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

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