Outputs a textual description of the provided BBC programme URL (or local file).

bbc_programme.py

#! /usr/bin/env python
import sys
from urllib import urlopen 
from bs4 import BeautifulSoup

def build_xmlns_alias_map( html_tag ):
    """
    :param html_tag: the `html` Tag that contains the `xmlns` attributes
    :return: a dict whose key is the namespace and whose value is the alias in this document
    """
    result = {}
    html_attrs = html_tag.attrs
    for a_name in html_attrs:
        a_value = html_attrs[ a_name ]
        if a_name.startswith('xmlns:'):
            ns_alias = a_name[ len('xmlns:'): ]
            result[ a_value ] = ns_alias
    return result

def main( argv ):
    """
    Scans the provided HTML document and enumerates the MusicSegment structures found therein.
    """
    XMLNS_DC = 'http://purl.org/dc/terms/'
    XMLNS_FOAF = 'http://xmlns.com/foaf/0.1/'
    XMLNS_MO = 'http://purl.org/ontology/mo/'
    XMLNS_PO = 'http://purl.org/ontology/po/'

    fh = urlopen( argv[1] )
    soup = BeautifulSoup( fh.read() )
    fh.close()
    del fh

    html = soup.find('html', recursive=False)
    del soup

    ns_alias_map = build_xmlns_alias_map( html )

    dc_ns = ns_alias_map[ XMLNS_DC ]
    mo_ns = ns_alias_map[ XMLNS_MO ]
    po_ns = ns_alias_map[ XMLNS_PO ]
    foaf_ns = ns_alias_map[ XMLNS_FOAF ]
    del ns_alias_map

    DcTitleNS = '%s:title' % dc_ns
    MusicSegmentNS = '%s:MusicSegment' % po_ns
    MusicArtistNS = '%s:MusicArtist' % mo_ns
    FoafNameNS = '%s:name' % foaf_ns

    segments = html.find( attrs= {'id':'segments'} )
    music_segs = segments.find_all( attrs= {'typeof':MusicSegmentNS} )
    del segments

    for seg in music_segs:
        art = seg.find( attrs= {'typeof': MusicArtistNS} )
        art_name = art.find( attrs= {'property': FoafNameNS} )
        tit = seg.find( attrs= {'property': DcTitleNS} )
        release = seg.find( attrs= {'class': r'release'} )

        release_label = None
        if release is not None:
            release_label = release.find( attrs= {'class': r'record-label'} )

        print('artist = %s' % art.string)
        print('title = %s' % tit.string)
        if release_label is not None:
            print('label = %s' % release_label.string)
        print('')

if __name__ == '__main__':
    main( sys.argv )