X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=yoweb-scrape;h=a7f4c9be10faab9bceef1439ee44e3fdb59789f1;hp=493ef53fcffc77825a31cf40dad8c9494cd80f15;hb=15799130f8d34856b9cefe3c71103564413fa04a;hpb=46fafd1982761ccb564d29a625528692abb4cd7c

diff --git a/yoweb-scrape b/yoweb-scrape
index 493ef53..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)
 
@@ -10,12 +12,14 @@ import urllib2
 import errno
 import sys
 import re as regexp
+import random
 from optparse import OptionParser
 
 from BeautifulSoup import BeautifulSoup
 
 opts = None
 
+#---------- YPP parameters and arrays ----------
 
 puzzles = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+
 	'/Battle Navigation/Gunning/Carpentry/Rumble/Treasure Haul'+
@@ -27,12 +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):
@@ -44,6 +52,10 @@ 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 = []
@@ -65,19 +77,23 @@ 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`)
 		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
+		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)
@@ -91,12 +107,12 @@ 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()
@@ -119,10 +135,13 @@ 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)
 
+#---------- logging assistance for troubled screenscrapers ----------
+
 class SoupLog:
 	def __init__(self):
 		self.msgs = [ ]
@@ -146,9 +165,12 @@ class SomethingSoupInfo(SoupLog):
 			convertEntities=BeautifulSoup.HTML_ENTITIES
 			)
 
+#---------- scraper for pirate pages ----------
+
 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 +178,7 @@ 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 +231,7 @@ u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-
 				skl.msg('puzzle "%s" multiple standings %s' %
 						(puzzle, `sl`))
 				continue
-			if not len(sl):
+			if not sl:
 				skl.msg('puzzle "%s" no standing found' % puzzle)
 				continue
 			standing = sl[0]
@@ -243,6 +266,8 @@ u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\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', ...]),
@@ -272,7 +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})
@@ -291,47 +316,338 @@ 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=puzzles):
+	def __init__(self, use_puzzles=None, col_width=6):
+		if use_puzzles is None:
+			if opts.ship_duty:
+				use_puzzles=[
+					'Navigating','Battle Navigation',
+					'Gunning',
+					['Sailing','Rigging'],
+					'Bilging',
+					'Carpentry',
+					'Treasure Haul'
+				]
+			else:
+				use_puzzles=puzzles
 		self._puzzles = use_puzzles
 		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' % (5,5, v)
+			self.s += ' %-*.*s' % (self._cw,self._cw, v)
+		if extra:
+			self.s += ' ' + extra
 		self.s += '\n'
 
 	def _puzstr(self, pi, puzzle):
 		if not isinstance(puzzle,list): puzzle = [puzzle]
 		try: standing = max([pi.standings[p] for p in puzzle])
 		except KeyError: return '?'
-		c1 = standingvals[standing][0]
-		if standing < 3: c1 = c1.lower() # 3 = Master
-		hashes = '*' * (standing / 2)
-		equals = '+' * (standing % 2)
-		return c1 + hashes + equals
+		if not standing: return ''
+		s = ''
+		if self._cw > 4:
+			c1 = standingvals[standing][0]
+			if standing < 3: c1 = c1.lower() # 3 = Master
+			s += `standing`
+		if self._cw > 5:
+			s += ' '
+		s += '*' * (standing / 2)
+		s += '+' * (standing % 2)
+		return s
 
 	def headings(self):
 		def puzn_redact(name):
 			if isinstance(name,list):
 				return '/'.join(
-					["%.2s" % puzn_redact(n)
+					["%.*s" % (self._cw/2, puzn_redact(n))
 					 for n in name])
 			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, pirate):
-		pi = PirateInfo(pirate, 600)
+	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(pirate, 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 = '''<ChatLogTracker
+ myself %s
+ vessel %s
+'''			% (self._myself.name, self._vessel)
+		assert ((self._v is None and self._vessel is None) or
+			(self._v is self._vl[self._vessel]))
+		if self._vessel is not None:
+			s += self._str_vessel(self._vessel, self._v)
+		for vn in sorted(self._vl.keys()):
+			if vn == self._vessel: continue
+			s += self._str_vessel(vn, self._vl[vn])
+		for p in self._pl:
+			pa = self._pl[p]
+			assert pa.v[p] is pa
+			assert pa.v in self._vl.values()
+		s += '>\n'
+		return s
+
+	def catchup(self, progress=None):
+		while True:
+			more = self._f.readline()
+			if not more: break
+
+			self._progress[0] += len(more)
+			if progress: progress.progress(*self._progress)
+
+			self._lbuf += more
+			if self._lbuf.endswith('\n'):
+				self.chatline(self._lbuf.rstrip())
+				self._lbuf = ''
+		if progress: progress.caughtup()
+
+	def changed(self):
+		rv = self._need_redisplay
+		self._need_redisplay = False
+		return rv
+	def myname(self):
+		# returns our pirate name
+		return self._myself.name
+	def vessel(self):
+		# returns the vessel we're aboard or None
+		return self._vessel
+	def aboard(self):
+		# returns a list of PirateAboard sorted by name
+		return [ self._v[pn]
+			 for pn in sorted(self._v.keys())
+			 if not pn.startswith('#') ]
+
+#---------- implementations of actual operation modes ----------
+
 def do_pirate(pirates, bu):
 	print '{'
 	for pirate in pirates:
@@ -341,8 +657,9 @@ 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)
+	if pi.crew is None: return None
+	return CrewInfo(pi.crew[0], max_age)
 
 def do_crew_of(args, bu):
 	ci = prep_crew_of(args, bu)
@@ -356,9 +673,103 @@ 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 ProgressPrintPercentage:
+	def __init__(self, f=sys.stdout): self._f = f
+	def progress(self,done,total):
+		self._f.write("scan chat logs %3d%%\r" % ((done*100) / total))
+		self._f.flush()
+	def caughtup(self):
+		self._f.write('                   \r')
+		self._f.flush()
+
+#----- modes which use the chat log parser are quite complex -----
+
+def prep_chat_log(args, bu,
+		progress=ProgressPrintPercentage(),
+		max_myself_age=3600):
+	if len(args) != 1: bu('this action takes only chat log filename')
+	logfn = args[0]
+	logfn_re = '(?:.*/)?([A-Z][a-z]+)_([a-z]+)_chat-log-\\w+$'
+	match = regexp.match(logfn_re, logfn)
+	if not match: bu('chat log filename is not in default format')
+	(pirate, fetcher.ocean) = match.groups()
+	
+	myself = PirateInfo(pirate,max_myself_age)
+	track = ChatLogTracker(myself, logfn)
+
+	opts.debug -= 1
+	track.catchup(progress)
+	opts.debug += 1
+
+	return (myself, track)
+
+def do_track_chat_log(args, bu):
+	(myself, track) = prep_chat_log(args, bu)
+	while True:
+		track.catchup()
+		if track.changed():
+			print track
+		time.sleep(1)
+
+def format_time_interval(ti):
+	if ti < 120: return '%d:%02d' % (ti / 60, ti % 60)
+	if ti < 7200: return '%2dm' % (ti / 60)
+	if ti < 86400: return '%dh' % (ti / 3600)
+	return '%dd' % (ti / 86400)
+
+def do_ship_aid(args, bu):
+	if opts.ship_duty is None: opts.ship_duty = True
+
+	(myself, track) = prep_chat_log(args, bu)
+
+	rotate_nya = '/-\\'
+
+	def timeevent(t,e):
+		if t is None: return ' ' * 22
+		return " %-4s %-16s" % (format_time_interval(now - t),e)
+
+	while True:
+		track.catchup()
+		now = time.time()
+
+		s = "%s" % track.myname()
+
+		vn = track.vessel()
+		if vn is None: print s + " ...?"; return
+
+		s += " on board the %s at %s\n" % (
+			vn, time.strftime("%Y-%m-%d %H:%M:%S"))
+
+		tbl = StandingsTable()
+		tbl.headings()
+
+		for pa in track.aboard():
+			pi = pa.pirate_info()
+
+			xs = ''
+			if pa.gunner: xs += 'G '
+			else: xs += '  '
+			xs += timeevent(pa.last_time, pa.last_event)
+			xs += timeevent(pa.last_chat_time, pa.last_chat_chan)
+
+			if pi is None:
+				tbl.pirate_dummy(pa.name, rotate_nya[0], xs)
+			else:
+				tbl.pirate(pi, xs)
+
+		s += tbl.results()
+
+		print '\n\n', s;
+
+		time.sleep(1)
+		rotate_nya = rotate_nya[1:2] + rotate_nya[0]
+
+#---------- main program ----------
+
 def main():
 	global opts, fetcher
 
@@ -368,19 +779,27 @@ actions:
  yoweb-scrape [--ocean OCEAN ...] pirate PIRATE
  yoweb-scrape [--ocean OCEAN ...] crew-of PIRATE
  yoweb-scrape [--ocean OCEAN ...] standings-crew-of PIRATE
+ yoweb-scrape [--ocean OCEAN ...] track-chat-log CHAT-LOG
+ yoweb-scrape [--ocean OCEAN ...] ship-aid CHAT-LOG
 ''')
 	ao = pa.add_option
-	ao('-O','--ocean',dest='ocean', metavar='OCEAN',
-		default='ice',
+	ao('-O','--ocean',dest='ocean', metavar='OCEAN', default=None,
 		help='select ocean OCEAN')
 	ao('--cache-dir', dest='cache_dir', metavar='DIR',
 		default='~/.yoweb-scrape-cache',
 		help='cache yoweb pages in DIR')
-	ao('-D','--debug', action='store_true', dest='debug', default=False,
+	ao('-D','--debug', action='count', dest='debug', default=0,
 		help='enable debugging output')
 	ao('-q','--quiet', action='store_true', dest='quiet',
 		help='suppress warning output')
+
+	ao('--ship-duty', action='store_true', dest='ship_duty',
+		help='show ship duty station puzzles')
+	ao('--all-puzzles', action='store_false', dest='ship_duty',
+		help='show all puzzles, not just ship duty stations')
+
 	(opts,args) = pa.parse_args()
+	random.seed()
 
 	if len(args) < 1:
 		pa.error('need a mode argument')
@@ -393,6 +812,7 @@ actions:
 	# fixed parameters
 	opts.min_max_age = 60
 	opts.expire_age = 3600
+	opts.ship_reboard_clearout = 3600
 
 	if opts.cache_dir.startswith('~/'):
 		opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]