10 from sqlalchemy
.orm
import join
11 import cStringIO
as StringIO
17 _msg_id_regex
= re
.compile(r
'<([^<>]+)>')
18 def _parse_msg_id(str):
19 m
= _msg_id_regex
.search(str)
23 parser
= email
.Parser
.Parser()
25 def _detect_reply_id(msg
):
26 if msg
['In-Reply-To']:
27 return _parse_msg_id(msg
['In-Reply-To'])
29 refs
= ' '.join(msg
.get_all('References'))
30 ref_ids
= [m
.group(1) for m
in _msg_id_regex
.finditer(refs
)]
33 def _get_text_payloads(msg
):
34 if not msg
.is_multipart():
35 yield msg
.get_payload()
37 for part
in email
.Iterators
.typed_subpart_iterator(msg
):
38 if part
.is_multipart():
39 yield part
.get_payload(0)
41 yield part
.get_payload()
43 _format_patch_regex
= re
.compile('.*^---$.*^diff --git', re
.MULTILINE|re
.DOTALL
)
44 _snip_patch_regex
= re
.compile('.*^-+ ?(?:8<|>8) ?-+\n(.*^diff --git.*)',
45 re
.MULTILINE|re
.DOTALL
)
46 def _guess_patch_contents(msg
):
47 for p
in _get_text_payloads(msg
):
48 if _format_patch_regex
.match(p
):
50 return msg
.as_string()
51 m
= _snip_patch_regex
.match(p
)
53 msg
.set_payload(m
.group(1))
54 return msg
.as_string()
58 blobtracker
= BlobTracker()
60 def try_patch(session
, m
, pp
, base_sha1
):
61 git('checkout', base_sha1
)
64 except patch
.PatchError
:
66 pipe
= git('show', ret_pipe
=True)
67 output
= git('patch-id', input_pipe
=pipe
)[0]
69 # this means the patch had no diff; e.g., a mode change
71 patch_id
, commit_id
= output
.split()
72 c
= session
.query(db
.Commit
).filter(db
.Commit
.sha1
== commit_id
).first()
74 output
= git('log', '-1', '--pretty=format:%ct %at')[0]
75 adate
, cdate
= [int(s
) for s
in output
.split()]
76 c
= db
.Commit(commit_id
, cdate
, adate
, patch_id
, False)
78 blobtracker
.scan_commit(commit_id
)
79 p
= db
.Patch(c
, m
.id, pp
.notes
)
83 def try_patch_anywhere(session
, msg
, m
):
84 print '*', m
.message_id
85 pdata
= _guess_patch_contents(msg
)
88 pp
= patch
.Patch(pdata
)
90 return # probably for another project
91 # first try on the commit given by the blobs
93 for prefix
in pp
.blobs_pre
:
94 ret
= (session
.query(db
.Blob
, db
.Commit
)
95 .filter(db
.Blob
.newest_commit_sha1
== db
.Commit
.sha1
)
96 .filter(db
.Blob
.sha1
.like(prefix
+'%'))
97 .filter(db
.Commit
.upstream
== True)
98 .order_by(db
.Commit
.cdate
.desc()).first())
100 print 'blob %s not found?!' % prefix
102 commits
.append(ret
[1])
107 print 'trying canonical commit %s' % cmt
.sha1
108 applied
= try_patch(session
, m
, pp
, cmt
.sha1
)
111 # this is just hopeless: it doesn't apply to the commit it should!
114 print "no canonical commit found"
115 # if we have a parent, try on the parent
116 parent
= session
.query(db
.Mail
).filter(db
.Mail
.message_id
==m
.in_reply_to
).first()
117 if parent
and parent
.has_patch
and parent
.patch_id
:
118 cmt
= (session
.query(db
.Commit
)
119 .filter(db
.Commit
.patch_id
==parent
.patch_id
)
120 .order_by(db
.Commit
.cdate
.desc()).first())
121 print 'trying to apply on parent %s' % cmt
.sha1
122 applied
= try_patch(session
, m
, pp
, cmt
.sha1
)
126 print "no parent commit found"
127 # try on origin/master
128 print 'trying on origin/master'
129 applied
= try_patch(session
, m
, pp
, 'origin/master')
132 # same for origin/next
133 print 'trying on origin/next'
134 applied
= try_patch(session
, m
, pp
, 'origin/next')
139 _whats_cooking_subject
= re
.compile(r
"^What's cooking in git\.git")
140 _whats_cooking_category
= re
.compile(r
"^\[(.*)\]$")
141 _whats_cooking_header
= re
.compile(r
"\* (../[a-zA-Z0-9-]+) \([^)]*\) \d+ commits?")
142 _whats_cooking_separator
= re
.compile(r
"^(-{5,}|-- )$")
144 def parse_whats_cooking(session
, msg
, mail
):
145 if not (msg
["Subject"] and _whats_cooking_subject
.match(msg
["Subject"])):
148 branch
= 'pu' # initial part goes on 'pu'
150 def _rotate_notes(category
, branch
, notes
):
152 t
= session
.query(db
.Topic
).filter(db
.Topic
.name
==branch
).first()
158 t
.cooking_notes
= '\n'.join(notes
)
161 notes
.append("[%s]" % category
)
163 for line
in _get_text_payload(msg
).splitlines():
164 if _whats_cooking_separator
.match(line
):
166 notes
= _rotate_notes(category
, branch
, notes
)
169 m
= _whats_cooking_category
.match(line
)
171 category
= m
.group(1)
172 notes
= _rotate_notes(category
, branch
, notes
)
174 m
= _whats_cooking_header
.match(line
)
176 notes
= _rotate_notes(category
, branch
, notes
)
183 def process_mail(session
, mail
):
184 msg
= parser
.parse(StringIO
.StringIO(mail
.data
))
185 parse_whats_cooking(session
, msg
, mail
)
186 patch
= try_patch_anywhere(session
, msg
, mail
)
188 mail
.patch_id
= patch
.commit
.patch_id
192 def _query_stale_mail(session
):
193 return (session
.query(db
.Mail
)
194 .filter(db
.Mail
.has_patch
== True)
195 .filter(db
.Mail
.stale
== True)
196 .order_by(db
.Mail
.post_date
.asc(), db
.Mail
.subject
.asc()))
198 def walk_stale_mail():
199 session
= db
.Session()
200 count
= _query_stale_mail(session
).count()
201 for mail
in _query_stale_mail(session
):
202 process_mail(session
, mail
)
204 sys
.stderr
.write("%6d\r" % count
)
205 sys
.stderr
.write("\n")
208 def get_mail_by_id(session
, msg_id
):
209 # Note: use first() because we don't know it exists. The DB
210 # guarantees uniqueness anyway.
211 return (session
.query(db
.Mail
)
212 .filter(db
.Mail
.message_id
== msg_id
)
215 _gmane_id_regex
= re
.compile(r
'<http://permalink\.gmane\.org/gmane\.comp\.version-control\.git/(\d+)>')
216 def insert_mail_into_db(msg
):
217 session
= db
.Session()
218 if (msg
.get('Message-Id', None)
219 and get_mail_by_id(session
, _parse_msg_id(msg
['Message-Id']))):
220 return # already exists
222 if msg
['Archived-At']:
223 m
= _gmane_id_regex
.match(msg
['Archived-At'])
225 mail
.gmane_id
= int(m
.group(1))
226 msgid
= msg
.get('Message-Id', None)
227 if not msgid
or not _parse_msg_id(msgid
):
229 msgid
= 'gmane-%d@mailnotes.trast.ch' % mail
.gmane_id
231 msgid
= 'fallback-%X@mailnotes.trast.ch' % random
.randrange(2**32)
233 msgid
= _parse_msg_id(msgid
)
234 mail
.message_id
= msgid
236 mail
.author
= msg
['From'][:255]
239 tm
= email
.utils
.parsedate(msg
['Date'])
246 mail
.subject
= msg
['Subject'][:255]
247 mail
.in_reply_to
= _detect_reply_id(msg
)
248 mail
.data
= msg
.as_string()
249 mail
.stale
= mail
.has_patch
= bool(_guess_patch_contents(msg
))
253 parent
= get_mail_by_id(session
, mail
.in_reply_to
)
256 # Flag all so-far unapplied patches downwards of this one as
257 # 'stale' so they'll be tried again. XXX should use an sql UPDATE
259 for child
in (session
.query(db
.Mail
)
260 .select_from(join(db
.Mail
, db
.Reference
,
261 db
.Mail
.id == db
.Reference
.mail_id
))
262 .filter(db
.Reference
.reference_id
== starter
.message_id
)
263 .filter(db
.Mail
.has_patch
== True)
264 .filter(db
.Mail
.patch_id
== None)
265 .filter(db
.Mail
.message_id
!= mail
.message_id
)
266 .filter(db
.Mail
.stale
== False)):
269 if msg
['References']:
270 for m
in _msg_id_regex
.finditer(' '.join(msg
.get_all('References'))):
271 session
.add(db
.Reference(mail
.id, m
.group(1)))
274 def parse_mbox(fname
):
275 session
= db
.Session()
276 mbox
= mailbox
.mbox(fname
, parser
.parse
)
277 mbox_parsed
= list(mbox
)
278 count
= len(mbox_parsed
)
279 for msg
in mbox_parsed
:
280 insert_mail_into_db(msg
)
282 sys
.stderr
.write("%6d\r" % count
)
283 sys
.stderr
.write("\n")
285 if __name__
== '__main__':
286 for mbox
in sys
.argv
[1:]: