from django.utils import simplejson as json
from django.utils.datastructures import SortedDict
- from philo.contrib.sobol import registry
+ from philo.contrib.sobol import registry, get_search_instance
from philo.contrib.sobol.forms import SearchForm
from philo.contrib.sobol.utils import HASH_REDIRECT_GET_KEY, URL_REDIRECT_GET_KEY, SEARCH_ARG_GET_KEY, check_redirect_hash, RegistryIterator
from philo.exceptions import ViewCanNotProvideSubpath
self._favored_results += subresults
else:
break
+ if len(self._favored_results) == len(results):
+ self._favored_results = []
return self._favored_results
class Meta:
get_latest_by = 'datetime'
-class RegistryChoiceField(SlugMultipleChoiceField):
- def _get_choices(self):
- if isinstance(self._choices, RegistryIterator):
- return self._choices.copy()
- elif hasattr(self._choices, 'next'):
- choices, self._choices = itertools.tee(self._choices)
- return choices
- else:
- return self._choices
- choices = property(_get_choices)
-
-
try:
from south.modelsinspector import add_introspection_rules
except ImportError:
pass
else:
- add_introspection_rules([], ["^philo\.contrib\.shipherd\.models\.RegistryChoiceField"])
+ add_introspection_rules([], ["^philo\.contrib\.sobol\.models\.RegistryChoiceField"])
class SearchView(MultiView):
"""Handles a view for the results of a search, anonymously tracks the selections made by end users, and provides an AJAX API for asynchronous search result loading. This can be particularly useful if some searches are slow."""
#: :class:`ForeignKey` to a :class:`.Page` which will be used to render the search results.
results_page = models.ForeignKey(Page, related_name='search_results_related')
- #: A :class:`.SlugMultipleChoiceField` whose choices are the contents of the :class:`.SearchRegistry`
- searches = RegistryChoiceField(choices=registry.iterchoices())
+ #: A :class:`.SlugMultipleChoiceField` whose choices are the contents of :obj:`.sobol.search.registry`
+ searches = SlugMultipleChoiceField(choices=registry.iterchoices())
#: A :class:`BooleanField` which controls whether or not the AJAX API is enabled.
#:
#: .. note:: If the AJAX API is enabled, a ``ajax_api_url`` attribute will be added to each search instance containing the url and get parameters for an AJAX request to retrieve results for that search.
)
return urlpatterns
- def get_search_instance(self, slug, search_string):
- """Gets the :class:`.BaseSearch` subclass registered with :obj:`.sobol.search.registry` as ``slug`` and instantiates it with ``search_string``."""
- return registry[slug](search_string.lower())
-
def results_view(self, request, extra_context=None):
"""
Renders :attr:`results_page` with a context containing an instance of :attr:`search_form`. If the form was submitted and was valid, then one of two things has happened:
search_instances = []
for slug in self.searches:
- search_instance = self.get_search_instance(slug, search_string)
- search_instances.append(search_instance)
+ if slug in registry:
+ search_instance = get_search_instance(slug, search_string)
+ search_instances.append(search_instance)
- if self.enable_ajax_api:
- search_instance.ajax_api_url = "%s?%s=%s" % (self.reverse('ajax_api_view', kwargs={'slug': slug}, node=request.node), SEARCH_ARG_GET_KEY, search_string)
+ if self.enable_ajax_api:
+ search_instance.ajax_api_url = "%s?%s=%s" % (self.reverse('ajax_api_view', kwargs={'slug': slug}, node=request.node), SEARCH_ARG_GET_KEY, search_string)
if eventlet and not self.enable_ajax_api:
pool = eventlet.GreenPool()
pool.waitall()
context.update({
- 'searches': search_instances
+ 'searches': search_instances,
+ 'favored_results': []
})
+
+ try:
+ search = Search.objects.get(string=search_string)
+ except Search.DoesNotExist:
+ pass
+ else:
+ context['favored_results'] = [r.url for r in search.get_favored_results()]
else:
form = SearchForm()
def ajax_api_view(self, request, slug, extra_context=None):
"""
- Returns a JSON string containing two keyed lists.
+ Returns a JSON object containing the following variables:
+ search
+ Contains the slug for the search.
results
Contains the results of :meth:`.Result.get_context` for each result.
rendered
hasMoreResults
``True`` or ``False`` whether the search has more results according to :meth:`BaseSearch.has_more_results`
moreResultsURL
- Contains None or a querystring which, once accessed, will note the :class:`Click` and redirect the user to a page containing more results.
+ Contains ``None`` or a querystring which, once accessed, will note the :class:`Click` and redirect the user to a page containing more results.
"""
search_string = request.GET.get(SEARCH_ARG_GET_KEY)
- if not request.is_ajax() or not self.enable_ajax_api or slug not in self.searches or search_string is None:
+ if not request.is_ajax() or not self.enable_ajax_api or slug not in registry or slug not in self.searches or search_string is None:
raise Http404
- search_instance = self.get_search_instance(slug, search_string)
+ search_instance = get_search_instance(slug, search_string)
return HttpResponse(json.dumps({
+ 'search': search_instance.slug,
'results': [result.get_context() for result in search_instance.results],
- 'rendered': [result.render() for result in search_instance.results],
- 'hasMoreResults': search.has_more_results(),
- 'moreResultsURL': (u"?%s" % search.more_results_querydict.urlencode()) if search.more_results_querydict else None,
+ 'hasMoreResults': search_instance.has_more_results,
+ 'moreResultsURL': search_instance.more_results_url,
}), mimetype="application/json")
#encoding: utf-8
import datetime
+ from hashlib import sha1
from django.conf import settings
from django.contrib.sites.models import Site
from django.utils.http import urlquote_plus
from django.utils.safestring import mark_safe
from django.utils.text import capfirst
- from django.template import loader, Context, Template
+ from django.template import loader, Context, Template, TemplateDoesNotExist
-from philo.contrib.sobol.utils import make_tracking_querydict, RegistryIterator
+from philo.contrib.sobol.utils import make_tracking_querydict
+from philo.utils.registry import Registry
if getattr(settings, 'SOBOL_USE_EVENTLET', False):
__all__ = (
- 'Result', 'BaseSearch', 'DatabaseSearch', 'URLSearch', 'JSONSearch', 'GoogleSearch', 'registry'
- 'Result', 'BaseSearch', 'DatabaseSearch', 'URLSearch', 'JSONSearch', 'GoogleSearch', 'SearchRegistry', 'registry', 'get_search_instance'
++ 'Result', 'BaseSearch', 'DatabaseSearch', 'URLSearch', 'JSONSearch', 'GoogleSearch', 'registry', 'get_search_instance'
)
- SEARCH_CACHE_KEY = 'philo_sobol_search_results'
- DEFAULT_RESULT_TEMPLATE_STRING = "{% if url %}<a href='{{ url }}'>{% endif %}{{ title }}{% if url %}</a>{% endif %}"
- DEFAULT_RESULT_TEMPLATE = Template(DEFAULT_RESULT_TEMPLATE_STRING)
-
- # Determines the timeout on the entire result cache.
- MAX_CACHE_TIMEOUT = 60*24*7
+ SEARCH_CACHE_SEED = 'philo_sobol_search_results'
+ USE_CACHE = getattr(settings, 'SOBOL_USE_SEARCH', True)
-class RegistrationError(Exception):
- """Raised if there is a problem registering a search with a :class:`SearchRegistry`"""
- pass
-
-
-class SearchRegistry(object):
- """Holds a registry of search types by slug."""
-
- def __init__(self):
- self._registry = {}
-
- def register(self, search, slug=None):
- """
- Register a search with the registry.
-
- :param search: The search class to register - generally a subclass of :class:`BaseSearch`
- :param slug: The slug which will be used to register the search class. If ``slug`` is ``None``, the search's default slug will be used.
- :raises: :class:`RegistrationError` if a different search is already registered with ``slug``.
-
- """
- slug = slug or search.slug
- if slug in self._registry:
- registered = self._registry[slug]
- if registered.__module__ != search.__module__:
- raise RegistrationError("A different search is already registered as `%s`" % slug)
- else:
- self._registry[slug] = search
-
- def unregister(self, search, slug=None):
- """
- Unregister a search from the registry.
-
- :param search: The search class to unregister - generally a subclass of :class:`BaseSearch`
- :param slug: If provided, the search will only be removed if it was registered with ``slug``. If not provided, the search class will be unregistered no matter what slug it was registered with.
- :raises: :class:`RegistrationError` if a slug is provided but the search registered with that slug is not ``search``.
-
- """
- if slug is not None:
- if slug in self._registry and self._registry[slug] == search:
- del self._registry[slug]
- raise RegistrationError("`%s` is not registered as `%s`" % (search, slug))
- else:
- for slug, search in self._registry.items():
- if search == search:
- del self._registry[slug]
-
- def items(self):
- """Returns a list of (slug, search) items in the registry."""
- return self._registry.items()
-
- def iteritems(self):
- """Returns an iterator over the (slug, search) pairs in the registry."""
- return RegistryIterator(self._registry, 'iteritems')
-
- def iterchoices(self):
- """Returns an iterator over (slug, search.verbose_name) pairs for the registry."""
- return RegistryIterator(self._registry, 'iteritems', lambda x: (x[0], x[1].verbose_name))
-
- def __getitem__(self, key):
- """Returns the search registered with ``key``."""
- return self._registry[key]
-
- def __iter__(self):
- """Returns an iterator over the keys in the registry."""
- return self._registry.__iter__()
-
-
-registry = SearchRegistry()
+#: A registry for :class:`BaseSearch` subclasses that should be available in the admin.
+registry = Registry()
-
+ def _make_cache_key(search, search_arg):
+ return sha1(SEARCH_CACHE_SEED + search.slug + search_arg).hexdigest()
+
+
+ def get_search_instance(slug, search_arg):
+ """Returns a search instance for the given slug, either from the cache or newly-instantiated."""
+ search = registry[slug]
+ search_arg = search_arg.lower()
+ if USE_CACHE:
+ key = _make_cache_key(search, search_arg)
+ cached = cache.get(key)
+ if cached:
+ return cached
+ instance = search(search_arg)
+ instance.slug = slug
+ return instance
+
+
class Result(object):
"""
:class:`Result` is a helper class that, given a search and a result of that search, is able to correctly render itself with a template defined by the search. Every :class:`Result` will pass a ``title``, a ``url`` (if applicable), and the raw ``result`` returned by the search into the template context when rendering.
return self.search.get_result_title(self.result)
def get_url(self):
- """Returns the url of the result or an empty string by calling :meth:`BaseSearch.get_result_querydict` on the raw result and then encoding the querydict returned."""
- qd = self.search.get_result_querydict(self.result)
- if qd is None:
- return ""
- return "?%s" % qd.urlencode()
+ """Returns the url of the result or ``None`` by calling :meth:`BaseSearch.get_result_url` on the raw result. This url will contain a querystring which, if used, will track a :class:`.Click` for the actual url."""
+ return self.search.get_result_url(self.result)
+
+ def get_actual_url(self):
+ """Returns the actual url of the result by calling :meth:`BaseSearch.get_actual_result_url` on the raw result."""
+ return self.search.get_actual_result_url(self.result)
+
+ def get_content(self):
+ """Returns the content of the result by calling :meth:`BaseSearch.get_result_content` on the raw result."""
+ return self.search.get_result_content(self.result)
def get_template(self):
- """Returns the template for the result by calling :meth:`BaseSearch.get_result_template` on the raw result."""
+ """Returns the template which will be used to render the :class:`Result` by calling :meth:`BaseSearch.get_result_template` on the raw result."""
return self.search.get_result_template(self.result)
- def get_extra_context(self):
- """Returns any extra context for the result by calling :meth:`BaseSearch.get_result_extra_context` on the raw result."""
- return self.search.get_result_extra_context(self.result)
-
def get_context(self):
"""
- Returns the context dictionary for the result. This is used both in rendering the result and in the AJAX return value for :meth:`.SearchView.ajax_api_view`. The context will contain everything from :meth:`get_extra_context` as well as the following keys:
+ Returns the context dictionary for the result. This is used both in rendering the result and in the AJAX return value for :meth:`.SearchView.ajax_api_view`. The context will contain the following keys:
title
The result of calling :meth:`get_title`
url
The result of calling :meth:`get_url`
- result
- The raw result which the :class:`Result` was instantiated with.
+ content
+ The result of calling :meth:`get_content`
"""
- context = self.get_extra_context()
- context.update({
- 'title': self.get_title(),
- 'url': self.get_url(),
- 'result': self.result
- })
- return context
+ if not hasattr(self, '_context'):
+ self._context = {
+ 'title': self.get_title(),
+ 'url': self.get_url(),
+ 'actual_url': self.get_actual_url(),
+ 'content': self.get_content()
+ }
+ return self._context
def render(self):
"""Returns the template from :meth:`get_template` rendered with the context from :meth:`get_context`."""
if 'verbose_name' not in attrs:
attrs['verbose_name'] = capfirst(' '.join(convert_camelcase(name).rsplit(' ', 1)[:-1]))
if 'slug' not in attrs:
- attrs['slug'] = name.lower()
+ attrs['slug'] = name[:-6].lower() if name.endswith("Search") else name.lower()
return super(BaseSearchMetaclass, cls).__new__(cls, name, bases, attrs)
"""
__metaclass__ = BaseSearchMetaclass
- #: The number of results to return from the complete list. Default: 10
- result_limit = 10
+ #: The number of results to return from the complete list. Default: 5
+ result_limit = 5
#: How long the items for the search should be cached (in minutes). Default: 48 hours.
_cache_timeout = 60*48
+ #: The path to the template which will be used to render the :class:`Result`\ s for this search. If this is ``None``, then the framework will try ``sobol/search/<slug>/result.html`` and ``sobol/search/result.html``.
+ result_template = None
+ #: The path to the template which will be used to generate the title of the :class:`Result`\ s for this search. If this is ``None``, then the framework will try ``sobol/search/<slug>/title.html`` and ``sobol/search/title.html``.
+ title_template = None
+ #: The path to the template which will be used to generate the content of the :class:`Result`\ s for this search. If this is ``None``, then the framework will try ``sobol/search/<slug>/content.html`` and ``sobol/search/content.html``.
+ content_template = None
def __init__(self, search_arg):
self.search_arg = search_arg
- def _get_cached_results(self):
- """Return the cached results if the results haven't timed out. Otherwise return None."""
- result_cache = cache.get(SEARCH_CACHE_KEY)
- if result_cache and self.__class__ in result_cache and self.search_arg.lower() in result_cache[self.__class__]:
- cached = result_cache[self.__class__][self.search_arg.lower()]
- if cached['timeout'] >= datetime.datetime.now():
- return cached['results']
- return None
-
- def _set_cached_results(self, results, timeout):
- """Sets the results to the cache for <timeout> minutes."""
- result_cache = cache.get(SEARCH_CACHE_KEY) or {}
- cached = result_cache.setdefault(self.__class__, {}).setdefault(self.search_arg.lower(), {})
- cached.update({
- 'results': results,
- 'timeout': datetime.datetime.now() + datetime.timedelta(minutes=timeout)
- })
- cache.set(SEARCH_CACHE_KEY, result_cache, MAX_CACHE_TIMEOUT)
-
@property
def results(self):
"""Retrieves cached results or initiates a new search via :meth:`get_results` and caches the results."""
if not hasattr(self, '_results'):
- results = self._get_cached_results()
- if results is None:
- try:
- # Cache one extra result so we can see if there are
- # more results to be had.
- limit = self.result_limit
- if limit is not None:
- limit += 1
- results = self.get_results(limit)
- except:
- if settings.DEBUG:
- raise
- # On exceptions, don't set any cache; just return.
- return []
+ try:
+ # Cache one extra result so we can see if there are
+ # more results to be had.
+ limit = self.result_limit
+ if limit is not None:
+ limit += 1
+ results = self.get_results(limit)
+ except:
+ if settings.DEBUG:
+ raise
+ # On exceptions, don't set any cache; just return.
+ return []
- self._set_cached_results(results, self._cache_timeout)
self._results = results
+
+ if USE_CACHE:
+ for result in results:
+ result.get_context()
+ key = _make_cache_key(self, self.search_arg)
+ cache.set(key, self, self._cache_timeout)
return self._results
"""Returns an iterable of up to ``limit`` results. The :meth:`get_result_title`, :meth:`get_result_url`, :meth:`get_result_template`, and :meth:`get_result_extra_context` methods will be used to interpret the individual items that this function returns, so the result can be an object with attributes as easily as a dictionary with keys. However, keep in mind that the raw results will be stored with django's caching mechanisms and will be converted to JSON."""
raise NotImplementedError
- def get_result_title(self, result):
- """Returns the title of the ``result``. Must be implemented by subclasses."""
- raise NotImplementedError
-
- def get_result_url(self, result):
+ def get_actual_result_url(self, result):
"""Returns the actual URL for the ``result`` or ``None`` if there is no URL. Must be implemented by subclasses."""
raise NotImplementedError
def get_result_querydict(self, result):
"""Returns a querydict for tracking selection of the result, or ``None`` if there is no URL for the result."""
- url = self.get_result_url(result)
+ url = self.get_actual_result_url(result)
if url is None:
return None
return make_tracking_querydict(self.search_arg, url)
+ def get_result_url(self, result):
+ """Returns ``None`` or a url which, when accessed, will register a :class:`.Click` for that url."""
+ qd = self.get_result_querydict(result)
+ if qd is None:
+ return None
+ return "?%s" % qd.urlencode()
+
+ def get_result_title(self, result):
+ """Returns the title of the ``result``. By default, renders ``sobol/search/<slug>/title.html`` or ``sobol/search/title.html`` with the result in the context. This can be overridden by setting :attr:`title_template` or simply overriding :meth:`get_result_title`. If no template can be found, this will raise :exc:`TemplateDoesNotExist`."""
+ return loader.render_to_string(self.title_template or [
+ 'sobol/search/%s/title.html' % self.slug,
+ 'sobol/search/title.html'
+ ], {'result': result})
+
+ def get_result_content(self, result):
+ """Returns the content for the ``result``. By default, renders ``sobol/search/<slug>/content.html`` or ``sobol/search/content.html`` with the result in the context. This can be overridden by setting :attr:`content_template` or simply overriding :meth:`get_result_content`. If no template is found, this will return an empty string."""
+ try:
+ return loader.render_to_string(self.content_template or [
+ 'sobol/search/%s/content.html' % self.slug,
+ 'sobol/search/content.html'
+ ], {'result': result})
+ except TemplateDoesNotExist:
+ return ""
+
def get_result_template(self, result):
- """Returns the template to be used for rendering the ``result``."""
- if hasattr(self, 'result_template'):
+ """Returns the template to be used for rendering the ``result``. For a search with slug ``google``, this would first try ``sobol/search/google/result.html``, then fall back on ``sobol/search/result.html``. Subclasses can override this by setting :attr:`result_template` to the path of another template."""
+ if self.result_template:
return loader.get_template(self.result_template)
- if not hasattr(self, '_result_template'):
- self._result_template = DEFAULT_RESULT_TEMPLATE
- return self._result_template
-
- def get_result_extra_context(self, result):
- """Returns any extra context to be used when rendering the ``result``."""
- return {}
+ return loader.select_template([
+ 'sobol/search/%s/result.html' % self.slug,
+ 'sobol/search/result.html'
+ ])
+ @property
def has_more_results(self):
"""Returns ``True`` if there are more results than :attr:`result_limit` and ``False`` otherwise."""
return len(self.results) > self.result_limit
- @property
- def more_results_url(self):
- """Returns the actual url for more results. This should be accessed through :attr:`more_results_querydict` in the template so that the click can be tracked."""
- raise NotImplementedError
+ def get_actual_more_results_url(self):
+ """Returns the actual url for more results. By default, simply returns ``None``."""
+ return None
- @property
- def more_results_querydict(self):
+ def get_more_results_querydict(self):
"""Returns a :class:`QueryDict` for tracking whether people click on a 'more results' link."""
- return make_tracking_querydict(self.search_arg, self.more_results_url)
+ url = self.get_actual_more_results_url()
+ if url:
+ return make_tracking_querydict(self.search_arg, url)
+ return None
+
+ @property
+ def more_results_url(self):
+ """Returns a URL which consists of a querystring which, when accessed, will log a :class:`.Click` for the actual URL."""
+ qd = self.get_more_results_querydict()
+ if qd is None:
+ return None
+ return "?%s" % qd.urlencode()
def __unicode__(self):
return self.verbose_name
def url(self):
"""The URL where the search gets its results. Composed from :attr:`search_url` and :attr:`query_format_str`."""
return self.search_url + self.query_format_str % urlquote_plus(self.search_arg)
-
- @property
- def more_results_url(self):
+
+ def get_actual_more_results_url(self):
return self.url
def parse_response(self, response, limit=None):
class GoogleSearch(JSONSearch):
"""An example implementation of a :class:`JSONSearch`."""
search_url = "http://ajax.googleapis.com/ajax/services/search/web"
- result_template = 'search/googlesearch.html'
_cache_timeout = 60
verbose_name = "Google search (current site)"
+ _more_results_url = None
@property
def query_format_str(self):
return True
return False
- @property
- def more_results_url(self):
+ def get_actual_more_results_url(self):
return self._more_results_url
+ def get_actual_result_url(self, result):
+ return result['unescapedUrl']
+
def get_result_title(self, result):
- return result['titleNoFormatting']
+ return mark_safe(result['titleNoFormatting'])
- def get_result_url(self, result):
- return result['unescapedUrl']
+ def get_result_content(self, result):
+ return mark_safe(result['content'])
registry.register(GoogleSearch)