X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/cb40f164b028883a2fc061307383d0aa98f8b609..0106fe8da9cfe46723150db88c25c7e0736619e0:/apps/search/index.py?ds=inline diff --git a/apps/search/index.py b/apps/search/index.py index 71c0ed236..e0f9de997 100644 --- a/apps/search/index.py +++ b/apps/search/index.py @@ -1,7 +1,7 @@ # -*- coding: utf-8 -*- from django.conf import settings -from lucene import SimpleFSDirectory, IndexWriter, CheckIndex, \ +from lucene import SimpleFSDirectory, IndexWriter, IndexWriterConfig, CheckIndex, \ File, Field, Integer, \ NumericField, Version, Document, JavaError, IndexSearcher, \ QueryParser, PerFieldAnalyzerWrapper, \ @@ -169,11 +169,13 @@ class BaseIndex(IndexStore): analyzer = WLAnalyzer() self.analyzer = analyzer - def open(self, analyzer=None): + def open(self, analyzer=None, timeout=None): if self.index: raise Exception("Index is already opened") - self.index = IndexWriter(self.store, self.analyzer,\ - IndexWriter.MaxFieldLength.LIMITED) + conf = IndexWriterConfig(Version.LUCENE_34, analyzer) + if timeout: + conf.setWriteLockTimeout(long(timeout)) + self.index = IndexWriter(self.store, conf) return self.index def optimize(self): @@ -331,11 +333,13 @@ class Index(BaseIndex): (dt.year, dt.month, dt.day), Field.Store.NO, Field.Index.NOT_ANALYZED) # get published date - source = book_info.source_name - if hasattr(book_info, 'source_name'): - match = self.published_date_re.search(source) + pd = None + if hasattr(book_info, 'source_name') and book_info.source_name: + match = self.published_date_re.search(book_info.source_name) if match is not None: - fields["published_date"] = Field("published_date", str(match.groups()[0]), Field.Store.YES, Field.Index.NOT_ANALYZED) + pd = str(match.groups()[0]) + if not pd: pd = "" + fields["published_date"] = Field("published_date", pd, Field.Store.YES, Field.Index.NOT_ANALYZED) return fields @@ -476,7 +480,7 @@ class Index(BaseIndex): is_footnote=Field("is_footnote", 'true', Field.Store.NO, Field.Index.NOT_ANALYZED)) self.index.addDocument(doc) - print "@ footnote text: %s" % footnote + #print "@ footnote text: %s" % footnote footnote = [] # handle fragments and themes. @@ -509,7 +513,7 @@ class Index(BaseIndex): fragment_anchor=fid, content=fix_format(frag['content']), themes=frag['themes']) - print '@ FRAG %s' % frag['content'] + #print '@ FRAG %s' % frag['content'] self.index.addDocument(doc) # Collect content. @@ -522,7 +526,7 @@ class Index(BaseIndex): # 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) + #print '@ CONTENT: %s' % fix_format(content) self.index.addDocument(doc) @@ -551,12 +555,12 @@ class ReusableIndex(Index): """ index = None - def open(self, analyzer=None, threads=4): - if ReusableIndex.index is not None: + def open(self, analyzer=None, **kw): + if ReusableIndex.index: self.index = ReusableIndex.index else: print("opening index") - Index.open(self, analyzer) + Index.open(self, analyzer, **kw) ReusableIndex.index = self.index atexit.register(ReusableIndex.close_reusable) @@ -566,13 +570,15 @@ class ReusableIndex(Index): @staticmethod def close_reusable(): - if ReusableIndex.index is not None: + if ReusableIndex.index: + print("closing index") ReusableIndex.index.optimize() ReusableIndex.index.close() ReusableIndex.index = None def close(self): - pass + if ReusableIndex.index: + ReusableIndex.index.commit() class JoinSearch(object): @@ -728,7 +734,7 @@ class SearchResult(object): for f in frags: try: - frag = catalogue.models.Fragment.objects.get(anchor=f[FRAGMENT]) + frag = catalogue.models.Fragment.objects.get(anchor=f[FRAGMENT], book__id=self.book_id) except catalogue.models.Fragment.DoesNotExist: # stale index continue @@ -1208,17 +1214,25 @@ class Search(IndexStore): if position is None or length is None: return None # locate content. - snippets = Snippets(stored.get('book_id')).open() + book_id = int(stored.get('book_id')) + snippets = Snippets(book_id).open() try: - text = snippets.get((int(position), - int(length))) - finally: - snippets.close() + try: + text = snippets.get((int(position), + int(length))) + finally: + snippets.close() - tokenStream = TokenSources.getAnyTokenStream(self.searcher.getIndexReader(), scoreDoc.doc, field, self.analyzer) - # highlighter.getBestTextFragments(tokenStream, text, False, 10) - snip = highlighter.getBestFragments(tokenStream, text, 3, "...") + tokenStream = TokenSources.getAnyTokenStream(self.searcher.getIndexReader(), scoreDoc.doc, field, self.analyzer) + # highlighter.getBestTextFragments(tokenStream, text, False, 10) + snip = highlighter.getBestFragments(tokenStream, text, 3, "...") + except Exception, e: + e2 = e + if hasattr(e, 'getJavaException'): + e2 = unicode(e.getJavaException()) + raise Exception("Problem fetching snippets for book %d, @%d len=%d" % (book_id, int(position), int(length)), + e2) return snip @staticmethod @@ -1298,7 +1312,7 @@ class Search(IndexStore): return only_term - def hint_tags(self, string, max_results=50, pdcounter=True, prefix=True): + def hint_tags(self, string, max_results=50, pdcounter=True, prefix=True, fuzzy=False): """ Return auto-complete hints for tags using prefix search. @@ -1310,14 +1324,14 @@ class Search(IndexStore): if prefix: q = self.make_prefix_phrase(toks, field) else: - q = self.make_term_query(toks, field) + q = self.make_term_query(toks, field, fuzzy=fuzzy) top.add(BooleanClause(q, BooleanClause.Occur.SHOULD)) no_book_cat = self.term_filter(Term("tag_category", "book"), inverse=True) return self.search_tags(top, no_book_cat, max_results=max_results, pdcounter=pdcounter) - def hint_books(self, string, max_results=50, prefix=True): + def hint_books(self, string, max_results=50, prefix=True, fuzzy=False): """ Returns auto-complete hints for book titles Because we do not index 'pseudo' title-tags. @@ -1328,7 +1342,7 @@ class Search(IndexStore): if prefix: q = self.make_prefix_phrase(toks, 'title') else: - q = self.make_term_query(toks, 'title') + q = self.make_term_query(toks, 'title', fuzzy=fuzzy) return self.search_books(q, self.term_filter(Term("is_book", "true")), max_results=max_results)