librarian update
[wolnelektury.git] / src / search / views.py
1 # -*- coding: utf-8 -*-
2 # This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
3 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
4 #
5 from django.conf import settings
6 from django.shortcuts import render_to_response
7 from django.template import RequestContext
8 from django.views.decorators import cache
9 from django.http import HttpResponse, JsonResponse
10 from django.utils.translation import ugettext as _
11
12 from catalogue.utils import split_tags
13 from catalogue.models import Book
14 from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook
15 from search.index import Search, SearchResult
16 from suggest.forms import PublishingSuggestForm
17 import re
18 import json
19
20
21 def match_word_re(word):
22     if 'sqlite' in settings.DATABASES['default']['ENGINE']:
23         return r"\b%s\b" % word
24     elif 'mysql' in settings.DATABASES['default']['ENGINE']:
25         return "[[:<:]]%s[[:>:]]" % word
26
27
28 query_syntax_chars = re.compile(r"[\\/*:(){}]")
29
30
31 def remove_query_syntax_chars(query, replace=' '):
32     return query_syntax_chars.sub(' ', query)
33
34
35 def did_you_mean(query, tokens):
36     return query
37     # change = {}
38     # for t in tokens:
39     #     authors = Tag.objects.filter(category='author', name__iregex=match_word_re(t))
40     #     if len(authors) > 0:
41     #         continue
42
43     #     if False:
44     #         if not dictionary.check(t):
45     #             try:
46     #                 change_to = dictionary.suggest(t)[0].lower()
47     #                 if change_to != t.lower():
48     #                     change[t] = change_to
49     #             except IndexError:
50     #                 pass
51
52     # if change == {}:
53     #     return None
54
55     # for frm, to in change.items():
56     #     query = query.replace(frm, to)
57
58     # return query
59
60
61 @cache.never_cache
62 def hint(request):
63     prefix = request.GET.get('term', '')
64     if len(prefix) < 2:
65         return JsonResponse([], safe=False)
66
67     prefix = remove_query_syntax_chars(prefix)
68
69     search = Search()
70     # tagi beda ograniczac tutaj
71     # ale tagi moga byc na ksiazce i na fragmentach
72     # jezeli tagi dot tylko ksiazki, to wazne zeby te nowe byly w tej samej ksiazce
73     # jesli zas dotycza themes, to wazne, zeby byly w tym samym fragmencie.
74
75     def is_dupe(tag):
76         if isinstance(tag, PDCounterAuthor):
77             if filter(lambda t: t.slug == tag.slug and t != tag, tags):
78                 return True
79         elif isinstance(tag, PDCounterBook):
80             if filter(lambda b: b.slug == tag.slug, tags):
81                 return True
82         return False
83
84     def category_name(c):
85         if c.startswith('pd_'):
86             c = c[len('pd_'):]
87         return _(c)
88
89     try:
90         limit = int(request.GET.get('max', ''))
91     except ValueError:
92         limit = -1
93     else:
94         if limit < 1:
95             limit = -1
96
97     data = []
98
99     tags = search.hint_tags(prefix, pdcounter=True)
100     tags = filter(lambda t: not is_dupe(t), tags)
101     for t in tags:
102         if not limit:
103             break
104         limit -= 1
105         data.append({
106             'label': t.name,
107             'category': category_name(t.category),
108             'id': t.id,
109             'url': t.get_absolute_url()
110             })
111     if limit:
112         books = search.hint_books(prefix)
113         for b in books:
114             if not limit:
115                 break
116             limit -= 1
117             data.append({
118                 'label': b.title,
119                 'category': _('book'),
120                 'id': b.id,
121                 'url': b.get_absolute_url()
122                 })
123
124     callback = request.GET.get('callback', None)
125     if callback:
126         return HttpResponse("%s(%s);" % (callback, json.dumps(data)),
127                             content_type="application/json; charset=utf-8")
128     else:
129         return JsonResponse(data, safe=False)
130
131
132 @cache.never_cache
133 def main(request):
134     query = request.GET.get('q', '')
135
136     if len(query) < 2:
137         return render_to_response(
138             'catalogue/search_too_short.html', {'prefix': query},
139             context_instance=RequestContext(request))
140     elif len(query) > 256:
141         return render_to_response(
142             'catalogue/search_too_long.html', {'prefix': query}, context_instance=RequestContext(request))
143
144     query = remove_query_syntax_chars(query)
145     
146     search = Search()
147
148     theme_terms = search.index.analyze(text=query, field="themes_pl") \
149         + search.index.analyze(text=query, field="themes")
150
151     # change hints
152     tags = search.hint_tags(query, pdcounter=True, prefix=False)
153     tags = split_tags(tags)
154
155     author_results = search.search_phrase(query, 'authors', book=True)
156     translator_results = search.search_phrase(query, 'translators', book=True)
157
158     title_results = search.search_phrase(query, 'title', book=True)
159
160     # Boost main author/title results with mixed search, and save some of its results for end of list.
161     # boost author, title results
162     author_title_mixed = search.search_some(query, ['authors', 'translators', 'title', 'tags'], query_terms=theme_terms)
163     author_title_rest = []
164
165     for b in author_title_mixed:
166         also_in_mixed = filter(lambda ba: ba.book_id == b.book_id, author_results + translator_results + title_results)
167         for b2 in also_in_mixed:
168             b2.boost *= 1.1
169         if also_in_mixed is []:
170             author_title_rest.append(b)
171
172     # Do a phrase search but a term search as well - this can give us better snippets then search_everywhere,
173     # Because the query is using only one field.
174     text_phrase = SearchResult.aggregate(
175         search.search_phrase(query, 'text', snippets=True, book=False),
176         search.search_some(query, ['text'], snippets=True, book=False, query_terms=theme_terms))
177
178     everywhere = search.search_everywhere(query, query_terms=theme_terms)
179
180     def already_found(results):
181         def f(e):
182             for r in results:
183                 if e.book_id == r.book_id:
184                     e.boost = 0.9
185                     results.append(e)
186                     return True
187             return False
188         return f
189     f = already_found(author_results + translator_results + title_results + text_phrase)
190     everywhere = filter(lambda x: not f(x), everywhere)
191
192     author_results = SearchResult.aggregate(author_results)
193     translator_results = SearchResult.aggregate(translator_results)
194     title_results = SearchResult.aggregate(title_results)
195
196     everywhere = SearchResult.aggregate(everywhere, author_title_rest)
197
198     for field, res in [('authors', author_results),
199                        ('translators', translator_results),
200                        ('title', title_results),
201                        ('text', text_phrase),
202                        ('text', everywhere)]:
203         res.sort(reverse=True)
204         for r in res:
205             search.get_snippets(r, query, field, 3)
206
207     suggestion = u''
208
209     def ensure_exists(r):
210         try:
211             return r.book
212         except Book.DoesNotExist:
213             return False
214
215     author_results = filter(ensure_exists, author_results)
216     translator_results = filter(ensure_exists, translator_results)
217     title_results = filter(ensure_exists, title_results)
218     text_phrase = filter(ensure_exists, text_phrase)
219     everywhere = filter(ensure_exists, everywhere)
220
221     results = author_results + translator_results + title_results + text_phrase + everywhere
222     # ensure books do exists & sort them
223     for res in (author_results, translator_results, title_results, text_phrase, everywhere):
224         res.sort(reverse=True)
225
226     # We don't want to redirect to book text, but rather display result page even with one result.
227     # if len(results) == 1:
228     #     fragment_hits = filter(lambda h: 'fragment' in h, results[0].hits)
229     #     if len(fragment_hits) == 1:
230     #         #anchor = fragment_hits[0]['fragment']
231     #         #frag = Fragment.objects.get(anchor=anchor)
232     #         return HttpResponseRedirect(fragment_hits[0]['fragment'].get_absolute_url())
233     #     return HttpResponseRedirect(results[0].book.get_absolute_url())
234     if len(results) == 0:
235         form = PublishingSuggestForm(initial={"books": query + ", "})
236         return render_to_response(
237             'catalogue/search_no_hits.html',
238             {
239                 'tags': tags,
240                 'prefix': query,
241                 'form': form,
242                 'did_you_mean': suggestion
243             },
244             context_instance=RequestContext(request))
245
246     return render_to_response(
247         'catalogue/search_multiple_hits.html',
248         {
249             'tags': tags,
250             'prefix': query,
251             'results': {
252                 'author': author_results,
253                 'translator': translator_results,
254                 'title': title_results,
255                 'content': text_phrase,
256                 'other': everywhere
257             },
258             'did_you_mean': suggestion
259         },
260         context_instance=RequestContext(request))