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

## @file\r
#  Check a patch for various format issues\r
#\r
#  Copyright (c) 2015 - 2018, Intel Corporation. All rights reserved.<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 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
    def check_contributed_under(self):\r
        cu_msg='Contributed-under: TianoCore Contribution Agreement 1.1'\r
        if self.msg.find(cu_msg) < 0:\r
            # Allow 1.0 for now while EDK II community transitions to 1.1\r
            cu_msg='Contributed-under: TianoCore Contribution Agreement 1.0'\r
            if self.msg.find(cu_msg) < 0:\r
                self.error('Missing Contributed-under! (Note: this must be ' +\r
                           'added 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* $''',\r
                   re.VERBOSE | re.MULTILINE)\r
\r
    def find_signatures(self, sig):\r
        if not sig.endswith('-by') and sig != 'Cc':\r
            sig += '-by'\r
        regex = self.make_signature_re(sig)\r
\r
        sigs = regex.findall(self.msg)\r
\r
        bad_case_sigs = filter(lambda m: m[0] != sig, sigs)\r
        for s in bad_case_sigs:\r
            self.error("'" +s[0] + "' should be '" + sig + "'")\r
\r
        for s in sigs:\r
            if s[1] != '':\r
                self.error('There should be no spaces between ' + sig +\r
                           " and the ':'")\r
            if s[2] != ' ':\r
                self.error("There should be a space after '" + sig + ":'")\r
\r
            self.check_email_address(s[3])\r
\r
        return sigs\r
\r
    email_re1 = re.compile(r'(?:\s*)(.*?)(\s*)<(.+)>\s*$',\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
    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]) >= 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'\ 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'\ 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
        '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', 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
5ac4548c	4	# Copyright (c) 2015 - 2018, Intel Corporation. All rights reserved.<BR>\r
a7e173b0	5	#\r
2e351cbe	6	# SPDX-License-Identifier: BSD-2-Clause-Patent\r
a7e173b0 JJ	7	#\r
	8	\r
	9	from __future__ import print_function\r
	10	\r
	11	VersionNumber = '0.1'\r
e6140670	12	__copyright__ = "Copyright (c) 2015 - 2016, Intel Corporation All rights reserved."\r
a7e173b0 JJ	13	\r
	14	import email\r
	15	import argparse\r
	16	import os\r
	17	import re\r
	18	import subprocess\r
	19	import sys\r
	20	\r
	21	class Verbose:\r
	22	SILENT, ONELINE, NORMAL = range(3)\r
	23	level = NORMAL\r
	24	\r
	25	class CommitMessageCheck:\r
	26	"""Checks the contents of a git commit message."""\r
	27	\r
	28	def __init__(self, subject, message):\r
	29	self.ok = True\r
	30	\r
	31	if subject is None and message is None:\r
	32	self.error('Commit message is missing!')\r
	33	return\r
	34	\r
	35	self.subject = subject\r
	36	self.msg = message\r
	37	\r
	38	self.check_contributed_under()\r
	39	self.check_signed_off_by()\r
	40	self.check_misc_signatures()\r
	41	self.check_overall_format()\r
	42	self.report_message_result()\r
	43	\r
	44	url = 'https://github.com/tianocore/tianocore.github.io/wiki/Commit-Message-Format'\r
	45	\r
	46	def report_message_result(self):\r
	47	if Verbose.level < Verbose.NORMAL:\r
	48	return\r
	49	if self.ok:\r
	50	# All checks passed\r
	51	return_code = 0\r
	52	print('The commit message format passed all checks.')\r
	53	else:\r
	54	return_code = 1\r
	55	if not self.ok:\r
	56	print(self.url)\r
	57	\r
	58	def error(self, *err):\r
	59	if self.ok and Verbose.level > Verbose.ONELINE:\r
	60	print('The commit message format is not valid:')\r
	61	self.ok = False\r
	62	if Verbose.level < Verbose.NORMAL:\r
	63	return\r
	64	count = 0\r
	65	for line in err:\r
	66	prefix = (' *', ' ')[count > 0]\r
	67	print(prefix, line)\r
	68	count += 1\r
	69	\r
	70	def check_contributed_under(self):\r
f2d1b866	71	cu_msg='Contributed-under: TianoCore Contribution Agreement 1.1'\r
a7e173b0	72	if self.msg.find(cu_msg) < 0:\r
f2d1b866 MK	73	# Allow 1.0 for now while EDK II community transitions to 1.1\r
	74	cu_msg='Contributed-under: TianoCore Contribution Agreement 1.0'\r
	75	if self.msg.find(cu_msg) < 0:\r
	76	self.error('Missing Contributed-under! (Note: this must be ' +\r
	77	'added by the code contributor!)')\r
a7e173b0 JJ	78	\r
	79	@staticmethod\r
	80	def make_signature_re(sig, re_input=False):\r
	81	if re_input:\r
	82	sub_re = sig\r
	83	else:\r
	84	sub_re = sig.replace('-', r'[-\s]+')\r
	85	re_str = (r'^(?P<tag>' + sub_re +\r
	86	r')(\s):(\s)(?P<value>\S.?)(?:\s)$')\r
	87	try:\r
	88	return re.compile(re_str, re.MULTILINE\|re.IGNORECASE)\r
	89	except Exception:\r
	90	print("Tried to compile re:", re_str)\r
	91	raise\r
	92	\r
	93	sig_block_re = \\r
	94	re.compile(r'''^\r
	95	(?: (?P<tag>[^:]+) \s* : \s*\r
	96	(?P<value>\S.*?) )\r
	97	\|\r
	98	(?: \[ (?P<updater>[^:]+) \s* : \s*\r
	99	(?P<note>.+?) \s* \] )\r
	100	\s* $''',\r
	101	re.VERBOSE \| re.MULTILINE)\r
	102	\r
	103	def find_signatures(self, sig):\r
	104	if not sig.endswith('-by') and sig != 'Cc':\r
	105	sig += '-by'\r
	106	regex = self.make_signature_re(sig)\r
	107	\r
	108	sigs = regex.findall(self.msg)\r
	109	\r
	110	bad_case_sigs = filter(lambda m: m[0] != sig, sigs)\r
	111	for s in bad_case_sigs:\r
	112	self.error("'" +s[0] + "' should be '" + sig + "'")\r
	113	\r
	114	for s in sigs:\r
	115	if s[1] != '':\r
	116	self.error('There should be no spaces between ' + sig +\r
	117	" and the ':'")\r
	118	if s[2] != ' ':\r
	119	self.error("There should be a space after '" + sig + ":'")\r
	120	\r
	121	self.check_email_address(s[3])\r
	122	\r
	123	return sigs\r
	124	\r
	125	email_re1 = re.compile(r'(?:\s)(.?)(\s)<(.+)>\s$',\r
	126	re.MULTILINE\|re.IGNORECASE)\r
	127	\r
	128	def check_email_address(self, email):\r
	129	email = email.strip()\r
	130	mo = self.email_re1.match(email)\r
	131	if mo is None:\r
	132	self.error("Email format is invalid: " + email.strip())\r
	133	return\r
	134	\r
	135	name = mo.group(1).strip()\r
	136	if name == '':\r
	137	self.error("Name is not provided with email address: " +\r
	138	email)\r
	139	else:\r
	140	quoted = len(name) > 2 and name[0] == '"' and name[-1] == '"'\r
	141	if name.find(',') >= 0 and not quoted:\r
142	self.error('Add quotes (") around name with a comma: ' +\r
143	name)\r
144	\r
145	if mo.group(2) == '':\r
146	self.error("There should be a space between the name and " +\r
147	"email address: " + email)\r
148	\r
149	if mo.group(3).find(' ') >= 0:\r
150	self.error("The email address cannot contain a space: " +\r
151	mo.group(3))\r
152	\r
153	def check_signed_off_by(self):\r
154	sob='Signed-off-by'\r
155	if self.msg.find(sob) < 0:\r
156	self.error('Missing Signed-off-by! (Note: this must be ' +\r
157	'added by the code contributor!)')\r
158	return\r
159	\r
160	sobs = self.find_signatures('Signed-off')\r
161	\r
162	if len(sobs) == 0:\r
163	self.error('Invalid Signed-off-by format!')\r
164	return\r
165	\r
166	sig_types = (\r
167	'Reviewed',\r
168	'Reported',\r
169	'Tested',\r
170	'Suggested',\r
171	'Acked',\r
172	'Cc'\r
173	)\r
174	\r
175	def check_misc_signatures(self):\r
176	for sig in self.sig_types:\r
177	self.find_signatures(sig)\r
178	\r
179	def check_overall_format(self):\r
180	lines = self.msg.splitlines()\r
181	\r
182	if len(lines) >= 1 and lines[0].endswith('\r\n'):\r
183	empty_line = '\r\n'\r
184	else:\r
185	empty_line = '\n'\r
186	\r
187	lines.insert(0, empty_line)\r
188	lines.insert(0, self.subject + empty_line)\r
189	\r
190	count = len(lines)\r
191	\r
192	if count <= 0:\r
193	self.error('Empty commit message!')\r
194	return\r
195	\r
e6140670	196	if count >= 1 and len(lines[0]) >= 72:\r
a7e173b0 JJ	197	self.error('First line of commit message (subject line) ' +\r
	198	'is too long.')\r
	199	\r
	200	if count >= 1 and len(lines[0].strip()) == 0:\r
	201	self.error('First line of commit message (subject line) ' +\r
	202	'is empty.')\r
	203	\r
	204	if count >= 2 and lines[1].strip() != '':\r
	205	self.error('Second line of commit message should be ' +\r
	206	'empty.')\r
	207	\r
	208	for i in range(2, count):\r
e6140670	209	if (len(lines[i]) >= 76 and\r
a7e173b0 JJ	210	len(lines[i].split()) > 1 and\r
	211	not lines[i].startswith('git-svn-id:')):\r
	212	self.error('Line %d of commit message is too long.' % (i + 1))\r
	213	\r
	214	last_sig_line = None\r
	215	for i in range(count - 1, 0, -1):\r
	216	line = lines[i]\r
	217	mo = self.sig_block_re.match(line)\r
	218	if mo is None:\r
	219	if line.strip() == '':\r
	220	break\r
	221	elif last_sig_line is not None:\r
	222	err2 = 'Add empty line before "%s"?' % last_sig_line\r
	223	self.error('The line before the signature block ' +\r
	224	'should be empty', err2)\r
	225	else:\r
	226	self.error('The signature block was not found')\r
	227	break\r
	228	last_sig_line = line.strip()\r
	229	\r
	230	(START, PRE_PATCH, PATCH) = range(3)\r
	231	\r
	232	class GitDiffCheck:\r
	233	"""Checks the contents of a git diff."""\r
	234	\r
	235	def __init__(self, diff):\r
	236	self.ok = True\r
	237	self.format_ok = True\r
	238	self.lines = diff.splitlines(True)\r
	239	self.count = len(self.lines)\r
	240	self.line_num = 0\r
	241	self.state = START\r
6a69dd49	242	self.new_bin = []\r
a7e173b0 JJ	243	while self.line_num < self.count and self.format_ok:\r
	244	line_num = self.line_num\r
	245	self.run()\r
	246	assert(self.line_num > line_num)\r
	247	self.report_message_result()\r
	248	\r
	249	def report_message_result(self):\r
	250	if Verbose.level < Verbose.NORMAL:\r
	251	return\r
	252	if self.ok:\r
	253	print('The code passed all checks.')\r
6a69dd49 HW	254	if self.new_bin:\r
	255	print('\nWARNING - The following binary files will be added ' +\r
	256	'into the repository:')\r
	257	for binary in self.new_bin:\r
	258	print(' ' + binary)\r
a7e173b0 JJ	259	\r
	260	def run(self):\r
	261	line = self.lines[self.line_num]\r
	262	\r
	263	if self.state in (PRE_PATCH, PATCH):\r
	264	if line.startswith('diff --git'):\r
	265	self.state = START\r
	266	if self.state == PATCH:\r
	267	if line.startswith('@@ '):\r
	268	self.state = PRE_PATCH\r
	269	elif len(line) >= 1 and line[0] not in ' -+' and \\r
aab57eff	270	not line.startswith(r'\ No newline ') and not self.binary:\r
a7e173b0 JJ	271	for line in self.lines[self.line_num + 1:]:\r
	272	if line.startswith('diff --git'):\r
	273	self.format_error('diff found after end of patch')\r
	274	break\r
	275	self.line_num = self.count\r
	276	return\r
	277	\r
	278	if self.state == START:\r
	279	if line.startswith('diff --git'):\r
	280	self.state = PRE_PATCH\r
ccaa7754	281	self.filename = line[13:].split(' ', 1)[0]\r
6a69dd49 HW	282	self.is_newfile = False\r
6a69dd49 HW	283	self.force_crlf = not self.filename.endswith('.sh')\r
a7e173b0 JJ	284	elif len(line.rstrip()) != 0:\r
	285	self.format_error("didn't find diff command")\r
	286	self.line_num += 1\r
	287	elif self.state == PRE_PATCH:\r
a7e173b0 JJ	288	if line.startswith('@@ '):\r
a7e173b0 JJ	289	self.state = PATCH\r
103733f8	290	self.binary = False\r
6a69dd49 HW	291	elif line.startswith('GIT binary patch') or \\r
6a69dd49 HW	292	line.startswith('Binary files'):\r
103733f8 JJ	293	self.state = PATCH\r
103733f8 JJ	294	self.binary = True\r
6a69dd49 HW	295	if self.is_newfile:\r
6a69dd49 HW	296	self.new_bin.append(self.filename)\r
a7e173b0 JJ	297	else:\r
a7e173b0 JJ	298	ok = False\r
6a69dd49	299	self.is_newfile = self.newfile_prefix_re.match(line)\r
a7e173b0 JJ	300	for pfx in self.pre_patch_prefixes:\r
	301	if line.startswith(pfx):\r
	302	ok = True\r
	303	if not ok:\r
	304	self.format_error("didn't find diff hunk marker (@@)")\r
	305	self.line_num += 1\r
	306	elif self.state == PATCH:\r
103733f8 JJ	307	if self.binary:\r
103733f8 JJ	308	pass\r
aab57eff	309	elif line.startswith('-'):\r
a7e173b0 JJ	310	pass\r
	311	elif line.startswith('+'):\r
	312	self.check_added_line(line[1:])\r
	313	elif line.startswith(r'\ No newline '):\r
	314	pass\r
	315	elif not line.startswith(' '):\r
	316	self.format_error("unexpected patch line")\r
	317	self.line_num += 1\r
	318	\r
	319	pre_patch_prefixes = (\r
	320	'--- ',\r
	321	'+++ ',\r
	322	'index ',\r
	323	'new file ',\r
	324	'deleted file ',\r
	325	'old mode ',\r
	326	'new mode ',\r
	327	'similarity index ',\r
	328	'rename ',\r
a7e173b0 JJ	329	)\r
	330	\r
	331	line_endings = ('\r\n', '\n\r', '\n', '\r')\r
	332	\r
6a69dd49 HW	333	newfile_prefix_re = \\r
	334	re.compile(r'''^\r
	335	index\ 0+\.\.\r
	336	''',\r
	337	re.VERBOSE)\r
a7e173b0 JJ	338	\r
	339	def added_line_error(self, msg, line):\r
	340	lines = [ msg ]\r
6a69dd49 HW	341	if self.filename is not None:\r
6a69dd49 HW	342	lines.append('File: ' + self.filename)\r
a7e173b0 JJ	343	lines.append('Line: ' + line)\r
	344	\r
	345	self.error(*lines)\r
	346	\r
c3926cdb YZ	347	old_debug_re = \\r
	348	re.compile(r'''\r
	349	DEBUG \s* \( \s* \( \s*\r
	350	(?: DEBUG_[A-Z_]+ \s* \\| \s)\r
	351	EFI_D_ ([A-Z_]+)\r
	352	''',\r
	353	re.VERBOSE)\r
	354	\r
a7e173b0 JJ	355	def check_added_line(self, line):\r
	356	eol = ''\r
	357	for an_eol in self.line_endings:\r
	358	if line.endswith(an_eol):\r
	359	eol = an_eol\r
	360	line = line[:-len(eol)]\r
	361	\r
	362	stripped = line.rstrip()\r
	363	\r
	364	if self.force_crlf and eol != '\r\n':\r
	365	self.added_line_error('Line ending (%s) is not CRLF' % repr(eol),\r
	366	line)\r
	367	if '\t' in line:\r
	368	self.added_line_error('Tab character used', line)\r
	369	if len(stripped) < len(line):\r
	370	self.added_line_error('Trailing whitespace found', line)\r
	371	\r
c3926cdb YZ	372	mo = self.old_debug_re.search(line)\r
	373	if mo is not None:\r
	374	self.added_line_error('EFI_D_' + mo.group(1) + ' was used, '\r
	375	'but DEBUG_' + mo.group(1) +\r
	376	' is now recommended', line)\r
	377	\r
a7e173b0 JJ	378	split_diff_re = re.compile(r'''\r
	379	(?P<cmd>\r
	380	^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
	381	)\r
	382	(?P<index>\r
	383	^ index \s+ .+ $\r
	384	)\r
	385	''',\r
	386	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
	387	\r
	388	def format_error(self, err):\r
	389	self.format_ok = False\r
	390	err = 'Patch format error: ' + err\r
	391	err2 = 'Line: ' + self.lines[self.line_num].rstrip()\r
	392	self.error(err, err2)\r
	393	\r
	394	def error(self, *err):\r
	395	if self.ok and Verbose.level > Verbose.ONELINE:\r
	396	print('Code format is not valid:')\r
	397	self.ok = False\r
	398	if Verbose.level < Verbose.NORMAL:\r
	399	return\r
	400	count = 0\r
	401	for line in err:\r
	402	prefix = (' *', ' ')[count > 0]\r
	403	print(prefix, line)\r
	404	count += 1\r
	405	\r
	406	class CheckOnePatch:\r
	407	"""Checks the contents of a git email formatted patch.\r
	408	\r
	409	Various checks are performed on both the commit message and the\r
	410	patch content.\r
	411	"""\r
	412	\r
	413	def __init__(self, name, patch):\r
	414	self.patch = patch\r
	415	self.find_patch_pieces()\r
	416	\r
	417	msg_check = CommitMessageCheck(self.commit_subject, self.commit_msg)\r
	418	msg_ok = msg_check.ok\r
	419	\r
	420	diff_ok = True\r
	421	if self.diff is not None:\r
	422	diff_check = GitDiffCheck(self.diff)\r
	423	diff_ok = diff_check.ok\r
	424	\r
	425	self.ok = msg_ok and diff_ok\r
	426	\r
	427	if Verbose.level == Verbose.ONELINE:\r
	428	if self.ok:\r
	429	result = 'ok'\r
	430	else:\r
	431	result = list()\r
	432	if not msg_ok:\r
	433	result.append('commit message')\r
	434	if not diff_ok:\r
	435	result.append('diff content')\r
	436	result = 'bad ' + ' and '.join(result)\r
	437	print(name, result)\r
	438	\r
	439	\r
	440	git_diff_re = re.compile(r'''\r
	441	^ diff \s+ --git \s+ a/.+ \s+ b/.+ $\r
442	''',\r
443	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
444	\r
445	stat_re = \\r
446	re.compile(r'''\r
447	(?P<commit_message> [\s\S\r\n]* )\r
448	(?P<stat>\r
449	^ --- $ [\r\n]+\r
450	(?: ^ \s+ .+ \s+ \\| \s+ \d+ \s+ \+* \-*\r
451	$ [\r\n]+ )+\r
452	[\s\S\r\n]+\r
453	)\r
454	''',\r
455	re.IGNORECASE \| re.VERBOSE \| re.MULTILINE)\r
456	\r
e709bbb1 YZ	457	subject_prefix_re = \\r
	458	re.compile(r'''^\r
	459	\s* (\[\r
	460	[^\[\]]* # Allow all non-brackets\r
	461	\])* \s*\r
	462	''',\r
	463	re.VERBOSE)\r
	464	\r
a7e173b0 JJ	465	def find_patch_pieces(self):\r
	466	if sys.version_info < (3, 0):\r
	467	patch = self.patch.encode('ascii', 'ignore')\r
	468	else:\r
	469	patch = self.patch\r
	470	\r
	471	self.commit_msg = None\r
	472	self.stat = None\r
	473	self.commit_subject = None\r
	474	self.commit_prefix = None\r
	475	self.diff = None\r
	476	\r
	477	if patch.startswith('diff --git'):\r
	478	self.diff = patch\r
	479	return\r
	480	\r
	481	pmail = email.message_from_string(patch)\r
	482	parts = list(pmail.walk())\r
	483	assert(len(parts) == 1)\r
	484	assert(parts[0].get_content_type() == 'text/plain')\r
	485	content = parts[0].get_payload(decode=True).decode('utf-8', 'ignore')\r
	486	\r
	487	mo = self.git_diff_re.search(content)\r
	488	if mo is not None:\r
	489	self.diff = content[mo.start():]\r
	490	content = content[:mo.start()]\r
	491	\r
	492	mo = self.stat_re.search(content)\r
	493	if mo is None:\r
	494	self.commit_msg = content\r
	495	else:\r
	496	self.stat = mo.group('stat')\r
	497	self.commit_msg = mo.group('commit_message')\r
	498	\r
	499	self.commit_subject = pmail['subject'].replace('\r\n', '')\r
	500	self.commit_subject = self.commit_subject.replace('\n', '')\r
e709bbb1	501	self.commit_subject = self.subject_prefix_re.sub('', self.commit_subject, 1)\r
a7e173b0 JJ	502	\r
	503	class CheckGitCommits:\r
	504	"""Reads patches from git based on the specified git revision range.\r
	505	\r
	506	The patches are read from git, and then checked.\r
	507	"""\r
	508	\r
	509	def __init__(self, rev_spec, max_count):\r
	510	commits = self.read_commit_list_from_git(rev_spec, max_count)\r
	511	if len(commits) == 1 and Verbose.level > Verbose.ONELINE:\r
	512	commits = [ rev_spec ]\r
	513	self.ok = True\r
	514	blank_line = False\r
	515	for commit in commits:\r
	516	if Verbose.level > Verbose.ONELINE:\r
	517	if blank_line:\r
	518	print()\r
	519	else:\r
	520	blank_line = True\r
	521	print('Checking git commit:', commit)\r
	522	patch = self.read_patch_from_git(commit)\r
	523	self.ok &= CheckOnePatch(commit, patch).ok\r
5ac4548c JC	524	if not commits:\r
5ac4548c JC	525	print("Couldn't find commit matching: '{}'".format(rev_spec))\r
a7e173b0 JJ	526	\r
	527	def read_commit_list_from_git(self, rev_spec, max_count):\r
	528	# Run git to get the commit patch\r
	529	cmd = [ 'rev-list', '--abbrev-commit', '--no-walk' ]\r
	530	if max_count is not None:\r
	531	cmd.append('--max-count=' + str(max_count))\r
	532	cmd.append(rev_spec)\r
	533	out = self.run_git(*cmd)\r
5ac4548c	534	return out.split() if out else []\r
a7e173b0 JJ	535	\r
	536	def read_patch_from_git(self, commit):\r
	537	# Run git to get the commit patch\r
	538	return self.run_git('show', '--pretty=email', commit)\r
	539	\r
	540	def run_git(self, *args):\r
	541	cmd = [ 'git' ]\r
	542	cmd += args\r
	543	p = subprocess.Popen(cmd,\r
	544	stdout=subprocess.PIPE,\r
	545	stderr=subprocess.STDOUT)\r
5ac4548c	546	Result = p.communicate()\r
6460513a	547	return Result[0].decode('utf-8', 'ignore') if Result[0] and Result[0].find(b"fatal")!=0 else None\r
a7e173b0 JJ	548	\r
	549	class CheckOnePatchFile:\r
	550	"""Performs a patch check for a single file.\r
	551	\r
	552	stdin is used when the filename is '-'.\r
	553	"""\r
	554	\r
	555	def __init__(self, patch_filename):\r
	556	if patch_filename == '-':\r
	557	patch = sys.stdin.read()\r
	558	patch_filename = 'stdin'\r
	559	else:\r
	560	f = open(patch_filename, 'rb')\r
	561	patch = f.read().decode('utf-8', 'ignore')\r
	562	f.close()\r
	563	if Verbose.level > Verbose.ONELINE:\r
	564	print('Checking patch file:', patch_filename)\r
	565	self.ok = CheckOnePatch(patch_filename, patch).ok\r
	566	\r
	567	class CheckOneArg:\r
	568	"""Performs a patch check for a single command line argument.\r
	569	\r
	570	The argument will be handed off to a file or git-commit based\r
	571	checker.\r
	572	"""\r
	573	\r
	574	def __init__(self, param, max_count=None):\r
	575	self.ok = True\r
	576	if param == '-' or os.path.exists(param):\r
	577	checker = CheckOnePatchFile(param)\r
	578	else:\r
	579	checker = CheckGitCommits(param, max_count)\r
	580	self.ok = checker.ok\r
	581	\r
	582	class PatchCheckApp:\r
	583	"""Checks patches based on the command line arguments."""\r
	584	\r
	585	def __init__(self):\r
	586	self.parse_options()\r
	587	patches = self.args.patches\r
	588	\r
	589	if len(patches) == 0:\r
	590	patches = [ 'HEAD' ]\r
	591	\r
	592	self.ok = True\r
	593	self.count = None\r
	594	for patch in patches:\r
	595	self.process_one_arg(patch)\r
	596	\r
	597	if self.count is not None:\r
	598	self.process_one_arg('HEAD')\r
	599	\r
	600	if self.ok:\r
	601	self.retval = 0\r
	602	else:\r
	603	self.retval = -1\r
	604	\r
	605	def process_one_arg(self, arg):\r
	606	if len(arg) >= 2 and arg[0] == '-':\r
	607	try:\r
	608	self.count = int(arg[1:])\r
	609	return\r
	610	except ValueError:\r
	611	pass\r
612	self.ok &= CheckOneArg(arg, self.count).ok\r
613	self.count = None\r
614	\r
615	def parse_options(self):\r
616	parser = argparse.ArgumentParser(description=__copyright__)\r
617	parser.add_argument('--version', action='version',\r
618	version='%(prog)s ' + VersionNumber)\r
619	parser.add_argument('patches', nargs='*',\r
620	help='[patch file \| git rev list]')\r
621	group = parser.add_mutually_exclusive_group()\r
622	group.add_argument("--oneline",\r
623	action="store_true",\r
624	help="Print one result per line")\r
625	group.add_argument("--silent",\r
626	action="store_true",\r
627	help="Print nothing")\r
628	self.args = parser.parse_args()\r
629	if self.args.oneline:\r
630	Verbose.level = Verbose.ONELINE\r
631	if self.args.silent:\r
632	Verbose.level = Verbose.SILENT\r
633	\r
634	if __name__ == "__main__":\r
635	sys.exit(PatchCheckApp().retval)\r