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=5b2fe9841f00a98c9320448a4b7f316813333e78;hp=2d08d269ea0e7c859e184b5c9caab1e7165ca472;hb=db96f9717ac132a6102c281bda32bd3dbb6cd12e;hpb=cbea021e8d7b7955cfe31ded5e832817c9c94edf diff --git a/yoweb-scrape b/yoweb-scrape index 2d08d26..5b2fe98 100755 --- a/yoweb-scrape +++ b/yoweb-scrape @@ -10,6 +10,7 @@ import urllib2 import errno import sys import re as regexp +import random from optparse import OptionParser from BeautifulSoup import BeautifulSoup @@ -27,12 +28,12 @@ standingvals = ('Able/Distinguished/Respected/Master'+ pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm') -max_pirate_namelen = 20 +max_pirate_namelen = 12 def debug(m): if opts.debug: - print >>sys.stderr, m + print m class Fetcher: def __init__(self, ocean, cachedir): @@ -44,6 +45,10 @@ class Fetcher: if oe.errno != errno.EEXIST: raise self._cache_scan(time.time()) + def _default_ocean(self): + if self.ocean is None: + self.ocean = 'ice' + def _cache_scan(self, now): # returns list of ages, unsorted ages = [] @@ -72,12 +77,12 @@ class Fetcher: min_age = 1 need_wait = 0 for age in ages: - if age < min_age: + if age < min_age and age < 300: debug('Fetcher morewait min=%d age=%d' % (min_age, age)) need_wait = max(need_wait, min_age - age) - min_age += 1 - min_age *= 1.5 + min_age += 3 + min_age *= 1.25 if need_wait > 0: debug('Fetcher wait %d' % need_wait) time.sleep(need_wait) @@ -91,12 +96,12 @@ class Fetcher: if oe.errno != errno.ENOENT: raise f = None now = time.time() - max_age = max(opts.min_max_age, max(max_age, opts.expire_age)) + max_age = max(opts.min_max_age, min(max_age, opts.expire_age)) if f is not None: s = os.fstat(f.fileno()) age = now - s.st_mtime if age > max_age: - debug('Fetcher stale') + debug('Fetcher stale %d < %d'% (max_age, age)) f = None if f is not None: data = f.read() @@ -119,6 +124,7 @@ class Fetcher: return data def yoweb(self, kind, tail, max_age): + self._default_ocean() url = 'http://%s.puzzlepirates.com/yoweb/%s%s' % ( self.ocean, kind, tail) return self.fetch(url, max_age) @@ -149,6 +155,7 @@ class SomethingSoupInfo(SoupLog): class PirateInfo(SomethingSoupInfo): # Public data members: # pi.standings = { 'Treasure Haul': 'Able' ... } + # pi.name = name # pi.crew = (id, name) # pi.flag = (id, name) # pi.msgs = [ 'message describing problem with scrape' ] @@ -156,6 +163,7 @@ class PirateInfo(SomethingSoupInfo): def __init__(self, pirate, max_age=300): SomethingSoupInfo.__init__(self, 'pirate.wm?target=', pirate, max_age) + self.name = pirate self._find_standings() self.crew = self._find_crewflag('crew', '^/yoweb/crew/info\\.wm') @@ -208,7 +216,7 @@ 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] @@ -310,7 +318,7 @@ class StandingsTable: self._cw = col_width-1 def _pline(self, pirate, puzstrs): - self.s += ' %-*s' % (max_pirate_namelen, pirate) + self.s += ' %-*s' % (max(max_pirate_namelen, 14), pirate) for v in puzstrs: self.s += ' %-*.*s' % (self._cw,self._cw, v) self.s += '\n' @@ -343,10 +351,9 @@ class StandingsTable: self._pline('', map(puzn_redact, self._puzzles)) def literalline(self, line): self.s += line + '\n' - def pirate(self, pirate): - pi = PirateInfo(pirate, 600) + def pirate(self, pi): puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles] - self._pline(pirate, puzstrs) + self._pline(pi.name, puzstrs) def results(self): return self.s @@ -360,8 +367,8 @@ def do_pirate(pirates, bu): def prep_crew_of(args, bu, max_age=300): if len(args) != 1: bu('crew-of takes one pirate name') - pi = PirateInfo(args[0]) - return CrewInfo(pi.crew[0]) + pi = PirateInfo(args[0], max_age) + return CrewInfo(pi.crew[0], max_age) def do_crew_of(args, bu): ci = prep_crew_of(args, bu) @@ -375,9 +382,271 @@ def do_standings_crew_of(args, bu): if not members: continue tab.literalline('%s:' % rank) for p in members: - tab.pirate(p) + pi = PirateInfo(p, random.randint(900,1800)) + tab.pirate(pi) print tab.results() +class PirateAboard: + # pa.v + # pa.last_time + # pa.last_event + # pa.gunner + # pa.last_chat_time + # pa.last_chat_chan + def __init__(pa, v, time, event): + pa.v = v + pa.last_time = time + pa.last_event = event + pa.last_chat_time = None + pa.last_chat_chan = None + pa.gunner = False + +class ChatLogTracker: + 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(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)\\.$') + if m: + (who,what) = m.groups() + pa = ob_x(who,'ordered '+what) + if what == 'Gunning': + pa.gunner = True + return d('duty order') + + m = rm('(\\w+) abandoned a (\\S.*\\S) station\\.$') + if m: oba('stopped'); return d('stopped') + + 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 = '''