4 signal.signal(signal.SIGINT, signal.SIG_DFL)
14 from optparse import OptionParser
16 from BeautifulSoup import BeautifulSoup
21 puzzles = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+
22 '/Battle Navigation/Gunning/Carpentry/Rumble/Treasure Haul'+
23 '/Drinking/Spades/Hearts/Treasure Drop/Poker/Distilling'+
24 '/Alchemistry/Shipwrightery/Blacksmithing/Foraging').split('/')
26 standingvals = ('Able/Distinguished/Respected/Master'+
27 '/Renowned/Grand-Master/Legendary/Ultimate').split('/')
29 pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm')
31 max_pirate_namelen = 12
39 def __init__(self, ocean, cachedir):
40 debug('Fetcher init %s' % cachedir)
42 self.cachedir = cachedir
43 try: os.mkdir(cachedir)
44 except (OSError,IOError), oe:
45 if oe.errno != errno.EEXIST: raise
46 self._cache_scan(time.time())
48 def _default_ocean(self):
49 if self.ocean is None:
52 def _cache_scan(self, now):
53 # returns list of ages, unsorted
55 debug('Fetcher scan_cache')
56 for leaf in os.listdir(self.cachedir):
57 if not leaf.startswith('#'): continue
58 path = self.cachedir + '/' + leaf
59 try: s = os.stat(path)
60 except (OSError,IOError), oe:
61 if oe.errno != errno.ENOENT: raise
63 age = now - s.st_mtime
64 if age > opts.expire_age:
65 debug('Fetcher expire %d %s' % (age, path))
67 except (OSError,IOError), oe:
68 if oe.errno != errno.ENOENT: raise
73 def need_wait(self, now):
74 ages = self._cache_scan(now)
76 debug('Fetcher ages ' + `ages`)
80 if age < min_age and age < 300:
81 debug('Fetcher morewait min=%d age=%d' %
83 need_wait = max(need_wait, min_age - age)
88 def _rate_limit_cache_clean(self, now):
89 need_wait = self.need_wait(now)
91 debug('Fetcher wait %d' % need_wait)
94 def fetch(self, url, max_age):
95 debug('Fetcher fetch %s' % url)
96 cache_corename = urllib.quote_plus(url)
97 cache_item = "%s/#%s#" % (self.cachedir, cache_corename)
98 try: f = file(cache_item, 'r')
99 except (OSError,IOError), oe:
100 if oe.errno != errno.ENOENT: raise
103 max_age = max(opts.min_max_age, min(max_age, opts.expire_age))
105 s = os.fstat(f.fileno())
106 age = now - s.st_mtime
108 debug('Fetcher stale %d < %d'% (max_age, age))
113 debug('Fetcher cached %d > %d' % (max_age, age))
116 debug('Fetcher fetch')
117 self._rate_limit_cache_clean(now)
119 stream = urllib2.urlopen(url)
121 cache_tmp = "%s/#%s~%d#" % (
122 self.cachedir, cache_corename, os.getpid())
123 f = file(cache_tmp, 'w')
126 os.rename(cache_tmp, cache_item)
127 debug('Fetcher stored')
130 def yoweb(self, kind, tail, max_age):
131 self._default_ocean()
132 url = 'http://%s.puzzlepirates.com/yoweb/%s%s' % (
133 self.ocean, kind, tail)
134 return self.fetch(url, max_age)
141 def soupm(self, obj, m):
142 self.msg(m + '; in ' + `obj`)
143 def needs_msgs(self, child_souplog):
144 self.msgs += child_souplog.msgs
145 child_souplog.msgs = [ ]
148 str = ''.join(obj.findAll(text=True))
151 class SomethingSoupInfo(SoupLog):
152 def __init__(self, kind, tail, max_age):
153 SoupLog.__init__(self)
154 html = fetcher.yoweb(kind, tail, max_age)
155 self._soup = BeautifulSoup(html,
156 convertEntities=BeautifulSoup.HTML_ENTITIES
159 class PirateInfo(SomethingSoupInfo):
160 # Public data members:
161 # pi.standings = { 'Treasure Haul': 'Able' ... }
163 # pi.crew = (id, name)
164 # pi.flag = (id, name)
165 # pi.msgs = [ 'message describing problem with scrape' ]
167 def __init__(self, pirate, max_age=300):
168 SomethingSoupInfo.__init__(self,
169 'pirate.wm?target=', pirate, max_age)
171 self._find_standings()
172 self.crew = self._find_crewflag('crew',
173 '^/yoweb/crew/info\\.wm')
174 self.flag = self._find_crewflag('flag',
175 '^/yoweb/flag/info\\.wm')
177 def _find_standings(self):
178 imgs = self._soup.findAll('img',
179 src=regexp.compile('/yoweb/images/stat.*'))
181 u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-Za-z]+)\\)\\s*$'
185 for skill in puzzles:
186 standings[skill] = [ ]
191 try: puzzle = img['alt']
192 except KeyError: continue
194 if not puzzle in puzzles:
195 skl.soupm(img, 'unknown puzzle: "%s"' % puzzle)
197 key = img.findParent('td')
199 skl.soupm(img, 'puzzle at root! "%s"' % puzzle)
201 valelem = key.findNextSibling('td')
203 skl.soupm(key, 'puzzle missing sibling "%s"'
206 valstr = soup_text(valelem)
207 match = re.match(valstr)
209 skl.soupm(key, ('puzzle "%s" unparseable'+
210 ' standing "%s"') % (puzzle, valstr))
212 standing = match.group(match.lastindex)
213 standings[puzzle].append(standing)
217 for puzzle in puzzles:
218 sl = standings[puzzle]
220 skl.msg('puzzle "%s" multiple standings %s' %
224 skl.msg('puzzle "%s" no standing found' % puzzle)
227 for i in range(0, len(standingvals)-1):
228 if standing == standingvals[i]:
229 self.standings[puzzle] = i
230 if not puzzle in self.standings:
231 skl.msg('puzzle "%s" unknown standing "%s"' %
234 all_standings_ok = True
235 for puzzle in puzzles:
236 if not puzzle in self.standings:
239 def _find_crewflag(self, cf, yoweb_re):
240 things = self._soup.findAll('a', href=regexp.compile(yoweb_re))
242 self.msg('zero or several %s id references found' % cf)
245 id_re = '\\b%sid\\=(\\w+)$' % cf
246 id_haystack = thing['href']
247 match = regexp.compile(id_re).search(id_haystack)
249 self.soupm(thing, ('incomprehensible %s id ref'+
250 ' (%s in %s)') % (cf, id_re, id_haystack))
252 name = soup_text(thing)
253 return (match.group(1), name)
256 return `(self.crew, self.flag, self.standings, self.msgs)`
258 class CrewInfo(SomethingSoupInfo):
259 # Public data members:
260 # ci.crew = [ ('Captain', ['Pirate', ...]),
261 # ('Senior Officer', [...]),
263 # pi.msgs = [ 'message describing problem with scrape' ]
265 def __init__(self, crewid, max_age=300):
266 SomethingSoupInfo.__init__(self,
267 'crew/info.wm?crewid=', crewid, max_age)
270 def _find_crew(self):
272 capts = self._soup.findAll('img',
273 src='/yoweb/images/crew-captain.png')
275 self.msg('crew members: no. of captain images != 1')
278 while not tbl.find('a', href=pirate_ref_re):
279 tbl = tbl.findParent('table')
281 self.msg('crew members: cannot find table')
283 current_rank_crew = None
284 crew_rank_re = regexp.compile('/yoweb/images/crew')
285 for row in tbl.contents:
286 # findAll(recurse=False)
287 if isinstance(row,basestring):
290 is_rank = row.find('img', attrs={'src': crew_rank_re})
292 rank = soup_text(row)
293 current_rank_crew = []
294 self.crew.append((rank, current_rank_crew))
296 for cell in row.findAll('a', href=pirate_ref_re):
297 if current_rank_crew is None:
298 self.soupm(cell, 'crew members: crew'
301 current_rank_crew.append(soup_text(cell))
304 return `(self.crew, self.msgs)`
306 class StandingsTable:
307 def __init__(self, use_puzzles=None, col_width=6):
308 if use_puzzles is None:
311 'Navigating','Battle Navigation',
313 ['Sailing','Rigging'],
320 self._puzzles = use_puzzles
322 self._cw = col_width-1
324 def _pline(self, pirate, puzstrs):
325 self.s += ' %-*s' % (max(max_pirate_namelen, 14), pirate)
327 self.s += ' %-*.*s' % (self._cw,self._cw, v)
330 def _puzstr(self, pi, puzzle):
331 if not isinstance(puzzle,list): puzzle = [puzzle]
332 try: standing = max([pi.standings[p] for p in puzzle])
333 except KeyError: return '?'
334 if isinstance(standing,basestring): return standing
335 if not standing: return ''
338 c1 = standingvals[standing][0]
339 if standing < 3: c1 = c1.lower() # 3 = Master
343 s += '*' * (standing / 2)
344 s += '+' * (standing % 2)
348 def puzn_redact(name):
349 if isinstance(name,list):
351 ["%.*s" % (self._cw/2, puzn_redact(n))
354 if spc < 0: return name
355 return name[0:min(4,spc)] + name[spc+1:]
356 self._pline('', map(puzn_redact, self._puzzles))
357 def literalline(self, line):
358 self.s += line + '\n'
359 def pirate(self, pi):
360 puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles]
361 self._pline(pi.name, puzstrs)
366 def do_pirate(pirates, bu):
368 for pirate in pirates:
369 info = PirateInfo(pirate)
370 print '%s: %s,' % (`pirate`, info)
373 def prep_crew_of(args, bu, max_age=300):
374 if len(args) != 1: bu('crew-of takes one pirate name')
375 pi = PirateInfo(args[0], max_age)
376 return CrewInfo(pi.crew[0], max_age)
378 def do_crew_of(args, bu):
379 ci = prep_crew_of(args, bu)
382 def do_standings_crew_of(args, bu):
383 ci = prep_crew_of(args, bu, 60)
384 tab = StandingsTable()
386 for (rank, members) in ci.crew:
387 if not members: continue
388 tab.literalline('%s:' % rank)
390 pi = PirateInfo(p, random.randint(900,1800))
403 def __init__(pa, pn, v, time, event):
407 pa.last_event = event
408 pa.last_chat_time = None
409 pa.last_chat_chan = None
413 def pirate_info(self):
414 if not pa.pi and not fetcher.need_wait(time.time):
415 pa.pi = PirateInfo(pa.name, 3600)
418 class ChatLogTracker:
419 def __init__(self, myself_pi, logfn):
420 self._pl = {} # self._pl['Pirate'] =
421 self._vl = {} # self._vl['Vessel']['Pirate'] = PirateAboard
422 # self._vl['Vessel']['#lastaboard']
423 self._v = None # self._v =
424 self._vessel = None # self._vl[self._vessel]
426 self._myself = myself_pi
427 self.need_redisplay = False
428 self._f = file(logfn)
430 self._progress = [0, os.fstat(self._f.fileno()).st_size]
433 self.need_redisplay = True
435 def _onboard_event(self,timestamp,pirate,event):
436 try: pa = self._pl[pirate]
437 except KeyError: pa = None
438 if pa is not None and pa.v is self._v:
439 pa.last_time = timestamp
440 pa.last_event = event
442 if pa is not None: del pa.v[pirate]
443 pa = PirateAboard(pirate, self._v, timestamp, event)
444 self._pl[pirate] = pa
446 self._v['#lastaboard'] = timestamp
450 def _trash_vessel(self, v):
452 if pn.startswith('#'): continue
456 def expire_garbage(self, timestamp):
457 for (vn,v) in list(self._vl.iteritems()):
458 la = v['#lastaboard']
459 if timestamp - la > opts.ship_reboard_clearout:
460 self._debug_line_disposition(timestamp,'',
462 self._trash_vessel(v)
465 def clear_vessel(self, timestamp):
466 if self._v is not None:
467 self._trash_vessel(self._v)
468 self._v = {'#lastaboard': timestamp}
469 self._vl[self._vessel] = self._v
471 def _debug_line_disposition(self,timestamp,l,m):
472 debug('CLT %13s %-30s %s' % (timestamp,m,l))
474 def chatline(self,l):
475 rm = lambda re: regexp.match(re,l)
476 d = lambda m: self._debug_line_disposition(timestamp,l,m)
479 m = rm('=+ (\\d+)/(\\d+)/(\\d+) =+$')
481 self._date = m.groups()
482 return d('date '+`self._date`)
484 if self._date is None:
485 return d('date unset')
487 m = rm('\\[(\d\d):(\d\d):(\d\d)\\] ')
489 return d('no timestamp')
491 time_tuple = [int(x) for x in self._date + m.groups()]
492 time_tuple += (-1,-1,-1)
493 timestamp = time.mktime(time_tuple)
494 l = l[l.find(' ')+1:]
497 return self._onboard_event(timestamp, who, event)
498 def ob1(did): ob_x(m.group(1), did); return d(did)
499 def oba(did): return ob1('%s %s' % (did, m.group(2)))
501 m = rm('Going aboard the (\\S.*\\S)\\.\\.\\.$')
503 pn = self._myself.name
504 self._vessel = m.group(1)
507 try: self._v = self._vl[self._vessel]
508 except KeyError: self._v = None; dm += ' new'
510 if self._v is not None: la = self._v['#lastaboard']
511 else: la = 0; dm += ' ?la'
513 if timestamp - la > opts.ship_reboard_clearout:
514 self.clear_vessel(timestamp)
517 ob_x(pn, 'we boarded')
518 self.expire_garbage(timestamp)
522 return d('no vessel')
524 m = rm('(\\w+) has come aboard\\.$')
525 if m: return ob1('boarded');
527 m = rm('You have ordered (\\w+) to do some (\\S.*\\S)\\.$')
529 (who,what) = m.groups()
530 pa = ob_x(who,'ordered '+what)
531 if what == 'Gunning':
533 return d('duty order')
535 m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$')
536 if m: oba('stopped'); return d('stopped')
540 try: pa = self._pl[who]
541 except KeyError: return d('chat mystery')
543 pa.last_chat_time = timestamp
544 pa.last_chat_chan = what
546 return d(what+' chat')
548 m = rm('(\\w+) (?:issued an order|ordered everyone) "')
549 if m: return ob1('general order');
551 m = rm('(\\w+) says, "')
552 if m: return chat('public')
554 m = rm('(\\w+) tells ye, "')
555 if m: return chat('private')
557 m = rm('(\\w+) flag officer chats, "')
558 if m: return chat('flag officer')
560 m = rm('(\\w+) officer chats, "')
561 if m: return chat('officer')
563 m = rm('Game over\\. Winners: ([A-Za-z, ]+)\\.$')
565 pl = m.group(1).split(', ')
566 if not self._myself.name in pl:
567 return d('lost boarding battle')
569 if ' ' in pn: continue
570 ob_x(pn,'won boarding battle')
571 return d('won boarding battle')
573 m = rm('(\\w+) is eliminated\\!')
574 if m: return ob1('eliminated in fray');
576 m = rm('(\\w+) has left the vessel\.')
579 ob_x(who, 'disembarked')
582 return d('disembarked')
584 return d('not matched')
586 def _str_vessel(self, vn, v):
587 s = ' vessel %s\n' % vn
588 s += ' '*20 + "%-*s %13s\n" % (
589 max_pirate_namelen, '#lastaboard',
591 for pn in sorted(v.keys()):
592 if pn.startswith('#'): continue
595 assert self._pl[pn] == pa
596 s += ' '*20 + "%s %-*s %13s %-30s %13s %s\n" % (
597 (' ','G')[pa.gunner],
598 max_pirate_namelen, pn,
599 pa.last_time, pa.last_event,
600 pa.last_chat_time, pa.last_chat_chan)
604 s = '''<ChatLogTracker
607 ''' % (self._myself.name, self._vessel)
608 assert ((self._v is None and self._vessel is None) or
609 (self._v is self._vl[self._vessel]))
610 if self._vessel is not None:
611 s += self._str_vessel(self._vessel, self._v)
612 for vn in sorted(self._vl.keys()):
613 if vn == self._vessel: continue
614 s += self._str_vessel(vn, self._vl[vn])
618 assert pa.v in self._vl.values()
622 def catchup(self, progress=None):
624 more = self._f.readline()
627 self._progress[0] += len(more)
628 if progress: progress.progress(*self._progress)
631 if self._lbuf.endswith('\n'):
632 self.chatline(self._lbuf.rstrip())
634 if progress: progress.caughtup()
636 class ProgressPrintPercentage:
637 def __init__(self, f=sys.stdout): self._f = f
638 def progress(self,done,total):
639 self._f.write("scan chat logs %3d%%\r" % ((done*100) / total))
645 def simple_printer(x): print x
647 def run_chat_log(args, bu, progress=ProgressPrintPercentage(),
648 display=simple_printer, max_myself_age=3600):
649 if len(args) != 1: bu('this action takes only chat log filename')
651 logfn_re = '(?:.*/)?([A-Z][a-z]+)_([a-z]+)_chat-log-\\w+$'
652 match = regexp.match(logfn_re, logfn)
653 if not match: bu('chat log filename is not in default format')
654 (pirate, fetcher.ocean) = match.groups()
656 myself_pi = PirateInfo(pirate,max_myself_age)
657 track = ChatLogTracker(myself_pi, logfn)
659 track.catchup(progress)
662 if track.need_redisplay:
664 track.need_redisplay = False
667 def do_track_chat_log(args, bu):
668 run_chat_log(args, bu)
670 #def do_ship_aid(args, bu):
676 '''usage: .../yoweb-scrape [OPTION...] ACTION [ARGS...]
678 yoweb-scrape [--ocean OCEAN ...] pirate PIRATE
679 yoweb-scrape [--ocean OCEAN ...] crew-of PIRATE
680 yoweb-scrape [--ocean OCEAN ...] standings-crew-of PIRATE
681 yoweb-scrape [--ocean OCEAN ...] track-chat-log CHAT-LOG
682 yoweb-scrape [--ocean OCEAN ...] ship-aid CHAT-LOG
685 ao('-O','--ocean',dest='ocean', metavar='OCEAN', default=None,
686 help='select ocean OCEAN')
687 ao('--cache-dir', dest='cache_dir', metavar='DIR',
688 default='~/.yoweb-scrape-cache',
689 help='cache yoweb pages in DIR')
690 ao('-D','--debug', action='store_true', dest='debug', default=False,
691 help='enable debugging output')
692 ao('-q','--quiet', action='store_true', dest='quiet',
693 help='suppress warning output')
695 ao('--ship-duty', action='store_true', dest='ship_duty',
696 help='show ship duty station puzzles')
697 ao('--all-puzzles', action='store_false', dest='ship_duty',
698 help='show all puzzles, not just ship duty stations')
700 (opts,args) = pa.parse_args()
704 pa.error('need a mode argument')
707 mode_fn_name = 'do_' + mode.replace('_','#').replace('-','_')
708 try: mode_fn = globals()[mode_fn_name]
709 except KeyError: pa.error('unknown mode "%s"' % mode)
712 opts.min_max_age = 60
713 opts.expire_age = 3600
714 opts.ship_reboard_clearout = 3600
716 if opts.cache_dir.startswith('~/'):
717 opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]
719 fetcher = Fetcher(opts.ocean, opts.cache_dir)
721 mode_fn(args[1:], pa.error)