revamp search
[wolnelektury.git] / src / search / views.py
index f7aa77c..8b055a4 100644 (file)
@@ -3,19 +3,18 @@
 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
 #
 from django.conf import settings
-from django.shortcuts import render_to_response, get_object_or_404
+from django.shortcuts import render_to_response
 from django.template import RequestContext
 from django.views.decorators import cache
-from django.http import HttpResponse, HttpResponseRedirect, Http404, HttpResponsePermanentRedirect, JsonResponse
+from django.http import HttpResponse, JsonResponse
 from django.utils.translation import ugettext as _
 
 from catalogue.utils import split_tags
-from catalogue.models import Book, Tag, Fragment
+from catalogue.models import Book
 from pdcounter.models import Author as PDCounterAuthor, BookStub as PDCounterBook
 from search.index import Search, SearchResult
 from suggest.forms import PublishingSuggestForm
 import re
-#import enchant
 import json
 
 
@@ -116,7 +115,7 @@ def hint(request):
                 break
             limit -= 1
             data.append({
-                'label': b.title,
+                'label': '<cite>%s</cite>, %s' % (b.title, b.author_unicode()),
                 'category': _('book'),
                 'id': b.id,
                 'url': b.get_absolute_url()
@@ -132,53 +131,49 @@ def hint(request):
 
 @cache.never_cache
 def main(request):
-    results = {}
-
-    results = None
-    query = None
-
     query = request.GET.get('q', '')
+    query = ' '.join(query.split())
+    # filter out private use characters
+    import unicodedata
+    query = ''.join(ch for ch in query if unicodedata.category(ch) != 'Co')
 
     if len(query) < 2:
-        return render_to_response('catalogue/search_too_short.html',
-                                  {'prefix': query},
+        return render_to_response(
+            'catalogue/search_too_short.html', {'prefix': query},
             context_instance=RequestContext(request))
+    elif len(query) > 256:
+        return render_to_response(
+            'catalogue/search_too_long.html', {'prefix': query}, context_instance=RequestContext(request))
 
     query = remove_query_syntax_chars(query)
-    
-    search = Search()
 
-    theme_terms = search.index.analyze(text=query, field="themes_pl") \
-        + search.index.analyze(text=query, field="themes")
+    words = query.split()
+    if len(words) > 10:
+        query = ' '.join(words[:10])
+
+    search = Search()
 
-            # change hints
+    # change hints
     tags = search.hint_tags(query, pdcounter=True, prefix=False)
     tags = split_tags(tags)
 
-    author_results = search.search_phrase(query, 'authors', book=True)
-    translator_results = search.search_phrase(query, 'translators', book=True)
+    author_results = search.search_words(words, ['authors'])
 
-    title_results = search.search_phrase(query, 'title', book=True)
+    title_results = search.search_words(words, ['title'])
 
-    # 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(query, ['authors', 'translators', 'title', 'tags'], query_terms=theme_terms)
+    author_title_mixed = search.search_words(words, ['authors', 'title', 'metadata'])
     author_title_rest = []
 
     for b in author_title_mixed:
-        also_in_mixed = filter(lambda ba: ba.book_id == b.book_id, author_results + translator_results + title_results)
+        also_in_mixed = filter(lambda ba: ba.book_id == b.book_id, author_results + title_results)
         for b2 in also_in_mixed:
             b2.boost *= 1.1
-        if also_in_mixed is []:
+        if not also_in_mixed:
             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(query, 'text', snippets=True, book=False),
-        search.search_some(query, ['text'], snippets=True, book=False, query_terms=theme_terms))
+    text_phrase = SearchResult.aggregate(search.search_words(words, ['text'], book=False))
 
-    everywhere = search.search_everywhere(query, query_terms=theme_terms)
+    everywhere = search.search_words(words, ['metadata', 'text', 'themes_pl'], book=False)
 
     def already_found(results):
         def f(e):
@@ -189,17 +184,15 @@ def main(request):
                     return True
             return False
         return f
-    f = already_found(author_results + translator_results + title_results + text_phrase)
+    f = already_found(author_results + title_results + text_phrase)
     everywhere = filter(lambda x: not f(x), everywhere)
 
-    author_results = SearchResult.aggregate(author_results)
-    translator_results = SearchResult.aggregate(translator_results)
+    author_results = SearchResult.aggregate(author_results, author_title_rest)
     title_results = SearchResult.aggregate(title_results)
 
     everywhere = SearchResult.aggregate(everywhere, author_title_rest)
 
     for field, res in [('authors', author_results),
-                       ('translators', translator_results),
                        ('title', title_results),
                        ('text', text_phrase),
                        ('text', everywhere)]:
@@ -216,40 +209,37 @@ def main(request):
             return False
 
     author_results = filter(ensure_exists, author_results)
-    translator_results = filter(ensure_exists, translator_results)
     title_results = filter(ensure_exists, title_results)
     text_phrase = filter(ensure_exists, text_phrase)
     everywhere = filter(ensure_exists, everywhere)
 
-    results = author_results + translator_results + title_results + text_phrase + everywhere
     # ensure books do exists & sort them
-    for res in (author_results, translator_results, title_results, text_phrase, everywhere):
+    for res in (author_results, title_results, text_phrase):
         res.sort(reverse=True)
 
-    # We don't want to redirect to book text, but rather display result page even with one result.
-    # 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())
-    if len(results) == 0:
+    if not (author_results or title_results or text_phrase or everywhere):
         form = PublishingSuggestForm(initial={"books": query + ", "})
-        return render_to_response('catalogue/search_no_hits.html',
-                                  {'tags': tags,
-                                   'prefix': query,
-                                   "form": form,
-                                   'did_you_mean': suggestion},
+        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('catalogue/search_multiple_hits.html',
-                              {'tags': tags,
-                               'prefix': query,
-                               'results': {'author': author_results,
-                                           'translator': translator_results,
-                                           'title': title_results,
-                                           'content': text_phrase,
-                                           'other': everywhere},
-                               'did_you_mean': suggestion},
+    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))