yoweb-scrape

   1 #!/usr/bin/python
   2
   3 import signal
   4 signal.signal(signal.SIGINT, signal.SIG_DFL)
   5
   6 import os
   7 import time
   8 import urllib
   9 import urllib2
  10 import errno
  11 import sys
  12 import re as regexp
  13 import random
  14 from optparse import OptionParser
  15
  16 from BeautifulSoup import BeautifulSoup
  17
  18 opts = None
  19
  20
  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('/')
  25
  26 standingvals = ('Able/Distinguished/Respected/Master'+
  27                 '/Renowned/Grand-Master/Legendary/Ultimate').split('/')
  28
  29 pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm')
  30
  31 max_pirate_namelen = 12
  32
  33
  34 def debug(m):
  35         if opts.debug > 0:
  36                 print m
  37
  38 class Fetcher:
  39         def __init__(self, ocean, cachedir):
  40                 debug('Fetcher init %s' % cachedir)
  41                 self.ocean = ocean
  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())
  47
  48         def _default_ocean(self):
  49                 if self.ocean is None:
  50                         self.ocean = 'ice'
  51
  52         def _cache_scan(self, now):
  53                 # returns list of ages, unsorted
  54                 ages = []
  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
  62                                 continue
  63                         age = now - s.st_mtime
  64                         if age > opts.expire_age:
  65                                 debug('Fetcher    expire %d %s' % (age, path))
  66                                 try: os.remove(path)
  67                                 except (OSError,IOError), oe:
  68                                         if oe.errno != errno.ENOENT: raise
  69                                 continue
  70                         ages.append(age)
  71                 return ages
  72
  73         def need_wait(self, now):
  74                 ages = self._cache_scan(now)
  75                 ages.sort()
  76                 debug('Fetcher   ages ' + `ages`)
  77                 min_age = 1
  78                 need_wait = 0
  79                 for age in ages:
  80                         if age < min_age and age < 300:
  81                                 debug('Fetcher   morewait min=%d age=%d' %
  82                                         (min_age, age))
  83                                 need_wait = max(need_wait, min_age - age)
  84                         min_age += 3
  85                         min_age *= 1.25
  86                 return need_wait
  87
  88         def _rate_limit_cache_clean(self, now):
  89                 need_wait = self.need_wait(now)
  90                 if need_wait > 0:
  91                         debug('Fetcher   wait %d' % need_wait)
  92                         time.sleep(need_wait)
  93
  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
 101                         f = None
 102                 now = time.time()
 103                 max_age = max(opts.min_max_age, min(max_age, opts.expire_age))
 104                 if f is not None:
 105                         s = os.fstat(f.fileno())
 106                         age = now - s.st_mtime
 107                         if age > max_age:
 108                                 debug('Fetcher  stale %d < %d'% (max_age, age))
 109                                 f = None
 110                 if f is not None:
 111                         data = f.read()
 112                         f.close()
 113                         debug('Fetcher  cached %d > %d' % (max_age, age))
 114                         return data
 115
 116                 debug('Fetcher  fetch')
 117                 self._rate_limit_cache_clean(now)
 118
 119                 stream = urllib2.urlopen(url)
 120                 data = stream.read()
 121                 cache_tmp = "%s/#%s~%d#" % (
 122                         self.cachedir, cache_corename, os.getpid())
 123                 f = file(cache_tmp, 'w')
 124                 f.write(data)
 125                 f.close()
 126                 os.rename(cache_tmp, cache_item)
 127                 debug('Fetcher  stored')
 128                 return data
 129
 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)
 135
 136 class SoupLog:
 137         def __init__(self):
 138                 self.msgs = [ ]
 139         def msg(self, m):
 140                 self.msgs.append(m)
 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 = [ ]
 146
 147 def soup_text(obj):
 148         str = ''.join(obj.findAll(text=True))
 149         return str.strip()
 150
 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
 157                         )
 158
 159 class PirateInfo(SomethingSoupInfo):
 160         # Public data members:
 161         #  pi.standings = { 'Treasure Haul': 'Able' ... }
 162         #  pi.name = name
 163         #  pi.crew = (id, name)
 164         #  pi.flag = (id, name)
 165         #  pi.msgs = [ 'message describing problem with scrape' ]
 166
 167         def __init__(self, pirate, max_age=300):
 168                 SomethingSoupInfo.__init__(self,
 169                         'pirate.wm?target=', pirate, max_age)
 170                 self.name = pirate
 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')
 176
 177         def _find_standings(self):
 178                 imgs = self._soup.findAll('img',
 179                         src=regexp.compile('/yoweb/images/stat.*'))
 180                 re = regexp.compile(
 181 u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-Za-z]+)\\)\\s*$'
 182                         )
 183                 standings = { }
 184
 185                 for skill in puzzles:
 186                         standings[skill] = [ ]
 187
 188                 skl = SoupLog()
 189
 190                 for img in imgs:
 191                         try: puzzle = img['alt']
 192                         except KeyError: continue
 193
 194                         if not puzzle in puzzles:
 195                                 skl.soupm(img, 'unknown puzzle: "%s"' % puzzle)
 196                                 continue
 197                         key = img.findParent('td')
 198                         if key is None:
 199                                 skl.soupm(img, 'puzzle at root! "%s"' % puzzle)
 200                                 continue
 201                         valelem = key.findNextSibling('td')
 202                         if valelem is None:
 203                                 skl.soupm(key, 'puzzle missing sibling "%s"'
 204                                         % puzzle)
 205                                 continue
 206                         valstr = soup_text(valelem)
 207                         match = re.match(valstr)
 208                         if match is None:
 209                                 skl.soupm(key, ('puzzle "%s" unparseable'+
 210                                         ' standing "%s"') % (puzzle, valstr))
 211                                 continue
 212                         standing = match.group(match.lastindex)
 213                         standings[puzzle].append(standing)
 214
 215                 self.standings = { }
 216
 217                 for puzzle in puzzles:
 218                         sl = standings[puzzle]
 219                         if len(sl) > 1:
 220                                 skl.msg('puzzle "%s" multiple standings %s' %
 221                                                 (puzzle, `sl`))
 222                                 continue
 223                         if not sl:
 224                                 skl.msg('puzzle "%s" no standing found' % puzzle)
 225                                 continue
 226                         standing = sl[0]
 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"' %
 232                                         (puzzle, standing))
 233
 234                 all_standings_ok = True
 235                 for puzzle in puzzles:
 236                         if not puzzle in self.standings:
 237                                 self.needs_msgs(skl)
 238
 239         def _find_crewflag(self, cf, yoweb_re):
 240                 things = self._soup.findAll('a', href=regexp.compile(yoweb_re))
 241                 if len(things) != 1:
 242                         self.msg('zero or several %s id references found' % cf)
 243                         return None
 244                 thing = things[0]
 245                 id_re = '\\b%sid\\=(\\w+)$' % cf
 246                 id_haystack = thing['href']
 247                 match = regexp.compile(id_re).search(id_haystack)
 248                 if match is None:
 249                         self.soupm(thing, ('incomprehensible %s id ref'+
 250                                 ' (%s in %s)') % (cf, id_re, id_haystack))
 251                         return None
 252                 name = soup_text(thing)
 253                 return (match.group(1), name)
 254
 255         def __str__(self):
 256                 return `(self.crew, self.flag, self.standings, self.msgs)`
 257
 258 class CrewInfo(SomethingSoupInfo):
 259         # Public data members:
 260         #  ci.crew = [ ('Captain',        ['Pirate', ...]),
 261         #              ('Senior Officer', [...]),
 262         #               ... ]
 263         #  pi.msgs = [ 'message describing problem with scrape' ]
 264
 265         def __init__(self, crewid, max_age=300):
 266                 SomethingSoupInfo.__init__(self,
 267                         'crew/info.wm?crewid=', crewid, max_age)
 268                 self._find_crew()
 269
 270         def _find_crew(self):
 271                 self.crew = []
 272                 capts = self._soup.findAll('img',
 273                         src='/yoweb/images/crew-captain.png')
 274                 if len(capts) != 1:
 275                         self.msg('crew members: no. of captain images != 1')
 276                         return
 277                 tbl = capts[0]
 278                 while not tbl.find('a', href=pirate_ref_re):
 279                         tbl = tbl.findParent('table')
 280                         if not tbl:
 281                                 self.msg('crew members: cannot find table')
 282                                 return
 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):
 288                                 continue
 289
 290                         is_rank = row.find('img', attrs={'src': crew_rank_re})
 291                         if is_rank:
 292                                 rank = soup_text(row)
 293                                 current_rank_crew = []
 294                                 self.crew.append((rank, current_rank_crew))
 295                                 continue
 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'
 299                                                 ' before rank')
 300                                         continue
 301                                 current_rank_crew.append(soup_text(cell))
 302
 303         def __str__(self):
 304                 return `(self.crew, self.msgs)`
 305
 306 class StandingsTable:
 307         def __init__(self, use_puzzles=None, col_width=6):
 308                 if use_puzzles is None:
 309                         if opts.ship_duty:
 310                                 use_puzzles=[
 311                                         'Navigating','Battle Navigation',
 312                                         'Gunning',
 313                                         ['Sailing','Rigging'],
 314                                         'Bilging',
 315                                         'Carpentry',
 316                                         'Treasure Haul'
 317                                 ]
 318                         else:
 319                                 use_puzzles=puzzles
 320                 self._puzzles = use_puzzles
 321                 self.s = ''
 322                 self._cw = col_width-1
 323
 324         def _pline(self, pirate, puzstrs, extra):
 325                 self.s += ' %-*s' % (max(max_pirate_namelen, 14), pirate)
 326                 for v in puzstrs:
 327                         self.s += ' %-*.*s' % (self._cw,self._cw, v)
 328                 if extra:
 329                         self.s += ' ' + extra
 330                 self.s += '\n'
 331
 332         def _puzstr(self, pi, puzzle):
 333                 if not isinstance(puzzle,list): puzzle = [puzzle]
 334                 try: standing = max([pi.standings[p] for p in puzzle])
 335                 except KeyError: return '?'
 336                 if not standing: return ''
 337                 s = ''
 338                 if self._cw > 4:
 339                         c1 = standingvals[standing][0]
 340                         if standing < 3: c1 = c1.lower() # 3 = Master
 341                         s += `standing`
 342                 if self._cw > 5:
 343                         s += ' '
 344                 s += '*' * (standing / 2)
 345                 s += '+' * (standing % 2)
 346                 return s
 347
 348         def headings(self):
 349                 def puzn_redact(name):
 350                         if isinstance(name,list):
 351                                 return '/'.join(
 352                                         ["%.*s" % (self._cw/2, puzn_redact(n))
 353                                          for n in name])
 354                         spc = name.find(' ')
 355                         if spc < 0: return name
 356                         return name[0:min(4,spc)] + name[spc+1:]
 357                 self._pline('', map(puzn_redact, self._puzzles), None)
 358         def literalline(self, line):
 359                 self.s += line + '\n'
 360         def pirate_dummy(self, name, standingstring, extra=None):
 361                 self._pline(name, standingstring * len(self._puzzles), extra)
 362         def pirate(self, pi, extra=None):
 363                 puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles]
 364                 self._pline(pi.name, puzstrs, extra)
 365
 366         def results(self):
 367                 return self.s
 368
 369 def do_pirate(pirates, bu):
 370         print '{'
 371         for pirate in pirates:
 372                 info = PirateInfo(pirate)
 373                 print '%s: %s,' % (`pirate`, info)
 374         print '}'
 375
 376 def prep_crew_of(args, bu, max_age=300):
 377         if len(args) != 1: bu('crew-of takes one pirate name')
 378         pi = PirateInfo(args[0], max_age)
 379         if pi.crew is None: return None
 380         return CrewInfo(pi.crew[0], max_age)
 381
 382 def do_crew_of(args, bu):
 383         ci = prep_crew_of(args, bu)
 384         print ci
 385
 386 def do_standings_crew_of(args, bu):
 387         ci = prep_crew_of(args, bu, 60)
 388         tab = StandingsTable()
 389         tab.headings()
 390         for (rank, members) in ci.crew:
 391                 if not members: continue
 392                 tab.literalline('%s:' % rank)
 393                 for p in members:
 394                         pi = PirateInfo(p, random.randint(900,1800))
 395                         tab.pirate(pi)
 396         print tab.results()
 397
 398 class PirateAboard:
 399         # pa.v
 400         # pa.name
 401         # pa.last_time
 402         # pa.last_event
 403         # pa.gunner
 404         # pa.last_chat_time
 405         # pa.last_chat_chan
 406         # pa.pi
 407
 408         def __init__(pa, pn, v, time, event):
 409                 pa.name = pn
 410                 pa.v = v
 411                 pa.last_time = time
 412                 pa.last_event = event
 413                 pa.last_chat_time = None
 414                 pa.last_chat_chan = None
 415                 pa.gunner = False
 416                 pa.pi = None
 417
 418         def pirate_info(pa):
 419                 if not pa.pi and not fetcher.need_wait(time.time()):
 420                         pa.pi = PirateInfo(pa.name, 3600)
 421                 return pa.pi
 422
 423 class ChatLogTracker:
 424         def __init__(self, myself_pi, logfn):
 425                 self._pl = {}   # self._pl['Pirate'] =
 426                 self._vl = {}   #   self._vl['Vessel']['Pirate'] = PirateAboard
 427                                 # self._vl['Vessel']['#lastaboard']
 428                 self._v = None          # self._v =
 429                 self._vessel = None     #       self._vl[self._vessel]
 430                 self._date = None
 431                 self._myself = myself_pi
 432                 self._need_redisplay = False
 433                 self._f = file(logfn)
 434                 self._lbuf = ''
 435                 self._progress = [0, os.fstat(self._f.fileno()).st_size]
 436
 437         def _refresh(self):
 438                 self._need_redisplay = True
 439
 440         def _onboard_event(self,timestamp,pirate,event):
 441                 try: pa = self._pl[pirate]
 442                 except KeyError: pa = None
 443                 if pa is not None and pa.v is self._v:
 444                         pa.last_time = timestamp
 445                         pa.last_event = event
 446                 else:
 447                         if pa is not None: del pa.v[pirate]
 448                         pa = PirateAboard(pirate, self._v, timestamp, event)
 449                         self._pl[pirate] = pa
 450                         self._v[pirate] = pa
 451                 self._v['#lastaboard'] = timestamp
 452                 self._refresh()
 453                 return pa
 454
 455         def _trash_vessel(self, v):
 456                 for pn in v:
 457                         if pn.startswith('#'): continue
 458                         del self._pl[pn]
 459                 self._refresh()
 460
 461         def expire_garbage(self, timestamp):
 462                 for (vn,v) in list(self._vl.iteritems()):
 463                         la = v['#lastaboard']
 464                         if timestamp - la > opts.ship_reboard_clearout:
 465                                 self._debug_line_disposition(timestamp,'',
 466                                         'stale reset '+vn)
 467                                 self._trash_vessel(v)
 468                                 del self._vl[vn]
 469
 470         def clear_vessel(self, timestamp):
 471                 if self._v is not None:
 472                         self._trash_vessel(self._v)
 473                 self._v = {'#lastaboard': timestamp}
 474                 self._vl[self._vessel] = self._v
 475
 476         def _debug_line_disposition(self,timestamp,l,m):
 477                 debug('CLT %13s %-30s %s' % (timestamp,m,l))
 478
 479         def chatline(self,l):
 480                 rm = lambda re: regexp.match(re,l)
 481                 d = lambda m: self._debug_line_disposition(timestamp,l,m)
 482                 timestamp = None
 483
 484                 m = rm('=+ (\\d+)/(\\d+)/(\\d+) =+$')
 485                 if m:
 486                         self._date = m.groups()
 487                         return d('date '+`self._date`)
 488
 489                 if self._date is None:
 490                         return d('date unset')
 491
 492                 m = rm('\\[(\d\d):(\d\d):(\d\d)\\] ')
 493                 if not m:
 494                         return d('no timestamp')
 495
 496                 time_tuple = [int(x) for x in self._date + m.groups()]
 497                 time_tuple += (-1,-1,-1)
 498                 timestamp = time.mktime(time_tuple)
 499                 l = l[l.find(' ')+1:]
 500
 501                 def ob_x(who,event):
 502                         return self._onboard_event(timestamp, who, event)
 503                 def ob1(did): ob_x(m.group(1), did); return d(did)
 504                 def oba(did): return ob1('%s %s' % (did, m.group(2)))
 505
 506                 m = rm('Going aboard the (\\S.*\\S)\\.\\.\\.$')
 507                 if m:
 508                         pn = self._myself.name
 509                         self._vessel = m.group(1)
 510                         dm = 'boarding'
 511
 512                         try:             self._v = self._vl[self._vessel]
 513                         except KeyError: self._v = None; dm += ' new'
 514
 515                         if self._v is not None:  la = self._v['#lastaboard']
 516                         else:                    la = 0; dm += ' ?la'
 517
 518                         if timestamp - la > opts.ship_reboard_clearout:
 519                                 self.clear_vessel(timestamp)
 520                                 dm += ' stale'
 521
 522                         ob_x(pn, 'we boarded')
 523                         self.expire_garbage(timestamp)
 524                         return d(dm)
 525
 526                 if self._v is None:
 527                         return d('no vessel')
 528
 529                 m = rm('(\\w+) has come aboard\\.$')
 530                 if m: return ob1('boarded');
 531
 532                 m = rm('You have ordered (\\w+) to do some (\\S.*\\S)\\.$')
 533                 if m:
 534                         (who,what) = m.groups()
 535                         pa = ob_x(who,'ord '+what)
 536                         if what == 'Gunning':
 537                                 pa.gunner = True
 538                         return d('duty order')
 539
 540                 m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$')
 541                 if m: oba('stopped'); return d("end")
 542
 543                 def chat(what):
 544                         who = m.group(1)
 545                         try: pa = self._pl[who]
 546                         except KeyError: return d('chat mystery')
 547                         if pa.v is self._v:
 548                                 pa.last_chat_time = timestamp
 549                                 pa.last_chat_chan = what
 550                                 self._refresh()
 551                                 return d(what+' chat')
 552
 553                 m = rm('(\\w+) (?:issued an order|ordered everyone) "')
 554                 if m: return ob1('general order');
 555
 556                 m = rm('(\\w+) says, "')
 557                 if m: return chat('public')
 558
 559                 m = rm('(\\w+) tells ye, "')
 560                 if m: return chat('private')
 561
 562                 m = rm('(\\w+) flag officer chats, "')
 563                 if m: return chat('flag officer')
 564
 565                 m = rm('(\\w+) officer chats, "')
 566                 if m: return chat('officer')
 567
 568                 m = rm('Game over\\.  Winners: ([A-Za-z, ]+)\\.$')
 569                 if m:
 570                         pl = m.group(1).split(', ')
 571                         if not self._myself.name in pl:
 572                                 return d('lost boarding battle')
 573                         for pn in pl:
 574                                 if ' ' in pn: continue
 575                                 ob_x(pn,'won boarding battle')
 576                         return d('won boarding battle')
 577
 578                 m = rm('(\\w+) is eliminated\\!')
 579                 if m: return ob1('eliminated in fray');
 580
 581                 m = rm('(\\w+) has left the vessel\.')
 582                 if m:
 583                         who = m.group(1)
 584                         ob_x(who, 'disembarked')
 585                         del self._v[who]
 586                         del self._pl[who]
 587                         return d('disembarked')
 588
 589                 return d('not matched')
 590
 591         def _str_vessel(self, vn, v):
 592                 s = ' vessel %s\n' % vn
 593                 s += ' '*20 + "%-*s   %13s\n" % (
 594                                 max_pirate_namelen, '#lastaboard',
 595                                 v['#lastaboard'])
 596                 for pn in sorted(v.keys()):
 597                         if pn.startswith('#'): continue
 598                         pa = v[pn]
 599                         assert pa.v == v
 600                         assert self._pl[pn] == pa
 601                         s += ' '*20 + "%s %-*s %13s %-30s %13s %s\n" % (
 602                                 (' ','G')[pa.gunner],
 603                                 max_pirate_namelen, pn,
 604                                 pa.last_time, pa.last_event,
 605                                 pa.last_chat_time, pa.last_chat_chan)
 606                 return s
 607
 608         def __str__(self):
 609                 s = '''<ChatLogTracker
 610  myself %s
 611  vessel %s
 612 '''                     % (self._myself.name, self._vessel)
 613                 assert ((self._v is None and self._vessel is None) or
 614                         (self._v is self._vl[self._vessel]))
 615                 if self._vessel is not None:
 616                         s += self._str_vessel(self._vessel, self._v)
 617                 for vn in sorted(self._vl.keys()):
 618                         if vn == self._vessel: continue
 619                         s += self._str_vessel(vn, self._vl[vn])
 620                 for p in self._pl:
 621                         pa = self._pl[p]
 622                         assert pa.v[p] is pa
 623                         assert pa.v in self._vl.values()
 624                 s += '>\n'
 625                 return s
 626
 627         def catchup(self, progress=None):
 628                 while True:
 629                         more = self._f.readline()
 630                         if not more: break
 631
 632                         self._progress[0] += len(more)
 633                         if progress: progress.progress(*self._progress)
 634
 635                         self._lbuf += more
 636                         if self._lbuf.endswith('\n'):
 637                                 self.chatline(self._lbuf.rstrip())
 638                                 self._lbuf = ''
 639                 if progress: progress.caughtup()
 640
 641         def changed(self):
 642                 rv = self._need_redisplay
 643                 self._need_redisplay = False
 644                 return rv
 645         def myname(self):
 646                 # returns our pirate name
 647                 return self._myself.name
 648         def vessel(self):
 649                 # returns the vessel we're aboard or None
 650                 return self._vessel
 651         def aboard(self):
 652                 # returns a list of PirateAboard sorted by name
 653                 return [ self._v[pn]
 654                          for pn in sorted(self._v.keys())
 655                          if not pn.startswith('#') ]
 656
 657 class ProgressPrintPercentage:
 658         def __init__(self, f=sys.stdout): self._f = f
 659         def progress(self,done,total):
 660                 self._f.write("scan chat logs %3d%%\r" % ((done*100) / total))
 661                 self._f.flush()
 662         def caughtup(self):
 663                 self._f.write('                   \r')
 664                 self._f.flush()
 665
 666 def prep_chat_log(args, bu,
 667                 progress=ProgressPrintPercentage(),
 668                 max_myself_age=3600):
 669         if len(args) != 1: bu('this action takes only chat log filename')
 670         logfn = args[0]
 671         logfn_re = '(?:.*/)?([A-Z][a-z]+)_([a-z]+)_chat-log-\\w+$'
 672         match = regexp.match(logfn_re, logfn)
 673         if not match: bu('chat log filename is not in default format')
 674         (pirate, fetcher.ocean) = match.groups()
 675
 676         myself = PirateInfo(pirate,max_myself_age)
 677         track = ChatLogTracker(myself, logfn)
 678
 679         opts.debug -= 1
 680         track.catchup(progress)
 681         opts.debug += 1
 682
 683         return (myself, track)
 684
 685 def do_track_chat_log(args, bu):
 686         (myself, track) = prep_chat_log(args, bu)
 687         while True:
 688                 track.catchup()
 689                 if track.changed():
 690                         print track
 691                 time.sleep(1)
 692
 693 def format_time_interval(ti):
 694         if ti < 120: return '%d:%02d' % (ti / 60, ti % 60)
 695         if ti < 7200: return '%2dm' % (ti / 60)
 696         if ti < 86400: return '%dh' % (ti / 3600)
 697         return '%dd' % (ti / 86400)
 698
 699 def do_ship_aid(args, bu):
 700         if opts.ship_duty is None: opts.ship_duty = True
 701
 702         (myself, track) = prep_chat_log(args, bu)
 703
 704         rotate_nya = '/-\\'
 705
 706         def timeevent(t,e):
 707                 if t is None: return ' ' * 22
 708                 return " %-4s %-16s" % (format_time_interval(now - t),e)
 709
 710         while True:
 711                 track.catchup()
 712                 now = time.time()
 713
 714                 s = "%s" % track.myname()
 715
 716                 vn = track.vessel()
 717                 if vn is None: print s + " ...?"; return
 718
 719                 s += " on board the %s at %s\n" % (
 720                         vn, time.strftime("%Y-%m-%d %H:%M:%S"))
 721
 722                 tbl = StandingsTable()
 723                 tbl.headings()
 724
 725                 for pa in track.aboard():
 726                         pi = pa.pirate_info()
 727
 728                         xs = ''
 729                         if pa.gunner: xs += 'G '
 730                         else: xs += '  '
 731                         xs += timeevent(pa.last_time, pa.last_event)
 732                         xs += timeevent(pa.last_chat_time, pa.last_chat_chan)
 733
 734                         if pi is None:
 735                                 tbl.pirate_dummy(pa.name, rotate_nya[0], xs)
 736                         else:
 737                                 tbl.pirate(pi, xs)
 738
 739                 s += tbl.results()
 740
 741                 print '\n\n', s;
 742
 743                 time.sleep(1)
 744                 rotate_nya = rotate_nya[1:2] + rotate_nya[0]
 745
 746 def main():
 747         global opts, fetcher
 748
 749         pa = OptionParser(
 750 '''usage: .../yoweb-scrape [OPTION...] ACTION [ARGS...]
 751 actions:
 752  yoweb-scrape [--ocean OCEAN ...] pirate PIRATE
 753  yoweb-scrape [--ocean OCEAN ...] crew-of PIRATE
 754  yoweb-scrape [--ocean OCEAN ...] standings-crew-of PIRATE
 755  yoweb-scrape [--ocean OCEAN ...] track-chat-log CHAT-LOG
 756  yoweb-scrape [--ocean OCEAN ...] ship-aid CHAT-LOG
 757 ''')
 758         ao = pa.add_option
 759         ao('-O','--ocean',dest='ocean', metavar='OCEAN', default=None,
 760                 help='select ocean OCEAN')
 761         ao('--cache-dir', dest='cache_dir', metavar='DIR',
 762                 default='~/.yoweb-scrape-cache',
 763                 help='cache yoweb pages in DIR')
 764         ao('-D','--debug', action='count', dest='debug', default=0,
 765                 help='enable debugging output')
 766         ao('-q','--quiet', action='store_true', dest='quiet',
 767                 help='suppress warning output')
 768
 769         ao('--ship-duty', action='store_true', dest='ship_duty',
 770                 help='show ship duty station puzzles')
 771         ao('--all-puzzles', action='store_false', dest='ship_duty',
 772                 help='show all puzzles, not just ship duty stations')
 773
 774         (opts,args) = pa.parse_args()
 775         random.seed()
 776
 777         if len(args) < 1:
 778                 pa.error('need a mode argument')
 779
 780         mode = args[0]
 781         mode_fn_name = 'do_' + mode.replace('_','#').replace('-','_')
 782         try: mode_fn = globals()[mode_fn_name]
 783         except KeyError: pa.error('unknown mode "%s"' % mode)
 784
 785         # fixed parameters
 786         opts.min_max_age = 60
 787         opts.expire_age = 3600
 788         opts.ship_reboard_clearout = 3600
 789
 790         if opts.cache_dir.startswith('~/'):
 791                 opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]
 792
 793         fetcher = Fetcher(opts.ocean, opts.cache_dir)
 794
 795         mode_fn(args[1:], pa.error)
 796
 797 main()