff --git'):\r
	293	self.state = START\r
	294	if self.state == PATCH:\r
	295	if line.startswith('@@ '):\r
	296	self.state = PRE_PATCH\r
	297	elif len(line) >= 1 and line[0] not in ' -+' and \\r
05579e51	298	not line.startswith('\r\n') and \\r
aab57eff	299	not line.startswith(r'\ No newline ') and not self.binary:\r
a7e173b0 JJ	300	for line in self.lines[self.line_num + 1:]:\r
	301	if line.startswith('diff --git'):\r
	302	self.format_error('diff found after end of patch')\r
	303	break\r
	304	self.line_num = self.count\r
	305	return\r
	306	\r
	307	if self.state == START:\r
	308	if line.startswith('diff --git'):\r
	309	self.state = PRE_PATCH\r
ccaa7754	310	self.filename = line[13:].split(' ', 1)[0]\r
6a69dd49 HW	311	self.is_newfile = False\r
6a69dd49 HW	312	self.force_crlf = not self.filename.endswith('.sh')\r
a7e173b0 JJ	313	elif len(line.rstrip()) != 0:\r
	314	self.format_error("didn't find diff command")\r
	315	self.line_num += 1\r
	316	elif self.state == PRE_PATCH:\r
a7e173b0 JJ	317	if line.startswith('@@ '):\r
a7e173b0 JJ	318	self.state = PATCH\r
103733f8	319	self.binary = False\r
6a69dd49 HW	320	elif line.startswith('GIT binary patch') or \\r
6a69dd49 HW	321	line.startswith('Binary files'):\r
103733f8 JJ	322	self.state = PATCH\r
103733f8 JJ	323	self.binary = True\r
6a69dd49 HW	324	if self.is_newfile:\r
6a69dd49 HW	325	self.new_bin.append(self.filename)\r
a7e173b0 JJ	326	else:\r
a7e173b0 JJ	327	ok = False\r
6a69dd49	328	self.is_newfile = self.newfile_prefix_re.match(line)\r
a7e173b0 JJ	329	for pfx in self.pre_patch_prefixes:\r
	330	if line.startswith(pfx):\r
	331	ok = True\r
	332	if not ok:\r
	333	self.format_error("didn't find diff hunk marker (@@)")\r
	334	self.line_num += 1\r
	335	elif self.state == PATCH:\r
103733f8 JJ	336	if self.binary:\r
103733f8 JJ	337	pass\r
aab57eff	338	elif line.startswith('-'):\r
a7e173b0 JJ	339	pass\r
	340	elif line.startswith('+'):\r
	341	self.check_added_line(line[1:])\r
05579e51 MK	342	elif line.startswith('\r\n'):\r
05579e51 MK	343	pass\r
a7e173b0 JJ	344	elif line.startswith(r'\ No newline '):\r
	345	pass\r
	346	elif not line.startswith(' '):\r
	347	self.format_error("unexpected patch line")\r
	348	self.line_num += 1\r
	349	\r
	350	pre_patch_prefixes = (\r
	351	'--- ',\r
	352	'+++ ',\r
	353	'index ',\r
	354	'new file ',\r
	355	'deleted file ',\r
	356	'old mode ',\r
	357	'new mode ',\r
	358	'similarity index ',\r
1eeb5ff1 MK	359	'copy from ',\r
1eeb5ff1 MK	360	'copy to ',\r
a7e173b0	361	'rename ',\r
a7e173b0 JJ	362	)\r
	363	\r
	364	line_endings = ('\r\n', '\n\r', '\n', '\r')\r
	365	\r
6a69dd49 HW	366	newfile_prefix_re = \\r
	367	re.compile(r'''^\r
	368	index\ 0+\.\.\r
	369	''',\r
	370	re.VERBOSE)\r
a7e173b0 JJ	371	\r
	372	def added_line_error(self, msg, line):\r
	373	lines = [ msg ]\r
6a69dd49 HW	374	if self.filename is not None:\r
6a69dd49 HW	375	lines.append('File: ' + self.filename)\r
a7e173b0 JJ	376	lines.append('Line: ' + line)\r
	377	\r
	378	self.error(*lines)\r
	379	\r
c3926cdb YZ	380	old_debug_re = \\r
	381	re.compile(r'''\r
	382	DEBUG \s* \( \s* \( \s*\r
	383	(?: DEBUG_[A-Z_]+ \s* \\| \s)\r
	384	EFI_D_ ([A-Z_]+)\r
	385	''',\r
	386	re.VERBOSE)\r
	387	\r
a7e173b0 JJ	388	def check_added_line(self, line):\r
	389	eol = ''\r
	390	for an_eol in self.line_endings:\r
	391	if line.endswith(an_eol):\r
	392	eol = an_eol\r
	393	line = line[:-len(eol)]\r
	394	\r
	395	stripped = line.rstrip()\r
	396	\r
	397	if self.force_crlf and eol != '\r\n':\r
	398	self.added_line_error('Line ending (%s) is not CRLF' % repr(eol),\r
	399	line)\r
	400	if '\t' in line:\r
	401	self.added_line_error('Tab character used', line)\r
	402	if len(stripped) < len(line):\r
	403	self.added_line_error('Trailing whitespace found', line)\r
	404	\r
c3926cdb YZ	405	mo = self.old_debug_re.search(line)\r
	406	if mo is not None:\r
	407	self.added_line_error('EFI_D_' + mo.group(1) + ' was used, '\r
	408	'but DEBUG_' + mo.group(1) +\r
	409	' is now recommended', line)\r
	410	\r
a7e173b0 JJ	411	split_diff_re = re.compile(r'''\r
	412	(?P<cmd>\r
	413	^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
	414	)\r
	415	(?P<index>\r
	416	^ index \s+ .+ $\r
	417	)\r
	418	''',\r
	419	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
	420	\r
	421	def format_error(self, err):\r
	422	self.format_ok = False\r
	423	err = 'Patch format error: ' + err\r
	424	err2 = 'Line: ' + self.lines[self.line_num].rstrip()\r
	425	self.error(err, err2)\r
	426	\r
	427	def error(self, *err):\r
	428	if self.ok and Verbose.level > Verbose.ONELINE:\r
	429	print('Code format is not valid:')\r
	430	self.ok = False\r
	431	if Verbose.level < Verbose.NORMAL:\r
	432	return\r
	433	count = 0\r
	434	for line in err:\r
	435	prefix = (' *', ' ')[count > 0]\r
	436	print(prefix, line)\r
	437	count += 1\r
	438	\r
	439	class CheckOnePatch:\r
	440	"""Checks the contents of a git email formatted patch.\r
	441	\r
	442	Various checks are performed on both the commit message and the\r
	443	patch content.\r
	444	"""\r
	445	\r
	446	def __init__(self, name, patch):\r
	447	self.patch = patch\r
	448	self.find_patch_pieces()\r
	449	\r
	450	msg_check = CommitMessageCheck(self.commit_subject, self.commit_msg)\r
	451	msg_ok = msg_check.ok\r
	452	\r
	453	diff_ok = True\r
	454	if self.diff is not None:\r
	455	diff_check = GitDiffCheck(self.diff)\r
	456	diff_ok = diff_check.ok\r
	457	\r
	458	self.ok = msg_ok and diff_ok\r
	459	\r
	460	if Verbose.level == Verbose.ONELINE:\r
	461	if self.ok:\r
	462	result = 'ok'\r
	463	else:\r
	464	result = list()\r
	465	if not msg_ok:\r
	466	result.append('commit message')\r
	467	if not diff_ok:\r
	468	result.append('diff content')\r
	469	result = 'bad ' + ' and '.join(result)\r
	470	print(name, result)\r
	471	\r
	472	\r
	473	git_diff_re = re.compile(r'''\r
	474	^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
475	''',\r
476	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
477	\r
478	stat_re = \\r
479	re.compile(r'''\r
480	(?P<commit_message> [\s\S\r\n]* )\r
481	(?P<stat>\r
482	^ --- $ [\r\n]+\r
483	(?: ^ \s+ .+ \s+ \\| \s+ \d+ \s+ \+* \-*\r
484	$ [\r\n]+ )+\r
485	[\s\S\r\n]+\r
486	)\r
487	''',\r
488	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
489	\r
e709bbb1 YZ	490	subject_prefix_re = \\r
	491	re.compile(r'''^\r
	492	\s* (\[\r
	493	[^\[\]]* # Allow all non-brackets\r
	494	\])* \s*\r
	495	''',\r
	496	re.VERBOSE)\r
	497	\r
a7e173b0 JJ	498	def find_patch_pieces(self):\r
	499	if sys.version_info < (3, 0):\r
	500	patch = self.patch.encode('ascii', 'ignore')\r
	501	else:\r
	502	patch = self.patch\r
	503	\r
	504	self.commit_msg = None\r
	505	self.stat = None\r
	506	self.commit_subject = None\r
	507	self.commit_prefix = None\r
	508	self.diff = None\r
	509	\r
	510	if patch.startswith('diff --git'):\r
	511	self.diff = patch\r
	512	return\r
	513	\r
	514	pmail = email.message_from_string(patch)\r
	515	parts = list(pmail.walk())\r
	516	assert(len(parts) == 1)\r
	517	assert(parts[0].get_content_type() == 'text/plain')\r
	518	content = parts[0].get_payload(decode=True).decode('utf-8', 'ignore')\r
	519	\r
	520	mo = self.git_diff_re.search(content)\r
	521	if mo is not None:\r
	522	self.diff = content[mo.start():]\r
	523	content = content[:mo.start()]\r
	524	\r
	525	mo = self.stat_re.search(content)\r
	526	if mo is None:\r
	527	self.commit_msg = content\r
	528	else:\r
	529	self.stat = mo.group('stat')\r
	530	self.commit_msg = mo.group('commit_message')\r
	531	\r
	532	self.commit_subject = pmail['subject'].replace('\r\n', '')\r
	533	self.commit_subject = self.commit_subject.replace('\n', '')\r
e709bbb1	534	self.commit_subject = self.subject_prefix_re.sub('', self.commit_subject, 1)\r
a7e173b0 JJ	535	\r
	536	class CheckGitCommits:\r
	537	"""Reads patches from git based on the specified git revision range.\r
	538	\r
	539	The patches are read from git, and then checked.\r
	540	"""\r
	541	\r
	542	def __init__(self, rev_spec, max_count):\r
	543	commits = self.read_commit_list_from_git(rev_spec, max_count)\r
	544	if len(commits) == 1 and Verbose.level > Verbose.ONELINE:\r
	545	commits = [ rev_spec ]\r
	546	self.ok = True\r
	547	blank_line = False\r
	548	for commit in commits:\r
	549	if Verbose.level > Verbose.ONELINE:\r
	550	if blank_line:\r
	551	print()\r
	552	else:\r
	553	blank_line = True\r
	554	print('Checking git commit:', commit)\r
	555	patch = self.read_patch_from_git(commit)\r
	556	self.ok &= CheckOnePatch(commit, patch).ok\r
5ac4548c JC	557	if not commits:\r
5ac4548c JC	558	print("Couldn't find commit matching: '{}'".format(rev_spec))\r
a7e173b0 JJ	559	\r
	560	def read_commit_list_from_git(self, rev_spec, max_count):\r
	561	# Run git to get the commit patch\r
	562	cmd = [ 'rev-list', '--abbrev-commit', '--no-walk' ]\r
	563	if max_count is not None:\r
	564	cmd.append('--max-count=' + str(max_count))\r
	565	cmd.append(rev_spec)\r
	566	out = self.run_git(*cmd)\r
5ac4548c	567	return out.split() if out else []\r
a7e173b0 JJ	568	\r
	569	def read_patch_from_git(self, commit):\r
	570	# Run git to get the commit patch\r
96603b4f	571	return self.run_git('show', '--pretty=email', '--no-textconv', commit)\r
a7e173b0 JJ	572	\r
	573	def run_git(self, *args):\r
	574	cmd = [ 'git' ]\r
	575	cmd += args\r
	576	p = subprocess.Popen(cmd,\r
	577	stdout=subprocess.PIPE,\r
	578	stderr=subprocess.STDOUT)\r
5ac4548c	579	Result = p.communicate()\r
6460513a	580	return Result[0].decode('utf-8', 'ignore') if Result[0] and Result[0].find(b"fatal")!=0 else None\r
a7e173b0 JJ	581	\r
	582	class CheckOnePatchFile:\r
	583	"""Performs a patch check for a single file.\r
	584	\r
	585	stdin is used when the filename is '-'.\r
	586	"""\r
	587	\r
	588	def __init__(self, patch_filename):\r
	589	if patch_filename == '-':\r
	590	patch = sys.stdin.read()\r
	591	patch_filename = 'stdin'\r
	592	else:\r
	593	f = open(patch_filename, 'rb')\r
	594	patch = f.read().decode('utf-8', 'ignore')\r
	595	f.close()\r
	596	if Verbose.level > Verbose.ONELINE:\r
	597	print('Checking patch file:', patch_filename)\r
	598	self.ok = CheckOnePatch(patch_filename, patch).ok\r
	599	\r
	600	class CheckOneArg:\r
	601	"""Performs a patch check for a single command line argument.\r
	602	\r
	603	The argument will be handed off to a file or git-commit based\r
	604	checker.\r
	605	"""\r
	606	\r
	607	def __init__(self, param, max_count=None):\r
	608	self.ok = True\r
	609	if param == '-' or os.path.exists(param):\r
	610	checker = CheckOnePatchFile(param)\r
	611	else:\r
	612	checker = CheckGitCommits(param, max_count)\r
	613	self.ok = checker.ok\r
	614	\r
	615	class PatchCheckApp:\r
	616	"""Checks patches based on the command line arguments."""\r
	617	\r
	618	def __init__(self):\r
	619	self.parse_options()\r
	620	patches = self.args.patches\r
	621	\r
	622	if len(patches) == 0:\r
	623	patches = [ 'HEAD' ]\r
	624	\r
	625	self.ok = True\r
	626	self.count = None\r
	627	for patch in patches:\r
	628	self.process_one_arg(patch)\r
	629	\r
	630	if self.count is not None:\r
	631	self.process_one_arg('HEAD')\r
	632	\r
	633	if self.ok:\r
	634	self.retval = 0\r
	635	else:\r
	636	self.retval = -1\r
	637	\r
	638	def process_one_arg(self, arg):\r
	639	if len(arg) >= 2 and arg[0] == '-':\r
	640	try:\r
	641	self.count = int(arg[1:])\r
	642	return\r
	643	except ValueError:\r
	644	pass\r
645	self.ok &= CheckOneArg(arg, self.count).ok\r
646	self.count = None\r
647	\r
648	def parse_options(self):\r
649	parser = argparse.ArgumentParser(description=__copyright__)\r
650	parser.add_argument('--version', action='version',\r
651	version='%(prog)s ' + VersionNumber)\r
652	parser.add_argument('patches', nargs='*',\r
653	help='[patch file \| git rev list]')\r
654	group = parser.add_mutually_exclusive_group()\r
655	group.add_argument("--oneline",\r
656	action="store_true",\r
657	help="Print one result per line")\r
658	group.add_argument("--silent",\r
659	action="store_true",\r
660	help="Print nothing")\r
661	self.args = parser.parse_args()\r
662	if self.args.oneline:\r
663	Verbose.level = Verbose.ONELINE\r
664	if self.args.silent:\r
665	Verbose.level = Verbose.SILENT\r
666	\r
667	if __name__ == "__main__":\r
668	sys.exit(PatchCheckApp().retval)\r