X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=yoweb-scrape;h=f66dda1f999f434d9c43d728cba07bd77b9967b4;hb=ca2f709bb8508768d4cfa91c700e329d25a777d6;hp=e14719210d443c96e7b65ae8a93dc4e37924be36;hpb=323e44d92cadb46171b6a0706a9c7ed35e0e3fde;p=ypp-sc-tools.db-test.git

diff --git a/yoweb-scrape b/yoweb-scrape
index e147192..f66dda1 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 = []
@@ -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)
@@ -210,7 +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]
@@ -312,7 +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'
@@ -376,10 +382,250 @@ def do_standings_crew_of(args, bu):
 		if not members: continue
 		tab.literalline('%s:' % rank)
 		for p in members:
-			pi = PirateInfo(p, 1800)
+			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):
+		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
+
+	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 = '''<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 do_ship_aid(args, bu):
+	if len(args) != 1: bu('ship-aid 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('ship-aid chat log filename is not in default format')
+	(pirate, fetcher.ocean) = match.groups()
+	myself_pi = PirateInfo(pirate,3600)
+	track = ChatLogTracker(myself_pi)
+	f = file(logfn)
+	l = ''
+	while True:
+		l += f.readline()
+		if l.endswith('\n'):
+			track.chatline(l.rstrip())
+			l = ''
+			continue
+		if l:
+			continue
+		if track.need_redisplay:
+			print track
+			track.need_redisplay = False
+		time.sleep(1)
+
 def main():
 	global opts, fetcher
 
@@ -389,10 +635,10 @@ 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 ...] 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',
@@ -406,6 +652,7 @@ actions:
 		help='show ship duty station puzzles')
 
 	(opts,args) = pa.parse_args()
+	random.seed()
 
 	if len(args) < 1:
 		pa.error('need a mode argument')
@@ -418,6 +665,7 @@ 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:]