dcd80b29fa6c8b314a9c3f6d7b382945b22b75f0
[wolnelektury.git] / apps / search / views.py
1 # -*- coding: utf-8 -*-
2
3 from django.conf import settings
4 from django.shortcuts import render_to_response, get_object_or_404
5 from django.template import RequestContext
6 from django.contrib.auth.decorators import login_required
7 from django.views.decorators import cache
8 from django.http import HttpResponse, HttpResponseRedirect, Http404, HttpResponsePermanentRedirect
9 from django.utils.translation import ugettext as _
10
11 from catalogue.utils import split_tags
12 from catalogue.models import Book, Tag, Fragment
13 from catalogue.fields import dumps
14 from catalogue.views import JSONResponse
15 from search import Search, JVM, SearchResult
16 from lucene import StringReader
17 from suggest.forms import PublishingSuggestForm
18 import re
19 import enchant
20
21 dictionary = enchant.Dict('pl_PL')
22
23
24 def match_word_re(word):
25     if 'sqlite' in settings.DATABASES['default']['ENGINE']:
26         return r"\b%s\b" % word
27     elif 'mysql' in settings.DATABASES['default']['ENGINE']:
28         return "[[:<:]]%s[[:>:]]" % word
29
30
31 def did_you_mean(query, tokens):
32     change = {}
33     for t in tokens:
34         authors = Tag.objects.filter(category='author', name__iregex=match_word_re(t))
35         if len(authors) > 0:
36             continue
37
38         if not dictionary.check(t):
39             try:
40                 change_to = dictionary.suggest(t)[0].lower()
41                 if change_to != t.lower():
42                     change[t] = change_to
43             except IndexError:
44                 pass
45
46     if change == {}:
47         return None
48
49     for frm, to in change.items():
50         query = query.replace(frm, to)
51
52     return query
53
54 JVM.attachCurrentThread()
55 search = Search()
56
57
58 def hint(request):
59     prefix = request.GET.get('term', '')
60     if len(prefix) < 2:
61         return JSONResponse([])
62     JVM.attachCurrentThread()
63
64     hint = search.hint()
65     try:
66         tags = request.GET.get('tags', '')
67         hint.tags(Tag.get_tag_list(tags))
68     except:
69         pass
70
71     # tagi beda ograniczac tutaj
72     # ale tagi moga byc na ksiazce i na fragmentach
73     # jezeli tagi dot tylko ksiazki, to wazne zeby te nowe byly w tej samej ksiazce
74     # jesli zas dotycza themes, to wazne, zeby byly w tym samym fragmencie.
75
76     tags = search.hint_tags(prefix, pdcounter=True)
77     books = search.hint_books(prefix)
78
79     def category_name(c):
80         if c.startswith('pd_'):
81             c = c[len('pd_'):]
82         return _(c)
83
84     return JSONResponse(
85         [{'label': t.name,
86           'category': category_name(t.category),
87           'id': t.id,
88           'url': t.get_absolute_url()}
89           for t in tags] + \
90           [{'label': b.title,
91             'category': _('book'),
92             'id': b.id,
93             'url': b.get_absolute_url()}
94             for b in books])
95
96
97 def main(request):
98     results = {}
99     JVM.attachCurrentThread()  # where to put this?
100
101     results = None
102     query = None
103     fuzzy = False #0.8
104
105     query = request.GET.get('q','')
106     # book_id = request.GET.get('book', None)
107     # book = None
108     # if book_id is not None:
109     #     book = get_object_or_404(Book, id=book_id)
110
111     # hint = search.hint()
112     # try:
113     #     tag_list = Tag.get_tag_list(tags)
114     # except:
115     #     tag_list = []
116
117     if len(query) < 2:
118         return render_to_response('catalogue/search_too_short.html', {'prefix': query},
119                                   context_instance=RequestContext(request))
120
121     # hint.tags(tag_list)
122     # if book:
123     #     hint.books(book)
124     tags = search.hint_tags(query, pdcounter=True, prefix=False, fuzzy=fuzzy)
125     tags = split_tags(tags)
126
127     toks = StringReader(query)
128     tokens_cache = {}
129
130     author_results = search.search_phrase(toks, 'authors', fuzzy=fuzzy, tokens_cache=tokens_cache)
131     title_results = search.search_phrase(toks, 'title', fuzzy=fuzzy, tokens_cache=tokens_cache)
132
133     # Boost main author/title results with mixed search, and save some of its results for end of list.
134     # boost author, title results
135     author_title_mixed = search.search_some(toks, ['authors', 'title', 'tags'], fuzzy=fuzzy, tokens_cache=tokens_cache)
136     author_title_rest = []
137     for b in author_title_mixed:
138         bks = filter(lambda ba: ba.book_id == b.book_id, author_results + title_results)
139         for b2 in bks:
140             b2.boost *= 1.1
141         if bks is []:
142             author_title_rest.append(b)
143
144     # Do a phrase search but a term search as well - this can give us better snippets then search_everywhere,
145     # Because the query is using only one field.
146     text_phrase = SearchResult.aggregate(
147         search.search_phrase(toks, 'content', fuzzy=fuzzy, tokens_cache=tokens_cache, snippets=True, book=False, slop=4),
148         search.search_some(toks, ['content'], tokens_cache=tokens_cache, snippets=True, book=False))
149
150     everywhere = search.search_everywhere(toks, fuzzy=fuzzy, tokens_cache=tokens_cache)
151
152     def already_found(results):
153         def f(e):
154             for r in results:
155                 if e.book_id == r.book_id:
156                     e.boost = 0.9
157                     results.append(e)
158                     return True
159             return False
160         return f
161     f = already_found(author_results + title_results + text_phrase)
162     everywhere = filter(lambda x: not f(x), everywhere)
163
164     author_results = SearchResult.aggregate(author_results)
165     title_results = SearchResult.aggregate(title_results)
166
167     everywhere = SearchResult.aggregate(everywhere, author_title_rest)
168
169     for res in [author_results, title_results, text_phrase, everywhere]:
170         res.sort(reverse=True)
171         for r in res:
172             for h in r.hits:
173                 h['snippets'] = map(lambda s:
174                                     re.subn(r"(^[ \t\n]+|[ \t\n]+$)", u"",
175                                             re.subn(r"[ \t\n]*\n[ \t\n]*", u"\n", s)[0])[0], h['snippets'])
176
177     suggestion = did_you_mean(query, search.get_tokens(toks, field="SIMPLE"))
178
179     def ensure_exists(r):
180         try:
181             return r.book
182         except Book.DoesNotExist:
183             return False
184
185     author_results = filter(ensure_exists, author_results)
186     title_results = filter(ensure_exists, title_results)
187     text_phrase = filter(ensure_exists, text_phrase)
188     everywhere = filter(ensure_exists, everywhere)
189
190     results = author_results + title_results + text_phrase + everywhere
191     # ensure books do exists & sort them
192     results.sort(reverse=True)
193
194     if len(results) == 1:
195         fragment_hits = filter(lambda h: 'fragment' in h, results[0].hits)
196         if len(fragment_hits) == 1:
197             #anchor = fragment_hits[0]['fragment']
198             #frag = Fragment.objects.get(anchor=anchor)
199             return HttpResponseRedirect(fragment_hits[0]['fragment'].get_absolute_url())
200         return HttpResponseRedirect(results[0].book.get_absolute_url())
201     elif len(results) == 0:
202         form = PublishingSuggestForm(initial={"books": query + ", "})
203         return render_to_response('catalogue/search_no_hits.html',
204                                   {'tags': tags,
205                                    'prefix': query,
206                                    "form": form,
207                                    'did_you_mean': suggestion},
208             context_instance=RequestContext(request))
209
210     return render_to_response('catalogue/search_multiple_hits.html',
211                               {'tags': tags,
212                                'prefix': query,
213                                'results': { 'author': author_results,
214                                             'title': title_results,
215                                             'content': text_phrase,
216                                             'other': everywhere},
217                                'did_you_mean': suggestion},
218         context_instance=RequestContext(request))