+class SearchResult(object):
+ def __init__(self, searcher, scoreDocs, score=None, how_found=None, snippets=None):
+ self.snippets = []
+
+ if score:
+ self.score = score
+ else:
+ self.score = scoreDocs.score
+
+ self.hits = []
+
+ stored = searcher.doc(scoreDocs.doc)
+ self.book_id = int(stored.get("book_id"))
+
+ header_type = stored.get("header_type")
+ if not header_type:
+ return
+
+ sec = (header_type, int(stored.get("header_index")))
+ header_span = stored.get('header_span')
+ header_span = header_span is not None and int(header_span) or 1
+
+ fragment = stored.get("fragment_anchor")
+
+ hit = (sec + (header_span,), fragment, scoreDocs.score, {'how_found': how_found, 'snippets': snippets})
+
+ self.hits.append(hit)
+
+ 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.hits += other.hits
+ if other.score > self.score:
+ self.score = other.score
+ return self
+
+ def get_book(self):
+ return catalogue.models.Book.objects.get(id=self.book_id)
+
+ book = property(get_book)
+
+ def process_hits(self):
+ frags = filter(lambda r: r[1] is not None, self.hits)
+ sect = filter(lambda r: r[1] is None, self.hits)
+ sect = filter(lambda s: 0 == len(filter(
+ lambda f: s[0][1] >= f[0][1] and s[0][1] < f[0][1] + f[0][2],
+ frags)), sect)
+
+ hits = []
+
+ for s in sect:
+ m = {'score': s[2],
+ 'header_index': s[0][1]
+ }
+ m.update(s[3])
+ hits.append(m)
+
+ for f in frags:
+ frag = catalogue.models.Fragment.objects.get(anchor=f[1])
+ m = {'score': f[2],
+ 'fragment': frag,
+ 'themes': frag.tags.filter(category='theme')
+ }
+ m.update(f[3])
+ hits.append(m)
+
+ hits.sort(lambda a, b: cmp(a['score'], b['score']), reverse=True)
+
+ print("--- %s" % hits)
+
+ return hits
+
+ 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 Hint(object):
+ def __init__(self, search):
+ self.search = search
+ self.book_tags = {}
+ self.part_tags = []
+ self._books = []
+
+ def books(self, *books):
+ self._books = books
+
+ def tags(self, tags):
+ for t in tags:
+ if t.category in ['author', 'title', 'epoch', 'genre', 'kind']:
+ lst = self.book_tags.get(t.category, [])
+ lst.append(t)
+ self.book_tags[t.category] = lst
+ if t.category in ['theme']:
+ self.part_tags.append(t)
+
+ def tag_filter(self, tags, field='tags'):
+ q = BooleanQuery()
+
+ for tag in tags:
+ toks = self.search.get_tokens(tag.name, field=field)
+ tag_phrase = PhraseQuery()
+ for tok in toks:
+ tag_phrase.add(Term(field, tok))
+ q.add(BooleanClause(tag_phrase, BooleanClause.Occur.MUST))
+
+ return QueryWrapperFilter(q)
+
+ def book_filter(self):
+ tags = reduce(lambda a, b: a + b, self.book_tags.values(), [])
+ if tags:
+ return self.tag_filter(tags)
+ else:
+ return None
+
+ def part_filter(self):
+ fs = []
+ if self.part_tags:
+ fs.append(self.tag_filter(self.part_tags, field='themes'))
+
+ if self._books != []:
+ bf = BooleanFilter()
+ for b in self._books:
+ id_filter = NumericRangeFilter.newIntRange('book_id', b.id, b.id, True, True)
+ bf.add(FilterClause(id_filter, BooleanClause.Occur.SHOULD))
+ fs.append(bf)
+
+ return MultiSearch.chain_filters(fs)
+
+ def should_search_for_book(self):
+ return self._books == []
+
+ def just_search_in(self, all):
+ """Holds logic to figure out which indexes should be search, when we have some hinst already"""
+ some = []
+ for field in all:
+ if field == 'author' and 'author' in self.book_tags:
+ continue
+ if field == 'title' and self._books != []:
+ continue
+ if (field == 'themes' or field == 'themes_pl') and self.part_tags:
+ continue
+ some.append(field)
+ return some
+
+
fnp / wolnelektury.git / blobdiff