X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/b80f646690239d9a90d708d57780f328328b26d5..162d876b6e05f4bacf2473148a989a4e2b93a2ed:/apps/search/views.py diff --git a/apps/search/views.py b/apps/search/views.py index 438989c01..ec8275b91 100644 --- a/apps/search/views.py +++ b/apps/search/views.py @@ -1,35 +1,50 @@ # -*- coding: utf-8 -*- +from django.conf import settings from django.shortcuts import render_to_response, get_object_or_404 from django.template import RequestContext from django.contrib.auth.decorators import login_required from django.views.decorators import cache from django.http import HttpResponse, HttpResponseRedirect, Http404, HttpResponsePermanentRedirect +from django.utils.translation import ugettext as _ -from catalogue.utils import get_random_hash -from catalogue.models import Book, Tag, Fragment, TAG_CATEGORIES -from catalogue.fields import dumps +from catalogue.utils import split_tags +from catalogue.models import Book, Tag, Fragment +from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook from catalogue.views import JSONResponse -from catalogue import forms -from search import MultiSearch, JVM, SearchResult +from search import Search, JVM, SearchResult from lucene import StringReader from suggest.forms import PublishingSuggestForm +from time import sleep +import re +#import enchant +import json -import enchant +#dictionary = enchant.Dict('en_US') -dictionary = enchant.Dict('pl_PL') + +def match_word_re(word): + if 'sqlite' in settings.DATABASES['default']['ENGINE']: + return r"\b%s\b" % word + elif 'mysql' in settings.DATABASES['default']['ENGINE']: + return "[[:<:]]%s[[:>:]]" % word def did_you_mean(query, tokens): change = {} - # sprawdzić, czy słowo nie jest aby autorem - proste szukanie termu w author! for t in tokens: - print("%s ok? %s, sug: %s" %(t, dictionary.check(t), dictionary.suggest(t))) - if not dictionary.check(t): - try: - change[t] = dictionary.suggest(t)[0] - except IndexError: - pass + authors = Tag.objects.filter(category='author', name__iregex=match_word_re(t)) + if len(authors) > 0: + continue + + if False: + if not dictionary.check(t): + try: + change_to = dictionary.suggest(t)[0].lower() + if change_to != t.lower(): + change[t] = change_to + except IndexError: + pass if change == {}: return None @@ -40,21 +55,30 @@ def did_you_mean(query, tokens): return query -def category_name(category): - try: - return filter(lambda c: c[0] == category, TAG_CATEGORIES)[0][1].encode('utf-8') - except IndexError: - raise KeyError("No category %s" % category) +JVM.attachCurrentThread() +_search = None + + +def get_search(): + global _search + + while _search is False: + sleep(1) + + if _search is None: + _search = False + _search = Search() + return _search def hint(request): prefix = request.GET.get('term', '') if len(prefix) < 2: - return JSONResponse(dumps(None)) + return JSONResponse([]) JVM.attachCurrentThread() - s = MultiSearch() - hint = s.hint() + search = get_search() + hint = search.hint() try: tags = request.GET.get('tags', '') hint.tags(Tag.get_tag_list(tags)) @@ -66,99 +90,165 @@ def hint(request): # jezeli tagi dot tylko ksiazki, to wazne zeby te nowe byly w tej samej ksiazce # jesli zas dotycza themes, to wazne, zeby byly w tym samym fragmencie. - tags = s.hint_tags(prefix) - books = s.hint_books(prefix) - - # TODO DODAC TU HINTY - - return JSONResponse( - [{'label': t.name, - 'category': category_name(t.category), - 'id': t.id, - 'url': t.get_absolute_url()} - for t in tags] + \ - [{'label': b.title, - 'category': category_name('book'), - 'id': b.id, - 'url': b.get_absolute_url()} - for b in books]) - - -def foo(s, q, tag_list=None): - hint = s.hint() - try: - tag_list = Tag.get_tag_list(tag_list) - hint.tags(tag_list) - except: - tag_list = None - - q = StringReader(q) - return (q, hint) + tags = search.hint_tags(prefix, pdcounter=True) + books = search.hint_books(prefix) + + + def is_dupe(tag): + if isinstance(tag, PDCounterAuthor): + if filter(lambda t: t.slug == tag.slug and t != tag, tags): + return True + elif isinstance(tag, PDCounterBook): + if filter(lambda b: b.slug == tag.slug, tags): + return True + return False + + tags = filter(lambda t: not is_dupe(t), tags) + + def category_name(c): + if c.startswith('pd_'): + c = c[len('pd_'):] + return _(c) + + callback = request.GET.get('callback', None) + data = [{'label': t.name, + 'category': category_name(t.category), + 'id': t.id, + 'url': t.get_absolute_url()} + for t in tags] + \ + [{'label': b.title, + 'category': _('book'), + 'id': b.id, + 'url': b.get_absolute_url()} + for b in books] + if callback: + return HttpResponse("%s(%s);" % (callback, json.dumps(data)), + content_type="application/json; charset=utf-8") + else: + return JSONResponse(data) + def main(request): results = {} JVM.attachCurrentThread() # where to put this? - srch = MultiSearch() results = None query = None - fuzzy = False - - if 'q' in request.GET: - tags = request.GET.get('tags', '') - query = request.GET['q'] - book_id = request.get('book', None) - book = None - if book_id is not None: - book = get_object_or_404(Book, id=book_id) - - hint = srch.hint() + fuzzy = False #0.8 + + query = request.GET.get('q','') + # book_id = request.GET.get('book', None) + # book = None + # if book_id is not None: + # book = get_object_or_404(Book, id=book_id) + + # hint = search.hint() + # try: + # tag_list = Tag.get_tag_list(tags) + # except: + # tag_list = [] + + if len(query) < 2: + return render_to_response('catalogue/search_too_short.html', {'prefix': query}, + context_instance=RequestContext(request)) + + search = get_search() + # hint.tags(tag_list) + # if book: + # hint.books(book) + tags = search.hint_tags(query, pdcounter=True, prefix=False, fuzzy=fuzzy) + tags = split_tags(tags) + + toks = StringReader(query) + tokens_cache = {} + + author_results = search.search_phrase(toks, 'authors', fuzzy=fuzzy, tokens_cache=tokens_cache) + title_results = search.search_phrase(toks, 'title', fuzzy=fuzzy, tokens_cache=tokens_cache) + + # Boost main author/title results with mixed search, and save some of its results for end of list. + # boost author, title results + author_title_mixed = search.search_some(toks, ['authors', 'title', 'tags'], fuzzy=fuzzy, tokens_cache=tokens_cache) + author_title_rest = [] + for b in author_title_mixed: + bks = filter(lambda ba: ba.book_id == b.book_id, author_results + title_results) + for b2 in bks: + b2.boost *= 1.1 + if bks is []: + author_title_rest.append(b) + + # Do a phrase search but a term search as well - this can give us better snippets then search_everywhere, + # Because the query is using only one field. + text_phrase = SearchResult.aggregate( + search.search_phrase(toks, 'content', fuzzy=fuzzy, tokens_cache=tokens_cache, snippets=True, book=False, slop=4), + search.search_some(toks, ['content'], tokens_cache=tokens_cache, snippets=True, book=False)) + + everywhere = search.search_everywhere(toks, fuzzy=fuzzy, tokens_cache=tokens_cache) + + def already_found(results): + def f(e): + for r in results: + if e.book_id == r.book_id: + e.boost = 0.9 + results.append(e) + return True + return False + return f + f = already_found(author_results + title_results + text_phrase) + everywhere = filter(lambda x: not f(x), everywhere) + + author_results = SearchResult.aggregate(author_results) + title_results = SearchResult.aggregate(title_results) + + everywhere = SearchResult.aggregate(everywhere, author_title_rest) + + for res in [author_results, title_results, text_phrase, everywhere]: + res.sort(reverse=True) + for r in res: + for h in r.hits: + h['snippets'] = map(lambda s: + re.subn(r"(^[ \t\n]+|[ \t\n]+$)", u"", + re.subn(r"[ \t\n]*\n[ \t\n]*", u"\n", s)[0])[0], h['snippets']) + + suggestion = did_you_mean(query, search.get_tokens(toks, field="SIMPLE")) + + def ensure_exists(r): try: - tag_list = Tag.get_tag_list(tags) - except: - tag_list = [] - - if len(query) < 2: - return render_to_response('catalogue/search_too_short.html', {'tags': tag_list, 'prefix': query}, - context_instance=RequestContext(request)) - - hint.tags(tag_list) - hint.book(book) - - toks = StringReader(query) - fuzzy = 'fuzzy' in request.GET - if fuzzy: - fuzzy = 0.7 - - results = SearchResult.aggregate(srch.search_perfect_book(toks, fuzzy=fuzzy, hint=hint), - srch.search_perfect_parts(toks, fuzzy=fuzzy, hint=hint), - srch.search_everywhere(toks, fuzzy=fuzzy, hint=hint)) - results.sort(reverse=True) - - for r in results: - print r.hits - - if len(results) == 1: - if len(results[0].hits) == 0: - return HttpResponseRedirect(results[0].book.get_absolute_url()) - elif len(results[0].hits) == 1 and results[0].hits[0] is not None: - frag = Fragment.objects.get(anchor=results[0].hits[0]) - return HttpResponseRedirect(frag.get_absolute_url()) - elif len(results) == 0: - form = PublishingSuggestForm(initial={"books": query + ", "}) - return render_to_response('catalogue/search_no_hits.html', - {'tags': tag_list, 'prefix': query, "pubsuggest_form": form, - 'form': forms.SearchForm()}, - context_instance=RequestContext(request)) - - return render_to_response('catalogue/search_multiple_hits.html', - {'tags': tag_list, 'prefix': query, - 'results': results, 'from': forms.SearchForm()}, + return r.book + except Book.DoesNotExist: + return False + + author_results = filter(ensure_exists, author_results) + title_results = filter(ensure_exists, title_results) + text_phrase = filter(ensure_exists, text_phrase) + everywhere = filter(ensure_exists, everywhere) + + results = author_results + title_results + text_phrase + everywhere + # ensure books do exists & sort them + results.sort(reverse=True) + + if len(results) == 1: + fragment_hits = filter(lambda h: 'fragment' in h, results[0].hits) + if len(fragment_hits) == 1: + #anchor = fragment_hits[0]['fragment'] + #frag = Fragment.objects.get(anchor=anchor) + return HttpResponseRedirect(fragment_hits[0]['fragment'].get_absolute_url()) + return HttpResponseRedirect(results[0].book.get_absolute_url()) + elif len(results) == 0: + form = PublishingSuggestForm(initial={"books": query + ", "}) + return render_to_response('catalogue/search_no_hits.html', + {'tags': tags, + 'prefix': query, + "form": form, + 'did_you_mean': suggestion}, context_instance=RequestContext(request)) - # return render_to_response('newsearch/search.html', {'results': results, - # 'did_you_mean': (query is not None) and - # did_you_mean(query, srch.get_tokens(query, field='SIMPLE')), - # 'fuzzy': fuzzy}, - # context_instance=RequestContext(request)) + return render_to_response('catalogue/search_multiple_hits.html', + {'tags': tags, + 'prefix': query, + 'results': { 'author': author_results, + 'title': title_results, + 'content': text_phrase, + 'other': everywhere}, + 'did_you_mean': suggestion}, + context_instance=RequestContext(request))