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