+ bks.append(catalogue.models.Book.objects.get(id=doc.get("book_id")))
+ return (bks, tops.totalHits)
+
+# TokenStream tokenStream = analyzer.tokenStream(fieldName, reader);
+# OffsetAttribute offsetAttribute = tokenStream.getAttribute(OffsetAttribute.class);
+# CharTermAttribute charTermAttribute = tokenStream.getAttribute(CharTermAttribute.class);
+
+# while (tokenStream.incrementToken()) {
+# int startOffset = offsetAttribute.startOffset();
+# int endOffset = offsetAttribute.endOffset();
+# String term = charTermAttribute.toString();
+# }
+
+
+class SearchResult(object):
+ def __init__(self, searcher, scoreDocs, score=None):
+ if score:
+ self.score = score
+ else:
+ self.score = scoreDocs.score
+
+ self.fragments = []
+ self.scores = {}
+ self.sections = []
+
+ stored = searcher.doc(scoreDocs.doc)
+ self.book_id = int(stored.get("book_id"))
+
+ fragment = stored.get("fragment_anchor")
+ if fragment:
+ self.fragments.append(fragment)
+ self.scores[fragment] = scoreDocs.score
+
+ header_type = stored.get("header_type")
+ if header_type:
+ sec = (header_type, int(stored.get("header_index")))
+ self.sections.append(sec)
+ self.scores[sec] = scoreDocs.score
+
+ def get_book(self):
+ return catalogue.models.Book.objects.get(id=self.book_id)
+
+ book = property(get_book)
+
+ def get_parts(self):
+ book = self.book
+ parts = [{"header": s[0], "position": s[1], '_score_key': s} for s in self.sections] \
+ + [{"fragment": book.fragments.get(anchor=f), '_score_key':f} for f in self.fragments]
+
+ parts.sort(lambda a, b: cmp(self.scores[a['_score_key']], self.scores[b['_score_key']]))
+ print("bookid: %d parts: %s" % (self.book_id, parts))
+ return parts
+
+ parts = property(get_parts)
+
+ def merge(self, other):
+ if self.book_id != other.book_id:
+ raise ValueError("this search result is or book %d; tried to merge with %d" % (self.book_id, other.book_id))
+ self.fragments += other.fragments
+ self.sections += other.sections
+ self.scores.update(other.scores)
+ if other.score > self.score:
+ self.score = other.score
+ return self
+
+ def __unicode__(self):
+ return u'SearchResult(book_id=%d, score=%d)' % (self.book_id, self.score)
+
+ @staticmethod
+ def aggregate(*result_lists):
+ books = {}
+ for rl in result_lists:
+ for r in rl:
+ if r.book_id in books:
+ books[r.book_id].merge(r)
+ #print(u"already have one with score %f, and this one has score %f" % (books[book.id][0], found.score))
+ else:
+ books[r.book_id] = r
+ return books.values()
+
+ def __cmp__(self, other):
+ return cmp(self.score, other.score)
+
+
+class MultiSearch(Search):
+ """Class capable of IMDb-like searching"""
+ def get_tokens(self, queryreader):
+ if isinstance(queryreader, str) or isinstance(queryreader, unicode):
+ queryreader = StringReader(queryreader)
+ queryreader.reset()
+ tokens = self.analyzer.reusableTokenStream('content', queryreader)
+ toks = []
+ while tokens.incrementToken():
+ cta = tokens.getAttribute(CharTermAttribute.class_)
+ toks.append(cta.toString())
+ return toks
+
+ def make_phrase(self, tokens, field='content', slop=2):
+ phrase = PhraseQuery()
+ phrase.setSlop(slop)
+ for t in tokens:
+ term = Term(field, t)
+ phrase.add(term)
+ return phrase
+
+ def make_term_query(self, tokens, field='content', modal=BooleanClause.Occur.SHOULD):
+ q = BooleanQuery()
+ for t in tokens:
+ term = Term(field, t)
+ q.add(BooleanClause(TermQuery(term), modal))
+ return q
+
+ def content_query(self, query):
+ return BlockJoinQuery(query, self.parent_filter,
+ BlockJoinQuery.ScoreMode.Total)
+
+ def search_perfect_book(self, tokens, max_results=20):
+ qrys = [self.make_phrase(tokens, field=fld) for fld in ['author', 'title']]
+
+ books = []
+ for q in qrys:
+ top = self.searcher.search(q, max_results)
+ for found in top.scoreDocs:
+ books.append(SearchResult(self.searcher, found))
+ return books
+
+ def search_perfect_parts(self, tokens, max_results=20):
+ qrys = [self.make_phrase(tokens, field=fld) for fld in ['content']]
+
+ books = []
+ for q in qrys:
+ top = self.searcher.search(q, max_results)
+ for found in top.scoreDocs:
+ books.append(SearchResult(self.searcher, found))
+
+ return books
+
+ def search_everywhere(self, tokens, max_results=20):
+ q = BooleanQuery()
+ in_meta = BooleanQuery()
+ in_content = BooleanQuery()
+
+ for fld in ['themes', 'content']:
+ in_content.add(BooleanClause(self.make_term_query(tokens, field=fld), BooleanClause.Occur.SHOULD))
+
+ for fld in ['author', 'title', 'epochs', 'genres', 'kinds']:
+ in_meta.add(BooleanClause(self.make_term_query(tokens, field=fld), BooleanClause.Occur.SHOULD))
+
+ q.add(BooleanClause(in_meta, BooleanClause.Occur.MUST))
+ in_content_join = self.content_query(in_content)
+ q.add(BooleanClause(in_content_join, BooleanClause.Occur.MUST))
+
+ collector = BlockJoinCollector(Sort.RELEVANCE, 100, True, True)
+
+ self.searcher.search(q, collector)
+
+ books = []
+
+ top_groups = collector.getTopGroups(in_content_join, Sort.RELEVANCE, 0, max_results, 0, True)
+ if top_groups:
+ for grp in top_groups.groups:
+ for part in grp.scoreDocs:
+ books.append(SearchResult(self.searcher, part, score=grp.maxScore))
+ return books
+
+ def multisearch(self, query, max_results=50):
+ """
+ Search strategy:
+ - (phrase) OR -> content
+ -> title
+ -> author
+ - (keywords) -> author
+ -> motyw
+ -> tags
+ -> content
+ """
+ # queryreader = StringReader(query)
+ # tokens = self.get_tokens(queryreader)
+
+ # top_level = BooleanQuery()
+ # Should = BooleanClause.Occur.SHOULD
+
+ # phrase_level = BooleanQuery()
+ # phrase_level.setBoost(1.3)
+
+ # p_content = self.make_phrase(tokens, joined=True)
+ # p_title = self.make_phrase(tokens, 'title')
+ # p_author = self.make_phrase(tokens, 'author')
+
+ # phrase_level.add(BooleanClause(p_content, Should))
+ # phrase_level.add(BooleanClause(p_title, Should))
+ # phrase_level.add(BooleanClause(p_author, Should))
+
+ # kw_level = BooleanQuery()
+
+ # kw_level.add(self.make_term_query(tokens, 'author'), Should)
+ # j_themes = self.make_term_query(tokens, 'themes', joined=True)
+ # kw_level.add(j_themes, Should)
+ # kw_level.add(self.make_term_query(tokens, 'tags'), Should)
+ # j_con = self.make_term_query(tokens, joined=True)
+ # kw_level.add(j_con, Should)
+
+ # top_level.add(BooleanClause(phrase_level, Should))
+ # top_level.add(BooleanClause(kw_level, Should))
+
+ return None
+
+
+ def do_search(self, query, max_results=50, collector=None):
+ tops = self.searcher.search(query, max_results)
+ #tops = self.searcher.search(p_content, max_results)
+
+ bks = []
+ for found in tops.scoreDocs:
+ doc = self.searcher.doc(found.doc)
+ b = catalogue.models.Book.objects.get(id=doc.get("book_id"))
+ bks.append(b)
+ print "%s (%d) -> %f" % (b, b.id, found.score)
fnp / wolnelektury.git / blobdiff