Пример #1
0
def advanced_search(request):
    """Elasticsearch-specific Advanced search view"""

    to_json = JSONRenderer().render
    template = 'search/results.html'

    # 1. Prep request.
    r = request.GET.copy()
    # TODO: Figure out how to get rid of 'a' and do it.
    # It basically is used to switch between showing the form or results.
    a = request.GET.get('a', '2')
    # TODO: This is so the 'a=1' stays in the URL for pagination.
    r['a'] = 1

    language = locale_or_default(
        request.GET.get('language', request.LANGUAGE_CODE))
    r['language'] = language
    lang = language.lower()
    lang_name = settings.LANGUAGES_DICT.get(lang) or ''

    # 2. Build form.
    search_form = AdvancedSearchForm(r, auto_id=False)
    search_form.set_allowed_forums(request.user)

    # 3. Validate request.
    # Note: a == 2 means "show the form"--that's all we use it for now.
    if a == '2' or not search_form.is_valid():
        if request.IS_JSON:
            return HttpResponse(json.dumps(
                {'error': _('Invalid search data.')}),
                                content_type=request.CONTENT_TYPE,
                                status=400)

        t = 'search/form.html'
        data = {
            'advanced': True,
            'request': request,
            'search_form': search_form
        }
        # get value for search input from last search term.
        last_search = request.COOKIES.get(settings.LAST_SEARCH_COOKIE)
        # If there is any cached input from last search, pass it to template
        if last_search and 'q' not in r:
            cached_field = urlquote(last_search)
            data.update({'cached_field': cached_field})

        return cache_control(render(request, t, data),
                             settings.SEARCH_CACHE_PERIOD)

    # 4. Generate search.
    cleaned = search_form.cleaned_data

    # We use a regular S here because we want to search across
    # multiple doctypes.
    searcher = (AnalyzerS().es(
        urls=settings.ES_URLS,
        timeout=settings.ES_TIMEOUT,
        use_ssl=settings.ES_USE_SSL,
        http_auth=settings.ES_HTTP_AUTH,
        connection_class=RequestsHttpConnection).indexes(
            es_utils.read_index('default')))

    doctypes = []
    final_filter = F()
    unix_now = int(time.time())
    interval_filters = (('created', cleaned['created'],
                         cleaned['created_date']),
                        ('updated', cleaned['updated'],
                         cleaned['updated_date']))

    # Start - wiki search configuration

    if cleaned['w'] & constants.WHERE_WIKI:
        wiki_f = F(model='wiki_document')

        # Category filter
        if cleaned['category']:
            wiki_f &= F(document_category__in=cleaned['category'])

        # Locale filter
        wiki_f &= F(document_locale=language)

        # Product filter
        products = cleaned['product']
        for p in products:
            wiki_f &= F(product=p)

        # Topics filter
        topics = cleaned['topics']
        for t in topics:
            wiki_f &= F(topic=t)

        # Archived bit
        if not cleaned['include_archived']:
            wiki_f &= F(document_is_archived=False)

        # Apply sortby
        sortby = cleaned['sortby_documents']
        try:
            searcher = searcher.order_by(*constants.SORT_DOCUMENTS[sortby])
        except IndexError:
            # Skip index errors because they imply the user is sending us sortby values
            # that aren't valid.
            pass

        doctypes.append(DocumentMappingType.get_mapping_type_name())
        final_filter |= wiki_f

    # End - wiki search configuration

    # Start - support questions configuration

    if cleaned['w'] & constants.WHERE_SUPPORT:
        question_f = F(model='questions_question')

        # These filters are ternary, they can be either YES, NO, or OFF
        ternary_filters = ('is_locked', 'is_solved', 'has_answers',
                           'has_helpful', 'is_archived')
        d = dict(('question_%s' % filter_name,
                  _ternary_filter(cleaned[filter_name]))
                 for filter_name in ternary_filters if cleaned[filter_name])
        if d:
            question_f &= F(**d)

        if cleaned['asked_by']:
            question_f &= F(question_creator=cleaned['asked_by'])

        if cleaned['answered_by']:
            question_f &= F(question_answer_creator=cleaned['answered_by'])

        q_tags = [t.strip() for t in cleaned['q_tags'].split(',')]
        for t in q_tags:
            if t:
                question_f &= F(question_tag=t)

        # Product filter
        products = cleaned['product']
        for p in products:
            question_f &= F(product=p)

        # Topics filter
        topics = cleaned['topics']
        for t in topics:
            question_f &= F(topic=t)

        # Note: num_voted (with a d) is a different field than num_votes
        # (with an s). The former is a dropdown and the latter is an
        # integer value.
        if cleaned['num_voted'] == constants.INTERVAL_BEFORE:
            question_f &= F(
                question_num_votes__lte=max(cleaned['num_votes'], 0))
        elif cleaned['num_voted'] == constants.INTERVAL_AFTER:
            question_f &= F(question_num_votes__gte=cleaned['num_votes'])

        # Apply sortby
        sortby = cleaned['sortby']
        try:
            searcher = searcher.order_by(*constants.SORT_QUESTIONS[sortby])
        except IndexError:
            # Skip index errors because they imply the user is sending us sortby values
            # that aren't valid.
            pass

        # Apply created and updated filters
        for filter_name, filter_option, filter_date in interval_filters:
            if filter_option == constants.INTERVAL_BEFORE:
                before = {
                    filter_name + '__gte': 0,
                    filter_name + '__lte': max(filter_date, 0)
                }

                question_f &= F(**before)

            elif filter_option == constants.INTERVAL_AFTER:
                after = {
                    filter_name + '__gte': min(filter_date, unix_now),
                    filter_name + '__lte': unix_now
                }

                question_f &= F(**after)

        doctypes.append(QuestionMappingType.get_mapping_type_name())
        final_filter |= question_f

    # End - support questions configuration

    # Start - discussion forum configuration

    if cleaned['w'] & constants.WHERE_DISCUSSION:
        discussion_f = F(model='forums_thread')

        if cleaned['author']:
            discussion_f &= F(post_author_ord=cleaned['author'])

        if cleaned['thread_type']:
            if constants.DISCUSSION_STICKY in cleaned['thread_type']:
                discussion_f &= F(post_is_sticky=1)

            if constants.DISCUSSION_LOCKED in cleaned['thread_type']:
                discussion_f &= F(post_is_locked=1)

        valid_forum_ids = [
            f.id for f in Forum.authorized_forums_for_user(request.user)
        ]

        forum_ids = None
        if cleaned['forum']:
            forum_ids = [f for f in cleaned['forum'] if f in valid_forum_ids]

        # If we removed all the forums they wanted to look at or if
        # they didn't specify, then we filter on the list of all
        # forums they're authorized to look at.
        if not forum_ids:
            forum_ids = valid_forum_ids

        discussion_f &= F(post_forum_id__in=forum_ids)

        # Apply created and updated filters
        for filter_name, filter_option, filter_date in interval_filters:
            if filter_option == constants.INTERVAL_BEFORE:
                before = {
                    filter_name + '__gte': 0,
                    filter_name + '__lte': max(filter_date, 0)
                }

                discussion_f &= F(**before)

            elif filter_option == constants.INTERVAL_AFTER:
                after = {
                    filter_name + '__gte': min(filter_date, unix_now),
                    filter_name + '__lte': unix_now
                }

                discussion_f &= F(**after)

        doctypes.append(ThreadMappingType.get_mapping_type_name())
        final_filter |= discussion_f

    # End - discussion forum configuration

    # Done with all the filtery stuff--time  to generate results

    searcher = searcher.doctypes(*doctypes)
    searcher = searcher.filter(final_filter)

    if 'explain' in request.GET and request.GET['explain'] == '1':
        searcher = searcher.explain()

    cleaned_q = cleaned['q']

    # Set up the highlights. Show the entire field highlighted.
    searcher = searcher.highlight(
        'question_content',  # support forum
        'document_summary',  # kb
        'post_content',  # contributor forum
        pre_tags=['<b>'],
        post_tags=['</b>'],
        number_of_fragments=0)

    searcher = apply_boosts(searcher)

    # Build the query
    if cleaned_q:
        query_fields = chain(*[
            cls.get_query_fields() for cls in
            [DocumentMappingType, ThreadMappingType, QuestionMappingType]
        ])
        query = {}
        # Create a simple_query_search query for every field we want to search.
        for field in query_fields:
            query['%s__sqs' % field] = cleaned_q

        # Transform the query to use locale aware analyzers.
        query = es_utils.es_query_with_analyzer(query, language)

        searcher = searcher.query(should=True, **query)

    searcher = searcher[:settings.SEARCH_MAX_RESULTS]

    # 5. Generate output
    pages = paginate(request, searcher, settings.SEARCH_RESULTS_PER_PAGE)

    if pages.paginator.count == 0:
        # If we know there aren't any results, show fallback_results.
        fallback_results = _fallback_results(language, cleaned['product'])
        results = []
    else:
        fallback_results = None
        results = build_results_list(pages, request.IS_JSON)

    items = [(k, v) for k in search_form.fields for v in r.getlist(k)
             if v and k != 'a']
    items.append(('a', '2'))

    product = Product.objects.filter(slug__in=cleaned['product'])
    if product:
        product_titles = [
            pgettext('DB: products.Product.title', p.title) for p in product
        ]
    else:
        product_titles = [_('All Products')]

    # FIXME: This is probably bad l10n.
    product_titles = ', '.join(product_titles)

    data = {
        'num_results': pages.paginator.count,
        'results': results,
        'fallback_results': fallback_results,
        'product_titles': product_titles,
        'q': cleaned['q'],
        'w': cleaned['w'],
        'lang_name': lang_name,
        'advanced': True,
        'products': Product.objects.filter(visible=True)
    }

    if request.IS_JSON:
        data['total'] = len(data['results'])
        data['products'] = [{
            'slug': p.slug,
            'title': p.title
        } for p in data['products']]

        if product:
            data['product'] = product[0].slug

        pages = Paginator(pages)
        data['pagination'] = dict(
            number=pages.pager.number,
            num_pages=pages.pager.paginator.num_pages,
            has_next=pages.pager.has_next(),
            has_previous=pages.pager.has_previous(),
            max=pages.max,
            span=pages.span,
            dotted_upper=pages.pager.dotted_upper,
            dotted_lower=pages.pager.dotted_lower,
            page_range=pages.pager.page_range,
            url=pages.pager.url,
        )
        if not results:
            data['message'] = _('No pages matched the search criteria')
        json_data = to_json(data)
        if request.JSON_CALLBACK:
            json_data = request.JSON_CALLBACK + '(' + json_data + ');'
        return HttpResponse(json_data, content_type=request.CONTENT_TYPE)

    data.update({
        'product': product,
        'pages': pages,
        'search_form': search_form
    })
    resp = cache_control(render(request, template, data),
                         settings.SEARCH_CACHE_PERIOD)
    resp.set_cookie(settings.LAST_SEARCH_COOKIE,
                    urlquote(cleaned['q']),
                    max_age=3600,
                    secure=False,
                    httponly=False)
    return resp
Пример #2
0
def advanced_search(request, template=None):
    """Elasticsearch-specific Advanced search view"""

    to_json = JSONRenderer().render

    # 1. Prep request.
    r = request.GET.copy()
    # TODO: Figure out how to get rid of 'a' and do it.
    # It basically is used to switch between showing the form or results.
    a = request.GET.get('a', '2')
    # TODO: This is so the 'a=1' stays in the URL for pagination.
    r['a'] = 1

    language = locale_or_default(request.GET.get('language', request.LANGUAGE_CODE))
    r['language'] = language
    lang = language.lower()
    lang_name = settings.LANGUAGES_DICT.get(lang) or ''

    # 2. Build form.
    search_form = AdvancedSearchForm(r, auto_id=False)
    search_form.set_allowed_forums(request.user)

    # 3. Validate request.
    # Note: a == 2 means "show the form"--that's all we use it for now.
    if a == '2' or not search_form.is_valid():
        if request.IS_JSON:
            return HttpResponse(
                json.dumps({'error': _('Invalid search data.')}),
                content_type=request.CONTENT_TYPE,
                status=400)

        t = template if request.MOBILE else 'search/form.html'
        data = {'advanced': True,
                'request': request,
                'search_form': search_form}
        # get value for search input from last search term.
        last_search = request.COOKIES.get(settings.LAST_SEARCH_COOKIE)
        # If there is any cached input from last search, pass it to template
        if last_search and 'q' not in r:
            cached_field = urlquote(last_search)
            data.update({'cached_field': cached_field})

        return cache_control(
            render(request, t, data),
            settings.SEARCH_CACHE_PERIOD)

    # 4. Generate search.
    cleaned = search_form.cleaned_data

    # On mobile, we default to just wiki results.
    if request.MOBILE and cleaned['w'] == constants.WHERE_BASIC:
        cleaned['w'] = constants.WHERE_WIKI

    # We use a regular S here because we want to search across
    # multiple doctypes.
    searcher = (AnalyzerS().es(urls=settings.ES_URLS)
                .indexes(es_utils.read_index('default')))

    doctypes = []
    final_filter = F()
    unix_now = int(time.time())
    interval_filters = (
        ('created', cleaned['created'], cleaned['created_date']),
        ('updated', cleaned['updated'], cleaned['updated_date'])
    )

    # Start - wiki search configuration

    if cleaned['w'] & constants.WHERE_WIKI:
        wiki_f = F(model='wiki_document')

        # Category filter
        if cleaned['category']:
            wiki_f &= F(document_category__in=cleaned['category'])

        # Locale filter
        wiki_f &= F(document_locale=language)

        # Product filter
        products = cleaned['product']
        for p in products:
            wiki_f &= F(product=p)

        # Topics filter
        topics = cleaned['topics']
        for t in topics:
            wiki_f &= F(topic=t)

        # Archived bit
        if not cleaned['include_archived']:
            wiki_f &= F(document_is_archived=False)

        # Apply sortby
        sortby = cleaned['sortby_documents']
        try:
            searcher = searcher.order_by(*constants.SORT_DOCUMENTS[sortby])
        except IndexError:
            # Skip index errors because they imply the user is sending us sortby values
            # that aren't valid.
            pass

        doctypes.append(DocumentMappingType.get_mapping_type_name())
        final_filter |= wiki_f

    # End - wiki search configuration

    # Start - support questions configuration

    if cleaned['w'] & constants.WHERE_SUPPORT:
        question_f = F(model='questions_question')

        # These filters are ternary, they can be either YES, NO, or OFF
        ternary_filters = ('is_locked', 'is_solved', 'has_answers',
                           'has_helpful', 'is_archived')
        d = dict(('question_%s' % filter_name,
                  _ternary_filter(cleaned[filter_name]))
                 for filter_name in ternary_filters if cleaned[filter_name])
        if d:
            question_f &= F(**d)

        if cleaned['asked_by']:
            question_f &= F(question_creator=cleaned['asked_by'])

        if cleaned['answered_by']:
            question_f &= F(question_answer_creator=cleaned['answered_by'])

        q_tags = [t.strip() for t in cleaned['q_tags'].split(',')]
        for t in q_tags:
            if t:
                question_f &= F(question_tag=t)

        # Product filter
        products = cleaned['product']
        for p in products:
            question_f &= F(product=p)

        # Topics filter
        topics = cleaned['topics']
        for t in topics:
            question_f &= F(topic=t)

        # Note: num_voted (with a d) is a different field than num_votes
        # (with an s). The former is a dropdown and the latter is an
        # integer value.
        if cleaned['num_voted'] == constants.INTERVAL_BEFORE:
            question_f &= F(question_num_votes__lte=max(cleaned['num_votes'], 0))
        elif cleaned['num_voted'] == constants.INTERVAL_AFTER:
            question_f &= F(question_num_votes__gte=cleaned['num_votes'])

        # Apply sortby
        sortby = cleaned['sortby']
        try:
            searcher = searcher.order_by(*constants.SORT_QUESTIONS[sortby])
        except IndexError:
            # Skip index errors because they imply the user is sending us sortby values
            # that aren't valid.
            pass

        # Apply created and updated filters
        for filter_name, filter_option, filter_date in interval_filters:
            if filter_option == constants.INTERVAL_BEFORE:
                before = {filter_name + '__gte': 0,
                          filter_name + '__lte': max(filter_date, 0)}

                question_f &= F(**before)

            elif filter_option == constants.INTERVAL_AFTER:
                after = {filter_name + '__gte': min(filter_date, unix_now),
                         filter_name + '__lte': unix_now}

                question_f &= F(**after)

        doctypes.append(QuestionMappingType.get_mapping_type_name())
        final_filter |= question_f

    # End - support questions configuration

    # Start - discussion forum configuration

    if cleaned['w'] & constants.WHERE_DISCUSSION:
        discussion_f = F(model='forums_thread')

        if cleaned['author']:
            discussion_f &= F(post_author_ord=cleaned['author'])

        if cleaned['thread_type']:
            if constants.DISCUSSION_STICKY in cleaned['thread_type']:
                discussion_f &= F(post_is_sticky=1)

            if constants.DISCUSSION_LOCKED in cleaned['thread_type']:
                discussion_f &= F(post_is_locked=1)

        valid_forum_ids = [f.id for f in Forum.authorized_forums_for_user(request.user)]

        forum_ids = None
        if cleaned['forum']:
            forum_ids = [f for f in cleaned['forum'] if f in valid_forum_ids]

        # If we removed all the forums they wanted to look at or if
        # they didn't specify, then we filter on the list of all
        # forums they're authorized to look at.
        if not forum_ids:
            forum_ids = valid_forum_ids

        discussion_f &= F(post_forum_id__in=forum_ids)

        # Apply created and updated filters
        for filter_name, filter_option, filter_date in interval_filters:
            if filter_option == constants.INTERVAL_BEFORE:
                before = {filter_name + '__gte': 0,
                          filter_name + '__lte': max(filter_date, 0)}

                discussion_f &= F(**before)

            elif filter_option == constants.INTERVAL_AFTER:
                after = {filter_name + '__gte': min(filter_date, unix_now),
                         filter_name + '__lte': unix_now}

                discussion_f &= F(**after)

        doctypes.append(ThreadMappingType.get_mapping_type_name())
        final_filter |= discussion_f

    # End - discussion forum configuration

    # Done with all the filtery stuff--time  to generate results

    searcher = searcher.doctypes(*doctypes)
    searcher = searcher.filter(final_filter)

    if 'explain' in request.GET and request.GET['explain'] == '1':
        searcher = searcher.explain()

    cleaned_q = cleaned['q']

    # Set up the highlights. Show the entire field highlighted.
    searcher = searcher.highlight(
        'question_content',  # support forum
        'document_summary',  # kb
        'post_content',  # contributor forum
        pre_tags=['<b>'],
        post_tags=['</b>'],
        number_of_fragments=0)

    searcher = apply_boosts(searcher)

    # Build the query
    if cleaned_q:
        query_fields = chain(*[
            cls.get_query_fields() for cls in [
                DocumentMappingType,
                ThreadMappingType,
                QuestionMappingType
            ]
        ])
        query = {}
        # Create a simple_query_search query for every field we want to search.
        for field in query_fields:
            query['%s__sqs' % field] = cleaned_q

        # Transform the query to use locale aware analyzers.
        query = es_utils.es_query_with_analyzer(query, language)

        searcher = searcher.query(should=True, **query)

    searcher = searcher[:settings.SEARCH_MAX_RESULTS]

    # 5. Generate output
    pages = paginate(request, searcher, settings.SEARCH_RESULTS_PER_PAGE)

    if pages.paginator.count == 0:
        # If we know there aren't any results, show fallback_results.
        fallback_results = _fallback_results(language, cleaned['product'])
        results = []
    else:
        fallback_results = None
        results = build_results_list(pages, request.IS_JSON)

    items = [(k, v) for k in search_form.fields for
             v in r.getlist(k) if v and k != 'a']
    items.append(('a', '2'))

    product = Product.objects.filter(slug__in=cleaned['product'])
    if product:
        product_titles = [pgettext('DB: products.Product.title', p.title) for p in product]
    else:
        product_titles = [_('All Products')]

    # FIXME: This is probably bad l10n.
    product_titles = ', '.join(product_titles)

    data = {
        'num_results': pages.paginator.count,
        'results': results,
        'fallback_results': fallback_results,
        'product_titles': product_titles,
        'q': cleaned['q'],
        'w': cleaned['w'],
        'lang_name': lang_name,
        'advanced': True,
        'products': Product.objects.filter(visible=True)
    }

    if request.IS_JSON:
        data['total'] = len(data['results'])
        data['products'] = [{'slug': p.slug, 'title': p.title}
                            for p in data['products']]

        if product:
            data['product'] = product[0].slug

        pages = Paginator(pages)
        data['pagination'] = dict(
            number=pages.pager.number,
            num_pages=pages.pager.paginator.num_pages,
            has_next=pages.pager.has_next(),
            has_previous=pages.pager.has_previous(),
            max=pages.max,
            span=pages.span,
            dotted_upper=pages.pager.dotted_upper,
            dotted_lower=pages.pager.dotted_lower,
            page_range=pages.pager.page_range,
            url=pages.pager.url,
        )
        if not results:
            data['message'] = _('No pages matched the search criteria')
        json_data = to_json(data)
        if request.JSON_CALLBACK:
            json_data = request.JSON_CALLBACK + '(' + json_data + ');'
        return HttpResponse(json_data, content_type=request.CONTENT_TYPE)

    data.update({
        'product': product,
        'pages': pages,
        'search_form': search_form
    })
    resp = cache_control(render(request, template, data), settings.SEARCH_CACHE_PERIOD)
    resp.set_cookie(settings.LAST_SEARCH_COOKIE, urlquote(cleaned['q']),
                    max_age=3600, secure=False, httponly=False)
    return resp