notes.py

   1 #!/usr/bin/python
   2
   3 import os
   4 import sys
   5 import time
   6 import re
   7 from collections import defaultdict
   8 from sqlalchemy.sql import and_
   9 from sqlalchemy.orm import join
  10
  11 import db
  12 from git import git
  13
  14 _eol_space_re = re.compile('[ \t]+$', re.MULTILINE)
  15
  16 class Notes(object):
  17
  18     def __init__(self, refname, indexname):
  19         self._indexname = indexname
  20         self._env = { 'GIT_INDEX_FILE': '.git/'+indexname }
  21         self._ref = refname
  22         self._reset()
  23
  24     def _reset(self):
  25         self._cache = defaultdict(list)
  26         self._index = {}
  27
  28     def extend(self, sha1, seq):
  29         for line in seq:
  30             self.append_line(sha1, line)
  31
  32     def append_line(self, sha1, line):
  33         self._cache[sha1].append(line)
  34
  35     def __contains__(self, sha1):
  36         return sha1 in self._cache
  37
  38     def flush(self):
  39         try:
  40             os.unlink(self._indexname)
  41         except OSError:
  42             pass
  43         count = len(self._cache)
  44         input = []
  45         for cmt_sha1 in self._cache.iterkeys():
  46             count = count - 1
  47             sys.stdout.write('%6d\r' % count)
  48             sys.stdout.flush()
  49             notes = ''.join(self._cache[cmt_sha1]).strip('\n')
  50             notes = _eol_space_re.sub('', notes) + '\n'
  51             blob_sha1 = git('hash-object', '-w', '--stdin', input=notes)[0].strip()
  52             input.append("100644 %s\t%s\n" % (blob_sha1, cmt_sha1))
  53         sys.stdout.write('\n')
  54         git('update-index', '--index-info', input=''.join(input), env=self._env)
  55         previous, ret = git('rev-parse', self._ref)
  56         if ret != 0:
  57             args = []
  58             previous_arg = []
  59         else:
  60             args = ['-p', previous.strip()]
  61             previous_arg = [previous.strip()]
  62         tree_sha1 = git('write-tree', env=self._env)[0].strip()
  63         head_sha1 = git('commit-tree', tree_sha1, *args,
  64                         **{'input':'Mass annotation by notes.py'})[0].strip()
  65         git('update-ref', '-m', 'Mass annotation by notes.py',
  66             self._ref, head_sha1, *previous_arg)
  67         self._reset()
  68
  69 def split_and_tab(buf):
  70     ret = []
  71     for line in str(buf).splitlines():
  72         ret.append('\t%s\n' % line)
  73     return ret
  74
  75 def compute_notes(commit, mail, guess_by=None):
  76     full = []
  77     terse = []
  78     if mail.author:
  79         full.append('From: %s\n' % mail.author)
  80     if mail.subject:
  81         full.append('Subject: %s\n' % mail.subject)
  82     full.append('Message-Id: <%s>\n' % mail.message_id)
  83     terse.append('Message-Id: <%s>\n' % mail.message_id)
  84     full.append('Date: %s\n'
  85                  % time.strftime("%c", time.localtime(mail.post_date)))
  86     if mail.in_reply_to:
  87         full.append('In-Reply-To: <%s>\n' % mail.in_reply_to)
  88     if mail.gmane_id:
  89         full.append('Archived-At: <http://permalink.gmane.org/gmane.comp.version-control.git/%d>\n' % mail.gmane_id)
  90         terse.append('Archived-At: <http://permalink.gmane.org/gmane.comp.version-control.git/%d>\n' % mail.gmane_id)
  91     if len(mail.patch)>0 and mail.patch[0].extra_notes:
  92         full.append('Extra-Notes:\n')
  93         full.extend(split_and_tab(mail.patch[0].extra_notes))
  94     full.append('\n')
  95     terse.append('\n')
  96     return full, terse
  97
  98 _merge_re = re.compile("^([a-f0-9]{40}) Merge branch '([^']+)'")
  99 def _redo_pu(full, terse):
 100     pu_ref = git('rev-parse', 'origin/pu')[0].strip()
 101     pu_topic = db.session.query(db.Topic).filter(db.Topic.name == 'pu').first()
 102     if pu_topic:
 103         full.append_line(pu_ref, 'Pu-Overview:\n')
 104         full.extend(pu_ref, split_and_tab(pu_topic.cooking_notes))
 105         full.append_line(pu_ref, '\n')
 106     for line in git('log', '--first-parent', '--pretty=tformat:%H %s',
 107                     'origin/master..origin/pu', ret_pipe=True):
 108         m = _merge_re.match(line)
 109         if not m:
 110             continue
 111         sha1 = m.group(1)
 112         branch = m.group(2)
 113         t = db.session.query(db.Topic).filter(db.Topic.name == branch).first()
 114         if t and t.cooking_notes:
 115             full.append_line(sha1, 'Pu-Topic:\n')
 116             full.extend(sha1, split_and_tab(t.cooking_notes))
 117             full.append_line(sha1, '\n')
 118
 119 def _redo_for_query(full, terse, query, skip_if_exists=False, guess_by=None):
 120     count = 0
 121     for cmt, mail in query:
 122         if skip_if_exists and cmt.sha1 in full:
 123             continue
 124         f, t = compute_notes(cmt, mail, guess_by=guess_by)
 125         if not (f or t):
 126             continue
 127         if f:
 128             full.extend(cmt.sha1, f)
 129         if t:
 130             terse.extend(cmt.sha1, t)
 131         sys.stdout.write('\r%6d' % count)
 132         sys.stdout.flush()
 133         count = count + 1
 134     sys.stdout.write('\n')
 135
 136 def _redo_patches(full, terse):
 137     # matching by patch-id
 138     query = (db.session.query(db.Commit, db.Mail)
 139              .select_from(join(db.Mail, db.Commit,
 140                                db.Mail.patch_id==db.Commit.patch_id))
 141              .filter(db.Commit.upstream==True)
 142              .filter(db.Mail.has_patch==True)
 143              .order_by(db.Mail.post_date))
 144     _redo_for_query(full, terse, query)
 145     # matching by author/date
 146     query = (db.session.query(db.Commit, db.Mail)
 147              .select_from(join(db.Mail, db.Commit,
 148                                and_(db.Mail.author==db.Commit.author,
 149                                     db.Mail.post_date==db.Commit.adate)))
 150              .filter(db.Commit.upstream==True)
 151              .filter(db.Mail.has_patch==True))
 152     _redo_for_query(full, terse, query, True, guess_by='author,date')
 153
 154 def _redo_all():
 155     count = 0
 156     full = Notes('refs/heads/notes/full', 'git-full-notes-index')
 157     terse = Notes('refs/heads/notes/terse', 'git-terse-notes-index')
 158     _redo_patches(full, terse)
 159     _redo_pu(full, terse)
 160     full.flush()
 161     terse.flush()
 162
 163 if __name__ == '__main__':
 164     if len(sys.argv) > 1:
 165         cmt = db.session.query(db.Commit).filter(db.Commit.sha1==sys.argv[1]).one()
 166         print compute_notes(cmt)
 167     else:
 168         _redo_all()