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=db5dc1338298a47a2c9120d2f93b168d07842ba5;hp=ba828f51c0fb1a125e6ef729f82caacf0f7919b4;hb=1d18909d73612e4c10c0079a1c886277312f45ad;hpb=a732139b9b5d7a0a7d132f3020101f52ceea00e2

diff --git a/yoweb-scrape b/yoweb-scrape
index ba828f5..db5dc13 100755
--- a/yoweb-scrape
+++ b/yoweb-scrape
@@ -110,19 +110,14 @@ def yppsc_dir():
 #---------- caching and rate-limiting data fetcher ----------
 
 class Fetcher:
-	def __init__(self, ocean, cachedir):
+	def __init__(self, cachedir):
 		debug('Fetcher init %s' % cachedir)
-		self.ocean = ocean
 		self.cachedir = cachedir
 		try: os.mkdir(cachedir)
 		except (OSError,IOError), oe:
 			if oe.errno != errno.EEXIST: raise
 		self._cache_scan(time.time())
 
-	def default_ocean(self, ocean='ice'):
-		if self.ocean is None:
-			self.ocean = ocean
-
 	def _cache_scan(self, now):
 		# returns list of ages, unsorted
 		ages = []
@@ -204,12 +199,38 @@ class Fetcher:
 		debug('Fetcher  stored')
 		return data
 
+class Yoweb(Fetcher):
+	def __init__(self, ocean, cachedir):
+		debug('Yoweb init %s' % cachedir)
+		self.ocean = ocean
+		Fetcher.__init__(self, cachedir)
+
+	def default_ocean(self, ocean='ice'):
+		if self.ocean is None:
+			self.ocean = ocean
+
 	def yoweb(self, kind, tail, max_age):
 		self.default_ocean()
+		assert(self.ocean)
 		url = 'http://%s.puzzlepirates.com/yoweb/%s%s' % (
 			self.ocean, kind, tail)
 		return self.fetch(url, max_age)
 
+class Yppedia(Fetcher):
+	def __init__(self, cachedir):
+		debug('Yoweb init %s' % cachedir)
+		self.base = 'http://yppedia.puzzlepirates.com/'
+		self.localhtml = opts.localhtml
+		Fetcher.__init__(self, cachedir)
+
+	def __call__(self, rhs):
+		if self.localhtml is None:
+			url = self.base + rhs
+			debug('Yppedia retrieving YPP '+url);
+			return self.fetch(url, 3000)
+		else:
+			return file(opts.localhtml + '/' + rhs, 'r')
+
 #---------- logging assistance for troubled screenscrapers ----------
 
 class SoupLog:
@@ -340,12 +361,14 @@ u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-
 
 class CrewInfo(SomethingSoupInfo):
 	# Public data members:
+	#  ci.crewid
 	#  ci.crew = [ ('Captain',        ['Pirate', ...]),
 	#	       ('Senior Officer', [...]),
 	#		... ]
 	#  pi.msgs = [ 'message describing problem with scrape' ]
 
 	def __init__(self, crewid, max_age=300):
+		self.crewid = crewid
 		SomethingSoupInfo.__init__(self,
 			'crew/info.wm?crewid=', crewid, max_age)
 		self._find_crew()
@@ -386,23 +409,38 @@ 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]
 	#
 	#   islands[n] = (islandname, islandid)
 	#
 	def __init__(self, flagid, max_age=600):
+		self.flagid = flagid
 		SomethingSoupInfo.__init__(self,
 			'flag/info.wm?flagid=', flagid, max_age)
 		self._find_flag()
@@ -447,7 +485,13 @@ 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
@@ -497,49 +541,133 @@ class FlagInfo(SomethingSoupInfo):
 
 #---------- scraper for ocean info incl. embargoes etc. ----------
 
-class IslandInfo():
+class IslandBasicInfo():
+	# Public data attributes:
+	#  ocean
+	#  name
+	# Public data attributes maybe set by caller:
+	#  arch
 	def __init__(self, ocean, islename):
 		self.ocean = ocean
 		self.name = islename
-	def collect(self):
-		pass
-	def yppedia_dataf(self):
+	def yppedia(self):
 		def q(x): return urllib.quote(x.replace(' ','_'))
 		url_rhs = q(self.name) + '_(' + q(self.ocean) + ')'
-		if opts.localhtml is None:
-			url = 'http://yppedia.puzzlepirates.com/' + url_rhs
-			debug('IslandInfo retrieving YPP '+url);
-			return urllib.urlopen(url)
-		else:
-			return file(opts.localhtml + '/' + url_rhs, 'r')
-	def yoweb_url(self):
-		soup = BeautifulSoup(self.yppedia_dataf())
+		return yppedia(url_rhs)
+	def __str__(self):
+		return `(self.ocean, self.name)`
+
+class IslandExtendedInfo(IslandBasicInfo):
+	# Public data attributes (inherited):
+	#  ocean
+	#  name
+	# Public data attributes (additional):
+	#  islandid
+	#  yoweb_url
+	#  flagid
+	def __init__(self, ocean, islename):
+		IslandBasicInfo.__init__(self, ocean, islename)
+		self.islandid = None
+		self.yoweb_url = None
+		self._collect_yoweb()
+		self._collect_flagid()
+
+	def _collect_yoweb(self):
+		debug('IEI COLLECT YOWEB '+`self.name`)
+		self.islandid = None
+		self.yoweb_url = None
+
+		soup = BeautifulSoup(self.yppedia())
 		content = soup.find('div', attrs = {'id': 'content'})
 		yoweb_re = regexp.compile('^http://\w+\.puzzlepirates\.com/'+
-			'yoweb/island/info\.wm\?islandid=\d+$')
+			'yoweb/island/info\.wm\?islandid=(\d+)$')
 		a = soup.find('a', attrs = { 'href': yoweb_re })
-		if a is None: return None
-		return a['href']
-	def ruling_flag_id(self):
-		yo = self.yoweb_url()
+		if a is None:
+			debug('IEI COLLECT YOWEB '+`self.name`+' NONE')
+			return
+
+		debug('IEI COLLECT YOWEB '+`self.name`+' GOT '+``a``)
+		self.yoweb_url = a['href']
+		m = yoweb_re.search(self.yoweb_url)
+		self.islandid = m.group(1)
+
+	def _collect_flagid(self):
+		self.flagid = None
+
+		yo = self.yoweb_url
+		debug('IEI COLLECT FLAGID '+`self.name`+' URL '+`yo`)
 		if yo is None: return None
-		dataf = fetcher.fetch(yo, 600)
+		dataf = fetcher.fetch(yo, 1800)
 		soup = BeautifulSoup(dataf)
-		ruler_re = regexp.compile('http://\w+\.puzzlepirates\.com/'+
-			'yoweb/flag/info\.wm\?flagid=(\d+)$')
+		ruler_re = regexp.compile(
+			'/yoweb/flag/info\.wm\?flagid=(\d+)$')
 		ruler = soup.find('a', attrs = { 'href': ruler_re })
-		if not ruler: return None
-		m = ruler_re.find(ruler['href'])
-		return m.group(1)
+		if not ruler: 
+			debug('IEI COLLECT FLAGID '+`self.name`+' NONE')
+			return
+		debug('IEI COLLECT FLAGID '+`self.name`+' GOT '+``ruler``)
+		m = ruler_re.search(ruler['href'])
+		self.flagid = m.group(1)
+
+	def __str__(self):
+		return `(self.ocean, self.islandid, self.name,
+			self.yoweb_url, self.flagid)`
+
+class IslandFlagInfo(IslandExtendedInfo):
+	# Public data attributes (inherited):
+	#  ocean
+	#  name
+	#  islandid
+	#  yoweb_url
+	#  flagid
+	# Public data attributes (additional):
+	#  flag
+	def __init__(self, ocean, islename):
+		IslandExtendedInfo.__init__(self, ocean, islename)
+		self.flag = None
+		self._collect_flag()
+
+	def _collect_flag(self):
+		if self.flagid is None: return
+		self.flag = FlagInfo(self.flagid, 1800)
+
+	def __str__(self):
+		return IslandExtendedInfo.__str__(self) + '; ' + str(self.flag)
+
+class NullProgressReporter():
+	def doing(self, msg): pass
+	def stop(self): pass
+
+class TypewriterProgressReporter():
+	def __init__(self):
+		self._l = 0
+	def doing(self,m):
+		self._doing(m + '...')
+	def _doing(self,m):
+		self._write('\r')
+		self._write(m)
+		less = self._l - len(m)
+		if less > 0:
+			self._write(' ' * less)
+			self._write('\b' * less)
+		self._l = len(m)
+		sys.stdout.flush()
+	def stop(self):
+		self._doing('')
+		self._l = 0
+	def _write(self,t):
+		sys.stdout.write(t)
 
 class OceanInfo():
-	# Public data attributes (valid after collect()):
+	# Public data attributes:
 	#   oi.islands[islename] = IslandInfo(...)
 	#   oi.arches[archname][islename] = IslandInfo(...)
-	def __init__(self):
-		self.isleclass = IslandInfo
+	def __init__(self, isleclass=IslandBasicInfo):
+		self.isleclass = isleclass
 		self.ocean = fetcher.ocean.lower().capitalize()
-	def collect(self):
+
+		progressreporter.doing('fetching ocean info')
+
 		cmdl = ['./yppedia-ocean-scraper']
 		if opts.localhtml is not None:
 			cmdl += ['--local-html-dir',opts.localhtml]
@@ -557,10 +685,16 @@ class OceanInfo():
 		arch_re = regexp.compile('^ (\S.*)')
 		island_re = regexp.compile('^  (\S.*)')
 
+		oscraper.wait()
+		assert(oscraper.returncode == 0)
+
 		self.islands = { }
 		self.arches = { }
 		archname = None
 
+		isles = [ ]
+		progressreporter.doing('parsing ocean info')
+
 		for l in oscraper.stdout:
 			debug('OceanInfo collect l '+`l`)
 			l = l.rstrip('\n')
@@ -568,10 +702,7 @@ class OceanInfo():
 			if m:
 				assert(archname is not None)
 				islename = m.group(1)
-				isle = self.isleclass(self.ocean, islename)
-				isle.arch = archname
-				self.islands[islename] = isle
-				self.arches[archname][islename] = isle
+				isles.append((archname, islename))
 				continue
 			m = arch_re.match(l)
 			if m:
@@ -580,8 +711,19 @@ class OceanInfo():
 				self.arches[archname] = { }
 				continue
 			assert(False)
-		oscraper.wait()
-		assert(oscraper.returncode == 0)
+
+		for i in xrange(0, len(isles)-1):
+			(archname, islename) = isles[i]
+			progressreporter.doing(
+				'fetching isle info %2d/%d (%s: %s)'
+				% (i, len(isles), archname, islename))
+			isle = self.isleclass(self.ocean, islename)
+			isle.arch = archname
+			self.islands[islename] = isle
+			self.arches[archname][islename] = isle
+
+	def __str__(self):
+		return `(self.islands, self.arches)`
 
 #---------- pretty-printer for tables of pirate puzzle standings ----------
 
@@ -1232,22 +1374,27 @@ def do_pirate(pirates, bu):
 		print '%s: %s,' % (`pirate`, info)
 	print '}'
 
-def prep_crew_of(args, bu, max_age=300):
-	if len(args) != 1: bu('crew-of takes one pirate name')
+def prep_crewflag_of(args, bu, max_age, selector, constructor):
+	if len(args) != 1: bu('crew-of etc. take one pirate name')
 	pi = PirateInfo(args[0], max_age)
-	if pi.crew is None: return None
-	return CrewInfo(pi.crew[0], max_age)
+	cf = selector(pi)
+	if cf is None: return None
+	return constructor(cf[0], max_age)
+
+def prep_crew_of(args, bu, max_age=300):
+	return prep_crewflag_of(args, bu, max_age,
+		(lambda pi: pi.crew), CrewInfo)
+
+def prep_flag_of(args, bu, max_age=300):
+	return prep_crewflag_of(args, bu, max_age,
+		(lambda pi: pi.flag), FlagInfo)
 
 def do_crew_of(args, bu):
 	ci = prep_crew_of(args, bu)
 	print ci
 
 def do_flag_of(args, bu):
-	if len(args) != 1: bu('flag-of takes one pirate name')
-	max_age = 300
-	pi = PirateInfo(args[0], max_age)
-	if pi.flag is None: fi = None
-	else: fi = FlagInfo(pi.flag[0], max_age)
+	fi = prep_flag_of(args, bu)
 	print fi
 
 def do_standings_crew_of(args, bu):
@@ -1265,12 +1412,95 @@ def do_standings_crew_of(args, bu):
 def do_ocean(args, bu):
 	if (len(args)): bu('ocean takes no further arguments')
 	fetcher.default_ocean()
-	oi = OceanInfo()
-	oi.collect()
+	oi = OceanInfo(IslandFlagInfo)
+	print oi
 	for islename in sorted(oi.islands.keys()):
 		isle = oi.islands[islename]
-		yoweb_url = isle.yoweb_url()
-		print " %s -- %s" % (islename, yoweb_url)
+		print isle
+
+def do_embargoes(args, bu):
+	if (len(args)): bu('ocean takes no further arguments')
+	fetcher.default_ocean()
+	oi = OceanInfo(IslandFlagInfo)
+	wr = sys.stdout.write
+	print ('EMBARGOES:  Island    | Owning flag'+
+		'                    | Embargoed flags')
+
+	def getflname(isle):
+		if isle.islandid is None: return 'uncolonisable'
+		if isle.flag is None: return 'uncolonised'
+		return isle.flag.name
+
+	progressreporter.stop()
+
+	for archname in sorted(oi.arches.keys()):
+		print 'ARCHIPELAGO: ',archname
+		for islename in sorted(oi.arches[archname].keys()):
+			isle = oi.islands[islename]
+			wr(' %-20s | ' % isle.name)
+			flname = getflname(isle)
+			wr('%-30s | ' % flname)
+			flag = isle.flag
+			if flag is None: print ''; continue
+			delim = ''
+			for rel in flag.relations:
+				if rel.this_declaring >= 0: continue
+				wr(delim)
+				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:
+			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)
+	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 -----
 
@@ -1593,7 +1823,7 @@ class KeystrokeReader(DummyKeystrokeReader):
 #---------- main program ----------
 
 def main():
-	global opts, fetcher
+	global opts, fetcher, yppedia, progressreporter
 
 	pa = OptionParser(
 '''usage: .../yoweb-scrape [OPTION...] ACTION [ARGS...]
@@ -1602,6 +1832,8 @@ actions:
  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 ...] ocean|embargoes
+ 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:
@@ -1675,7 +1907,13 @@ display modes (for --display) apply to ship-aid:
 		else:
 			opts.display = 'overwrite'
 
-	fetcher = Fetcher(opts.ocean, opts.cache_dir)
+	fetcher = Yoweb(opts.ocean, opts.cache_dir)
+	yppedia = Yppedia(opts.cache_dir)
+
+	if opts.debug or not os.isatty(0):
+		 progressreporter = NullProgressReporter()
+	else:
+		progressreporter = TypewriterProgressReporter()
 
 	mode_fn(args[1:], pa.error)