X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/77c65b39b5a28706889b2468c266c8e085a42e3c..9771b4edcbb556927312f624de6263395f82e0ba:/apps/search/index.py diff --git a/apps/search/index.py b/apps/search/index.py index 46fac4832..10069e4e4 100644 --- a/apps/search/index.py +++ b/apps/search/index.py @@ -53,8 +53,9 @@ class WLAnalyzer(PerFieldAnalyzerWrapper): self.addAnalyzer("source_url", keyword) self.addAnalyzer("source_name", simple) self.addAnalyzer("publisher", simple) - self.addAnalyzer("author", simple) + self.addAnalyzer("authors", simple) self.addAnalyzer("is_book", keyword) + # shouldn't the title have two forms? _pl and simple? self.addAnalyzer("themes", simple) self.addAnalyzer("themes_pl", polish) @@ -62,6 +63,8 @@ class WLAnalyzer(PerFieldAnalyzerWrapper): self.addAnalyzer("tag_name", simple) self.addAnalyzer("tag_name_pl", polish) + self.addAnalyzer("translators", simple) + self.addAnalyzer("KEYWORD", keyword) self.addAnalyzer("SIMPLE", simple) self.addAnalyzer("POLISH", polish) @@ -117,13 +120,11 @@ class Snippets(object): self.file.write(txt) pos = (self.position, l) self.position += l - print "Snip<%s>%s" %(pos, txt) return pos def get(self, pos): self.file.seek(pos[0], 0) txt = self.file.read(pos[1]).decode('utf-8') - print "got from snippets %d bytes from %s:" % (len(txt), pos) return txt def close(self): @@ -173,12 +174,12 @@ class Index(IndexStore): q = NumericRangeQuery.newIntRange("book_id", book.id, book.id, True, True) self.index.deleteDocuments(q) - def index_book(self, book, overwrite=True): + def index_book(self, book, book_info=None, overwrite=True): if overwrite: self.remove_book(book) book_doc = self.create_book_doc(book) - meta_fields = self.extract_metadata(book) + meta_fields = self.extract_metadata(book, book_info) for f in meta_fields.values(): if isinstance(f, list) or isinstance(f, tuple): for elem in f: @@ -189,7 +190,7 @@ class Index(IndexStore): self.index.addDocument(book_doc) del book_doc - self.index_content(book, book_fields=[meta_fields['title'], meta_fields['author']]) + self.index_content(book, book_fields=[meta_fields['title'], meta_fields['authors']]) master_tags = [ 'opowiadanie', @@ -212,11 +213,11 @@ class Index(IndexStore): doc.add(NumericField("parent_id", Field.Store.YES, True).setIntValue(book.parent.id)) return doc - def extract_metadata(self, book): + def extract_metadata(self, book, book_info=None): fields = {} - book_info = dcparser.parse(book.xml_file) - print("extract metadata for book %s id=%d, thread%d" % (book.slug, book.id, current_thread().ident)) + if book_info is None: + book_info = dcparser.parse(open(book.xml_file.path)) fields['slug'] = Field("slug", book.slug, Field.Store.NO, Field.Index.ANALYZED_NO_NORMS) fields['tags'] = self.add_gaps([Field("tags", t.name, Field.Store.NO, Field.Index.ANALYZED) for t in book.tags], 'tags') @@ -248,6 +249,7 @@ class Index(IndexStore): dt = getattr(book_info, field.name) fields[field.name] = Field(field.name, "%04d%02d%02d" %\ (dt.year, dt.month, dt.day), Field.Store.NO, Field.Index.NOT_ANALYZED) + return fields def get_master(self, root): @@ -262,7 +264,7 @@ class Index(IndexStore): return reduce(lambda a, b: a + b, zip(fields, gap()))[0:-1] def index_content(self, book, book_fields=[]): - wld = WLDocument.from_file(book.xml_file.path) + wld = WLDocument.from_file(book.xml_file.path, parse_dublincore=False) root = wld.edoc.getroot() master = self.get_master(root) @@ -317,6 +319,13 @@ class Index(IndexStore): return doc + def give_me_utf8(s): + if isinstance(s, unicode): + return s.encode('utf-8') + else: + return s + + fragments = {} snippets = Snippets(book.id).open('w') try: @@ -340,7 +349,7 @@ class Index(IndexStore): elif start is not None and start.tag == 'motyw': fid = start.attrib['id'][1:] if start.text is not None: - fragments[fid]['themes'] += map(unicode.strip, start.text.split(',')) + fragments[fid]['themes'] += map(str.strip, map(give_me_utf8, start.text.split(','))) fragments[fid]['content'].append(start.tail) elif start is not None and start.tag == 'end': fid = start.attrib['id'][1:] @@ -403,35 +412,23 @@ class ReusableIndex(Index): if you cannot rely on atexit, use ReusableIndex.close_reusable() yourself. """ index = None - pool = None - pool_jobs = None def open(self, analyzer=None, threads=4): if ReusableIndex.index is not None: self.index = ReusableIndex.index else: print("opening index") - ReusableIndex.pool = ThreadPool(threads, initializer=lambda: JVM.attachCurrentThread() ) - ReusableIndex.pool_jobs = [] Index.open(self, analyzer) ReusableIndex.index = self.index atexit.register(ReusableIndex.close_reusable) - def index_book(self, *args, **kw): - job = ReusableIndex.pool.apply_async(log_exception_wrapper(Index.index_book), (self,) + args, kw) - ReusableIndex.pool_jobs.append(job) + # def index_book(self, *args, **kw): + # job = ReusableIndex.pool.apply_async(log_exception_wrapper(Index.index_book), (self,) + args, kw) + # ReusableIndex.pool_jobs.append(job) @staticmethod def close_reusable(): if ReusableIndex.index is not None: - print("wait for indexing to finish") - for job in ReusableIndex.pool_jobs: - job.get() - sys.stdout.write('.') - sys.stdout.flush() - print("done.") - ReusableIndex.pool.close() - ReusableIndex.index.optimize() ReusableIndex.index.close() ReusableIndex.index = None @@ -489,6 +486,7 @@ class Search(IndexStore): bks.append(catalogue.models.Book.objects.get(id=doc.get("book_id"))) return (bks, tops.totalHits) + def search(self, query, max_results=50): query = self.query(query) query = self.wrapjoins(query, ["content", "themes"]) @@ -618,10 +616,10 @@ class Hint(object): self.search = search self.book_tags = {} self.part_tags = [] - self._book = None + self._books = [] - def book(self, book): - self._book = book + def books(self, *books): + self._books = books def tags(self, tags): for t in tags: @@ -655,20 +653,26 @@ class Hint(object): fs = [] if self.part_tags: fs.append(self.tag_filter(self.part_tags, field='themes')) - if self._book is not None: - fs.append(NumericRangeFilter.newIntRange('book_id', self._book.id, self._book.id, True, True)) + + 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._book is None + 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: + if field == 'authors' and 'author' in self.book_tags: continue - if field == 'title' and self._book is not None: + if field == 'title' and self._books != []: continue if (field == 'themes' or field == 'themes_pl') and self.part_tags: continue @@ -741,12 +745,12 @@ class MultiSearch(Search): q.add(BooleanClause(term, modal)) return q - def content_query(self, query): - return BlockJoinQuery(query, self.parent_filter, - BlockJoinQuery.ScoreMode.Total) + # def content_query(self, query): + # return BlockJoinQuery(query, self.parent_filter, + # BlockJoinQuery.ScoreMode.Total) def search_perfect_book(self, searched, max_results=20, fuzzy=False, hint=None): - fields_to_search = ['author', 'title'] + fields_to_search = ['authors', 'title'] only_in = None if hint: if not hint.should_search_for_book(): @@ -808,10 +812,10 @@ class MultiSearch(Search): # query themes/content x author/title/tags q = BooleanQuery() - in_meta = BooleanQuery() + # in_meta = BooleanQuery() in_content = BooleanQuery() - for fld in ['themes', 'content', 'tags', 'author', 'title']: + for fld in ['themes', 'content', 'tags', 'authors', 'title']: in_content.add(BooleanClause(self.make_term_query(tokens, field=fld, fuzzy=False), BooleanClause.Occur.SHOULD)) topDocs = self.searcher.search(q, only_in, max_results) @@ -819,15 +823,14 @@ class MultiSearch(Search): books.append(SearchResult(self.searcher, found)) return books - def multisearch(self, query, max_results=50): """ Search strategy: - (phrase) OR -> content -> title - -> author - - (keywords) -> author + -> authors + - (keywords) -> authors -> motyw -> tags -> content @@ -893,7 +896,6 @@ class MultiSearch(Search): # highlighter.getBestTextFragments(tokenStream, text, False, 10) # import pdb; pdb.set_trace() snip = highlighter.getBestFragments(tokenStream, text, 3, "...") - print('snips: %s' % snip) return [snip] @@ -926,6 +928,14 @@ class MultiSearch(Search): return tags + def search_books(self, query, filter=None, max_results=10): + bks = [] + tops = self.searcher.search(query, filter, max_results) + for found in tops.scoreDocs: + doc = self.searcher.doc(found.doc) + bks.append(catalogue.models.Book.objects.get(id=doc.get("book_id"))) + return bks + def create_prefix_phrase(self, toks, field): q = MultiPhraseQuery() for i in range(len(toks)):