X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=yoweb-scrape;h=a7f4c9be10faab9bceef1439ee44e3fdb59789f1;hp=4e12d7f6da3a57835e2f6a7952613b18a101c44e;hb=15799130f8d34856b9cefe3c71103564413fa04a;hpb=2943c3282218303bfaaf76ec6efc41089adf9e6d diff --git a/yoweb-scrape b/yoweb-scrape index 4e12d7f..a7f4c9b 100755 --- a/yoweb-scrape +++ b/yoweb-scrape @@ -1,5 +1,7 @@ #!/usr/bin/python +#---------- setup ---------- + import signal signal.signal(signal.SIGINT, signal.SIG_DFL) @@ -17,6 +19,7 @@ from BeautifulSoup import BeautifulSoup opts = None +#---------- YPP parameters and arrays ---------- puzzles = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+ '/Battle Navigation/Gunning/Carpentry/Rumble/Treasure Haul'+ @@ -28,12 +31,16 @@ standingvals = ('Able/Distinguished/Respected/Master'+ pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm') -max_pirate_namelen = 20 +max_pirate_namelen = 12 + +#---------- general utilities ---------- def debug(m): - if opts.debug: - print >>sys.stderr, m + if opts.debug > 0: + print m + +#---------- caching and rate-limiting data fetcher ---------- class Fetcher: def __init__(self, ocean, cachedir): @@ -70,7 +77,7 @@ class Fetcher: ages.append(age) return ages - def _rate_limit_cache_clean(self, now): + def need_wait(self, now): ages = self._cache_scan(now) ages.sort() debug('Fetcher ages ' + `ages`) @@ -83,6 +90,10 @@ class Fetcher: need_wait = max(need_wait, min_age - age) min_age += 3 min_age *= 1.25 + return need_wait + + def _rate_limit_cache_clean(self, now): + need_wait = self.need_wait(now) if need_wait > 0: debug('Fetcher wait %d' % need_wait) time.sleep(need_wait) @@ -129,6 +140,8 @@ class Fetcher: self.ocean, kind, tail) return self.fetch(url, max_age) +#---------- logging assistance for troubled screenscrapers ---------- + class SoupLog: def __init__(self): self.msgs = [ ] @@ -152,6 +165,8 @@ class SomethingSoupInfo(SoupLog): convertEntities=BeautifulSoup.HTML_ENTITIES ) +#---------- scraper for pirate pages ---------- + class PirateInfo(SomethingSoupInfo): # Public data members: # pi.standings = { 'Treasure Haul': 'Able' ... } @@ -216,7 +231,7 @@ u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A- skl.msg('puzzle "%s" multiple standings %s' % (puzzle, `sl`)) continue - if not len(sl): + if not sl: skl.msg('puzzle "%s" no standing found' % puzzle) continue standing = sl[0] @@ -251,6 +266,8 @@ u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A- def __str__(self): return `(self.crew, self.flag, self.standings, self.msgs)` +#---------- scraper for crew pages ---------- + class CrewInfo(SomethingSoupInfo): # Public data members: # ci.crew = [ ('Captain', ['Pirate', ...]), @@ -280,7 +297,7 @@ class CrewInfo(SomethingSoupInfo): crew_rank_re = regexp.compile('/yoweb/images/crew') for row in tbl.contents: # findAll(recurse=False) - if isinstance(row, unicode): + if isinstance(row,basestring): continue is_rank = row.find('img', attrs={'src': crew_rank_re}) @@ -299,6 +316,8 @@ class CrewInfo(SomethingSoupInfo): def __str__(self): return `(self.crew, self.msgs)` +#---------- pretty-printer for tables of pirate puzzle standings ---------- + class StandingsTable: def __init__(self, use_puzzles=None, col_width=6): if use_puzzles is None: @@ -317,10 +336,12 @@ class StandingsTable: self.s = '' self._cw = col_width-1 - def _pline(self, pirate, puzstrs): - self.s += ' %-*s' % (max_pirate_namelen, pirate) + def _pline(self, pirate, puzstrs, extra): + self.s += ' %-*s' % (max(max_pirate_namelen, 14), pirate) for v in puzstrs: self.s += ' %-*.*s' % (self._cw,self._cw, v) + if extra: + self.s += ' ' + extra self.s += '\n' def _puzstr(self, pi, puzzle): @@ -348,16 +369,285 @@ class StandingsTable: spc = name.find(' ') if spc < 0: return name return name[0:min(4,spc)] + name[spc+1:] - self._pline('', map(puzn_redact, self._puzzles)) + self._pline('', map(puzn_redact, self._puzzles), None) def literalline(self, line): self.s += line + '\n' - def pirate(self, pi): + def pirate_dummy(self, name, standingstring, extra=None): + self._pline(name, standingstring * len(self._puzzles), extra) + def pirate(self, pi, extra=None): puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles] - self._pline(pi.name, puzstrs) + self._pline(pi.name, puzstrs, extra) def results(self): return self.s +#---------- chat log parser ---------- + +class PirateAboard: + # This is essentially a transparent, dumb, data class. + # pa.v + # pa.name + # pa.last_time + # pa.last_event + # pa.gunner + # pa.last_chat_time + # pa.last_chat_chan + # pa.pi + + def __init__(pa, pn, v, time, event): + pa.name = pn + pa.v = v + pa.last_time = time + pa.last_event = event + pa.last_chat_time = None + pa.last_chat_chan = None + pa.gunner = False + pa.pi = None + + def pirate_info(pa): + if not pa.pi and not fetcher.need_wait(time.time()): + pa.pi = PirateInfo(pa.name, 3600) + return pa.pi + +class ChatLogTracker: + # This is quite complex so we make it opaque. Use the + # official invokers, accessors etc. + + def __init__(self, myself_pi, logfn): + self._pl = {} # self._pl['Pirate'] = + self._vl = {} # self._vl['Vessel']['Pirate'] = PirateAboard + # self._vl['Vessel']['#lastaboard'] + self._v = None # self._v = + self._vessel = None # self._vl[self._vessel] + self._date = None + self._myself = myself_pi + self._need_redisplay = False + self._f = file(logfn) + self._lbuf = '' + self._progress = [0, os.fstat(self._f.fileno()).st_size] + + def _refresh(self): + self._need_redisplay = True + + def _onboard_event(self,timestamp,pirate,event): + try: pa = self._pl[pirate] + except KeyError: pa = None + if pa is not None and pa.v is self._v: + pa.last_time = timestamp + pa.last_event = event + else: + if pa is not None: del pa.v[pirate] + pa = PirateAboard(pirate, self._v, timestamp, event) + self._pl[pirate] = pa + self._v[pirate] = pa + self._v['#lastaboard'] = timestamp + self._refresh() + return pa + + def _trash_vessel(self, v): + for pn in v: + if pn.startswith('#'): continue + del self._pl[pn] + self._refresh() + + def expire_garbage(self, timestamp): + for (vn,v) in list(self._vl.iteritems()): + la = v['#lastaboard'] + if timestamp - la > opts.ship_reboard_clearout: + self._debug_line_disposition(timestamp,'', + 'stale reset '+vn) + self._trash_vessel(v) + del self._vl[vn] + + def clear_vessel(self, timestamp): + if self._v is not None: + self._trash_vessel(self._v) + self._v = {'#lastaboard': timestamp} + self._vl[self._vessel] = self._v + + def _debug_line_disposition(self,timestamp,l,m): + debug('CLT %13s %-30s %s' % (timestamp,m,l)) + + def chatline(self,l): + rm = lambda re: regexp.match(re,l) + d = lambda m: self._debug_line_disposition(timestamp,l,m) + timestamp = None + + m = rm('=+ (\\d+)/(\\d+)/(\\d+) =+$') + if m: + self._date = m.groups() + return d('date '+`self._date`) + + if self._date is None: + return d('date unset') + + m = rm('\\[(\d\d):(\d\d):(\d\d)\\] ') + if not m: + return d('no timestamp') + + time_tuple = [int(x) for x in self._date + m.groups()] + time_tuple += (-1,-1,-1) + timestamp = time.mktime(time_tuple) + l = l[l.find(' ')+1:] + + def ob_x(who,event): + return self._onboard_event(timestamp, who, event) + def ob1(did): ob_x(m.group(1), did); return d(did) + def oba(did): return ob1('%s %s' % (did, m.group(2))) + + m = rm('Going aboard the (\\S.*\\S)\\.\\.\\.$') + if m: + pn = self._myself.name + self._vessel = m.group(1) + dm = 'boarding' + + try: self._v = self._vl[self._vessel] + except KeyError: self._v = None; dm += ' new' + + if self._v is not None: la = self._v['#lastaboard'] + else: la = 0; dm += ' ?la' + + if timestamp - la > opts.ship_reboard_clearout: + self.clear_vessel(timestamp) + dm += ' stale' + + ob_x(pn, 'we boarded') + self.expire_garbage(timestamp) + return d(dm) + + if self._v is None: + return d('no vessel') + + m = rm('(\\w+) has come aboard\\.$') + if m: return ob1('boarded'); + + m = rm('You have ordered (\\w+) to do some (\\S.*\\S)\\.$') + if m: + (who,what) = m.groups() + pa = ob_x(who,'ord '+what) + if what == 'Gunning': + pa.gunner = True + return d('duty order') + + m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$') + if m: oba('stopped'); return d("end") + + def chat(what): + who = m.group(1) + try: pa = self._pl[who] + except KeyError: return d('chat mystery') + if pa.v is self._v: + pa.last_chat_time = timestamp + pa.last_chat_chan = what + self._refresh() + return d(what+' chat') + + m = rm('(\\w+) (?:issued an order|ordered everyone) "') + if m: return ob1('general order'); + + m = rm('(\\w+) says, "') + if m: return chat('public') + + m = rm('(\\w+) tells ye, "') + if m: return chat('private') + + m = rm('(\\w+) flag officer chats, "') + if m: return chat('flag officer') + + m = rm('(\\w+) officer chats, "') + if m: return chat('officer') + + m = rm('Game over\\. Winners: ([A-Za-z, ]+)\\.$') + if m: + pl = m.group(1).split(', ') + if not self._myself.name in pl: + return d('lost boarding battle') + for pn in pl: + if ' ' in pn: continue + ob_x(pn,'won boarding battle') + return d('won boarding battle') + + m = rm('(\\w+) is eliminated\\!') + if m: return ob1('eliminated in fray'); + + m = rm('(\\w+) has left the vessel\.') + if m: + who = m.group(1) + ob_x(who, 'disembarked') + del self._v[who] + del self._pl[who] + return d('disembarked') + + return d('not matched') + + def _str_vessel(self, vn, v): + s = ' vessel %s\n' % vn + s += ' '*20 + "%-*s %13s\n" % ( + max_pirate_namelen, '#lastaboard', + v['#lastaboard']) + for pn in sorted(v.keys()): + if pn.startswith('#'): continue + pa = v[pn] + assert pa.v == v + assert self._pl[pn] == pa + s += ' '*20 + "%s %-*s %13s %-30s %13s %s\n" % ( + (' ','G')[pa.gunner], + max_pirate_namelen, pn, + pa.last_time, pa.last_event, + pa.last_chat_time, pa.last_chat_chan) + return s + + def __str__(self): + s = '''