X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=yoweb-scrape;h=2d968921ab1eed4023fea7344ce90951f2c7c21d;hb=31cc3a65d81a3b9edd09372950ffbf3da5ab283b;hp=f66dda1f999f434d9c43d728cba07bd77b9967b4;hpb=ca2f709bb8508768d4cfa91c700e329d25a777d6;p=ypp-sc-tools.db-live.git

diff --git a/yoweb-scrape b/yoweb-scrape
index f66dda1..2d96892 100755
--- a/yoweb-scrape
+++ b/yoweb-scrape
@@ -70,7 +70,7 @@ 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 +83,10 @@ 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)
@@ -280,7 +284,7 @@ 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})
@@ -327,6 +331,7 @@ class StandingsTable:
 		if not isinstance(puzzle,list): puzzle = [puzzle]
 		try: standing = max([pi.standings[p] for p in puzzle])
 		except KeyError: return '?'
+		if isinstance(standing,basestring): return standing
 		if not standing: return ''
 		s = ''
 		if self._cw > 4:
@@ -393,16 +398,25 @@ class PirateAboard:
 	# pa.gunner
 	# pa.last_chat_time
 	# pa.last_chat_chan
-	def __init__(pa, v, time, event):
+	# pa.pi
+
+	def __init__(pa, pn, v, time, event):
+		pa.pn = 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(self):
+		if not pa.pi and not fetcher.need_wait(time.time):
+			pa.pi = PirateInfo(pa.name, 3600)
+		return pa.pi
 
 class ChatLogTracker:
-	def __init__(self, myself_pi):
+	def __init__(self, myself_pi, logfn):
 		self._pl = {}	# self._pl['Pirate'] =
 		self._vl = {}	#   self._vl['Vessel']['Pirate'] = PirateAboard
 				# self._vl['Vessel']['#lastaboard']
@@ -411,6 +425,9 @@ class ChatLogTracker:
 		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
@@ -423,7 +440,7 @@ class ChatLogTracker:
 			pa.last_event = event
 		else:
 			if pa is not None: del pa.v[pirate]
-			pa = PirateAboard(self._v, timestamp, event)
+			pa = PirateAboard(pirate, self._v, timestamp, event)
 			self._pl[pirate] = pa
 			self._v[pirate] = pa
 		self._v['#lastaboard'] = timestamp
@@ -602,30 +619,56 @@ class ChatLogTracker:
 		s += '>\n'
 		return s
 
-def do_ship_aid(args, bu):
-	if len(args) != 1: bu('ship-aid takes only chat log filename')
+	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()
+
+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()
+
+def simple_printer(x): print x
+
+def run_chat_log(args, bu, progress=ProgressPrintPercentage(),
+		display=simple_printer, 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('ship-aid chat log filename is not in default format')
+	if not match: bu('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 = ''
+	
+	myself_pi = PirateInfo(pirate,max_myself_age)
+	track = ChatLogTracker(myself_pi, logfn)
+
+	track.catchup(progress)
 	while True:
-		l += f.readline()
-		if l.endswith('\n'):
-			track.chatline(l.rstrip())
-			l = ''
-			continue
-		if l:
-			continue
+		track.catchup()
 		if track.need_redisplay:
-			print track
+			display(track)
 			track.need_redisplay = False
 		time.sleep(1)
 
+def do_track_chat_log(args, bu):
+	run_chat_log(args, bu)
+
+#def do_ship_aid(args, bu):
+
 def main():
 	global opts, fetcher
 
@@ -635,6 +678,7 @@ 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
@@ -650,6 +694,8 @@ actions:
 
 	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()