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:
  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):
 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                 self.s += '\n'
 329
 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 ''
 336                 s = ''
 337                 if self._cw > 4:
 338                         c1 = standingvals[standing][0]
 339                         if standing < 3: c1 = c1.lower() # 3 = Master
 340                         s += `standing`
 341                 if self._cw > 5:
 342                         s += ' '
 343                 s += '*' * (standing / 2)
 344                 s += '+' * (standing % 2)
 345                 return s
 346
 347         def headings(self):
 348                 def puzn_redact(name):
 349                         if isinstance(name,list):
 350                                 return '/'.join(
 351                                         ["%.*s" % (self._cw/2, puzn_redact(n))
 352                                          for n in name])
 353                         spc = name.find(' ')
 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)
 362
 363         def results(self):
 364                 return self.s
 365
 366 def do_pirate(pirates, bu):
 367         print '{'
 368         for pirate in pirates:
 369                 info = PirateInfo(pirate)
 370                 print '%s: %s,' % (`pirate`, info)
 371         print '}'
 372
 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)
 377
 378 def do_crew_of(args, bu):
 379         ci = prep_crew_of(args, bu)
 380         print ci
 381
 382 def do_standings_crew_of(args, bu):
 383         ci = prep_crew_of(args, bu, 60)
 384         tab = StandingsTable()
 385         tab.headings()
 386         for (rank, members) in ci.crew:
 387                 if not members: continue
 388                 tab.literalline('%s:' % rank)
 389                 for p in members:
 390                         pi = PirateInfo(p, random.randint(900,1800))
 391                         tab.pirate(pi)
 392         print tab.results()
 393
 394 class PirateAboard:
 395         # pa.v
 396         # pa.last_time
 397         # pa.last_event
 398         # pa.gunner
 399         # pa.last_chat_time
 400         # pa.last_chat_chan
 401         # pa.pi
 402
 403         def __init__(pa, pn, v, time, event):
 404                 pa.pn = pn
 405                 pa.v = v
 406                 pa.last_time = time
 407                 pa.last_event = event
 408                 pa.last_chat_time = None
 409                 pa.last_chat_chan = None
 410                 pa.gunner = False
 411                 pa.pi = None
 412
 413         def pirate_info(self):
 414                 if not pa.pi and not fetcher.need_wait(time.time):
 415                         pa.pi = PirateInfo(pa.name, 3600)
 416                 return pa.pi
 417
 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]
 425                 self._date = None
 426                 self._myself = myself_pi
 427                 self.need_redisplay = False
 428                 self._f = file(logfn)
 429                 self._lbuf = ''
 430                 self._progress = [0, os.fstat(self._f.fileno()).st_size]
 431
 432         def _refresh(self):
 433                 self.need_redisplay = True
 434
 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
 441                 else:
 442                         if pa is not None: del pa.v[pirate]
 443                         pa = PirateAboard(pirate, self._v, timestamp, event)
 444                         self._pl[pirate] = pa
 445                         self._v[pirate] = pa
 446                 self._v['#lastaboard'] = timestamp
 447                 self._refresh()
 448                 return pa
 449
 450         def _trash_vessel(self, v):
 451                 for pn in v:
 452                         if pn.startswith('#'): continue
 453                         del self._pl[pn]
 454                 self._refresh()
 455
 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,'',
 461                                         'stale reset '+vn)
 462                                 self._trash_vessel(v)
 463                                 del self._vl[vn]
 464
 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
 470
 471         def _debug_line_disposition(self,timestamp,l,m):
 472                 debug('CLT %13s %-30s %s' % (timestamp,m,l))
 473
 474         def chatline(self,l):
 475                 rm = lambda re: regexp.match(re,l)
 476                 d = lambda m: self._debug_line_disposition(timestamp,l,m)
 477                 timestamp = None
 478
 479                 m = rm('=+ (\\d+)/(\\d+)/(\\d+) =+$')
 480                 if m:
 481                         self._date = m.groups()
 482                         return d('date '+`self._date`)
 483
 484                 if self._date is None:
 485                         return d('date unset')
 486
 487                 m = rm('\\[(\d\d):(\d\d):(\d\d)\\] ')
 488                 if not m:
 489                         return d('no timestamp')
 490
 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:]
 495
 496                 def ob_x(who,event):
 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)))
 500
 501                 m = rm('Going aboard the (\\S.*\\S)\\.\\.\\.$')
 502                 if m:
 503                         pn = self._myself.name
 504                         self._vessel = m.group(1)
 505                         dm = 'boarding'
 506
 507                         try:             self._v = self._vl[self._vessel]
 508                         except KeyError: self._v = None; dm += ' new'
 509
 510                         if self._v is not None:  la = self._v['#lastaboard']
 511                         else:                    la = 0; dm += ' ?la'
 512
 513                         if timestamp - la > opts.ship_reboard_clearout:
 514                                 self.clear_vessel(timestamp)
 515                                 dm += ' stale'
 516
 517                         ob_x(pn, 'we boarded')
 518                         self.expire_garbage(timestamp)
 519                         return d(dm)
 520
 521                 if self._v is None:
 522                         return d('no vessel')
 523
 524                 m = rm('(\\w+) has come aboard\\.$')
 525                 if m: return ob1('boarded');
 526
 527                 m = rm('You have ordered (\\w+) to do some (\\S.*\\S)\\.$')
 528                 if m:
 529                         (who,what) = m.groups()
 530                         pa = ob_x(who,'ordered '+what)
 531                         if what == 'Gunning':
 532                                 pa.gunner = True
 533                         return d('duty order')
 534
 535                 m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$')
 536                 if m: oba('stopped'); return d('stopped')
 537
 538                 def chat(what):
 539                         who = m.group(1)
 540                         try: pa = self._pl[who]
 541                         except KeyError: return d('chat mystery')
 542                         if pa.v is self._v:
 543                                 pa.last_chat_time = timestamp
 544                                 pa.last_chat_chan = what
 545                                 self._refresh()
 546                                 return d(what+' chat')
 547
 548                 m = rm('(\\w+) (?:issued an order|ordered everyone) "')
 549                 if m: return ob1('general order');
 550
 551                 m = rm('(\\w+) says, "')
 552                 if m: return chat('public')
 553
 554                 m = rm('(\\w+) tells ye, "')
 555                 if m: return chat('private')
 556
 557                 m = rm('(\\w+) flag officer chats, "')
 558                 if m: return chat('flag officer')
 559
 560                 m = rm('(\\w+) officer chats, "')
 561                 if m: return chat('officer')
 562
 563                 m = rm('Game over\\.  Winners: ([A-Za-z, ]+)\\.$')
 564                 if m:
 565                         pl = m.group(1).split(', ')
 566                         if not self._myself.name in pl:
 567                                 return d('lost boarding battle')
 568                         for pn in pl:
 569                                 if ' ' in pn: continue
 570                                 ob_x(pn,'won boarding battle')
 571                         return d('won boarding battle')
 572
 573                 m = rm('(\\w+) is eliminated\\!')
 574                 if m: return ob1('eliminated in fray');
 575
 576                 m = rm('(\\w+) has left the vessel\.')
 577                 if m:
 578                         who = m.group(1)
 579                         ob_x(who, 'disembarked')
 580                         del self._v[who]
 581                         del self._pl[who]
 582                         return d('disembarked')
 583
 584                 return d('not matched')
 585
 586         def _str_vessel(self, vn, v):
 587                 s = ' vessel %s\n' % vn
 588                 s += ' '*20 + "%-*s   %13s\n" % (
 589                                 max_pirate_namelen, '#lastaboard',
 590                                 v['#lastaboard'])
 591                 for pn in sorted(v.keys()):
 592                         if pn.startswith('#'): continue
 593                         pa = v[pn]
 594                         assert pa.v == v
 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)
 601                 return s
 602
 603         def __str__(self):
 604                 s = '''<ChatLogTracker
 605  myself %s
 606  vessel %s
 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])
 615                 for p in self._pl:
 616                         pa = self._pl[p]
 617                         assert pa.v[p] is pa
 618                         assert pa.v in self._vl.values()
 619                 s += '>\n'
 620                 return s
 621
 622         def catchup(self, progress=None):
 623                 while True:
 624                         more = self._f.readline()
 625                         if not more: break
 626
 627                         self._progress[0] += len(more)
 628                         if progress: progress.progress(*self._progress)
 629
 630                         self._lbuf += more
 631                         if self._lbuf.endswith('\n'):
 632                                 self.chatline(self._lbuf.rstrip())
 633                                 self._lbuf = ''
 634                 if progress: progress.caughtup()
 635
 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))
 640                 self._f.flush()
 641         def caughtup(self):
 642                 self._f.write('                   \r')
 643                 self._f.flush()
 644
 645 def simple_printer(x): print x
 646
 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')
 650         logfn = args[0]
 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()
 655
 656         myself_pi = PirateInfo(pirate,max_myself_age)
 657         track = ChatLogTracker(myself_pi, logfn)
 658
 659         track.catchup(progress)
 660         while True:
 661                 track.catchup()
 662                 if track.need_redisplay:
 663                         display(track)
 664                         track.need_redisplay = False
 665                 time.sleep(1)
 666
 667 def do_track_chat_log(args, bu):
 668         run_chat_log(args, bu)
 669
 670 #def do_ship_aid(args, bu):
 671
 672 def main():
 673         global opts, fetcher
 674
 675         pa = OptionParser(
 676 '''usage: .../yoweb-scrape [OPTION...] ACTION [ARGS...]
 677 actions:
 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
 683 ''')
 684         ao = pa.add_option
 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')
 694
 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')
 699
 700         (opts,args) = pa.parse_args()
 701         random.seed()
 702
 703         if len(args) < 1:
 704                 pa.error('need a mode argument')
 705
 706         mode = args[0]
 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)
 710
 711         # fixed parameters
 712         opts.min_max_age = 60
 713         opts.expire_age = 3600
 714         opts.ship_reboard_clearout = 3600
 715
 716         if opts.cache_dir.startswith('~/'):
 717                 opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]
 718
 719         fetcher = Fetcher(opts.ocean, opts.cache_dir)
 720
 721         mode_fn(args[1:], pa.error)
 722
 723 main()