From: Marcin Koziej Date: Mon, 30 Jan 2012 16:44:44 +0000 (+0100) Subject: Merge branch 'pretty' of github.com:fnp/wolnelektury into pretty X-Git-Url: https://git.mdrn.pl/wolnelektury.git/commitdiff_plain/cb40f164b028883a2fc061307383d0aa98f8b609?hp=01d7c5f6d52a56a9800ea7eccfeb25762261813d Merge branch 'pretty' of github.com:fnp/wolnelektury into pretty Conflicts: wolnelektury/templates/catalogue/book_searched.html --- diff --git a/apps/search/index.py b/apps/search/index.py index 47583ec32..71c0ed236 100644 --- a/apps/search/index.py +++ b/apps/search/index.py @@ -370,11 +370,18 @@ class Index(BaseIndex): return [] def walker(node, ignore_tags=[]): - yield node, None - for child in filter(lambda n: n.tag not in ignore_tags, list(node)): - for b, e in walker(child): - yield b, e - yield None, node + + if node.tag not in ignore_tags: + yield node, None, None + if node.text is not None: + yield None, node.text, None + for child in list(node): + for b, t, e in walker(child): + yield b, t, e + yield None, None, node + + if node.tail is not None: + yield None, node.tail, None return def fix_format(text): @@ -446,35 +453,50 @@ class Index(BaseIndex): # section content content = [] - footnote = None - - for start, end in walker(header, ignore_tags=self.ignore_content_tags): - # handle footnotes - # if start is not None and start.tag in self.footnote_tags: - # footnote = ' '.join(start.itertext()) - # elif end is not None and footnote is not None and end.tag in self.footnote_tags: - # doc = add_part(snippets, header_index=position, header_type=header.tag, - # content=footnote) + footnote = [] - # self.index.addDocument(doc) + def all_content(text): + for frag in fragments.values(): + frag['content'].append(text) + content.append(text) + handle_text = [all_content] - # footnote = None + for start, text, end in walker(header, ignore_tags=self.ignore_content_tags): + # handle footnotes + if start is not None and start.tag in self.footnote_tags: + footnote = [] + def collect_footnote(t): + footnote.append(t) + handle_text.append(collect_footnote) + elif end is not None and footnote is not [] and end.tag in self.footnote_tags: + handle_text.pop() + doc = add_part(snippets, header_index=position, header_type=header.tag, + content=u''.join(footnote), + is_footnote=Field("is_footnote", 'true', Field.Store.NO, Field.Index.NOT_ANALYZED)) + + self.index.addDocument(doc) + print "@ footnote text: %s" % footnote + footnote = [] + # handle fragments and themes. if start is not None and start.tag == 'begin': fid = start.attrib['id'][1:] fragments[fid] = {'content': [], 'themes': [], 'start_section': position, 'start_header': header.tag} + # themes for this fragment elif start is not None and start.tag == 'motyw': fid = start.attrib['id'][1:] + handle_text.append(None) if start.text is not None: fragments[fid]['themes'] += map(str.strip, map(give_me_utf8, start.text.split(','))) + elif end is not None and end.tag == 'motyw': + handle_text.pop() elif start is not None and start.tag == 'end': fid = start.attrib['id'][1:] if fid not in fragments: continue # a broken node, skip it - # import pdb; pdb.set_trace() frag = fragments[fid] if frag['themes'] == []: continue # empty themes list. @@ -487,22 +509,20 @@ class Index(BaseIndex): fragment_anchor=fid, content=fix_format(frag['content']), themes=frag['themes']) - + print '@ FRAG %s' % frag['content'] self.index.addDocument(doc) # Collect content. - elif start is not None: - for frag in fragments.values(): - frag['content'].append(start.text) - content.append(start.text) - elif end is not None: - for frag in fragments.values(): - frag['content'].append(end.tail) - content.append(end.tail) + + if text is not None and handle_text is not []: + hdl = handle_text[-1] + if hdl is not None: + hdl(text) # in the end, add a section text. doc = add_part(snippets, header_index=position, header_type=header.tag, content=fix_format(content)) + print '@ CONTENT: %s' % fix_format(content) self.index.addDocument(doc)