2 # Copyright (c) 2012 Eric S. Raymond <esr@thyrsus.com>
3 # Distributed under BSD terms.
5 # This script contains git porcelain and porcelain byproducts.
6 # Requires Python 2.6, or 2.5 with the simplejson library installed.
8 # usage: irkerhook.py [-V] [-n] [--variable=value...] [commit_id...]
10 # This script is meant to be run in an update or post-commit hook.
11 # Try it with -n to see the notification dumped to stdout and verify
12 # that it looks sane. With -V this script dumps its version and exits.
14 # See the irkerhook manual page in the distribution for a detailed
15 # explanation of how to configure this hook.
17 # The default location of the irker proxy, if the project configuration
18 # does not override it.
19 default_server = "localhost"
22 # The default service used to turn your web-view URL into a tinyurl so it
23 # will take up less space on the IRC notification line.
24 default_tinyifier = "http://tinyurl.com/api-create.php?url="
26 # Map magic urlprefix values to actual URL prefixes.
28 "viewcvs": "http://%(host)s/viewcvs/%(repo)s?view=revision&revision=",
29 "gitweb": "http://%(host)s/cgi-bin/gitweb.cgi?p=%(repo)s;a=commit;h=",
30 "cgit": "http://%(host)s/cgi-bin/cgit.cgi/%(repo)s/commit/?id=",
33 # By default, ship to the freenode #commits list
34 default_channels = "irc://chat.freenode.net/#commits"
37 # No user-serviceable parts below this line:
42 import os, sys, commands, socket, urllib, subprocess, locale, datetime, re
43 from pipes import quote as shellquote
45 import simplejson as json # Faster, also makes us Python-2.5-compatible
50 return unicode(commands.getstatusoutput(command)[1], locale.getlocale()[1] or 'UTF-8').encode(locale.getlocale()[1] or 'UTF-8')
53 def __init__(self, extractor, commit):
63 self.author_date = None
64 self.commit_date = None
65 self.__dict__.update(extractor.__dict__)
66 def __unicode__(self):
67 "Produce a notification string from this commit."
68 if self.urlprefix.lower() == "none":
71 urlprefix = urlprefixmap.get(self.urlprefix, self.urlprefix)
72 webview = (urlprefix % self.__dict__) + self.commit
74 if urllib.urlopen(webview).getcode() == 404:
76 if self.tinyifier and self.tinyifier.lower() != "none":
78 # Didn't get a retrieval error or 404 on the web
79 # view, so try to tinyify a reference to it.
80 self.url = open(urllib.urlretrieve(self.tinyifier + webview)[0]).read()
82 self.url = self.url.decode('UTF-8')
91 res = self.template % self.__dict__
92 return unicode(res, 'UTF-8') if not isinstance(res, unicode) else res
94 class GenericExtractor:
95 "Generic class for encapsulating data from a VCS."
97 numerics = ["maxchannels"]
99 def __init__(self, arguments):
100 self.arguments = arguments
103 # These aren't really repo data but they belong here anyway...
106 self.tinyifier = default_tinyifier
111 self.urlprefix = None
112 self.host = socket.getfqdn()
114 self.filtercmd = None
115 # Color highlighting is disabled by default.
117 self.bold = self.green = self.blue = self.yellow = ""
118 self.brown = self.magenta = self.cyan = self.reset = ""
119 def activate_color(self, style):
122 # mIRC colors are mapped as closely to the ANSI colors as
123 # possible. However, bright colors (green, blue, red,
124 # yellow) have been made their dark counterparts since
125 # ChatZilla does not properly darken mIRC colors in the
126 # Light Motif color scheme.
128 self.green = '\x0303'
131 self.yellow = '\x0307'
132 self.brown = '\x0305'
133 self.magenta = '\x0306'
137 self.bold = '\x1b[1m'
138 self.green = '\x1b[1;32m'
139 self.blue = '\x1b[1;34m'
140 self.red = '\x1b[1;31m'
141 self.yellow = '\x1b[1;33m'
142 self.brown = '\x1b[33m'
143 self.magenta = '\x1b[35m'
144 self.cyan = '\x1b[36m'
145 self.reset = '\x1b[0m'
146 def load_preferences(self, conf):
147 "Load preferences from a file in the repository root."
148 if not os.path.exists(conf):
151 for line in open(conf):
153 if line.startswith("#") or not line.strip():
155 elif line.count('=') != 1:
156 sys.stderr.write('"%s", line %d: missing = in config line\n' \
159 fields = line.split('=')
161 sys.stderr.write('"%s", line %d: too many fields in config line\n' \
164 variable = fields[0].strip()
165 value = fields[1].strip()
166 if value.lower() == "true":
168 elif value.lower() == "false":
170 # User cannot set maxchannels - only a command-line arg can do that.
171 if variable == "maxchannels":
173 setattr(self, variable, value)
174 def do_overrides(self):
175 "Make command-line overrides possible."
176 for tok in self.arguments:
177 for key in self.__dict__:
178 if tok.startswith("--" + key + "="):
179 val = tok[len(key)+3:]
180 setattr(self, key, val)
181 for (key, val) in self.__dict__.items():
182 if key in GenericExtractor.booleans:
183 if type(val) == type("") and val.lower() == "true":
184 setattr(self, key, True)
185 elif type(val) == type("") and val.lower() == "false":
186 setattr(self, key, False)
187 elif key in GenericExtractor.numerics:
188 setattr(self, key, int(val))
189 elif key in GenericExtractor.strings:
190 setattr(self, key, val)
192 sys.stderr.write("irkerhook.py: no project name set!\n")
195 self.repo = self.project.lower()
196 if not self.channels:
197 self.channels = default_channels % self.__dict__
198 if self.color and self.color.lower() != "none":
199 self.activate_color(self.color)
201 def has(dirname, paths):
202 "Test for existence of a list of paths."
203 # all() is a python2.5 construct
204 for exists in [os.path.exists(os.path.join(dirname, x)) for x in paths]:
209 # VCS-dependent code begins here
211 class GitExtractor(GenericExtractor):
212 "Metadata extraction for the git version control system."
214 def is_repository(dirname):
215 # Must detect both ordinary and bare repositories
216 return has(dirname, [".git"]) or \
217 has(dirname, ["HEAD", "refs", "objects"])
218 def __init__(self, arguments):
219 GenericExtractor.__init__(self, arguments)
220 # Get all global config variables
221 self.project = do("git config --get irker.project")
222 self.repo = do("git config --get irker.repo")
223 self.server = do("git config --get irker.server")
224 self.channels = do("git config --get irker.channels")
225 self.email = do("git config --get irker.email")
226 self.tcp = do("git config --bool --get irker.tcp")
227 self.template = '%(bold)s%(project)s:%(reset)s %(green)s%(author)s%(reset)s %(repo)s:%(yellow)s%(branch)s%(reset)s * %(bold)s%(rev)s%(reset)s / %(bold)s%(files)s%(reset)s: %(logmsg)s %(brown)s%(url)s%(reset)s'
228 self.tinyifier = do("git config --get irker.tinyifier") or default_tinyifier
229 self.color = do("git config --get irker.color")
230 self.urlprefix = do("git config --get irker.urlprefix") or "gitweb"
231 self.cialike = do("git config --get irker.cialike")
232 self.filtercmd = do("git config --get irker.filtercmd")
233 # These are git-specific
234 self.refname = do("git symbolic-ref HEAD 2>/dev/null")
235 self.revformat = do("git config --get irker.revformat")
236 # The project variable defaults to the name of the repository toplevel.
238 bare = do("git config --bool --get core.bare")
239 if bare.lower() == "true":
242 keyfile = ".git/HEAD"
245 if os.path.exists(os.path.join(here, keyfile)):
246 self.project = os.path.basename(here)
247 if self.project.endswith('.git'):
248 self.project = self.project[0:-4]
251 sys.stderr.write("irkerhook.py: no git repo below root!\n")
253 here = os.path.dirname(here)
257 "Return a symbolic reference to the tip commit of the current branch."
259 def commit_factory(self, commit_id):
260 "Make a Commit object holding data for a specified commit ID."
261 commit = Commit(self, commit_id)
262 commit.branch = re.sub(r"^refs/[^/]*/", "", self.refname)
263 # Compute a description for the revision
264 if self.revformat == 'raw':
265 commit.rev = commit.commit
266 elif self.revformat == 'short':
268 else: # self.revformat == 'describe'
269 commit.rev = do("git describe %s 2>/dev/null" % shellquote(commit.commit))
271 commit.rev = commit.commit[:12]
272 # Extract the meta-information for the commit
273 commit.files = do("git diff-tree -r --name-only " + shellquote(commit.commit))
274 commit.files = " ".join(commit.files.strip().split("\n")[1:])
275 # Design choice: for git we ship only the first message line, which is
276 # conventionally supposed to be a summary of the commit. Under
277 # other VCSes a different choice may be appropriate.
278 commit.author_name, commit.mail, commit.logmsg = \
279 do("git log -1 '--pretty=format:%an%n%ae%n%s' " + shellquote(commit.commit)).split("\n")
280 # This discards the part of the author's address after @.
281 # Might be be nice to ship the full email address, if not
282 # for spammers' address harvesters - getting this wrong
283 # would make the freenode #commits channel into harvester heaven.
284 commit.author = commit.mail.split("@")[0]
285 commit.author_date, commit.commit_date = \
286 do("git log -1 '--pretty=format:%ai|%ci' " + shellquote(commit.commit)).split("|")
289 class SvnExtractor(GenericExtractor):
290 "Metadata extraction for the svn version control system."
292 def is_repository(dirname):
293 return has(dirname, ["format", "hooks", "locks"])
294 def __init__(self, arguments):
295 GenericExtractor.__init__(self, arguments)
296 # Some things we need to have before metadata queries will work
297 self.repository = '.'
298 for tok in arguments:
299 if tok.startswith("--repository="):
300 self.repository = tok[13:]
301 self.project = os.path.basename(self.repository)
302 self.template = '%(bold)s%(project)s%(reset)s: %(green)s%(author)s%(reset)s %(repo)s * %(bold)s%(rev)s%(reset)s / %(bold)s%(files)s%(reset)s: %(logmsg)s %(brown)s%(url)s%(reset)s'
303 self.urlprefix = "viewcvs"
304 self.load_preferences(os.path.join(self.repository, "irker.conf"))
307 sys.stderr.write("irker: under svn, hook requires a commit argument.\n")
309 def commit_factory(self, commit_id):
311 commit = Commit(self, commit_id)
313 commit.rev = "r%s" % self.id
314 commit.author = self.svnlook("author")
315 commit.commit_date = self.svnlook("date").partition('(')[0]
316 commit.files = self.svnlook("dirs-changed").strip().replace("\n", " ")
317 commit.logmsg = self.svnlook("log").strip()
319 def svnlook(self, info):
320 return do("svnlook %s %s --revision %s" % (shellquote(info), shellquote(self.repository), shellquote(self.id)))
322 class HgExtractor(GenericExtractor):
323 "Metadata extraction for the Mercurial version control system."
325 def is_repository(directory):
326 return has(directory, [".hg"])
327 def __init__(self, arguments):
328 # This fiddling with arguments is necessary since the Mercurial hook can
329 # be run in two different ways: either directly via Python (in which
330 # case hg should be pointed to the hg_hook function below) or as a
331 # script (in which case the normal __main__ block at the end of this
332 # file is exercised). In the first case, we already get repository and
333 # ui objects from Mercurial, in the second case, we have to create them
334 # from the root path.
335 self.repository = None
336 if arguments and type(arguments[0]) == type(()):
337 # Called from hg_hook function
338 ui, self.repository = arguments[0]
339 arguments = [] # Should not be processed further by do_overrides
341 # Called from command line: create repo/ui objects
342 from mercurial import hg, ui as uimod
345 for tok in arguments:
346 if tok.startswith('--repository='):
349 ui.readconfig(os.path.join(repopath, '.hg', 'hgrc'), repopath)
350 self.repository = hg.repository(ui, repopath)
352 GenericExtractor.__init__(self, arguments)
353 # Extract global values from the hg configuration file(s)
354 self.project = ui.config('irker', 'project')
355 self.repo = ui.config('irker', 'repo')
356 self.server = ui.config('irker', 'server')
357 self.channels = ui.config('irker', 'channels')
358 self.email = ui.config('irker', 'email')
359 self.tcp = str(ui.configbool('irker', 'tcp')) # converted to bool again in do_overrides
360 self.template = '%(bold)s%(project)s:%(reset)s %(green)s%(author)s%(reset)s %(repo)s:%(yellow)s%(branch)s%(reset)s * %(bold)s%(rev)s%(reset)s / %(bold)s%(files)s%(reset)s: %(logmsg)s %(brown)s%(url)s%(reset)s'
361 self.tinyifier = ui.config('irker', 'tinyifier') or default_tinyifier
362 self.color = ui.config('irker', 'color')
363 self.urlprefix = (ui.config('irker', 'urlprefix') or
364 ui.config('web', 'baseurl') or '')
366 # self.commit is appended to this by do_overrides
367 self.urlprefix = self.urlprefix.rstrip('/') + '/rev/'
368 self.cialike = ui.config('irker', 'cialike')
369 self.filtercmd = ui.config('irker', 'filtercmd')
371 self.project = os.path.basename(self.repository.root.rstrip('/'))
374 "Return a symbolic reference to the tip commit of the current branch."
376 def commit_factory(self, commit_id):
377 "Make a Commit object holding data for a specified commit ID."
378 from mercurial.node import short
379 from mercurial.templatefilters import person
380 node = self.repository.lookup(commit_id)
381 commit = Commit(self, short(node))
382 # Extract commit-specific values from a "context" object
383 ctx = self.repository.changectx(node)
384 commit.rev = '%d:%s' % (ctx.rev(), commit.commit)
385 commit.branch = ctx.branch()
386 commit.author = person(ctx.user())
387 commit.author_date = \
388 datetime.datetime.fromtimestamp(ctx.date()[0]).strftime('%Y-%m-%d %H:%M:%S')
389 commit.logmsg = ctx.description()
390 # Extract changed files from status against first parent
391 st = self.repository.status(ctx.p1().node(), ctx.node())
392 commit.files = ' '.join(st[0] + st[1] + st[2])
395 def hg_hook(ui, repo, **kwds):
396 # To be called from a Mercurial "commit", "incoming" or "changegroup" hook.
397 # Example configuration:
399 # incoming.irker = python:/path/to/irkerhook.py:hg_hook
400 extractor = HgExtractor([(ui, repo)])
401 start = repo[kwds['node']].rev()
404 # changegroup with multiple commits, so we generate a notification
406 for rev in range(start, end):
407 ship(extractor, rev, False)
409 ship(extractor, kwds['node'], False)
411 # The files we use to identify a Subversion repo might occur as content
412 # in a git or hg repo, but the special subdirectories for those are more
413 # reliable indicators. So test for Subversion last.
414 extractors = [GitExtractor, HgExtractor, SvnExtractor]
416 # VCS-dependent code ends here
418 def ship(extractor, commit, debug):
419 "Ship a notification for the specified commit."
420 metadata = extractor.commit_factory(commit)
422 # This is where we apply filtering
423 if extractor.filtercmd:
424 cmd = '%s %s' % (shellquote(extractor.filtercmd),
425 shellquote(json.dumps(metadata.__dict__)))
426 data = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE).stdout.read()
428 metadata.__dict__.update(json.loads(data))
430 sys.stderr.write("irkerhook.py: could not decode JSON: %s\n" % data)
433 # Rewrite the file list if too long. The objective here is only
434 # to be easier on the eyes.
435 if extractor.cialike \
436 and extractor.cialike.lower() != "none" \
437 and len(metadata.files) > int(extractor.cialike):
438 files = metadata.files.split()
439 dirs = set([d.rpartition('/')[0] for d in files])
441 metadata.files = "(%s files)" % (len(files),)
443 metadata.files = "(%s files in %s dirs)" % (len(files), len(dirs))
444 # Message reduction. The assumption here is that IRC can't handle
445 # lines more than 510 characters long. If we exceed that length, we
446 # try knocking out the file list, on the theory that for notification
447 # purposes the commit text is more important. If it's still too long
448 # there's nothing much can be done other than ship it expecting the IRC
449 # server to truncate.
450 privmsg = unicode(metadata)
451 if len(privmsg) > 510:
453 privmsg = unicode(metadata)
455 # Anti-spamming guard. It's deliberate that we get maxchannels not from
456 # the user-filtered metadata but from the extractor data - means repo
457 # administrators can lock in that setting.
458 channels = metadata.channels.split(",")
459 if extractor.maxchannels != 0:
460 channels = channels[:extractor.maxchannels]
463 message = json.dumps({"to": channels, "privmsg": privmsg})
469 # We can't really figure out what our SF username is without
470 # exploring our environment. The mail pipeline doesn't care
471 # about who sent the mail, other than being from sourceforge.
472 # A better way might be to simply call mail(1)
473 sender = "irker@users.sourceforge.net"
474 msg = """From: %(sender)s
477 %(message)s""" % {"sender":sender, "message":message}
479 smtp = smtplib.SMTP()
481 smtp.sendmail(sender, extractor.email, msg)
485 sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
486 sock.connect((extractor.server or default_server, IRKER_PORT))
487 sock.sendall(message + "\n")
492 sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
493 sock.sendto(message + "\n", (extractor.server or default_server, IRKER_PORT))
496 except socket.error, e:
497 sys.stderr.write("%s\n" % e)
499 if __name__ == "__main__":
501 repository = os.getcwd()
503 for arg in sys.argv[1:]:
507 print "irkerhook.py: version", version
509 elif arg.startswith("--repository="):
510 repository = arg[13:]
511 elif not arg.startswith("--"):
514 # Figure out which extractor we should be using
515 for candidate in extractors:
516 if candidate.is_repository(repository):
520 sys.stderr.write("irkerhook: cannot identify a repository type.\n")
522 extractor = cls(sys.argv[1:])
526 commits = [extractor.head()]
527 for commit in commits:
528 ship(extractor, commit, not notify)