simplify search results list
[wolnelektury.git] / src / search / views.py
index c6759ea..a659329 100644 (file)
@@ -115,7 +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()
@@ -132,74 +132,56 @@ def hint(request):
 @cache.never_cache
 def main(request):
     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},
             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
     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)
-
-    title_results = search.search_phrase(query, 'title', book=True)
-
-    # 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_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)
-        for b2 in also_in_mixed:
-            b2.boost *= 1.1
-        if also_in_mixed 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(query, 'text', snippets=True, book=False),
-        search.search_some(query, ['text'], snippets=True, book=False, query_terms=theme_terms))
-
-    everywhere = search.search_everywhere(query, query_terms=theme_terms)
-
-    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 + translator_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)
-    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)]:
-        res.sort(reverse=True)
-        for r in res:
-            search.get_snippets(r, query, field, 3)
+    results_parts = []
+
+    search_fields = []
+    fieldsets = (
+        (['authors'], True),
+        (['title'], True),
+        (['metadata'], True),
+        (['text', 'themes_pl'], False),
+    )
+    for fieldset, is_book in fieldsets:
+        search_fields += fieldset
+        results_parts.append(search.search_words(words, search_fields, book=is_book))
+
+    results = []
+    ids_results = {}
+    for results_part in results_parts:
+        for result in sorted(SearchResult.aggregate(results_part), reverse=True):
+            book_id = result.book_id
+            if book_id in ids_results:
+                ids_results[book_id].merge(result)
+            else:
+                results.append(result)
+                ids_results[book_id] = result
+
+    for result in results:
+        search.get_snippets(result, query, num=3)
 
     suggestion = u''
 
@@ -209,26 +191,9 @@ def main(request):
         except Book.DoesNotExist:
             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):
-        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:
+    results = filter(ensure_exists, results)
+
+    if not results:
         form = PublishingSuggestForm(initial={"books": query + ", "})
         return render_to_response(
             'catalogue/search_no_hits.html',
@@ -243,15 +208,9 @@ def main(request):
     return render_to_response(
         'catalogue/search_multiple_hits.html',
         {
-            'tags': tags,
+            'tags': tags['author'] + tags['kind'] + tags['genre'] + tags['epoch'] + tags['theme'],
             'prefix': query,
-            'results': {
-                'author': author_results,
-                'translator': translator_results,
-                'title': title_results,
-                'content': text_phrase,
-                'other': everywhere
-            },
+            'results': results,
             'did_you_mean': suggestion
         },
         context_instance=RequestContext(request))