MythTV: bliptvXSL_api.py Source File

00001 #!/usr/bin/env python
00002 # -*- coding: UTF-8 -*-
00003 # ----------------------
00004 # Name: bliptvXSL_api - XPath and XSLT functions for the Blip.tv RSS/HTML itmes
00005 # Python Script
00006 # Author:   R.D. Vaughan
00007 # Purpose:  This python script is intended to perform a variety of utility functions
00008 #           for the conversion of data to the MNV standard RSS output format.
00009 #           See this link for the specifications:
00010 #           http://www.mythtv.org/wiki/MythNetvision_Grabber_Script_Format
00011 #
00012 # License:Creative Commons GNU GPL v2
00013 # (http://creativecommons.org/licenses/GPL/2.0/)
00014 #-------------------------------------
00015 __title__ ="bliptvXSL_api - XPath and XSLT functions for the Blip.tv RSS/HTML"
00016 __author__="R.D. Vaughan"
00017 __purpose__='''
00018 This python script is intended to perform a variety of utility functions
00019 for the conversion of data to the MNV standard RSS output format.
00020 See this link for the specifications:
00021 http://www.mythtv.org/wiki/MythNetvision_Grabber_Script_Format
00022 '''
00023 
00024 __version__="v0.1.1"
00025 # 0.1.0 Initial development
00026 # 0.1.1 Fixed a bug when an autoplay link cannot be created
00027 #       Added MP4 as an acceptable downloadable video file type
00028 
00029 
00030 # Specify the class names that have XPath extention functions
00031 __xpathClassList__ = ['xpathFunctions', ]
00032 
00033 # Specify the XSLT extention class names. Each class is a stand lone extention function
00034 #__xsltExtentionList__ = ['xsltExtExample', ]
00035 __xsltExtentionList__ = []
00036 
00037 import os, sys, re, time, datetime, shutil, urllib, string
00038 from copy import deepcopy
00039 
00040 
00041 class OutStreamEncoder(object):
00042     """Wraps a stream with an encoder"""
00043     def __init__(self, outstream, encoding=None):
00044         self.out = outstream
00045         if not encoding:
00046             self.encoding = sys.getfilesystemencoding()
00047         else:
00048             self.encoding = encoding
00049 
00050     def write(self, obj):
00051         """Wraps the output stream, encoding Unicode strings with the specified encoding"""
00052         if isinstance(obj, unicode):
00053             try:
00054                 self.out.write(obj.encode(self.encoding))
00055             except IOError:
00056                 pass
00057         else:
00058             try:
00059                 self.out.write(obj)
00060             except IOError:
00061                 pass
00062 
00063     def __getattr__(self, attr):
00064         """Delegate everything but write to the stream"""
00065         return getattr(self.out, attr)
00066 sys.stdout = OutStreamEncoder(sys.stdout, 'utf8')
00067 sys.stderr = OutStreamEncoder(sys.stderr, 'utf8')
00068 
00069 try:
00070     from StringIO import StringIO
00071     from lxml import etree
00072 except Exception, e:
00073     sys.stderr.write(u'\n! Error - Importing the "lxml" and "StringIO" python libraries failed on error(%s)\n' % e)
00074     sys.exit(1)
00075 
00076 # Check that the lxml library is current enough
00077 # From the lxml documents it states: (http://codespeak.net/lxml/installation.html)
00078 # "If you want to use XPath, do not use libxml2 2.6.27. We recommend libxml2 2.7.2 or later"
00079 # Testing was performed with the Ubuntu 9.10 "python-lxml" version "2.1.5-1ubuntu2" repository package
00080 version = ''
00081 for digit in etree.LIBXML_VERSION:
00082     version+=str(digit)+'.'
00083 version = version[:-1]
00084 if version < '2.7.2':
00085     sys.stderr.write(u'''
00086 ! Error - The installed version of the "lxml" python library "libxml" version is too old.
00087           At least "libxml" version 2.7.2 must be installed. Your version is (%s).
00088 ''' % version)
00089     sys.exit(1)
00090 
00091 
00092 class xpathFunctions(object):
00093     """Functions specific extending XPath
00094     """
00095     def __init__(self):
00096         self.functList = ['bliptvFlvLinkGeneration', 'bliptvDownloadLinkGeneration', 'bliptvEpisode', 'bliptvIsCustomHTML', ]
00097         self.episodeRegex = [
00098             re.compile(u'''TERRA\\ (?P<episodeno>[0-9]+).*$''', re.UNICODE),
00099             ]
00100         self.namespaces = {
00101             'xsi': u"http://www.w3.org/2001/XMLSchema-instance",
00102             'media': u"http://search.yahoo.com/mrss/",
00103             'xhtml': u"http://www.w3.org/1999/xhtml",
00104             'atm': u"http://www.w3.org/2005/Atom",
00105             'mythtv': "http://www.mythtv.org/wiki/MythNetvision_Grabber_Script_Format",
00106             'itunes':"http://www.itunes.com/dtds/podcast-1.0.dtd",
00107             'creativeCommons': "http://backend.userland.com/creativeCommonsRssModule",
00108             'geo': "http://www.w3.org/2003/01/geo/wgs84_pos#",
00109             'blip': "http://blip.tv/dtd/blip/1.0",
00110             'wfw': "http://wellformedweb.org/CommentAPI/",
00111             'amp': "http://www.adobe.com/amp/1.0",
00112             'dcterms': "http://purl.org/dc/terms",
00113             'gm': "http://www.google.com/schemas/gm/1.1",
00114             'mediaad': "http://blip.tv/dtd/mediaad/1.0",
00115             }
00116         self.flvFilter = etree.XPath(".//media:content[@type='video/x-flv']", namespaces=self.namespaces)
00117         self.m4vFilter = etree.XPath(".//media:content[@type='video/mp4' or @type='video/quicktime' or @type='video/x-m4v']", namespaces=self.namespaces)
00118         self.durationFilter = etree.XPath(".//blip:runtime/text()", namespaces=self.namespaces)
00119         self.linkFilter = etree.XPath("./link/text()", namespaces=self.namespaces)
00120         self.languageFilter = etree.XPath("../language/text()", namespaces=self.namespaces)
00121     # end __init__()
00122 
00123 ######################################################################################################
00124 #
00125 # Start of XPath extension functions
00126 #
00127 ######################################################################################################
00128 
00129     def bliptvFlvLinkGeneration(self, context, arg):
00130         '''Generate a link for the Blip.tv site.
00131         Call example: 'mnvXpath:bliptvFlvLinkGeneration(.)'
00132         return the url link
00133         '''
00134         flvFile = self.flvFilter(arg[0])
00135         if len(flvFile):
00136             flvFileLink = flvFile[0].attrib['url']
00137             return u'%s%s' % (common.linkWebPage('dummy', 'bliptv'), flvFileLink.replace(u'.flv', u'').replace(u'http://blip.tv/file/get/', u''))
00138         else:
00139             return self.linkFilter(arg[0])[0]
00140     # end bliptvXSLLinkGeneration()
00141 
00142     def bliptvDownloadLinkGeneration(self, context, arg):
00143         '''Generate a download link for the Blip.tv site.
00144         Call example: 'mnvXpath:bliptvDownloadLinkGeneration(.)'
00145         return an array of one download link element
00146         '''
00147         downloadLink = etree.XML(u'<link></link>')
00148         flvFile = self.flvFilter(arg[0])
00149         m4vFile = self.m4vFilter(arg[0])
00150         if len(m4vFile):
00151             downloadLink.attrib['url'] = m4vFile[0].attrib['url']
00152             if m4vFile[0].attrib.get('width'):
00153                 downloadLink.attrib['width'] = m4vFile[0].attrib['width']
00154             if m4vFile[0].attrib.get('height'):
00155                 downloadLink.attrib['height'] = m4vFile[0].attrib['height']
00156             if m4vFile[0].attrib.get('fileSize'):
00157                 downloadLink.attrib['length'] = m4vFile[0].attrib['fileSize']
00158             if len(self.durationFilter(arg[0])):
00159                 downloadLink.attrib['duration'] = self.durationFilter(arg[0])[0]
00160             downloadLink.attrib['lang'] = self.languageFilter(arg[0])[0]
00161             return [downloadLink]
00162         elif len(flvFile):
00163             downloadLink.attrib['url'] = flvFile[0].attrib['url']
00164             if flvFile[0].attrib.get('width'):
00165                 downloadLink.attrib['width'] = flvFile[0].attrib['width']
00166             if flvFile[0].attrib.get('height'):
00167                 downloadLink.attrib['height'] = flvFile[0].attrib['height']
00168             if flvFile[0].attrib.get('fileSize'):
00169                 downloadLink.attrib['length'] = flvFile[0].attrib['fileSize']
00170             if len(self.durationFilter(arg[0])):
00171                 downloadLink.attrib['duration'] = self.durationFilter(arg[0])[0]
00172             downloadLink.attrib['lang'] = self.languageFilter(arg[0])[0]
00173             return [downloadLink]
00174         else:
00175             downloadLink.attrib['url'] = self.linkFilter(arg[0])[0]
00176             if len(self.durationFilter(arg[0])):
00177                 downloadLink.attrib['duration'] = self.durationFilter(arg[0])[0]
00178             downloadLink.attrib['lang'] = self.languageFilter(arg[0])[0]
00179             return [downloadLink]
00180     # end bliptvDownloadLinkGeneration()
00181 
00182     def bliptvEpisode(self, context, arg):
00183         '''Parse the title string and extract an episode number
00184         Call example: 'mnvXpath:bliptvEpisode(./title/text())'
00185         return the url link
00186         '''
00187         episodeNumber = u''
00188         for index in range(len(self.episodeRegex)):
00189             match = self.episodeRegex[index].match(arg[0])
00190             if match:
00191                 episodeNumber = match.groups()
00192                 break
00193         return etree.XML(u'<episode>%s</episode>' % episodeNumber)
00194     # end bliptvEpisode()
00195 
00196     def bliptvIsCustomHTML(self, context, arg):
00197         '''Parse the item element and deternmine if there is a flv file
00198         Call example: 'mnvXpath:bliptvIsCustomHTML(.)'
00199         return True is there is a '.flv' file
00200         return False if there is no .flv' file
00201         '''
00202         if len(self.flvFilter(arg[0])):
00203             return True
00204         return False
00205     # end bliptvIsCustomHTML()
00206 
00207 ######################################################################################################
00208 #
00209 # End of XPath extension functions
00210 #
00211 ######################################################################################################
00212 
00213 ######################################################################################################
00214 #
00215 # Start of XSLT extension functions
00216 #
00217 ######################################################################################################
00218 
00219 ######################################################################################################
00220 #
00221 # End of XSLT extension functions
00222 #
00223 ######################################################################################################