tools/patman/patchstream.py

   1 # Copyright (c) 2011 The Chromium OS Authors.
   2 #
   3 # SPDX-License-Identifier:      GPL-2.0+
   4 #
   5
   6 import os
   7 import re
   8 import shutil
   9 import tempfile
  10
  11 import command
  12 import commit
  13 import gitutil
  14 from series import Series
  15
  16 # Tags that we detect and remove
  17 re_remove = re.compile('^BUG=|^TEST=|^BRANCH=|^Change-Id:|^Review URL:'
  18     '|Reviewed-on:|Commit-\w*:')
  19
  20 # Lines which are allowed after a TEST= line
  21 re_allowed_after_test = re.compile('^Signed-off-by:')
  22
  23 # Signoffs
  24 re_signoff = re.compile('^Signed-off-by: *(.*)')
  25
  26 # The start of the cover letter
  27 re_cover = re.compile('^Cover-letter:')
  28
  29 # A cover letter Cc
  30 re_cover_cc = re.compile('^Cover-letter-cc: *(.*)')
  31
  32 # Patch series tag
  33 re_series_tag = re.compile('^Series-([a-z-]*): *(.*)')
  34
  35 # Commit series tag
  36 re_commit_tag = re.compile('^Commit-([a-z-]*): *(.*)')
  37
  38 # Commit tags that we want to collect and keep
  39 re_tag = re.compile('^(Tested-by|Acked-by|Reviewed-by|Patch-cc): (.*)')
  40
  41 # The start of a new commit in the git log
  42 re_commit = re.compile('^commit ([0-9a-f]*)$')
  43
  44 # We detect these since checkpatch doesn't always do it
  45 re_space_before_tab = re.compile('^[+].* \t')
  46
  47 # States we can be in - can we use range() and still have comments?
  48 STATE_MSG_HEADER = 0        # Still in the message header
  49 STATE_PATCH_SUBJECT = 1     # In patch subject (first line of log for a commit)
  50 STATE_PATCH_HEADER = 2      # In patch header (after the subject)
  51 STATE_DIFFS = 3             # In the diff part (past --- line)
  52
  53 class PatchStream:
  54     """Class for detecting/injecting tags in a patch or series of patches
  55
  56     We support processing the output of 'git log' to read out the tags we
  57     are interested in. We can also process a patch file in order to remove
  58     unwanted tags or inject additional ones. These correspond to the two
  59     phases of processing.
  60     """
  61     def __init__(self, series, name=None, is_log=False):
  62         self.skip_blank = False          # True to skip a single blank line
  63         self.found_test = False          # Found a TEST= line
  64         self.lines_after_test = 0        # MNumber of lines found after TEST=
  65         self.warn = []                   # List of warnings we have collected
  66         self.linenum = 1                 # Output line number we are up to
  67         self.in_section = None           # Name of start...END section we are in
  68         self.notes = []                  # Series notes
  69         self.section = []                # The current section...END section
  70         self.series = series             # Info about the patch series
  71         self.is_log = is_log             # True if indent like git log
  72         self.in_change = 0               # Non-zero if we are in a change list
  73         self.blank_count = 0             # Number of blank lines stored up
  74         self.state = STATE_MSG_HEADER    # What state are we in?
  75         self.tags = []                   # Tags collected, like Tested-by...
  76         self.signoff = []                # Contents of signoff line
  77         self.commit = None               # Current commit
  78
  79     def AddToSeries(self, line, name, value):
  80         """Add a new Series-xxx tag.
  81
  82         When a Series-xxx tag is detected, we come here to record it, if we
  83         are scanning a 'git log'.
  84
  85         Args:
  86             line: Source line containing tag (useful for debug/error messages)
  87             name: Tag name (part after 'Series-')
  88             value: Tag value (part after 'Series-xxx: ')
  89         """
  90         if name == 'notes':
  91             self.in_section = name
  92             self.skip_blank = False
  93         if self.is_log:
  94             self.series.AddTag(self.commit, line, name, value)
  95
  96     def AddToCommit(self, line, name, value):
  97         """Add a new Commit-xxx tag.
  98
  99         When a Commit-xxx tag is detected, we come here to record it.
 100
 101         Args:
 102             line: Source line containing tag (useful for debug/error messages)
 103             name: Tag name (part after 'Commit-')
 104             value: Tag value (part after 'Commit-xxx: ')
 105         """
 106         if name == 'notes':
 107             self.in_section = 'commit-' + name
 108             self.skip_blank = False
 109
 110     def CloseCommit(self):
 111         """Save the current commit into our commit list, and reset our state"""
 112         if self.commit and self.is_log:
 113             self.series.AddCommit(self.commit)
 114             self.commit = None
 115
 116     def FormatTags(self, tags):
 117         out_list = []
 118         for tag in sorted(tags):
 119             if tag.startswith('Cc:'):
 120                 tag_list = tag[4:].split(',')
 121                 out_list += gitutil.BuildEmailList(tag_list, 'Cc:')
 122             else:
 123                 out_list.append(tag)
 124         return out_list
 125
 126     def ProcessLine(self, line):
 127         """Process a single line of a patch file or commit log
 128
 129         This process a line and returns a list of lines to output. The list
 130         may be empty or may contain multiple output lines.
 131
 132         This is where all the complicated logic is located. The class's
 133         state is used to move between different states and detect things
 134         properly.
 135
 136         We can be in one of two modes:
 137             self.is_log == True: This is 'git log' mode, where most output is
 138                 indented by 4 characters and we are scanning for tags
 139
 140             self.is_log == False: This is 'patch' mode, where we already have
 141                 all the tags, and are processing patches to remove junk we
 142                 don't want, and add things we think are required.
 143
 144         Args:
 145             line: text line to process
 146
 147         Returns:
 148             list of output lines, or [] if nothing should be output
 149         """
 150         # Initially we have no output. Prepare the input line string
 151         out = []
 152         line = line.rstrip('\n')
 153         if self.is_log:
 154             if line[:4] == '    ':
 155                 line = line[4:]
 156
 157         # Handle state transition and skipping blank lines
 158         series_tag_match = re_series_tag.match(line)
 159         commit_tag_match = re_commit_tag.match(line)
 160         commit_match = re_commit.match(line) if self.is_log else None
 161         cover_cc_match = re_cover_cc.match(line)
 162         signoff_match = re_signoff.match(line)
 163         tag_match = None
 164         if self.state == STATE_PATCH_HEADER:
 165             tag_match = re_tag.match(line)
 166         is_blank = not line.strip()
 167         if is_blank:
 168             if (self.state == STATE_MSG_HEADER
 169                     or self.state == STATE_PATCH_SUBJECT):
 170                 self.state += 1
 171
 172             # We don't have a subject in the text stream of patch files
 173             # It has its own line with a Subject: tag
 174             if not self.is_log and self.state == STATE_PATCH_SUBJECT:
 175                 self.state += 1
 176         elif commit_match:
 177             self.state = STATE_MSG_HEADER
 178
 179         # If we are in a section, keep collecting lines until we see END
 180         if self.in_section:
 181             if line == 'END':
 182                 if self.in_section == 'cover':
 183                     self.series.cover = self.section
 184                 elif self.in_section == 'notes':
 185                     if self.is_log:
 186                         self.series.notes += self.section
 187                 elif self.in_section == 'commit-notes':
 188                     if self.is_log:
 189                         self.commit.notes += self.section
 190                 else:
 191                     self.warn.append("Unknown section '%s'" % self.in_section)
 192                 self.in_section = None
 193                 self.skip_blank = True
 194                 self.section = []
 195             else:
 196                 self.section.append(line)
 197
 198         # Detect the commit subject
 199         elif not is_blank and self.state == STATE_PATCH_SUBJECT:
 200             self.commit.subject = line
 201
 202         # Detect the tags we want to remove, and skip blank lines
 203         elif re_remove.match(line) and not commit_tag_match:
 204             self.skip_blank = True
 205
 206             # TEST= should be the last thing in the commit, so remove
 207             # everything after it
 208             if line.startswith('TEST='):
 209                 self.found_test = True
 210         elif self.skip_blank and is_blank:
 211             self.skip_blank = False
 212
 213         # Detect the start of a cover letter section
 214         elif re_cover.match(line):
 215             self.in_section = 'cover'
 216             self.skip_blank = False
 217
 218         elif cover_cc_match:
 219             value = cover_cc_match.group(1)
 220             self.AddToSeries(line, 'cover-cc', value)
 221
 222         # If we are in a change list, key collected lines until a blank one
 223         elif self.in_change:
 224             if is_blank:
 225                 # Blank line ends this change list
 226                 self.in_change = 0
 227             elif line == '---':
 228                 self.in_change = 0
 229                 out = self.ProcessLine(line)
 230             else:
 231                 if self.is_log:
 232                     self.series.AddChange(self.in_change, self.commit, line)
 233             self.skip_blank = False
 234
 235         # Detect Series-xxx tags
 236         elif series_tag_match:
 237             name = series_tag_match.group(1)
 238             value = series_tag_match.group(2)
 239             if name == 'changes':
 240                 # value is the version number: e.g. 1, or 2
 241                 try:
 242                     value = int(value)
 243                 except ValueError as str:
 244                     raise ValueError("%s: Cannot decode version info '%s'" %
 245                         (self.commit.hash, line))
 246                 self.in_change = int(value)
 247             else:
 248                 self.AddToSeries(line, name, value)
 249                 self.skip_blank = True
 250
 251         # Detect Commit-xxx tags
 252         elif commit_tag_match:
 253             name = commit_tag_match.group(1)
 254             value = commit_tag_match.group(2)
 255             if name == 'notes':
 256                 self.AddToCommit(line, name, value)
 257                 self.skip_blank = True
 258
 259         # Detect the start of a new commit
 260         elif commit_match:
 261             self.CloseCommit()
 262             # TODO: We should store the whole hash, and just display a subset
 263             self.commit = commit.Commit(commit_match.group(1)[:8])
 264
 265         # Detect tags in the commit message
 266         elif tag_match:
 267             # Remove Tested-by self, since few will take much notice
 268             if (tag_match.group(1) == 'Tested-by' and
 269                     tag_match.group(2).find(os.getenv('USER') + '@') != -1):
 270                 self.warn.append("Ignoring %s" % line)
 271             elif tag_match.group(1) == 'Patch-cc':
 272                 self.commit.AddCc(tag_match.group(2).split(','))
 273             else:
 274                 self.tags.append(line);
 275
 276         # Suppress duplicate signoffs
 277         elif signoff_match:
 278             if (self.is_log or
 279                 self.commit.CheckDuplicateSignoff(signoff_match.group(1))):
 280                 out = [line]
 281
 282         # Well that means this is an ordinary line
 283         else:
 284             pos = 1
 285             # Look for ugly ASCII characters
 286             for ch in line:
 287                 # TODO: Would be nicer to report source filename and line
 288                 if ord(ch) > 0x80:
 289                     self.warn.append("Line %d/%d ('%s') has funny ascii char" %
 290                         (self.linenum, pos, line))
 291                 pos += 1
 292
 293             # Look for space before tab
 294             m = re_space_before_tab.match(line)
 295             if m:
 296                 self.warn.append('Line %d/%d has space before tab' %
 297                     (self.linenum, m.start()))
 298
 299             # OK, we have a valid non-blank line
 300             out = [line]
 301             self.linenum += 1
 302             self.skip_blank = False
 303             if self.state == STATE_DIFFS:
 304                 pass
 305
 306             # If this is the start of the diffs section, emit our tags and
 307             # change log
 308             elif line == '---':
 309                 self.state = STATE_DIFFS
 310
 311                 # Output the tags (signeoff first), then change list
 312                 out = []
 313                 log = self.series.MakeChangeLog(self.commit)
 314                 out += self.FormatTags(self.tags)
 315                 out += [line] + self.commit.notes + [''] + log
 316             elif self.found_test:
 317                 if not re_allowed_after_test.match(line):
 318                     self.lines_after_test += 1
 319
 320         return out
 321
 322     def Finalize(self):
 323         """Close out processing of this patch stream"""
 324         self.CloseCommit()
 325         if self.lines_after_test:
 326             self.warn.append('Found %d lines after TEST=' %
 327                     self.lines_after_test)
 328
 329     def ProcessStream(self, infd, outfd):
 330         """Copy a stream from infd to outfd, filtering out unwanting things.
 331
 332         This is used to process patch files one at a time.
 333
 334         Args:
 335             infd: Input stream file object
 336             outfd: Output stream file object
 337         """
 338         # Extract the filename from each diff, for nice warnings
 339         fname = None
 340         last_fname = None
 341         re_fname = re.compile('diff --git a/(.*) b/.*')
 342         while True:
 343             line = infd.readline()
 344             if not line:
 345                 break
 346             out = self.ProcessLine(line)
 347
 348             # Try to detect blank lines at EOF
 349             for line in out:
 350                 match = re_fname.match(line)
 351                 if match:
 352                     last_fname = fname
 353                     fname = match.group(1)
 354                 if line == '+':
 355                     self.blank_count += 1
 356                 else:
 357                     if self.blank_count and (line == '-- ' or match):
 358                         self.warn.append("Found possible blank line(s) at "
 359                                 "end of file '%s'" % last_fname)
 360                     outfd.write('+\n' * self.blank_count)
 361                     outfd.write(line + '\n')
 362                     self.blank_count = 0
 363         self.Finalize()
 364
 365
 366 def GetMetaDataForList(commit_range, git_dir=None, count=None,
 367                        series = Series()):
 368     """Reads out patch series metadata from the commits
 369
 370     This does a 'git log' on the relevant commits and pulls out the tags we
 371     are interested in.
 372
 373     Args:
 374         commit_range: Range of commits to count (e.g. 'HEAD..base')
 375         git_dir: Path to git repositiory (None to use default)
 376         count: Number of commits to list, or None for no limit
 377         series: Series object to add information into. By default a new series
 378             is started.
 379     Returns:
 380         A Series object containing information about the commits.
 381     """
 382     params = gitutil.LogCmd(commit_range,reverse=True, count=count,
 383                             git_dir=git_dir)
 384     stdout = command.RunPipe([params], capture=True).stdout
 385     ps = PatchStream(series, is_log=True)
 386     for line in stdout.splitlines():
 387         ps.ProcessLine(line)
 388     ps.Finalize()
 389     return series
 390
 391 def GetMetaData(start, count):
 392     """Reads out patch series metadata from the commits
 393
 394     This does a 'git log' on the relevant commits and pulls out the tags we
 395     are interested in.
 396
 397     Args:
 398         start: Commit to start from: 0=HEAD, 1=next one, etc.
 399         count: Number of commits to list
 400     """
 401     return GetMetaDataForList('HEAD~%d' % start, None, count)
 402
 403 def FixPatch(backup_dir, fname, series, commit):
 404     """Fix up a patch file, by adding/removing as required.
 405
 406     We remove our tags from the patch file, insert changes lists, etc.
 407     The patch file is processed in place, and overwritten.
 408
 409     A backup file is put into backup_dir (if not None).
 410
 411     Args:
 412         fname: Filename to patch file to process
 413         series: Series information about this patch set
 414         commit: Commit object for this patch file
 415     Return:
 416         A list of errors, or [] if all ok.
 417     """
 418     handle, tmpname = tempfile.mkstemp()
 419     outfd = os.fdopen(handle, 'w')
 420     infd = open(fname, 'r')
 421     ps = PatchStream(series)
 422     ps.commit = commit
 423     ps.ProcessStream(infd, outfd)
 424     infd.close()
 425     outfd.close()
 426
 427     # Create a backup file if required
 428     if backup_dir:
 429         shutil.copy(fname, os.path.join(backup_dir, os.path.basename(fname)))
 430     shutil.move(tmpname, fname)
 431     return ps.warn
 432
 433 def FixPatches(series, fnames):
 434     """Fix up a list of patches identified by filenames
 435
 436     The patch files are processed in place, and overwritten.
 437
 438     Args:
 439         series: The series object
 440         fnames: List of patch files to process
 441     """
 442     # Current workflow creates patches, so we shouldn't need a backup
 443     backup_dir = None  #tempfile.mkdtemp('clean-patch')
 444     count = 0
 445     for fname in fnames:
 446         commit = series.commits[count]
 447         commit.patch = fname
 448         result = FixPatch(backup_dir, fname, series, commit)
 449         if result:
 450             print '%d warnings for %s:' % (len(result), fname)
 451             for warn in result:
 452                 print '\t', warn
 453             print
 454         count += 1
 455     print 'Cleaned %d patches' % count
 456     return series
 457
 458 def InsertCoverLetter(fname, series, count):
 459     """Inserts a cover letter with the required info into patch 0
 460
 461     Args:
 462         fname: Input / output filename of the cover letter file
 463         series: Series object
 464         count: Number of patches in the series
 465     """
 466     fd = open(fname, 'r')
 467     lines = fd.readlines()
 468     fd.close()
 469
 470     fd = open(fname, 'w')
 471     text = series.cover
 472     prefix = series.GetPatchPrefix()
 473     for line in lines:
 474         if line.startswith('Subject:'):
 475             # TODO: if more than 10 patches this should save 00/xx, not 0/xx
 476             line = 'Subject: [%s 0/%d] %s\n' % (prefix, count, text[0])
 477
 478         # Insert our cover letter
 479         elif line.startswith('*** BLURB HERE ***'):
 480             # First the blurb test
 481             line = '\n'.join(text[1:]) + '\n'
 482             if series.get('notes'):
 483                 line += '\n'.join(series.notes) + '\n'
 484
 485             # Now the change list
 486             out = series.MakeChangeLog(None)
 487             line += '\n' + '\n'.join(out)
 488         fd.write(line)
 489     fd.close()