def next(self): start = time.time() searchers = [] postingLists = [] try: # get a searcher and posting list for each index for index in self._indices: # we create a copy of the original query, which can possibly be optimized # with index-specific knowledge. query = copy.deepcopy(self._query) # get the posting list to iterate through searcher = yield index.newSearcher() if not ISearcher.providedBy(searcher): raise TypeError("searcher does not implement ISearcher") query = yield query.optimizeMatcher(searcher) logger.debug("optimized query for index '%s': %s" % (index.name, str(query))) # if the query optimized out entirely, then skip to the next index if query == None: yield searcher.close() continue postingList = yield query.iterMatches(searcher, self._startId, self._endId) if not IPostingList.providedBy(postingList): raise TypeError( "posting list does not implement IPostingList") searchers.append(searcher) postingLists.append(postingList) if len(postingLists) == 0: raise StopIteration() # loop forever until we reach the search limit, we exhaust all of our # posting lists, or we encounter an exception currPostings = [(None, None, None) for i in range(len(postingLists))] smallestList = 0 lastId = None compar = (lambda x, y: cmp(y, x)) if self._reverse else cmp while True: # if we have reached our limit if len(self.events) == self._limit: self.runtime = time.time() - start raise StopIteration() # check each child iter for the lowest evid for currList in range(len(postingLists)): if currList == None: smallestList = 0 # if the postingList is None, then its closed if postingLists[currList] == None: # FIXME: close the posting list and searcher continue # if current posting for this posting list was not consumed if currPostings[currList] != (None, None, None): continue # otherwise get the next posting from the posting list currPostings[currList] = yield postingLists[ currList].nextPosting() # if the next posting for this posting list is (None,None,None), # then we are done with this posting list if currPostings[currList] == (None, None, None): postingList = postingLists[currList] yield postingList.close() postingLists[currList] = None continue # if the evid equals the last evid returned, then ignore it if lastId != None and currPostings[currList][0] == lastId: currPostings[currList] = (None, None, None) continue # we don't compare the first evid with itself if currList == 0: continue # if the evid is the current smallest evid, then remember it if (currPostings[smallestList] == (None, None, None) or compar(currPostings[currList][0], currPostings[smallestList][0]) < 0): smallestList = currList # get the next posting currList = None evid, _, store = currPostings[smallestList] # stop iterating if there are no more results if evid == None: self.runtime = time.time() - start raise StopIteration() # remember the last evid lastId = evid # forget the evid so we don't return it again currPostings[smallestList] = (None, None, None) # retrieve the event if not IEventStore.providedBy(store): raise TypeError("store does not implement IEventStore") event = yield store.getEvent(evid) defaultfield, defaultvalue, fields = event if defaultfield not in self.fields: self.fields.append(defaultfield) # keep a record of all field names found in the results for fieldname in fields.keys(): if fieldname not in self.fields: self.fields.append(fieldname) # filter out unwanted fields if self._fields != None: fields = dict([(k, v) for k, v in fields.items() if k in self._fields]) self.events.append(((evid.ts, evid.offset), defaultfield, defaultvalue, fields)) logger.trace("added event %s to results" % evid) finally: for postingList in postingLists: if postingList != None: yield postingList.close() for searcher in searchers: yield searcher.close()
def next(self): start = time.time() searchers = [] postingLists = [] try: # get a searcher and posting list for each index for index in self._indices: # we create a copy of the original query, which can possibly be optimized # with index-specific knowledge. query = copy.deepcopy(self._query) # get the posting list to iterate through searcher = yield index.newSearcher() if not ISearcher.providedBy(searcher): raise TypeError("searcher does not implement ISearcher") query = yield query.optimizeMatcher(searcher) logger.debug("optimized query for index '%s': %s" % (index.name,str(query))) # if the query optimized out entirely, then skip to the next index if query == None: yield searcher.close() continue postingList = yield query.iterMatches(searcher, self._startId, self._endId) if not IPostingList.providedBy(postingList): raise TypeError("posting list does not implement IPostingList") searchers.append(searcher) postingLists.append(postingList) if len(postingLists) == 0: raise StopIteration() # loop forever until we reach the search limit, we exhaust all of our # posting lists, or we encounter an exception currPostings = [(None,None,None) for i in range(len(postingLists))] smallestList = 0 lastId = None compar = (lambda x,y: cmp(y,x)) if self._reverse else cmp while True: # if we have reached our limit if len(self.events) == self._limit: self.runtime = time.time() - start raise StopIteration() # check each child iter for the lowest evid for currList in range(len(postingLists)): if currList == None: smallestList = 0 # if the postingList is None, then its closed if postingLists[currList] == None: # FIXME: close the posting list and searcher continue # if current posting for this posting list was not consumed if currPostings[currList] != (None,None,None): continue # otherwise get the next posting from the posting list currPostings[currList] = yield postingLists[currList].nextPosting() # if the next posting for this posting list is (None,None,None), # then we are done with this posting list if currPostings[currList] == (None,None,None): postingList = postingLists[currList] yield postingList.close() postingLists[currList] = None continue # if the evid equals the last evid returned, then ignore it if lastId != None and currPostings[currList][0] == lastId: currPostings[currList] = (None,None,None) continue # we don't compare the first evid with itself if currList == 0: continue # if the evid is the current smallest evid, then remember it if (currPostings[smallestList] == (None,None,None) or compar(currPostings[currList][0], currPostings[smallestList][0]) < 0): smallestList = currList # get the next posting currList = None evid,_,store = currPostings[smallestList] # stop iterating if there are no more results if evid == None: self.runtime = time.time() - start raise StopIteration() # remember the last evid lastId = evid # forget the evid so we don't return it again currPostings[smallestList] = (None,None,None) # retrieve the event if not IEventStore.providedBy(store): raise TypeError("store does not implement IEventStore") event = yield store.getEvent(evid) defaultfield, defaultvalue, fields = event if defaultfield not in self.fields: self.fields.append(defaultfield) # keep a record of all field names found in the results for fieldname in fields.keys(): if fieldname not in self.fields: self.fields.append(fieldname) # filter out unwanted fields if self._fields != None: fields = dict([(k,v) for k,v in fields.items() if k in self._fields]) self.events.append(((evid.ts,evid.offset), defaultfield, defaultvalue, fields)) logger.trace("added event %s to results" % evid) finally: for postingList in postingLists: if postingList != None: yield postingList.close() for searcher in searchers: yield searcher.close()
def next(self): try: nlists = len(self._postingLists) if nlists == 0: raise StopIteration() # check each child iter for the lowest evid curr = 0 for i in range(nlists): # if the postingList is None, then its closed if self._postingLists[i] == None: continue # if current posting for this posting list was consumed, then # get the next posting from the posting list if self._currPostings[i] == (None, None, None): self._currPostings[i] = self._postingLists[i].nextPosting() # if the next posting for this posting list is (None,None,None), # then we are done with this posting list if self._currPostings[i] == (None, None, None): self._postingLists[i] = None continue # if the evid equals the last evid returned, then ignore it if self._lastId != None and self._currPostings[i][0] == self._lastId: self._currPostings[i] = (None, None, None) continue # we don't compare the first evid with itself if i == 0: continue # if the evid is the current smallest evid, then remember it if ( self._currPostings[curr] == (None, None, None) or self._cmp(self._currPostings[i][0], self._currPostings[curr][0]) < 0 ): curr = i # stop iterating if there are no more results if self._currPostings[curr] == (None, None, None): raise StopIteration() evid, tvalue, store = self._currPostings[curr] # remember the last evid self._lastId = evid # forget the evid so we don't return it again self._currPostings[curr] = (None, None, None) # retrieve the event if not IEventStore.providedBy(store): raise TypeError("store does not implement IEventStore") defaultfield, defaultvalue, fields = store.getEvent(evid) if defaultfield not in self.fields: self.fields.append(defaultfield) # keep a record of all field names found in the results. for fieldname in fields.keys(): if fieldname not in self.fields: self.fields.append(fieldname) # filter out unwanted fields if self._fields != None: fields = dict([(k, v) for k, v in fields.items() if k in self._fields]) self.events.append(((evid.ts, evid.offset), defaultfield, defaultvalue, fields)) logger.trace("added event %s to resultset" % evid) # if we have reached our limit self._count += 1 if self._count == self._limit: raise StopIteration() except: self.close() self.runtime = time.time() - self._start logger.trace("retrieved %i events in %f seconds" % (len(self.events), self.runtime)) raise