chiark / gitweb /
but prettier like this
[ypp-sc-tools.db-test.git] / yoweb-scrape
index 54ef0fe14bd62ade32cf38a1b9c2b9c78f88ffeb..a8a1a1169b0ee5f4ddb4b5548575d53b3c9ae348 100755 (executable)
@@ -10,22 +10,27 @@ import urllib2
 import errno
 import sys
 import re as regexp
+import random
 from optparse import OptionParser
 
 from BeautifulSoup import BeautifulSoup
 
 opts = None
 
-duties = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+
+
+puzzles = ('Swordfighting/Bilging/Sailing/Rigging/Navigating'+
        '/Battle Navigation/Gunning/Carpentry/Rumble/Treasure Haul'+
        '/Drinking/Spades/Hearts/Treasure Drop/Poker/Distilling'+
        '/Alchemistry/Shipwrightery/Blacksmithing/Foraging').split('/')
 
-standingvals = ('Able/Distinguished/Respected/Master/Renowned'+
-               '/Grand-Master/Legendary/Ultimate').split('/')
+standingvals = ('Able/Distinguished/Respected/Master'+
+               '/Renowned/Grand-Master/Legendary/Ultimate').split('/')
 
 pirate_ref_re = regexp.compile('^/yoweb/pirate\\.wm')
 
+max_pirate_namelen = 12
+
+
 def debug(m):
        if opts.debug:
                print >>sys.stderr, m
@@ -40,6 +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 = []
@@ -52,7 +61,7 @@ class Fetcher:
                                if oe.errno != errno.ENOENT: raise
                                continue
                        age = now - s.st_mtime
-                       if age > opts.max_age:
+                       if age > opts.expire_age:
                                debug('Fetcher    expire %d %s' % (age, path))
                                try: os.remove(path)
                                except (OSError,IOError), oe:
@@ -68,17 +77,17 @@ class Fetcher:
                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 *= 2
-                       min_age += 1
+                       min_age += 3
+                       min_age *= 1.25
                if need_wait > 0:
                        debug('Fetcher   wait %d' % need_wait)
                        time.sleep(need_wait)
 
-       def fetch(self, url):
+       def fetch(self, url, max_age):
                debug('Fetcher fetch %s' % url)
                cache_corename = urllib.quote_plus(url)
                cache_item = "%s/#%s#" % (self.cachedir, cache_corename)
@@ -87,15 +96,17 @@ class Fetcher:
                        if oe.errno != errno.ENOENT: raise
                        f = None
                now = time.time()
+               max_age = max(opts.min_max_age, min(max_age, opts.expire_age))
                if f is not None:
                        s = os.fstat(f.fileno())
-                       if now > s.st_mtime + opts.max_age:
-                               debug('Fetcher  stale')
+                       age = now - s.st_mtime
+                       if age > max_age:
+                               debug('Fetcher  stale %d < %d'% (max_age, age))
                                f = None
                if f is not None:
                        data = f.read()
                        f.close()
-                       debug('Fetcher  cached')
+                       debug('Fetcher  cached %d > %d' % (max_age, age))
                        return data
 
                debug('Fetcher  fetch')
@@ -112,10 +123,11 @@ class Fetcher:
                debug('Fetcher  stored')
                return data
 
-       def yoweb(self, kind, tail):
+       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)
+               return self.fetch(url, max_age)
 
 class SoupLog:
        def __init__(self):
@@ -133,23 +145,25 @@ def soup_text(obj):
        return str.strip()
 
 class SomethingSoupInfo(SoupLog):
-       def __init__(self, kind, tail):
+       def __init__(self, kind, tail, max_age):
                SoupLog.__init__(self)
-               html = fetcher.yoweb(kind, tail)
-               self.soup = BeautifulSoup(html,
+               html = fetcher.yoweb(kind, tail, max_age)
+               self._soup = BeautifulSoup(html,
                        convertEntities=BeautifulSoup.HTML_ENTITIES
                        )
 
 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' ]
                
-       def __init__(self, pirate):
+       def __init__(self, pirate, max_age=300):
                SomethingSoupInfo.__init__(self,
-                       'pirate.wm?target=', pirate)
+                       'pirate.wm?target=', pirate, max_age)
+               self.name = pirate
                self._find_standings()
                self.crew = self._find_crewflag('crew',
                        '^/yoweb/crew/info\\.wm')
@@ -157,69 +171,69 @@ class PirateInfo(SomethingSoupInfo):
                        '^/yoweb/flag/info\\.wm')
 
        def _find_standings(self):
-               imgs = self.soup.findAll('img',
+               imgs = self._soup.findAll('img',
                        src=regexp.compile('/yoweb/images/stat.*'))
                re = regexp.compile(
 u'\\s*\\S*/([-A-Za-z]+)\\s*$|\\s*\\S*/\\S*\\s*\\(ocean\\-wide(?:\\s|\\xa0)+([-A-Za-z]+)\\)\\s*$'
                        )
                standings = { }
 
-               for skill in duties:
+               for skill in puzzles:
                        standings[skill] = [ ]
 
                skl = SoupLog()
 
                for img in imgs:
-                       try: duty = img['alt']
+                       try: puzzle = img['alt']
                        except KeyError: continue
 
-                       if not duty in duties:
-                               skl.soupm(img, 'unknown duty: "%s"' % duty)
+                       if not puzzle in puzzles:
+                               skl.soupm(img, 'unknown puzzle: "%s"' % puzzle)
                                continue
                        key = img.findParent('td')
                        if key is None:
-                               skl.soupm(img, 'duty at root! "%s"' % duty)
+                               skl.soupm(img, 'puzzle at root! "%s"' % puzzle)
                                continue
                        valelem = key.findNextSibling('td')
                        if valelem is None:
-                               skl.soupm(key, 'duty missing sibling "%s"'
-                                       % duty)
+                               skl.soupm(key, 'puzzle missing sibling "%s"'
+                                       % puzzle)
                                continue
                        valstr = soup_text(valelem)
                        match = re.match(valstr)
                        if match is None:
-                               skl.soupm(key, ('duty "%s" unparseable'+
-                                       ' standing "%s"') % (duty, valstr))
+                               skl.soupm(key, ('puzzle "%s" unparseable'+
+                                       ' standing "%s"') % (puzzle, valstr))
                                continue
                        standing = match.group(match.lastindex)
-                       standings[duty].append(standing)
+                       standings[puzzle].append(standing)
 
                self.standings = { }
 
-               for duty in duties:
-                       sl = standings[duty]
+               for puzzle in puzzles:
+                       sl = standings[puzzle]
                        if len(sl) > 1:
-                               skl.msg('duty "%s" multiple standings %s' %
-                                               (duty, `sl`))
+                               skl.msg('puzzle "%s" multiple standings %s' %
+                                               (puzzle, `sl`))
                                continue
                        if not len(sl):
-                               skl.msg('duty "%s" no standing found' % duty)
+                               skl.msg('puzzle "%s" no standing found' % puzzle)
                                continue
                        standing = sl[0]
                        for i in range(0, len(standingvals)-1):
                                if standing == standingvals[i]:
-                                       self.standings[duty] = i
-                       if not duty in self.standings:
-                               skl.msg('duty "%s" unknown standing "%s"' %
-                                       (duty, standing))
+                                       self.standings[puzzle] = i
+                       if not puzzle in self.standings:
+                               skl.msg('puzzle "%s" unknown standing "%s"' %
+                                       (puzzle, standing))
 
                all_standings_ok = True
-               for duty in duties:
-                       if not duty in self.standings:
+               for puzzle in puzzles:
+                       if not puzzle in self.standings:
                                self.needs_msgs(skl)
 
        def _find_crewflag(self, cf, yoweb_re):
-               things = self.soup.findAll('a', href=regexp.compile(yoweb_re))
+               things = self._soup.findAll('a', href=regexp.compile(yoweb_re))
                if len(things) != 1:
                        self.msg('zero or several %s id references found' % cf)
                        return None
@@ -244,14 +258,14 @@ class CrewInfo(SomethingSoupInfo):
        #               ... ]
        #  pi.msgs = [ 'message describing problem with scrape' ]
 
-       def __init__(self, crewid):
+       def __init__(self, crewid, max_age=300):
                SomethingSoupInfo.__init__(self,
-                       'crew/info.wm?crewid=', crewid)
+                       'crew/info.wm?crewid=', crewid, max_age)
                self._find_crew()
 
        def _find_crew(self):
                self.crew = []
-               capts = self.soup.findAll('img',
+               capts = self._soup.findAll('img',
                        src='/yoweb/images/crew-captain.png')
                if len(capts) != 1:
                        self.msg('crew members: no. of captain images != 1')
@@ -285,6 +299,65 @@ class CrewInfo(SomethingSoupInfo):
        def __str__(self):
                return `(self.crew, self.msgs)`
 
+class StandingsTable:
+       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(max_pirate_namelen, 14), pirate)
+               for v in puzstrs:
+                       self.s += ' %-*.*s' % (self._cw,self._cw, v)
+               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 '?'
+               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(
+                                       ["%.*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))
+       def literalline(self, line):
+               self.s += line + '\n'
+       def pirate(self, pi):
+               puzstrs = [self._puzstr(pi,puz) for puz in self._puzzles]
+               self._pline(pi.name, puzstrs)
+
+       def results(self):
+               return self.s
+
 def do_pirate(pirates, bu):
        print '{'
        for pirate in pirates:
@@ -292,17 +365,26 @@ def do_pirate(pirates, bu):
                print '%s: %s,' % (`pirate`, info)
        print '}'
 
-def prep_crew_of(args, 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)
+       return CrewInfo(pi.crew[0], max_age)
 
 def do_crew_of(args, bu):
        ci = prep_crew_of(args, bu)
        print ci
 
-#def do_dutytab_crew_of(pirates, badusage):
-#      if len(pirates) != 1: badusage('dutytab-crew-of takes one pirate name')
+def do_standings_crew_of(args, bu):
+       ci = prep_crew_of(args, bu, 60)
+       tab = StandingsTable()
+       tab.headings()
+       for (rank, members) in ci.crew:
+               if not members: continue
+               tab.literalline('%s:' % rank)
+               for p in members:
+                       pi = PirateInfo(p, random.randint(900,1800))
+                       tab.pirate(pi)
+       print tab.results()
 
 def main():
        global opts, fetcher
@@ -312,11 +394,11 @@ def main():
 actions:
  yoweb-scrape [--ocean OCEAN ...] pirate PIRATE
  yoweb-scrape [--ocean OCEAN ...] crew-of PIRATE
- yoweb-scrape [--ocean OCEAN ...] dutytab-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',
@@ -325,7 +407,12 @@ actions:
                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')
+
        (opts,args) = pa.parse_args()
+       random.seed()
 
        if len(args) < 1:
                pa.error('need a mode argument')
@@ -336,7 +423,8 @@ actions:
        except KeyError: pa.error('unknown mode "%s"' % mode)
 
        # fixed parameters
-       opts.max_age = 240
+       opts.min_max_age = 60
+       opts.expire_age = 3600
 
        if opts.cache_dir.startswith('~/'):
                opts.cache_dir = os.getenv('HOME') + opts.cache_dir[1:]