2 # This is part of ypp-sc-tools, a set of third-party tools for assisting
3 # players of Yohoho Puzzle Pirates.
5 # Copyright (C) 2009 Ian Jackson <ijackson@chiark.greenend.org.uk>
7 # This program is free software: you can redistribute it and/or modify
8 # it under the terms of the GNU General Public License as published by
9 # the Free Software Foundation, either version 3 of the License, or
10 # (at your option) any later version.
12 # This program is distributed in the hope that it will be useful,
13 # but WITHOUT ANY WARRANTY; without even the implied warranty of
14 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 # GNU General Public License for more details.
17 # You should have received a copy of the GNU General Public License
18 # along with this program. If not, see <http://www.gnu.org/licenses/>.
20 # Yohoho and Puzzle Pirates are probably trademarks of Three Rings and
21 # are used without permission. This program is not endorsed or
22 # sponsored by Three Rings.
25 yoweb-scrape is part of ypp-sc-tools Copyright (C) 2009 Ian Jackson
26 This program comes with ABSOLUTELY NO WARRANTY; this is free software,
27 and you are welcome to redistribute it under certain conditions.
28 For details, read the top of the yoweb-scrape file.
31 #---------- setup ----------
34 signal.signal(signal.SIGINT, signal.SIG_DFL)
48 from optparse import OptionParser
49 from StringIO import StringIO
51 from BeautifulSoup import BeautifulSoup
55 #---------- YPP parameters and arrays ----------
57 puzzles = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+
58 '/Battle Navigation/Gunning/Carpentry/Rumble/Treasure Haul'+
59 '/Drinking/Spades/Hearts/Treasure Drop/Poker/Distilling'+
60 '/Alchemistry/Shipwrightery/Blacksmithing/Foraging').split('/')
64 ['Sailing','Rigging'],
69 duty_puzzles = ([ 'Navigating', 'Battle Navigation' ] +
73 standingvals = ('Able/Proficient/Distinguished/Respected/Master'+
74 '/Renowned/Grand-Master/Legendary/Ultimate').split('/')
75 standing_limit = len(standingvals)
77 pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm')
79 max_pirate_namelen = 12
82 #---------- general utilities ----------
86 print >>opts.debug_file, m
90 opts.debug_file.flush()
96 def format_time_interval(ti):
97 if ti < 120: return '%d:%02d' % (ti / 60, ti % 60)
98 if ti < 7200: return '%2dm' % (ti / 60)
99 if ti < 86400: return '%dh' % (ti / 3600)
100 return '%dd' % (ti / 86400)
103 lib = os.getenv("YPPSC_YARRG_SRCBASE")
104 if lib is not None: return lib
106 lib = regexp.sub('/[^/]+$', '', lib)
107 os.environ["YPPSC_YARRG_SRCBASE"] = lib
110 #---------- caching and rate-limiting data fetcher ----------
113 def __init__(self, ocean, cachedir):
114 debug('Fetcher init %s' % cachedir)
116 self.cachedir = cachedir
117 try: os.mkdir(cachedir)
118 except (OSError,IOError), oe:
119 if oe.errno != errno.EEXIST: raise
120 self._cache_scan(time.time())
122 def default_ocean(self, ocean='ice'):
123 if self.ocean is None:
126 def _cache_scan(self, now):
127 # returns list of ages, unsorted
129 debug('Fetcher scan_cache')
130 for leaf in os.listdir(self.cachedir):
131 if not leaf.startswith('#'): continue
132 path = self.cachedir + '/' + leaf
133 try: s = os.stat(path)
134 except (OSError,IOError), oe:
135 if oe.errno != errno.ENOENT: raise
137 age = now - s.st_mtime
138 if age > opts.expire_age:
139 debug('Fetcher expire %d %s' % (age, path))
141 except (OSError,IOError), oe:
142 if oe.errno != errno.ENOENT: raise
147 def need_wait(self, now, imaginary=[]):
148 ages = self._cache_scan(now)
151 debug('Fetcher ages ' + `ages`)
155 if age < min_age and age <= 5:
156 debug('Fetcher morewait min=%d age=%d' %
158 need_wait = max(need_wait, min_age - age)
162 need_wait += random.random() - 0.5
165 def _rate_limit_cache_clean(self, now):
166 need_wait = self.need_wait(now)
168 debug('Fetcher wait %d' % need_wait)
171 def fetch(self, url, max_age):
172 debug('Fetcher fetch %s' % url)
173 cache_corename = urllib.quote_plus(url)
174 cache_item = "%s/#%s#" % (self.cachedir, cache_corename)
175 try: f = file(cache_item, 'r')
176 except (OSError,IOError), oe:
177 if oe.errno != errno.ENOENT: raise
180 max_age = max(opts.min_max_age, min(max_age, opts.expire_age))
182 s = os.fstat(f.fileno())
183 age = now - s.st_mtime
185 debug('Fetcher stale %d < %d'% (max_age, age))
190 debug('Fetcher cached %d > %d' % (max_age, age))
193 debug('Fetcher fetch')
194 self._rate_limit_cache_clean(now)
196 stream = urllib2.urlopen(url)
198 cache_tmp = "%s/#%s~%d#" % (
199 self.cachedir, cache_corename, os.getpid())
200 f = file(cache_tmp, 'w')
203 os.rename(cache_tmp, cache_item)
204 debug('Fetcher stored')
207 def yoweb(self, kind, tail, max_age):
209 url = 'http://%s.puzzlepirates.com/yoweb/%s%s' % (
210 self.ocean, kind, tail)
211 return self.fetch(url, max_age)
213 #---------- logging assistance for troubled screenscrapers ----------
220 def soupm(self, obj, m):
221 self.msg(m + '; in ' + `obj`)
222 def needs_msgs(self, child_souplog):
223 self.msgs += child_souplog.msgs
224 child_souplog.msgs = [ ]
227 str = ''.join(obj.findAll(text=True))
230 class SomethingSoupInfo(SoupLog):
231 def __init__(self, kind, tail, max_age):
232 SoupLog.__init__(self)
233 html = fetcher.yoweb(kind, tail, max_age)
234 self._soup = BeautifulSoup(html,
235 convertEntities=BeautifulSoup.HTML_ENTITIES
238 #---------- scraper for pirate pages ----------
240 class PirateInfo(SomethingSoupInfo):
241 # Public data members:
242 # pi.standings = { 'Treasure Haul': 'Able' ... }
244 # pi.crew = (id, name)
245 # pi.flag = (id, name)
246 # pi.msgs = [ 'message describing problem with scrape' ]
248 def __init__(self, pirate, max_age=300):
249 SomethingSoupInfo.__init__(self,
250 'pirate.wm?target=', pirate, max_age)
252 self._find_standings()
253 self.crew = self._find_crewflag('crew',
254 '^/yoweb/crew/info\\.wm')
255 self.flag = self._find_crewflag('flag',
256 '^/yoweb/flag/info\\.wm')
258 def _find_standings(self):
259 imgs = self._soup.findAll('img',
260 src=regexp.compile('/yoweb/images/stat.*'))
262 u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-Za-z]+)\\)\\s*$'
266 for skill in puzzles:
267 standings[skill] = [ ]
272 try: puzzle = img['alt']
273 except KeyError: continue
275 if not puzzle in puzzles:
276 skl.soupm(img, 'unknown puzzle: "%s"' % puzzle)
278 key = img.findParent('td')
280 skl.soupm(img, 'puzzle at root! "%s"' % puzzle)
282 valelem = key.findNextSibling('td')
284 skl.soupm(key, 'puzzle missing sibling "%s"'
287 valstr = soup_text(valelem)
288 match = re.match(valstr)
290 skl.soupm(key, ('puzzle "%s" unparseable'+
291 ' standing "%s"') % (puzzle, valstr))
293 standing = match.group(match.lastindex)
294 standings[puzzle].append(standing)
298 for puzzle in puzzles:
299 sl = standings[puzzle]
301 skl.msg('puzzle "%s" multiple standings %s' %
305 skl.msg('puzzle "%s" no standing found' % puzzle)
308 for i in range(0, standing_limit):
309 if standing == standingvals[i]:
310 self.standings[puzzle] = i
311 if not puzzle in self.standings:
312 skl.msg('puzzle "%s" unknown standing "%s"' %
315 all_standings_ok = True
316 for puzzle in puzzles:
317 if not puzzle in self.standings:
320 def _find_crewflag(self, cf, yoweb_re):
321 things = self._soup.findAll('a', href=regexp.compile(yoweb_re))
323 self.msg('zero or several %s id references found' % cf)
326 id_re = '\\b%sid\\=(\\w+)$' % cf
327 id_haystack = thing['href']
328 match = regexp.compile(id_re).search(id_haystack)
330 self.soupm(thing, ('incomprehensible %s id ref'+
331 ' (%s in %s)') % (cf, id_re, id_haystack))
333 name = soup_text(thing)
334 return (match.group(1), name)
337 return `(self.crew, self.flag, self.standings, self.msgs)`
339 #---------- scraper for crew pages ----------
341 class CrewInfo(SomethingSoupInfo):
342 # Public data members:
343 # ci.crew = [ ('Captain', ['Pirate', ...]),
344 # ('Senior Officer', [...]),
346 # pi.msgs = [ 'message describing problem with scrape' ]
348 def __init__(self, crewid, max_age=300):
349 SomethingSoupInfo.__init__(self,
350 'crew/info.wm?crewid=', crewid, max_age)
353 def _find_crew(self):
355 capts = self._soup.findAll('img',
356 src='/yoweb/images/crew-captain.png')
358 self.msg('crew members: no. of captain images != 1')
361 while not tbl.find('a', href=pirate_ref_re):
362 tbl = tbl.findParent('table')
364 self.msg('crew members: cannot find table')
366 current_rank_crew = None
367 crew_rank_re = regexp.compile('/yoweb/images/crew')
368 for row in tbl.contents:
369 # findAll(recurse=False)
370 if isinstance(row,basestring):
373 is_rank = row.find('img', attrs={'src': crew_rank_re})
375 rank = soup_text(row)
376 current_rank_crew = []
377 self.crew.append((rank, current_rank_crew))
379 for cell in row.findAll('a', href=pirate_ref_re):
380 if current_rank_crew is None:
381 self.soupm(cell, 'crew members: crew'
384 current_rank_crew.append(soup_text(cell))
387 return `(self.crew, self.msgs)`
389 class FlagInfo(SomethingSoupInfo):
390 # Public data members (after init):
394 # relations[n] = (otherflagname, otherflagid, [stringfromyoweb],
395 # thisdeclaring, otherdeclaringmin, otherdeclaringmax)
396 # # where {this,other}declaring{,min,max} are:
397 # # -1 {this,other} is declaring war
398 # # 0 {this,other} is not doing either
399 # # +1 {this,other} is allying
400 # relation_byname[otherflagname] = relations[some_n]
401 # relation_byid[otherflagname] = relations[some_n]
403 # islands[n] = (islandname, islandid)
405 def __init__(self, flagid, max_age=600):
406 SomethingSoupInfo.__init__(self,
407 'flag/info.wm?flagid=', flagid, max_age)
410 def _find_flag(self):
411 font2 = self._soup.find('font',{'size':'+2'})
412 self.name = font2.find('b').contents[0]
415 self.relation_byname = { }
416 self.relation_byid = { }
419 magnate = self._soup.find('img',{'src':
420 '/yoweb/images/repute-MAGNATE.png'})
421 warinfo = (magnate.findParent('table').findParent('tr').
422 findNextSibling('tr').findNext('td',{'align':'left'}))
425 print >>sys.stderr, 'WARNING: '+m
427 def wi_warn(head, waritem):
428 warn('unknown warmap item: %s: %s' %
429 (`head`, ``waritem``))
431 def wihelp_item(waritem, thing):
432 if waritem.name == 'a':
433 url = waritem.get('href', None)
435 return ('no url for '+thing,None,None)
437 hr = waritem.find('a',{'href':True})
438 if not hr: return ('no a for '+thing,None,None)
440 m = regexp.search('\?'+thing+'id=(\d+)$', url)
441 if not m: return ('no '+thing+'id',None,None)
445 return (thing+' name not just string',None,None)
446 return (None,tid,tname)
448 def wi_alwar(head, waritem, thisdecl, othermin, othermax):
449 (err,flagid,flagname) = wihelp_item(waritem,'flag')
451 rel = self.relation_byid.get(flagid, None)
452 if rel: return 'flag id twice!'
453 if flagname in self.relation_byname:
454 return 'flag name twice!'
455 rel = (flagname,flagid,[], thisdecl,othermin,othermax)
456 self.relations.append(rel)
457 self.relation_byid[flagid] = rel
458 self.relation_byname[flagid] = rel
460 def wi_isle(head, waritem):
461 (err,isleid,islename) = wihelp_item(waritem,'island')
463 self.islands.append((isleid,islename))
466 'Allied with': (wi_alwar,+1,+1,+1),
467 'Declaring war against': (wi_alwar,-1, 0,+1),
468 'At war with': (wi_alwar,-1,-1,-1),
469 'Trying to form an alliance with': (wi_alwar,+1,-1,0),
470 'Islands controlled by this flag': (wi_isle,),
473 how = (wi_warn, None)
475 for waritem in warinfo.contents:
476 debug('WARITEM '+``waritem``)
477 if isinstance(waritem, unicode):
478 waritem = waritem.strip()
479 if waritem: warn('unknown waritem '+``waritem``)
481 if waritem.name == 'br':
483 if waritem.name == 'b':
484 head = ''.join(waritem.findAll(text=True))
485 head = regexp.sub('\\s+', ' ', head).strip()
486 head = head.rstrip(':')
487 how = (head,) + warmap.get(head, (wi_warn,))
489 debug('WARHOW %s(%s, waritem, *%s)' %
490 (how[1], `how[0]`, `how[2:]`))
491 bad = how[1](how[0], waritem, *how[2:])
493 warn('bad waritem %s: %s: %s' % (`how[0]`,
497 return `(self.name, self.islands, self.relations)`
499 #---------- scraper for ocean info incl. embargoes etc. ----------
502 def __init__(self, ocean, islename):
507 def yppedia_dataf(self):
508 def q(x): return urllib.quote(x.replace(' ','_'))
509 url_rhs = q(self.name) + '_(' + q(self.ocean) + ')'
510 if opts.localhtml is None:
511 url = 'http://yppedia.puzzlepirates.com/' + url_rhs
512 debug('IslandInfo retrieving YPP '+url);
513 return urllib.urlopen(url)
515 return file(opts.localhtml + '/' + url_rhs, 'r')
517 soup = BeautifulSoup(self.yppedia_dataf())
518 content = soup.find('div', attrs = {'id': 'content'})
519 yoweb_re = regexp.compile('^http://\w+\.puzzlepirates\.com/'+
520 'yoweb/island/info\.wm\?islandid=\d+$')
521 a = soup.find('a', attrs = { 'href': yoweb_re })
522 if a is None: return None
524 def ruling_flag_id(self):
525 yo = self.yoweb_url()
526 if yo is None: return None
527 dataf = fetcher.fetch(yo, 600)
528 soup = BeautifulSoup(dataf)
529 ruler_re = regexp.compile('http://\w+\.puzzlepirates\.com/'+
530 'yoweb/flag/info\.wm\?flagid=(\d+)$')
531 ruler = soup.find('a', attrs = { 'href': ruler_re })
532 if not ruler: return None
533 m = ruler_re.find(ruler['href'])
537 # Public data attributes (valid after collect()):
538 # oi.islands[islename] = IslandInfo(...)
539 # oi.arches[archname][islename] = IslandInfo(...)
541 self.isleclass = IslandInfo
542 self.ocean = fetcher.ocean.lower().capitalize()
544 cmdl = ['./yppedia-ocean-scraper']
545 if opts.localhtml is not None:
546 cmdl += ['--local-html-dir',opts.localhtml]
548 debug('OceanInfo collect running ' + `cmdl`)
549 oscraper = subprocess.Popen(
551 stdout = subprocess.PIPE,
552 cwd = yppsc_dir()+'/yarrg',
553 shell=False, stderr=None,
555 h = oscraper.stdout.readline()
556 debug('OceanInfo collect h '+`h`)
557 assert(regexp.match('^ocean ', h))
558 arch_re = regexp.compile('^ (\S.*)')
559 island_re = regexp.compile('^ (\S.*)')
565 for l in oscraper.stdout:
566 debug('OceanInfo collect l '+`l`)
568 m = island_re.match(l)
570 assert(archname is not None)
571 islename = m.group(1)
572 isle = self.isleclass(self.ocean, islename)
574 self.islands[islename] = isle
575 self.arches[archname][islename] = isle
579 archname = m.group(1)
580 assert(archname not in self.arches)
581 self.arches[archname] = { }
585 assert(oscraper.returncode == 0)
587 #---------- pretty-printer for tables of pirate puzzle standings ----------
589 class StandingsTable:
590 def __init__(self, f, use_puzzles=None, col_width=6, gap_every=5):
591 if use_puzzles is None:
593 use_puzzles=duty_puzzles
596 self._puzzles = use_puzzles
598 self._cw = col_width-1
599 self._gap_every = gap_every
603 def _nl(self): self._o('\n')
605 def _pline(self, lhs, puzstrs, extra):
606 if (self._linecount > 0
607 and self._gap_every is not None
608 and not (self._linecount % self._gap_every)):
610 self._o('%-*s' % (max(max_pirate_namelen+1, 15), lhs))
612 self._o(' %-*.*s' % (self._cw,self._cw, v))
618 def _puzstr(self, pi, puzzle):
619 if not isinstance(puzzle,list): puzzle = [puzzle]
620 try: standing = max([pi.standings[p] for p in puzzle])
621 except KeyError: return '?'
622 if not standing: return ''
625 c1 = standingvals[standing][0]
626 if standing < 3: c1 = c1.lower() # 3 = Master
630 s += '*' * (standing / 2)
631 s += '+' * (standing % 2)
634 def headings(self, lhs='', rhs=None):
635 def puzn_redact(name):
636 if isinstance(name,list):
638 ["%.*s" % (self._cw/2, puzn_redact(n))
641 if spc < 0: return name
642 return name[0:min(4,spc)] + name[spc+1:]
644 self._pline(lhs, map(puzn_redact, self._puzzles), rhs)
646 def literalline(self, line):
650 def pirate_dummy(self, name, standingstring, extra=None):
651 standings = standingstring * len(self._puzzles)
652 self._pline(' '+name, standings, extra)
653 def pirate(self, pi, extra=None):
654 puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles]
655 self._pline(' '+pi.name, puzstrs, extra)
658 #---------- chat log parser ----------
661 # This is essentially a transparent, dumb, data class.
671 # Also used for jobbing applicants:
672 # happens when expires (to "-")
673 # - disembark, leaves crew no
674 # aboard evidence of them being aboard no
675 # applied "has applied for the job" 120s, configurable
676 # ashore "has taken a job" 30min, configurable
677 # declined "declined the job offer" 30s, configurable
678 # invited "has been invited to job" 120s, configurable
680 # pa.jobber None, 'ashore', 'applied', 'invited', 'declined'
681 # pa.expires expiry time time
683 def __init__(pa, pn, v, time, event):
687 pa.last_event = event
688 pa.last_chat_time = None
689 pa.last_chat_chan = None
698 age = now - pa.pi_fetched
699 guide = random.randint(120,240)
702 debug('PirateAboard refresh %d > %d %s' % (
703 age, guide, pa.name))
707 wait = fetcher.need_wait(now, imaginary)
709 debug('PirateAboard fetcher not ready %d' % wait)
711 pa.pi = PirateInfo(pa.name, 600)
715 class ChatLogTracker:
716 # This is quite complex so we make it opaque. Use the
717 # official invokers, accessors etc.
719 def __init__(self, myself_pi, logfn):
720 self._pl = {} # self._pl['Pirate'] =
721 self._vl = {} # self._vl['Vessel']['Pirate'] = PirateAboard
722 # self._vl['Vessel']['#lastinfo']
723 # self._vl['Vessel']['#name']
724 # self._v = self._vl[self._vessel]
726 self._myself = myself_pi
728 self._f = file(logfn)
729 flen = os.fstat(self._f.fileno()).st_size
731 if flen > max_backlog:
732 startpos = flen - max_backlog
733 self._f.seek(startpos)
735 self._progress = [0, flen - self._f.tell()]
736 self._disembark_myself()
737 self._need_redisplay = False
738 self._lastvessel = None
740 def _disembark_myself(self):
743 self.force_redisplay()
745 def force_redisplay(self):
746 self._need_redisplay = True
748 def _vessel_updated(self, v, timestamp):
750 v['#lastinfo'] = timestamp
751 self.force_redisplay()
753 def _onboard_event(self,v,timestamp,pirate,event,jobber=None):
754 pa = self._pl.get(pirate, None)
755 if pa is not None and pa.v is v:
756 pa.last_time = timestamp
757 pa.last_event = event
759 if pa is not None and pa.v is not None:
761 pa = PirateAboard(pirate, v, timestamp, event)
762 self._pl[pirate] = pa
763 if v is not None: v[pirate] = pa
766 if jobber is None: timeout = None
767 else: timeout = getattr(opts, 'timeout_'+jobber)
768 if timeout is None: pa.expires = None
769 else: pa.expires = timestamp + timeout
770 self._vessel_updated(v, timestamp)
773 def _expire_jobbers(self, now):
774 for pa in self._pl.values():
775 if pa.expires is None: continue
776 if pa.expires >= now: continue
778 del self._pl[pa.name]
779 if v is not None: del v[pa.name]
780 self.force_redisplay()
782 def _trash_vessel(self, v):
784 if pn.startswith('#'): continue
788 if v is self._v: self._disembark_myself()
789 self.force_redisplay()
791 def _vessel_stale(self, v, timestamp):
792 return timestamp - v['#lastinfo'] > opts.ship_reboard_clearout
794 def _vessel_check_expire(self, v, timestamp):
795 if not self._vessel_stale(v, timestamp):
797 self._debug_line_disposition(timestamp,'',
798 'stale-reset ' + v['#name'])
799 self._trash_vessel(v)
802 def expire_garbage(self, timestamp):
803 for v in self._vl.values():
804 self._vessel_check_expire(v, timestamp)
806 def _vessel_lookup(self, vn, timestamp, dml=[], create=False):
807 v = self._vl.get(vn, None)
809 v = self._vessel_check_expire(v, timestamp)
816 self._vl[vn] = v = { '#name': vn }
817 self._vessel_updated(v, timestamp)
820 def _find_matching_vessel(self, pattern, timestamp, cmdr,
821 dml=[], create=False):
822 # use when a commander pirate `cmdr' specified a vessel
823 # by name `pattern' (either may be None)
824 # if create is true, will create the vessel
825 # record if an exact name is specified
827 if (pattern is not None and
829 and len(pattern.split(' ')) == 2):
832 return self._vessel_lookup(
833 vn, timestamp, dml=dml, create=create)
836 pattern_check = lambda vn: True
838 re = '(?:.* )?%s$' % pattern.lower().replace('*','.+')
839 pattern_check = regexp.compile(re, regexp.I).match
843 cmdr_pa = self._pl.get(cmdr, None)
844 if cmdr_pa: tries.append((cmdr_pa.v, 'cmdr'))
846 tries.append((self._v, 'here'))
849 for (v, dm) in tries:
850 if v is None: dml.append(dm+'?'); continue
853 if not pattern_check(vn):
861 if pattern is not None and '*' in pattern:
864 for (vn,v) in self._vl.iteritems()
865 if not self._vessel_stale(v, timestamp)
868 #debug('CLT-RE /%s/ wanted (%s) searched (%s)' % (
870 # '/'.join(tried_vns),
871 # '/'.join([vn for (vn,v) in search])))
881 def _debug_line_disposition(self,timestamp,l,m):
882 debug('CLT %13s %-40s %s' % (timestamp,m,l))
884 def _rm_crew_l(self,re,l):
885 m = regexp.match(re,l)
886 if m and m.group(2) == self._myself.crew[1]:
891 def local_command(self, metacmd):
892 # returns None if all went well, or problem message
893 return self._command(self._myself.name, metacmd,
894 "local", time.time(),
895 (lambda m: debug('CMD %s' % metacmd)))
897 def _command(self, cmdr, metacmd, chan, timestamp, d):
898 # returns None if all went well, or problem message
899 metacmd = regexp.sub('\\s+', ' ', metacmd).strip()
901 '/([adj]) (?:([A-Za-z* ]+)\\s*:)?([A-Za-z ]+)$',
903 if not m2: return "unknown syntax or command"
905 (cmd, pattern, targets) = m2.groups()
906 dml = ['cmd', chan, cmd]
908 if cmd == 'a': each = self._onboard_event
909 elif cmd == 'd': each = disembark
910 else: each = lambda *l: self._onboard_event(*l,
911 **{'jobber':'applied'})
913 if cmdr == self._myself.name:
915 how = 'cmd: %s' % cmd
918 how = 'cmd: %s %s' % (cmd,cmdr)
921 if pattern is not None:
922 return "/j command does not take a vessel"
925 v = self._find_matching_vessel(
926 pattern, timestamp, cmdr,
929 if cmd == 'j' or v is not None:
930 targets = targets.strip().split(' ')
931 dml.append(`len(targets)`)
932 for target in targets:
933 each(v, timestamp, target.title(), how)
934 self._vessel_updated(v, timestamp)
941 def chatline(self,l):
942 rm = lambda re: regexp.match(re,l)
943 d = lambda m: self._debug_line_disposition(timestamp,l,m)
944 rm_crew = lambda re: self._rm_crew_l(re,l)
947 m = rm('=+ (\\d+)/(\\d+)/(\\d+) =+$')
949 self._date = [int(x) for x in m.groups()]
950 self._previous_timestamp = None
951 return d('date '+`self._date`)
953 if self._date is None:
954 return d('date unset')
956 m = rm('\\[(\d\d):(\d\d):(\d\d)\\] ')
958 return d('no timestamp')
961 time_tuple = (self._date +
962 [int(x) for x in m.groups()] +
964 timestamp = time.mktime(time_tuple)
965 if timestamp >= self._previous_timestamp: break
967 self._debug_line_disposition(timestamp,'',
968 'new date '+`self._date`)
970 self._previous_timestamp = timestamp
972 l = l[l.find(' ')+1:]
974 def ob_x(pirate,event):
975 return self._onboard_event(
976 self._v, timestamp, pirate, event)
977 def ob1(did): ob_x(m.group(1), did); return d(did)
978 def oba(did): return ob1('%s %s' % (did, m.group(2)))
980 def jb(pirate,jobber):
981 return self._onboard_event(
982 None, timestamp, pirate,
983 ("jobber %s" % jobber),
987 def disembark(v, timestamp, pirate, event):
989 v, timestamp, pirate, 'leaving '+event)
993 def disembark_me(why):
994 self._disembark_myself()
995 return d('disembark-me '+why)
997 m = rm('Going aboard the (\\S.*\\S)\\.\\.\\.$')
1000 pn = self._myself.name
1002 v = self._vessel_lookup(vn, timestamp, dm, create=True)
1003 self._lastvessel = self._vessel = vn
1005 ob_x(pn, 'we boarded')
1006 self.expire_garbage(timestamp)
1007 return d(' '.join(dm))
1010 return d('no vessel')
1012 m = rm('(\\w+) has come aboard\\.$')
1013 if m: return ob1('boarded');
1015 m = rm('You have ordered (\\w+) to do some (\\S.*\\S)\\.$')
1017 (who,what) = m.groups()
1018 pa = ob_x(who,'ord '+what)
1019 if what == 'Gunning':
1021 return d('duty order')
1023 m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$')
1024 if m: oba('stopped'); return d("end")
1026 def chat_core(speaker, chan):
1027 try: pa = self._pl[speaker]
1028 except KeyError: return 'mystery'
1029 if pa.v is not None and pa.v is not self._v:
1031 pa.last_chat_time = timestamp
1032 pa.last_chat_chan = chan
1033 self.force_redisplay()
1037 speaker = m.group(1)
1038 dm = chat_core(speaker, chan)
1039 return d('chat %s %s' % (chan, dm))
1041 def chat_metacmd(chan):
1042 (cmdr, metacmd) = m.groups()
1043 whynot = self._command(
1044 cmdr, metacmd, chan, timestamp, d)
1045 if whynot is not None:
1048 chat_core(cmdr, 'cmd '+chan)
1050 m = rm('(\\w+) (?:issued an order|ordered everyone) "')
1051 if m: return ob1('general order');
1053 m = rm('(\\w+) says, "')
1054 if m: return chat('public')
1056 m = rm('(\\w+) tells ye, "')
1057 if m: return chat('private')
1059 m = rm('Ye told (\\w+), "(.*)"$')
1060 if m: return chat_metacmd('private')
1062 m = rm('(\\w+) flag officer chats, "')
1063 if m: return chat('flag officer')
1065 m = rm('(\\w+) officer chats, "(.*)"$')
1066 if m: return chat_metacmd('officer')
1068 m = rm('Ye accepted the offer to job with ')
1069 if m: return disembark_me('jobbing')
1071 m = rm('Ye hop on the ferry and are whisked away ')
1072 if m: return disembark_me('ferry')
1074 m = rm('Whisking away to yer home on the magical winds')
1075 if m: return disembark_me('home')
1077 m = rm('Game over\\. Winners: ([A-Za-z, ]+)\\.$')
1079 pl = m.group(1).split(', ')
1080 if not self._myself.name in pl:
1081 return d('lost melee')
1083 if ' ' in pn: continue
1084 ob_x(pn,'won melee')
1085 return d('won melee')
1087 m = rm('(\\w+) is eliminated\\!')
1088 if m: return ob1('eliminated in fray');
1090 m = rm('(\\w+) has driven \w+ from the ship\\!')
1091 if m: return ob1('boarder repelled');
1093 m = rm('\w+ has bested (\\w+), and turns'+
1094 ' to the rest of the ship\\.')
1095 if m: return ob1('boarder unrepelled');
1097 pirate = rm_crew("(\\w+) has taken a job with '(.*)'\\.")
1098 if pirate: return jb(pirate, 'ashore')
1100 pirate = rm_crew("(\\w+) has left '(.*)'\\.")
1102 disembark(self._v, timestamp, pirate, 'left crew')
1103 return d('left crew')
1105 m = rm('(\w+) has applied for the posted job\.')
1106 if m: return jb(m.group(1), 'applied')
1108 pirate= rm_crew("(\\w+) has been invited to job for '(.*)'\\.")
1109 if pirate: return jb(pirate, 'invited')
1111 pirate = rm_crew("(\\w+) declined the job offer for '(.*)'\\.")
1112 if pirate: return jb(pirate, 'declined')
1114 m = rm('(\\w+) has left the vessel\.')
1117 disembark(self._v, timestamp, pirate, 'disembarked')
1118 return d('disembarked')
1120 return d('not-matched')
1122 def _str_pa(self, pn, pa):
1123 assert self._pl[pn] == pa
1124 s = ' '*20 + "%s %-*s %13s %-30s %13s %-20s %13s" % (
1125 (' ','G')[pa.gunner],
1126 max_pirate_namelen, pn,
1127 pa.last_time, pa.last_event,
1128 pa.last_chat_time, pa.last_chat_chan,
1130 if pa.expires is not None:
1131 s += " %-5d" % (pa.expires - pa.last_time)
1135 def _str_vessel(self, vn, v):
1136 s = ' vessel %s\n' % vn
1137 s += ' '*20 + "%-*s %13s\n" % (
1138 max_pirate_namelen, '#lastinfo',
1140 assert v['#name'] == vn
1141 for pn in sorted(v.keys()):
1142 if pn.startswith('#'): continue
1145 s += self._str_pa(pn,pa)
1149 s = '''<ChatLogTracker
1152 ''' % (self._myself.name, self._vessel)
1153 assert ((self._v is None and self._vessel is None) or
1154 (self._v is self._vl[self._vessel]))
1155 if self._vessel is not None:
1156 s += self._str_vessel(self._vessel, self._v)
1157 for vn in sorted(self._vl.keys()):
1158 if vn == self._vessel: continue
1159 s += self._str_vessel(vn, self._vl[vn])
1163 if pa.v is not None:
1164 assert pa.v[p] is pa
1165 assert pa.v in self._vl.values()
1167 s += self._str_pa(pa.name, pa)
1171 def catchup(self, progress=None):
1173 more = self._f.readline()
1176 self._progress[0] += len(more)
1177 if progress: progress.progress(*self._progress)
1180 if self._lbuf.endswith('\n'):
1181 self.chatline(self._lbuf.rstrip())
1184 debug(self.__str__())
1185 self._expire_jobbers(time.time())
1187 if progress: progress.caughtup()
1190 rv = self._need_redisplay
1191 self._need_redisplay = False
1194 # returns our pirate name
1195 return self._myself.name
1196 def vesselname(self):
1197 # returns the vessel name we're aboard or None
1199 def lastvesselname(self):
1200 # returns the last vessel name we were aboard or None
1201 return self._lastvessel
1202 def aboard(self, vesselname=True):
1203 # returns a list of PirateAboard the vessel
1204 # sorted by pirate name
1205 # you can pass this None and you'll get []
1206 # or True for the current vessel (which is the default)
1207 # the returned value is a fresh list of persistent
1208 # PirateAboard objects
1209 if vesselname is True: v = self._v
1210 else: v = self._vl.get(vesselname.title())
1211 if v is None: return []
1213 for pn in sorted(v.keys())
1214 if not pn.startswith('#') ]
1216 # returns a the jobbers' PirateAboards,
1217 # sorted by jobber class and reverse of expiry time
1219 for pa in self._pl.values()
1220 if pa.jobber is not None
1223 return (pa.jobber, -pa.expires)
1224 l.sort(key = compar_key)
1227 #---------- implementations of actual operation modes ----------
1229 def do_pirate(pirates, bu):
1231 for pirate in pirates:
1232 info = PirateInfo(pirate)
1233 print '%s: %s,' % (`pirate`, info)
1236 def prep_crew_of(args, bu, max_age=300):
1237 if len(args) != 1: bu('crew-of takes one pirate name')
1238 pi = PirateInfo(args[0], max_age)
1239 if pi.crew is None: return None
1240 return CrewInfo(pi.crew[0], max_age)
1242 def do_crew_of(args, bu):
1243 ci = prep_crew_of(args, bu)
1246 def do_flag_of(args, bu):
1247 if len(args) != 1: bu('flag-of takes one pirate name')
1249 pi = PirateInfo(args[0], max_age)
1250 if pi.flag is None: fi = None
1251 else: fi = FlagInfo(pi.flag[0], max_age)
1254 def do_standings_crew_of(args, bu):
1255 ci = prep_crew_of(args, bu, 60)
1256 tab = StandingsTable(sys.stdout)
1258 for (rank, members) in ci.crew:
1259 if not members: continue
1261 tab.literalline('%s:' % rank)
1263 pi = PirateInfo(p, random.randint(900,1800))
1266 def do_ocean(args, bu):
1267 if (len(args)): bu('ocean takes no further arguments')
1268 fetcher.default_ocean()
1271 for islename in sorted(oi.islands.keys()):
1272 isle = oi.islands[islename]
1273 yoweb_url = isle.yoweb_url()
1274 print " %s -- %s" % (islename, yoweb_url)
1276 #----- modes which use the chat log parser are quite complex -----
1278 class ProgressPrintPercentage:
1279 def __init__(self, f=sys.stdout):
1281 def progress_string(self,done,total):
1282 return "scan chat logs %3d%%\r" % ((done*100) / total)
1283 def progress(self,*a):
1284 self._f.write(self.progress_string(*a))
1286 def show_init(self, pirate, ocean):
1287 print >>self._f, 'Starting up, %s on the %s ocean' % (
1290 self._f.write(' \r')
1293 def prep_chat_log(args, bu,
1294 progress=ProgressPrintPercentage(),
1295 max_myself_age=3600):
1296 if len(args) != 1: bu('this action takes only chat log filename')
1298 logfn_re = '(?:.*/)?([A-Z][a-z]+)_([a-z]+)_'
1299 match = regexp.match(logfn_re, logfn)
1300 if not match: bu('chat log filename is not in expected format')
1301 (pirate, ocean) = match.groups()
1302 fetcher.default_ocean(ocean)
1304 progress.show_init(pirate, fetcher.ocean)
1305 myself = PirateInfo(pirate,max_myself_age)
1306 track = ChatLogTracker(myself, logfn)
1309 track.catchup(progress)
1312 track.force_redisplay()
1314 return (myself, track)
1316 def do_track_chat_log(args, bu):
1317 (myself, track) = prep_chat_log(args, bu)
1322 sleep(0.5 + 0.5 * random.random())
1324 #----- ship management aid -----
1326 class Display_dumb(ProgressPrintPercentage):
1328 ProgressPrintPercentage.__init__(self)
1331 def realstart(self):
1334 class Display_overwrite(ProgressPrintPercentage):
1336 ProgressPrintPercentage.__init__(self)
1338 null = file('/dev/null','w')
1339 curses.setupterm(fd=null.fileno())
1341 self._clear = curses.tigetstr('clear')
1343 self._debug('missing clear!')
1344 self.show = Display_dumb.show
1347 self._t = {'el':'', 'ed':''}
1348 if not self._init_sophisticated():
1349 for k in self._t.keys(): self._t[k] = ''
1350 self._t['ho'] = self._clear
1352 def _debug(self,m): debug('display overwrite: '+m)
1354 def _init_sophisticated(self):
1355 for k in self._t.keys():
1356 s = curses.tigetstr(k)
1358 self._t['ho'] = curses.tigetstr('ho')
1359 if not self._t['ho']:
1360 cup = curses.tigetstr('cup')
1361 self._t['ho'] = curses.tparm(cup,0,0)
1362 missing = [k for k in self._t.keys() if not self._t[k]]
1364 self.debug('missing '+(' '.join(missing)))
1369 w = sys.stdout.write
1370 def wti(k): w(self._t[k])
1374 for l in s.rstrip().split('\n'):
1383 def realstart(self):
1384 sys.stdout.write(self._clear)
1388 def do_ship_aid(args, bu):
1389 if opts.ship_duty is None: opts.ship_duty = True
1391 displayer = globals()['Display_'+opts.display]()
1393 (myself, track) = prep_chat_log(args, bu, progress=displayer)
1395 displayer.realstart()
1397 if os.isatty(0): kr_create = KeystrokeReader
1398 else: kr_create = DummyKeystrokeReader
1401 kreader = kr_create(0, 10)
1402 ship_aid_core(myself, track, displayer, kreader)
1407 class KeyBasedSorter:
1408 def compar_key_pa(self, pa):
1409 pi = pa.pirate_info()
1410 if pi is None: return None
1411 return self.compar_key(pi)
1412 def lsort_pa(self, l):
1413 l.sort(key = self.compar_key_pa)
1415 class NameSorter(KeyBasedSorter):
1416 def compar_key(self, pi): return pi.name
1417 def desc(self): return 'name'
1419 class SkillSorter(NameSorter):
1420 def __init__(self, relevant):
1421 self._want = frozenset(relevant.split('/'))
1423 for p in core_duty_puzzles:
1424 if isinstance(p,basestring): self._avoid.add(p)
1425 else: self._avoid |= set(p)
1426 self._avoid -= self._want
1427 self._desc = '%s' % relevant
1429 def desc(self): return self._desc
1431 def compar_key(self, pi):
1433 pi.standings.get(puz,-1)
1434 for puz in self._want
1437 -pi.standings.get(puz,standing_limit)
1438 for puz in self._avoid
1441 def negate(x): return -x
1442 debug('compar_key %s bw=%s ba=%s' % (pi.name, `best_want`,
1444 return (-best_want, map(negate, best_avoid), pi.name)
1446 def ship_aid_core(myself, track, displayer, kreader):
1449 vn = track.vesselname()
1450 if vn: return (vn, " on board the %s" % vn)
1451 vn = track.lastvesselname()
1452 if vn: return (vn, " ashore from the %s" % vn)
1453 return (None, " not on a vessel")
1456 if t is None: return ' ' * 22
1457 return " %-4s %-16s" % (format_time_interval(now - t),e)
1459 displayer.show(track.myname() + find_vessel()[1] + '...')
1472 (vn, vs) = find_vessel()
1475 if cliexec is not None:
1477 elif clierr is not None:
1478 s += 'Error: '+clierr
1479 elif clicmd is not None:
1482 s = track.myname() + vs
1483 s += " at %s" % time.strftime("%Y-%m-%d %H:%M:%S")
1488 tbl = StandingsTable(tbl_s)
1490 aboard = track.aboard(vn)
1491 sort.lsort_pa(aboard)
1493 jobbers = track.jobbers()
1495 if track.vesselname(): howmany = 'aboard: %2d' % len(aboard)
1498 tbl.headings(howmany, ' sorted by '+sort.desc())
1502 for pa in aboard + jobbers:
1503 if pa.jobber != last_jobber:
1504 last_jobber = pa.jobber
1506 tbl.literalline('jobbers '+last_jobber)
1508 pi = pa.pirate_info()
1511 if pa.gunner: xs += 'G '
1513 xs += timeevent(pa.last_time, pa.last_event)
1514 xs += timeevent(pa.last_chat_time, pa.last_chat_chan)
1517 tbl.pirate_dummy(pa.name, rotate_nya[0], xs)
1521 s += tbl_s.getvalue()
1525 if cliexec is not None:
1526 clierr = track.local_command("/"+cliexec.strip())
1532 rotate_nya = rotate_nya[1:3] + rotate_nya[0]
1535 if clierr is not None:
1539 if clicmd is not None:
1540 if k == '\r' or k == '\n':
1543 elif k == '\e' and clicmd != "":
1547 elif k == '\b' or k == '\177':
1548 clicmd = clicmd[ 0 : len(clicmd)-1 ]
1554 elif k == 'g': sort = SkillSorter('Gunning')
1555 elif k == 'c': sort = SkillSorter('Carpentry')
1556 elif k == 's': sort = SkillSorter('Sailing/Rigging')
1557 elif k == 'b': sort = SkillSorter('Bilging')
1558 elif k == 'n': sort = SkillSorter('Navigating')
1559 elif k == 'd': sort = SkillSorter('Battle Navigation')
1560 elif k == 't': sort = SkillSorter('Treasure Haul')
1561 elif k == 'a': sort = NameSorter()
1562 elif k == '/': clicmdbase = ""; clicmd = clicmdbase
1563 elif k == '+': clicmdbase = "a "; clicmd = clicmdbase
1564 else: pass # unknown key command
1566 #---------- individual keystroke input ----------
1568 class DummyKeystrokeReader:
1569 def __init__(self,fd,timeout_dummy): pass
1570 def stop(self): pass
1571 def getch(self): sleep(1); return None
1572 def info(self): return ' [noninteractive]'
1574 class KeystrokeReader(DummyKeystrokeReader):
1575 def __init__(self, fd, timeout_decisec=0):
1577 self._saved = termios.tcgetattr(fd)
1578 a = termios.tcgetattr(fd)
1579 a[3] &= ~(termios.ECHO | termios.ECHONL |
1580 termios.ICANON | termios.IEXTEN)
1581 a[6][termios.VMIN] = 0
1582 a[6][termios.VTIME] = timeout_decisec
1583 termios.tcsetattr(fd, termios.TCSANOW, a)
1585 termios.tcsetattr(self._fd, termios.TCSANOW, self._saved)
1588 byte = os.read(self._fd, 1)
1589 if not len(byte): return None
1594 #---------- main program ----------
1597 global opts, fetcher
1600 '''usage: .../yoweb-scrape [OPTION...] ACTION [ARGS...]
1602 yoweb-scrape [--ocean OCEAN ...] pirate PIRATE
1603 yoweb-scrape [--ocean OCEAN ...] crew-of PIRATE
1604 yoweb-scrape [--ocean OCEAN ...] standings-crew-of PIRATE
1605 yoweb-scrape [--ocean OCEAN ...] track-chat-log CHAT-LOG
1606 yoweb-scrape [options] ship-aid CHAT-LOG (must be .../PIRATE_OCEAN_chat-log*)
1608 display modes (for --display) apply to ship-aid:
1609 --display=dumb just print new information, scrolling the screen
1610 --display=overwrite use cursor motion, selective clear, etc. to redraw at top''')
1612 ao('-O','--ocean',dest='ocean', metavar='OCEAN', default=None,
1613 help='select ocean OCEAN')
1614 ao('--cache-dir', dest='cache_dir', metavar='DIR',
1615 default='~/.yoweb-scrape-cache',
1616 help='cache yoweb pages in DIR')
1617 ao('-D','--debug', action='count', dest='debug', default=0,
1618 help='enable debugging output')
1619 ao('--debug-fd', type='int', dest='debug_fd',
1620 help='write any debugging output to specified fd')
1621 ao('-q','--quiet', action='store_true', dest='quiet',
1622 help='suppress warning output')
1623 ao('--display', action='store', dest='display',
1624 type='choice', choices=['dumb','overwrite'],
1625 help='how to display ship aid')
1626 ao('--local-ypp-dir', action='store', dest='localhtml',
1627 help='get yppedia pages from local directory LOCALHTML'+
1628 ' instead of via HTTP')
1630 ao_jt = lambda wh, t: ao(
1631 '--timeout-sa-'+wh, action='store', dest='timeout_'+wh,
1632 default=t, help=('set timeout for expiring %s jobbers' % wh))
1633 ao_jt('applied', 120)
1634 ao_jt('invited', 120)
1635 ao_jt('declined', 30)
1636 ao_jt('ashore', 1800)
1638 ao('--ship-duty', action='store_true', dest='ship_duty',
1639 help='show ship duty station puzzles')
1640 ao('--all-puzzles', action='store_false', dest='ship_duty',
1641 help='show all puzzles, not just ship duty stations')
1643 ao('--min-cache-reuse', type='int', dest='min_max_age',
1644 metavar='SECONDS', default=60,
1645 help='always reuse cache yoweb data if no older than this')
1647 (opts,args) = pa.parse_args()
1651 print >>sys.stderr, copyright_info
1652 pa.error('need a mode argument')
1654 if opts.debug_fd is not None:
1655 opts.debug_file = os.fdopen(opts.debug_fd, 'w')
1657 opts.debug_file = sys.stdout
1660 mode_fn_name = 'do_' + mode.replace('_','#').replace('-','_')
1661 try: mode_fn = globals()[mode_fn_name]
1662 except KeyError: pa.error('unknown mode "%s"' % mode)
1665 opts.expire_age = max(3600, opts.min_max_age)
1667 opts.ship_reboard_clearout = 3600
1669 if opts.cache_dir.startswith('~/'):
1670 opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]
1672 if opts.display is None:
1673 if ((opts.debug > 0 and opts.debug_fd is None)
1674 or not os.isatty(sys.stdout.fileno())):
1675 opts.display = 'dumb'
1677 opts.display = 'overwrite'
1679 fetcher = Fetcher(opts.ocean, opts.cache_dir)
1681 mode_fn(args[1:], pa.error)