X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/9362aac47564f9b03a3fcad625b312e85dfdf6db..3fc3ab8693fc63c350e26484e876a8092f6b5d90:/apps/search/index.py diff --git a/apps/search/index.py b/apps/search/index.py index 1c79ea2a8..5ebae2c42 100644 --- a/apps/search/index.py +++ b/apps/search/index.py @@ -1,4 +1,5 @@ # -*- coding: utf-8 -*- + from django.conf import settings from lucene import SimpleFSDirectory, IndexWriter, File, Field, \ NumericField, Version, Document, JavaError, IndexSearcher, \ @@ -8,8 +9,10 @@ from lucene import SimpleFSDirectory, IndexWriter, File, Field, \ BlockJoinQuery, BlockJoinCollector, TermsFilter, \ HashSet, BooleanClause, Term, CharTermAttribute, \ PhraseQuery, StringReader, TermQuery, BlockJoinQuery, \ - Sort, Integer + Sort, Integer, \ + initVM, CLASSPATH # KeywordAnalyzer +JVM = initVM(CLASSPATH) import sys import os import errno @@ -85,7 +88,6 @@ class Index(IndexStore): if overwrite: self.remove_book(book) - doc = self.extract_metadata(book) parts = self.extract_content(book) block = ArrayList().of_(Document) @@ -262,14 +264,14 @@ class ReusableIndex(Index): self.index = ReusableIndex.index else: print("opening index") - ReusableIndex.pool = ThreadPool(threads) + 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(Index.index_book, (self,)+ args, kw) + job = ReusableIndex.pool.apply_async(Index.index_book, (self,) + args, kw) ReusableIndex.pool_jobs.append(job) @staticmethod @@ -277,7 +279,9 @@ class ReusableIndex(Index): if ReusableIndex.index is not None: print("closing index") for job in ReusableIndex.pool_jobs: - job.wait() + job.get() + sys.stdout.write('.') + sys.stdout.flush() ReusableIndex.pool.close() ReusableIndex.index.optimize()