X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.web-test.git;a=blobdiff_plain;f=yoweb-scrape;h=717b539131d8ffb0e4ba6abba5f6cddc3b38b199;hp=f9976b8341a14bf96aa5d851a7691c8b13df0ca2;hb=59824f48d09192d1c535adecf5ee2e4ee2ed47d5;hpb=d3bdf1a84d868f1eb2127cbea55d56f2fed8ecbd

diff --git a/yoweb-scrape b/yoweb-scrape
index f9976b8..717b539 100755
--- a/yoweb-scrape
+++ b/yoweb-scrape
@@ -45,6 +45,7 @@ import curses
 import termios
 import random
 import subprocess
+import copy
 from optparse import OptionParser
 from StringIO import StringIO
 
@@ -107,6 +108,18 @@ def yppsc_dir():
 	os.environ["YPPSC_YARRG_SRCBASE"] = lib
 	return lib
 
+soup_massage = copy.copy(BeautifulSoup.MARKUP_MASSAGE)
+soup_massage.append(
+		(regexp.compile('(\<td.*") ("center")'),
+		 lambda m: m.group(1)+' align='+m.group(2))
+	)
+
+def make_soup(*args, **kwargs):
+	return BeautifulSoup(*args,
+		convertEntities=BeautifulSoup.HTML_ENTITIES,
+		markupMassage=soup_massage,
+			 **kwargs)
+
 #---------- caching and rate-limiting data fetcher ----------
 
 class Fetcher:
@@ -118,12 +131,24 @@ class Fetcher:
 			if oe.errno != errno.EEXIST: raise
 		self._cache_scan(time.time())
 
-	def _cache_scan(self, now):
+	def _match_url_normalise(self, url):
+		without_scheme = regexp.sub('^[a-z]+://', '', url)
+		without_tail = regexp.sub('/.*', '', without_scheme)
+		return without_tail
+
+	def _cache_scan(self, now, match_url=None):
 		# returns list of ages, unsorted
+		if match_url is not None:
+			match_url = self._match_url_normalise(match_url)
 		ages = []
 		debug('Fetcher   scan_cache')
 		for leaf in os.listdir(self.cachedir):
 			if not leaf.startswith('#'): continue
+			if match_url is not None:
+				leaf_url = urllib.unquote_plus(leaf.strip('#'))
+				leaf_url = self._match_url_normalise(leaf_url)
+				if leaf_url != match_url:
+					continue
 			path = self.cachedir + '/' + leaf
 			try: s = os.stat(path)
 			except (OSError,IOError), oe:
@@ -139,8 +164,8 @@ class Fetcher:
 			ages.append(age)
 		return ages
 
-	def need_wait(self, now, imaginary=[]):
-		ages = self._cache_scan(now)
+	def need_wait(self, now, imaginary=[], next_url=None):
+		ages = self._cache_scan(now, match_url=next_url)
 		ages += imaginary
 		ages.sort()
 		debug('Fetcher   ages ' + `ages`)
@@ -152,15 +177,14 @@ class Fetcher:
 					(min_age, age))
 				need_wait = max(need_wait, min_age - age)
 			min_age += 3
-			min_age *= 1.25
 		if need_wait > 0:
 			need_wait += random.random() - 0.5
 		return need_wait
 
-	def _rate_limit_cache_clean(self, now):
-		need_wait = self.need_wait(now)
+	def _rate_limit_cache_clean(self, now, next_url=None):
+		need_wait = self.need_wait(now, next_url=next_url)
 		if need_wait > 0:
-			debug('Fetcher   wait %d' % need_wait)
+			debug('Fetcher   wait %f' % need_wait)
 			sleep(need_wait)
 
 	def fetch(self, url, max_age):
@@ -186,7 +210,7 @@ class Fetcher:
 			return data
 
 		debug('Fetcher  fetch')
-		self._rate_limit_cache_clean(now)
+		self._rate_limit_cache_clean(now, next_url=url)
 
 		stream = urllib2.urlopen(url)
 		data = stream.read()
@@ -252,9 +276,7 @@ class SomethingSoupInfo(SoupLog):
 	def __init__(self, kind, tail, max_age):
 		SoupLog.__init__(self)
 		html = fetcher.yoweb(kind, tail, max_age)
-		self._soup = BeautifulSoup(html,
-			convertEntities=BeautifulSoup.HTML_ENTITIES
-			)
+		self._soup = make_soup(html)
 
 #---------- scraper for pirate pages ----------
 
@@ -409,18 +431,31 @@ class CrewInfo(SomethingSoupInfo):
 	def __str__(self):
 		return `(self.crew, self.msgs)`
 
+class FlagRelation():
+	# Public data members (put there by hand by creater)
+	#	other_flagname
+	#	other_flagid
+	#	yoweb_heading
+	#	this_declaring
+	#	other_declaring_min
+	#	other_declaring_max
+	# where {this,other}_declaring{,_min,_max} are:
+	#	-1	{this,other} is declaring war
+	#	 0	{this,other} is not doing either
+	#	+1	{this,other} is allying
+	def __repr__(self):
+		return '<FlagRelation %s %d/%d..%d %s %s>' % (
+			self.yoweb_heading, self.this_declaring,
+			self.other_declaring_min, self.other_declaring_max,
+			self.other_flagname, self.other_flagid)
+
 class FlagInfo(SomethingSoupInfo):
 	# Public data members (after init):
 	#
 	#   flagid
 	#   name	#		string
 	#
-	#   relations[n] = (otherflagname, otherflagid, [stringfromyoweb],
-	#		thisdeclaring, otherdeclaringmin, otherdeclaringmax)
-	#		# where {this,other}declaring{,min,max} are:
-	#		#	-1	{this,other} is declaring war
-	#		#	 0	{this,other} is not doing either
-	#		#	+1	{this,other} is allying
+	#   relations[n] = FlagRelation
 	#   relation_byname[otherflagname] = relations[some_n]
 	#   relation_byid[otherflagname] = relations[some_n]
 	#
@@ -472,7 +507,13 @@ class FlagInfo(SomethingSoupInfo):
 			if rel: return 'flag id twice!'
 			if flagname in self.relation_byname:
 				return 'flag name twice!'
-			rel = (flagname,flagid,head, thisdecl,othermin,othermax)
+			rel = FlagRelation()
+			rel.other_flagname = flagname
+			rel.other_flagid = flagid
+			rel.yoweb_heading = head
+			rel.this_declaring = thisdecl
+			rel.other_declaring_min = othermin
+			rel.other_declaring_max = othermax
 			self.relations.append(rel)
 			self.relation_byid[flagid] = rel
 			self.relation_byname[flagid] = rel
@@ -558,7 +599,7 @@ class IslandExtendedInfo(IslandBasicInfo):
 		self.islandid = None
 		self.yoweb_url = None
 
-		soup = BeautifulSoup(self.yppedia())
+		soup = make_soup(self.yppedia())
 		content = soup.find('div', attrs = {'id': 'content'})
 		yoweb_re = regexp.compile('^http://\w+\.puzzlepirates\.com/'+
 			'yoweb/island/info\.wm\?islandid=(\d+)$')
@@ -579,7 +620,7 @@ class IslandExtendedInfo(IslandBasicInfo):
 		debug('IEI COLLECT FLAGID '+`self.name`+' URL '+`yo`)
 		if yo is None: return None
 		dataf = fetcher.fetch(yo, 1800)
-		soup = BeautifulSoup(dataf)
+		soup = make_soup(dataf)
 		ruler_re = regexp.compile(
 			'/yoweb/flag/info\.wm\?flagid=(\d+)$')
 		ruler = soup.find('a', attrs = { 'href': ruler_re })
@@ -1376,6 +1417,7 @@ def do_crew_of(args, bu):
 
 def do_flag_of(args, bu):
 	fi = prep_flag_of(args, bu)
+	print fi
 
 def do_standings_crew_of(args, bu):
 	ci = prep_crew_of(args, bu, 60)
@@ -1424,34 +1466,63 @@ def do_embargoes(args, bu):
 			if flag is None: print ''; continue
 			delim = ''
 			for rel in flag.relations:
-				(oname, oid, dummy, thisdeclaring,
-					odeclaringmin,odeclaringmax) = rel
-				if thisdeclaring >= 0: continue
+				if rel.this_declaring >= 0: continue
 				wr(delim)
-				wr(oname)
+				wr(rel.other_flagname)
 				delim = '; '
 			print ''
 
 def do_embargoes_flag_of(args, bu):
 	progressreporter.doing('fetching flag info')
 	fi = prep_flag_of(args, bu)
+	if fi is None:
+		progressreporter.stop()
+		print 'Pirate is not in a flag.'
+		return
+
 	oi = OceanInfo(IslandFlagInfo)
+
 	progressreporter.stop()
+	print ''
+
 	any = False
 	for islename in sorted(oi.islands.keys()):
 		isle = oi.islands[islename]
 		flag = isle.flag
 		if flag is None: continue
 		for rel in flag.relations:
-			(oname, oid, dummy, thisdeclaring,
-				odeclaringmin,odeclaringmax) = rel
-			if thisdeclaring >= 0: continue
-			if oid != fi.flagid: continue
+			if rel.this_declaring >= 0: continue
+			if rel.other_flagid != fi.flagid: continue
 			if not any: print 'EMBARGOED:'
 			any = True
-			print " %-30s (%s)" % (islename, flag.name)
+			print "  %-30s (%s)" % (islename, flag.name)
 	if not any:
 		print 'No embargoes.'
+	print ''
+
+	war_flag(fi)
+	print ''
+
+def do_war_flag_of(args, bu):
+	fi = prep_flag_of(args, bu)
+	war_flag(fi)
+
+def war_flag(fi):
+	any = False
+	for certain in [True, False]:
+		anythis = False
+		for rel in fi.relations:
+			if rel.this_declaring >= 0: continue
+			if (rel.other_declaring_max < 0) != certain: continue
+			if not anythis:
+				if certain: m = 'SINKING PvP'
+				else: m = 'RISK OF SINKING PvP'
+				print '%s (%s):' % (m, rel.yoweb_heading)
+			anythis = True
+			any = True
+			print " ", rel.other_flagname
+	if not any:
+		print 'No sinking PvP.'
 
 #----- modes which use the chat log parser are quite complex -----
 
@@ -1784,7 +1855,7 @@ actions:
  yoweb-scrape [--ocean OCEAN ...] standings-crew-of PIRATE
  yoweb-scrape [--ocean OCEAN ...] track-chat-log CHAT-LOG
  yoweb-scrape [--ocean OCEAN ...] ocean|embargoes
- yoweb-scrape [--ocean OCEAN ...] embargoes-flag-of PIRATE
+ yoweb-scrape [--ocean OCEAN ...] war-flag-of|embargoes-flag-of PIRATE
  yoweb-scrape [options] ship-aid CHAT-LOG  (must be .../PIRATE_OCEAN_chat-log*)
 
 display modes (for --display) apply to ship-aid: