Source Code for Module oort.util._genshifilters

 1  # TODO: rewrite using only elementtree..(?) 
 2  # TODO: cannot handle ordered fallbacks (will filter all others or none) 
 3  #        - solutions may have to expand all literals, store their language and 
 4  #          fallback accordingly, or use a rather complex simultaneous iteraror 
 5  #          loop on all values. 
 6   
 7  from genshi import XML 
 8  from genshi.core import Stream 
 9   
10  START, END = Stream.START, Stream.END 
11  XML_LANG = u'{http://www.w3.org/XML/1998/namespace}lang' 
12  RDF_WRAPPER = u'rdf-wrapper' 
13   
14 -def language_filtered_xml(valueOrList, lang, fragment=True, encoding=None): 
15      if isinstance(valueOrList, unicode): 
16          return langXML(valueOrList, lang, fragment, encoding) 
17      else: 
18          # TODO: use flattened iterator instead..(?) 
19          events = [] 
20          for value in valueOrList: 
21              if value: 
22                  events.extend( langXML(value, lang, fragment, encoding).events ) 
23          return Stream(events) 
24   
25 -def langXML(text, lang, fragment=True, encoding=None): 
26      if text.startswith('<?xml ') or text.startswith('<!DOCTYPE '): 
27          fragment = False 
28      if fragment: 
29          text = '<xml>%s</xml>' % text 
30      if isinstance(text, unicode): 
31          encoding = 'utf-16' 
32          text = text.encode(encoding) 
33      stream = XML(text) 
34      lang_filter = filter_language(lang) 
35      if fragment: 
36          return stream | skip_outer | lang_filter 
37      else: 
38          return stream | lang_filter 
39   
40 -def skip_outer(stream): 
41      """A filter that doesn't actually do anything with the stream.""" 
42      istream = iter(stream) 
43      istream.next() 
44      last = None 
45      for content in istream: 
46          if last: yield last 
47          last = content 
48   
49 -def filter_language(lang): 
50      def filter_lang(stream): 
51          depth = 0 
52          eating = False 
53          for kind, data, pos in stream: 
54              if kind == START and lang: 
55                  elemLang = get_elem_lang(data) 
56                  if elemLang: 
57                      # TODO: remove? remove_lang_attrib(data) 
58                      if elemLang != lang: 
59                          eating = True 
60              if eating == True: 
61                  if kind == START: 
62                      depth += 1 
63                  elif kind == END: 
64                      depth -= 1 
65                      if depth == 0: eating = False 
66                  continue 
67   
68              if kind == START and data[0] == RDF_WRAPPER: 
69                  continue 
70              elif kind == END and data == RDF_WRAPPER: 
71                  continue 
72              else: 
73                  yield kind, data, pos 
74   
75      return filter_lang 
76   
77 -def get_elem_lang(data): 
78      # FIXME: Remove 'lang', it's a temporary solution due 
79      # to rdflib xml/rdf XMLLiteral with embedded xml:lang 
80      # parsing problems.. 
81      for name, value in data[1]: 
82          if name in (XML_LANG, 'lang'): 
83              return value 
84