dee2.py

from urllib import urlencode
from urllib2 import urlopen,HTTPError,URLError
import socket
import re
import json

try:
    from unescape_charref import unescape_charref
except ImportError:
    unescape_charref=None

class DEE2JSONDecoder(json.JSONDecoder):
    def __init__(self,encoding=None,**kwgs):
        # some control sequences are not escaped in string
        # ex. eventid=105(headline) line 319 u'\t\t\t"Info" : "...\t(^^)"'
        # toavoid this set strict=False
        json.JSONDecoder.__init__(self,encoding,strict=False,**kwgs)

class DEE2EventData(object):
    DEE2_URL='http://manbow.nothing.sh/event/event.cgi'
    GETJSON_PARAM=(('action','JSONList'),)

    def __init__(self,eventid,name=None):
        self._eventid=int(eventid)
        self._name=name if isinstance(name,basestring) else None
        self._songlist=[]
        self._headline=[]

    def getsonglist(self,utf8=False):
        tmpjson=self._getjson(utf8)
        if self._name is None:
            self._name=tmpjson.get('EventName')
        self._songlist=tmpjson['Data']

    def getheadline(self,utf8=False):
        tmpjson=self._getjson(utf8,(('output','headline'),))
        if self._name is None:
            self._name=tmpjson.get('EventName')
        self._headline=tmpjson['Data']

    def get(self,utf8=False):
        self.getsonglist(utf8)
        self.getheadline(utf8)

    def _getjson(self,utf8,iterable=None):
        param=self.GETJSON_PARAM
        if utf8==True:
            param+=(('utf-8','on'),)
        if iterable:
            param+=iterable
        query=urlencode(param+(('event',self._eventid),))
        try:
            ret=urlopen(self.DEE2_URL+'?'+query).read()
        except URLError,e:
            if isinstance(e.reason, socket.timeout):
                raise Exception("timeout: %s" % e)
            else:
                raise
        except socket.timeout,e:
            raise Exception("timeout: %s" % e)

        # fix structure error: [{...},{...},...,{...},] -> [{...},{...},...,{...}]
        # this may be happen when last registered entry is deleted
        # ex. eventid=104 BOFU2015
        ret=re.sub(
            r'}\s*,\s*\]\s*}\s*\Z',
            r'}]}',
            ret
        )

        if not utf8:
            # remove extra escapes for Damemoji: [cp932 1st byte]\\ -> [cp932 1st byte]\
            ret=re.sub(
                '([\x81-\x9f\xe0-\xfc]'+r'\\)\\',
                r'\1',
                ret
            )

            # for output=headline
            # truncated multibyte character involves '"' when decoded
            # to avoid this, first, inserting ' ' before '"'
            ret=re.sub(
                r'^(\s*\".*?\"\s*:\s*\".*)\"(\s*[,\]}]?\s*)$',
                r'\1 "\2',
                ret,flags=re.MULTILINE
            )
            # second, decode this with errors='replace'
            # truncated character('[\x81-\x9f\xe0-\xfc] ') generates '\ufffd'
            # valid string does not change ' '
            ret=ret.decode('cp932',errors='replace')
            # and last, remove ' ' or '\ufffd' from end of each string
            ret=re.sub(
                ur'^(\s*\".*?\"\s*:\s*\".*)[ \ufffd]\"(\s*[,\]}]?\s*)$',
                ur'\1"\2',
                ret,flags=re.MULTILINE
            )

        else:
            # if utf8 mode, only to do is decoding string
            ret=ret.decode('cp932',errors='replace')

        if unescape_charref:
            # unescape character reference
            ret=unescape_charref(ret)

        # in some cases strings contain unescaped '"'
        # ex. eventid=26 line 2 '\t"EventName" : "B-1 "Unrestricted"" ,'
        # add escapes to those
        ret=re.sub(
            ur'(^\s*\".*?\"\s*:\s*\")(.*\".*)(?=\"\s*[,\]}]?\s*$)',
            lambda s:s.group(1)+re.sub(
                ur'((^|[^\\])(?:\\\\)*)(?=\")',
                ur'\1'+ur'\\',
                s.group(2)
            ),
            ret,flags=re.MULTILINE
        )

        return json.loads(ret,cls=DEE2JSONDecoder)

    def clear(self):
        self._eventid=None
        self._name=None
        self._songlist=[]
        self._headline=[]

    @property
    def eventid(self):
        return self._eventid
    @eventid.setter
    def eventid(self,evid):
        if not getattr(evid,'__int__') or int(evid)<=0:
            raise ValueError('eventid must be positive integer')
        elif int(evid)==self.eventid:
            self._eventid=int(evid)
        else:
            self.clear()
            self._eventid=int(evid)
    @eventid.deleter
    def eventid(self):
        self.clear()

    @property
    def eventname(self):
        return self._name
    @eventname.setter
    def eventname(self,name):
        self._name=name
    @eventname.deleter
    def eventname(self):
        self._name=None

    @property
    def songlist(self):
        return self._songlist
    @songlist.setter
    def songlist(self,sl):
        if getattr(sl, '__iter__'):
            self._songlist=list(sl)
        else:
            self._songlist=[]
    @songlist.deleter
    def songlist(self):
        self._songlist=[]

    @property
    def headline(self):
        return self._headline
    @headline.setter
    def headline(self,hl):
        if getattr(hl, '__iter__'):
            self._headline=list(hl)
        else:
            self._headline=[]
    @headline.deleter
    def headline(self):
        self._headline=[]

if __name__=='__main__':
    import sys
    import time
    args=sys.argv
    if len(args)>=2:
        for evid in args[1:]:
            try:
                ev=DEE2EventData(evid)
            except:
                continue
            ev.get()
            print u'EventID={} EventName={} songs={}'.format(
                ev.eventid,ev.eventname,len(ev.songlist)
            )
            ev.clear()
            if len(args)>2:time.sleep(1)
    else:
        print u'Usage: dee2.py ID1 ID2 ...'