Dead links reporting.
[wolnelektury.git] / apps / opds / views.py
index ec93da1..c529e02 100644 (file)
@@ -12,15 +12,18 @@ from django.utils.feedgenerator import Atom1Feed
 from django.conf import settings
 from django.http import Http404
 from django.contrib.sites.models import Site
+from django.utils.functional import lazy
 
 from basicauth import logged_in_or_basicauth, factory_decorator
 from catalogue.models import Book, Tag
 
-from search import Search, SearchResult, JVM
-from lucene import Term, QueryWrapperFilter, TermQuery
-
+from search.views import Search
+import operator
+import logging
 import re
 
+log = logging.getLogger('opds')
+
 from stats.utils import piwik_track
 
 _root_feeds = (
@@ -62,21 +65,22 @@ _root_feeds = (
 )
 
 
+current_domain = lazy(lambda: Site.objects.get_current().domain, str)()
 def full_url(url):
-    return urljoin("http://%s" % Site.objects.get_current().domain, url)
+    return urljoin("http://%s" % current_domain, url)
 
 
 class OPDSFeed(Atom1Feed):
     link_rel = u"subsection"
     link_type = u"application/atom+xml"
 
-    _book_parent_img = full_url(os.path.join(settings.STATIC_URL, "img/book-parent.png"))
+    _book_parent_img = lazy(lambda: full_url(os.path.join(settings.STATIC_URL, "img/book-parent.png")), str)()
     try:
         _book_parent_img_size = unicode(os.path.getsize(os.path.join(settings.STATIC_ROOT, "img/book-parent.png")))
     except:
         _book_parent_img_size = ''
 
-    _book_img = full_url(os.path.join(settings.STATIC_URL, "img/book.png"))
+    _book_img = lazy(lambda: full_url(os.path.join(settings.STATIC_URL, "img/book.png")), str)()
     try:
         _book_img_size = unicode(os.path.getsize(os.path.join(settings.STATIC_ROOT, "img/book.png")))
     except:
@@ -89,7 +93,7 @@ class OPDSFeed(Atom1Feed):
                                 {u"href": reverse("opds_authors"),
                                  u"rel": u"start",
                                  u"type": u"application/atom+xml"})
-        handler.addQuickElement(u"link", None, 
+        handler.addQuickElement(u"link", None,
                                 {u"href": full_url(os.path.join(settings.STATIC_URL, "opensearch.xml")),
                                  u"rel": u"search",
                                  u"type": u"application/opensearchdescription+xml"})
@@ -109,6 +113,7 @@ class OPDSFeed(Atom1Feed):
                  u"length": self._book_parent_img_size,
                  u"type": u"image/png"})
         if item['pubdate'] is not None:
+            # FIXME: rfc3339_date is undefined, is this ever run?
             handler.addQuickElement(u"updated", rfc3339_date(item['pubdate']).decode('utf-8'))
 
         # Author information.
@@ -125,6 +130,7 @@ class OPDSFeed(Atom1Feed):
         if item['unique_id'] is not None:
             unique_id = item['unique_id']
         else:
+            # FIXME: get_tag_uri is undefined, is this ever run?
             unique_id = get_tag_uri(item['link'], item['pubdate'])
         handler.addQuickElement(u"id", unique_id)
 
@@ -220,7 +226,7 @@ class ByCategoryFeed(Feed):
     author_link = u"http://wolnelektury.pl/"
 
     def get_object(self, request, category):
-        feed = [feed for feed in _root_feeds if feed['category']==category]
+        feed = [feed for feed in _root_feeds if feed['category'] == category]
         if feed:
             feed = feed[0]
         else:
@@ -232,7 +238,7 @@ class ByCategoryFeed(Feed):
         return feed['title']
 
     def items(self, feed):
-        return Tag.objects.filter(category=feed['category']).exclude(book_count=0)
+        return Tag.objects.filter(category=feed['category']).exclude(items=None)
 
     def item_title(self, item):
         return item.name
@@ -258,13 +264,7 @@ class ByTagFeed(AcquisitionFeed):
         return get_object_or_404(Tag, category=category, slug=slug)
 
     def items(self, tag):
-        books = Book.tagged.with_any([tag])
-        l_tags = Tag.objects.filter(category='book', slug__in=[book.book_tag_slug() for book in books.iterator()])
-        descendants_keys = [book.pk for book in Book.tagged.with_any(l_tags)]
-        if descendants_keys:
-            books = books.exclude(pk__in=descendants_keys)
-
-        return books
+        return Book.tagged_top_level([tag])
 
 
 @factory_decorator(logged_in_or_basicauth())
@@ -283,7 +283,7 @@ class UserFeed(Feed):
         return u"Półki użytkownika %s" % user.username
 
     def items(self, user):
-        return Tag.objects.filter(category='set', user=user).exclude(book_count=0)
+        return Tag.objects.filter(category='set', user=user).exclude(items=None)
 
     def item_title(self, item):
         return item.name
@@ -294,9 +294,6 @@ class UserFeed(Feed):
     def item_description(self):
         return u''
 
-# no class decorators in python 2.5
-#UserFeed = factory_decorator(logged_in_or_basicauth())(UserFeed)
-
 
 @factory_decorator(logged_in_or_basicauth())
 @piwik_track
@@ -316,17 +313,41 @@ class UserSetFeed(AcquisitionFeed):
     def items(self, tag):
         return Book.tagged.with_any([tag])
 
-# no class decorators in python 2.5
-#UserSetFeed = factory_decorator(logged_in_or_basicauth())(UserSetFeed)
-
 
 @piwik_track
 class SearchFeed(AcquisitionFeed):
     description = u"Wyniki wyszukiwania na stronie WolneLektury.pl"
     title = u"Wyniki wyszukiwania"
 
-    INLINE_QUERY_RE = re.compile(r"(author:(?P<author>[^ ]+)|title:(?P<title>[^ ]+)|categories:(?P<categories>[^ ]+)|description:(?P<description>[^ ]+))")
-    
+    QUOTE_OR_NOT = r'(?:(?=["])"([^"]+)"|([^ ]+))'
+    INLINE_QUERY_RE = re.compile(
+        r"author:" + QUOTE_OR_NOT +
+        "|translator:" + QUOTE_OR_NOT +
+        "|title:" + QUOTE_OR_NOT +
+        "|categories:" + QUOTE_OR_NOT +
+        "|description:" + QUOTE_OR_NOT +
+        "|text:" + QUOTE_OR_NOT
+        )
+    MATCHES = {
+        'author': (0, 1),
+        'translator': (2, 3),
+        'title': (4, 5),
+        'categories': (6, 7),
+        'description': (8, 9),
+        'text': (10, 11),
+        }
+
+    PARAMS_TO_FIELDS = {
+        'author': 'authors',
+        'translator': 'translators',
+        #        'title': 'title',
+        'categories': 'tag_name_pl',
+        'description': 'text',
+        #        'text': 'text',
+        }
+
+    ATOM_PLACEHOLDER = re.compile(r"^{(atom|opds):\w+}$")
+
     def get_object(self, request):
         """
         For OPDS 1.1 We should handle a query for search terms
@@ -334,7 +355,7 @@ class SearchFeed(AcquisitionFeed):
         OpenSearch defines fields: atom:author, atom:contributor (treated as translator),
         atom:title. Inline query provides author, title, categories (treated as book tags),
         description (treated as content search terms).
-        
+
         if search terms are provided, we shall search for books
         according to Hint information (from author & contributror & title).
 
@@ -342,99 +363,67 @@ class SearchFeed(AcquisitionFeed):
         (perhaps for is_book=True)
 
         """
-        JVM.attachCurrentThread()
 
         query = request.GET.get('q', '')
 
         inline_criteria = re.findall(self.INLINE_QUERY_RE, query)
         if inline_criteria:
-            def get_criteria(criteria, name, position):
-                e = filter(lambda el: el[0][0:len(name)] == name, criteria)
-                print e
-                if not e:
-                    return None
-                c = e[0][position]
-                print c
-                if c[0] == '"' and c[-1] == '"':
-                    c = c[1:-1]
-                    c = c.replace('+', ' ')
-                return c
-
-            #import pdb; pdb.set_trace()
-            author = get_criteria(inline_criteria, 'author', 1)
-            title = get_criteria(inline_criteria, 'title', 2)
-            translator = None
-            categories = get_criteria(inline_criteria, 'categories', 3)
-            query = get_criteria(inline_criteria, 'description', 4)
+            remains = re.sub(self.INLINE_QUERY_RE, '', query)
+            remains = re.sub(r'[ \t]+', ' ', remains)
+
+            def get_criteria(criteria, name):
+                for c in criteria:
+                    for p in self.MATCHES[name]:
+                        if c[p]:
+                            if p % 2 == 0:
+                                return c[p].replace('+', ' ')
+                            return c[p]
+                return None
+
+            criteria = dict(map(
+                lambda cn: (cn, get_criteria(inline_criteria, cn)),
+                ['author', 'translator', 'title', 'categories',
+                 'description', 'text']))
+            query = remains
+            # empty query and text set case?
+            log.debug("Inline query = [%s], criteria: %s" % (query, criteria))
         else:
-            author = request.GET.get('author', '')
-            title = request.GET.get('title', '')
-            translator = request.GET.get('translator', '')
-
-            # Our client didn't handle the opds placeholders
-            if author == '{atom:author}': author = ''       
-            if title == '{atom:title}': title = ''
-            if translator == '{atom:contributor}': translator = ''
-            categories = None
-            fuzzy = False
+            def remove_dump_data(val):
+                """Some clients don't get opds placeholders and just send them."""
+                if self.ATOM_PLACEHOLDER.match(val):
+                    return ''
+                return val
 
+            criteria = dict([(cn, remove_dump_data(request.GET.get(cn, '')))
+                        for cn in self.MATCHES.keys()])
+            # query is set above.
+            log.debug("Inline query = [%s], criteria: %s" % (query, criteria))
 
         srch = Search()
-        hint = srch.hint()
 
-        # Scenario 1: full search terms provided.
-        # Use auxiliarry information to narrow it and make it better.
+        book_hit_filter = srch.index.Q(book_id__any=True)
+        filters = [book_hit_filter] + [srch.index.Q(
+            **{self.PARAMS_TO_FIELDS.get(cn, cn): criteria[cn]}
+            ) for cn in self.MATCHES.keys() if cn in criteria
+            if criteria[cn]]
+
         if query:
-            filters = []
-
-            if author:
-                print "narrow to author %s" % author
-                hint.tags(srch.search_tags(author, filt=srch.term_filter(Term('tag_category', 'author'))))
-
-            if translator:
-                print "filter by translator %s" % translator
-                filters.append(QueryWrapperFilter(
-                    srch.make_phrase(srch.get_tokens(translator, field='translators'),
-                                     field='translators')))
-
-            if categories:
-                filters.append(QueryWrapperFilter(
-                    srch.make_phrase(srch.get_tokens(categories, field="tag_name_pl"),
-                                     field='tag_name_pl')))
-
-            flt = srch.chain_filters(filters)
-            if title:
-                print "hint by book title %s" % title
-                q = srch.make_phrase(srch.get_tokens(title, field='title'), field='title')
-                hint.books(*srch.search_books(q, filt=flt))
-
-            toks = srch.get_tokens(query)
-            print "tokens: %s" % toks
-            #            import pdb; pdb.set_trace()
-            results = SearchResult.aggregate(srch.search_perfect_book(toks, fuzzy=fuzzy, hint=hint),
-                srch.search_perfect_parts(toks, fuzzy=fuzzy, hint=hint),
-                srch.search_everywhere(toks, fuzzy=fuzzy, hint=hint))
-            results.sort(reverse=True)
-            return [r.book for r in results]
+            q = srch.index.query(
+                reduce(operator.or_,
+                       [srch.index.Q(**{self.PARAMS_TO_FIELDS.get(cn, cn): query})
+                        for cn in self.MATCHES.keys()],
+                srch.index.Q()))
         else:
-            # Scenario 2: since we no longer have to figure out what the query term means to the user,
-            # we can just use filters and not the Hint class.
-            filters = []
-
-            fields = {
-                'author': author,
-                'translators': translator,
-                'title': title
-                }
-
-            for fld, q in fields.items():
-                if q:
-                    filters.append(QueryWrapperFilter(
-                        srch.make_phrase(srch.get_tokens(q, field=fld), field=fld)))
-
-            flt = srch.chain_filters(filters)
-            books = srch.search_books(TermQuery(Term('is_book', 'true')), filt=flt)
-            return books
+            q = srch.index.query(srch.index.Q())
+
+        q = srch.apply_filters(q, filters).field_limit(score=True, fields=['book_id'])
+        results = q.execute()
+
+        book_scores = dict([(r['book_id'], r['score']) for r in results])
+        books = Book.objects.filter(id__in=set([r['book_id'] for r in results]))
+        books = list(books)
+        books.sort(reverse=True, key=lambda book: book_scores[book.id])
+        return books
 
     def get_link(self, query):
         return "%s?q=%s" % (reverse('search'), query)