X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/f931a5a3c86e677475ae7f3fc0074b4940bee85c..08f83172c482208d35cd75c704cebd1d70a73af2:/apps/search/views.py diff --git a/apps/search/views.py b/apps/search/views.py index 4db4bb8fb..ec8275b91 100644 --- a/apps/search/views.py +++ b/apps/search/views.py @@ -10,15 +10,17 @@ from django.utils.translation import ugettext as _ from catalogue.utils import split_tags from catalogue.models import Book, Tag, Fragment -from catalogue.fields import dumps +from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook from catalogue.views import JSONResponse from search import Search, JVM, SearchResult from lucene import StringReader from suggest.forms import PublishingSuggestForm +from time import sleep import re -import enchant +#import enchant +import json -dictionary = enchant.Dict('pl_PL') +#dictionary = enchant.Dict('en_US') def match_word_re(word): @@ -35,13 +37,14 @@ def did_you_mean(query, tokens): if len(authors) > 0: continue - 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 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 @@ -51,7 +54,22 @@ def did_you_mean(query, tokens): return query -search = Search() + +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', '') @@ -59,6 +77,7 @@ def hint(request): return JSONResponse([]) JVM.attachCurrentThread() + search = get_search() hint = search.hint() try: tags = request.GET.get('tags', '') @@ -74,22 +93,40 @@ def hint(request): 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_'):] + c = c[len('pd_'):] return _(c) - 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': _('book'), - 'id': b.id, - 'url': b.get_absolute_url()} - for b in books]) + 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): @@ -100,109 +137,118 @@ def main(request): query = None fuzzy = False #0.8 - if 'q' in request.GET: - # tags = request.GET.get('tags', '') - query = request.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)) - - # 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")) - print "dym? %s" % repr(suggestion).encode('utf-8') - - results = author_results + title_results + text_phrase + everywhere - 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)) - - print "TAGS: %s" % tags - return render_to_response('catalogue/search_multiple_hits.html', + 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: + 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, - 'results': { 'author': author_results, - 'title': title_results, - 'content': text_phrase, - 'other': everywhere}, + "form": form, 'did_you_mean': suggestion}, 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))