Implemented site-specific search.

Added site logos to searches across all sites for easier identification.
Added hints to make it more obvious which site you are searching.
Minor CSS tweaks.

python/media/css/main.css

@@ -105,13 +105,13 @@
 
 #search-results li {
     clear: both;
-    border-top: 1px solid #CCCCCC;
+    border-bottom: 1px solid #CCCCCC;
     margin-bottom: 10px;
     padding-right: 10px;
 }
 
 #search-results li:hover {
-    background-color: #F2F2F2;
+    background-color: #FAFAFA;
 }
 
 #search-results li .clearfix {
@@ -119,6 +119,15 @@
     height: 0;
 }
 
+#search-results li:last-child {
+    border-bottom: none;
+}
+
+.post-logo {
+    float: left;
+    width: 48px;
+}
+
 .post-stats-vertical {
     float: left;
     width: 64px;
@@ -149,6 +158,10 @@
     padding-left: 72px; /* 64px for post-stats-vertical + 8px for padding */
 }
 
+.post-summary-with-logo {
+    padding-left: 120px; /* 48px for post-logo + 64px for post-stats-vertical + 8px for padding */
+}
+
 .post-summary h3 {
     line-height: normal;
     padding: 7px 0;

python/src/stackdump/app.py

@@ -217,80 +217,31 @@ def site_index(site_key):
 @uses_solr
 @uses_db
 def search():
-    # TODO: scrub this first to avoid Solr injection attacks?
-    query = request.GET.get('q')
-    if not query:
-        redirect(settings.APP_URL_ROOT)
-    
-    # the page GET parameter is zero-based
-    page = int(request.GET.get('p', 0))
-    if page < 0: page = 0
-    
-    rows_per_page = int(request.GET.get('r', 10))
-    rows_per_page = (rows_per_page > 0) and rows_per_page or 10
-    
-    sort_args = {
-        'newest' : 'creationDate desc',
-        'votes' : 'votes desc',
-        'relevance' : 'score desc' # score is the special keyword for the
-                                   # relevancy score in Lucene
-    }
-    sort_by = request.GET.get('s', 'relevance').lower()
-    # default to sorting by relevance
-    if sort_by not in sort_args.keys():
-        sort_by = 'relevance'
-    
-    # perform search
-    results = solr_conn().search(query,
-                                 start=page*rows_per_page,
-                                 rows=rows_per_page,
-                                 sort=sort_args[sort_by])
-    decode_json_fields(results)
-    retrieve_users(results, question_only=True, ignore_comments=True)
-    
     context = { }
     context['site_root_path'] = ''
     context['sites'] = Site.select()
     
-    # TODO: scrub this first to avoid HTML injection attacks?
-    context['query'] = query
-    context['results'] = results
-    context['total_hits'] = results.hits
-    context['current_page'] = page + 1 # page template var is ones-based
-    context['rows_per_page'] = rows_per_page
-    context['total_pages'] =  int(math.ceil(float(results.hits) / rows_per_page))
-    context['sort_by'] = sort_by
+    context.update(perform_search())
     
     return render_template('results.html', context)
 
 @route('/:site_key#[\w\.]+#/search')
 @uses_templates
 @uses_solr
+@uses_db
 def site_search(site_key):
     context = { }
     context['site_root_path'] = '%s/' % site_key
+    # the template uses this to allow searching on other sites
+    context['sites'] = Site.select()
     
     try:
         context['site'] = Site.selectBy(key=site_key).getOne()
     except SQLObjectNotFound:
         raise HTTPError(code=404, output='No site exists with the key %s.' % site_key)
     
-    # TODO: scrub this first to avoid Solr injection attacks?
-    query = request.GET.get('q')
-    if not query:
-        redirect(settings.APP_URL_ROOT)
-    
-    page = request.GET.get('p', 0)
-    rows_per_page = request.GET.get('r', 10)
-    
-    # perform search
-    results = solr_conn().search(query, start=page*rows_per_page, rows=rows_per_page)
-    decode_json_fields(results)
-    retrieve_users(results)
-    
-    # TODO: scrub this first to avoid HTML injection attacks?
-    context['query'] = query
-    context['results'] = results
+    # perform the search limited by this site
+    context.update(perform_search(site_key))
     
     return render_template('site_results.html', context)
 
@@ -374,27 +325,27 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
     # get a list of all the user IDs
     user_ids_by_site = { }
     for r in results:
-        site_name = r['siteName']
-        if site_name not in user_ids_by_site.keys():
-            user_ids_by_site[site_name] = set()
+        site_key = r['siteKey']
+        if site_key not in user_ids_by_site.keys():
+            user_ids_by_site[site_key] = set()
         
         # the search result object itself
         for k in r.keys():
             if k.lower().endswith('userid'):
-                user_ids_by_site[site_name].add(r[k])
+                user_ids_by_site[site_key].add(r[k])
         
         # the question object
         question = r['question']
         for k in question.keys():
             if k.lower().endswith('userid'):
-                user_ids_by_site[site_name].add(question[k])
+                user_ids_by_site[site_key].add(question[k])
             
             comments = question.get('comments')
             if not ignore_comments and comments:
                 for c in comments:
                     for ck in c.keys():
                         if ck.lower().endswith('userid'):
-                            user_ids_by_site[site_name].add(c[ck])
+                            user_ids_by_site[site_key].add(c[ck])
         
         # the answers
         answers = r.get('answers')
@@ -402,21 +353,21 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
             for a in answers:
                 for k in a.keys():
                     if k.lower().endswith('userid'):
-                        user_ids_by_site[site_name].add(a[k])
+                        user_ids_by_site[site_key].add(a[k])
                 
                 comments = a.get('comments')
                 if not ignore_comments and comments:
                     for c in comments:
                         for ck in c.keys():
                             if ck.lower().endswith('userid'):
-                                user_ids_by_site[site_name].add(c[ck])
+                                user_ids_by_site[site_key].add(c[ck])
     
     # retrieve the user objects from the database by site
     users_by_site = { }
-    for site_name in user_ids_by_site.keys():
-        site = Site.select(Site.q.name == site_name).getOne()
+    for site_key in user_ids_by_site.keys():
+        site = Site.select(Site.q.key == site_key).getOne()
         user_objects = User.select(AND(User.q.site == site,
-                                       IN(User.q.sourceId, list(user_ids_by_site[site_name]))
+                                       IN(User.q.sourceId, list(user_ids_by_site[site_key]))
                                   ))
         
         # convert results into a dict with user id as the key
@@ -424,24 +375,24 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
         for u in user_objects:
             users[u.sourceId] = u
         
-        users_by_site[site_name] = users
+        users_by_site[site_key] = users
     
     # place user objects into the dict
     for r in results:
-        site_name = r['siteName']
+        site_key = r['siteKey']
         
         # the search result object itself
         for k in r.keys():
             if k.lower().endswith('userid'):
                 # use the same field name, minus the 'Id' on the end.
-                r[k[:-2]] = users_by_site[site_name].get(r[k])
+                r[k[:-2]] = users_by_site[site_key].get(r[k])
         
         # the question object
         question = r['question']
         for k in question.keys():
             if k.lower().endswith('userid'):
                 # use the same field name, minus the 'Id' on the end.
-                question[k[:-2]] = users_by_site[site_name].get(question[k])
+                question[k[:-2]] = users_by_site[site_key].get(question[k])
             
         comments = question.get('comments')
         if not ignore_comments and comments:
@@ -449,9 +400,7 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
                 for ck in c.keys():
                     if ck.lower().endswith('userid'):
                         # use the same field name, minus the 'Id' on the end.
-                        c[ck[:-2]] = users_by_site[site_name].get(c[ck])
-            
-            
+                        c[ck[:-2]] = users_by_site[site_key].get(c[ck])
         
         # the answers
         answers = r.get('answers')
@@ -460,7 +409,7 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
                 for k in a.keys():
                     if k.lower().endswith('userid'):
                         # use the same field name, minus the 'Id' on the end.
-                        a[k[:-2]] = users_by_site[site_name].get(a[k])
+                        a[k[:-2]] = users_by_site[site_key].get(a[k])
                 
                 comments = a.get('comments')
                 if not ignore_comments and comments:
@@ -468,7 +417,84 @@ def retrieve_users(results, question_only=False, ignore_comments=False):
                         for ck in c.keys():
                             if ck.lower().endswith('userid'):
                                 # use the same field name, minus the 'Id' on the end.
-                                c[ck[:-2]] = users_by_site[site_name].get(c[ck])
+                                c[ck[:-2]] = users_by_site[site_key].get(c[ck])
+
+def retrieve_sites(results):
+    '''\
+    Retrieves the site objects associated with the results.
+    '''
+    # get a list of all the site keys
+    site_keys = set()
+    for r in results:
+        site_keys.add(r['siteKey'])
+    
+    # retrieve the site objects from the database
+    sites = { }
+    for site_key in site_keys:
+        sites[site_key] = Site.select(Site.q.key == site_key).getOne()
+    
+    # place site objects into the dict
+    for r in results:
+        site_key = r['siteKey']
+        r['site'] = sites[site_key]
+
+def perform_search(site_key=None):
+    '''\
+    Common code for performing a search and returning the context for template
+    rendering.
+    
+    If a site_key was provided, the search will be limited to that particular
+    site.
+    '''
+    # TODO: scrub this first to avoid Solr injection attacks?
+    query = request.GET.get('q')
+    if not query:
+        redirect(settings.APP_URL_ROOT)
+    # this query string contains any special bits we add that we don't want
+    # the user to see.
+    int_query = query
+    if site_key:
+        int_query += ' AND siteKey:%s' % site_key
+    
+    # the page GET parameter is zero-based
+    page = int(request.GET.get('p', 0))
+    if page < 0: page = 0
+    
+    rows_per_page = int(request.GET.get('r', 10))
+    rows_per_page = (rows_per_page > 0) and rows_per_page or 10
+    
+    sort_args = {
+        'newest' : 'creationDate desc',
+        'votes' : 'votes desc',
+        'relevance' : 'score desc' # score is the special keyword for the
+                                   # relevancy score in Lucene
+    }
+    sort_by = request.GET.get('s', 'relevance').lower()
+    # default to sorting by relevance
+    if sort_by not in sort_args.keys():
+        sort_by = 'relevance'
+    
+    # perform search
+    results = solr_conn().search(int_query,
+                                 start=page*rows_per_page,
+                                 rows=rows_per_page,
+                                 sort=sort_args[sort_by])
+    decode_json_fields(results)
+    retrieve_users(results, question_only=True, ignore_comments=True)
+    retrieve_sites(results)
+    
+    context = { }
+    
+    # TODO: scrub this first to avoid HTML injection attacks?
+    context['query'] = query
+    context['results'] = results
+    context['total_hits'] = results.hits
+    context['current_page'] = page + 1 # page template var is ones-based
+    context['rows_per_page'] = rows_per_page
+    context['total_pages'] =  int(math.ceil(float(results.hits) / rows_per_page))
+    context['sort_by'] = sort_by
+    
+    return context
 
 # END VIEW HELPERS
 

python/src/stackdump/templates/base.html

@@ -24,7 +24,7 @@
                 
                 
                 <form method="get" action="{{ SETTINGS.APP_URL_ROOT }}{{ site_root_path }}search" class="pull-right"> 
-                    <input type="text" placeholder="Search" />
+                    <input type="text" placeholder="Search {% if site %}{{ site.name }}{% endif %}" />
                 </form> 
             </div> 
         </div> 

python/src/stackdump/templates/includes/results.html.inc

@@ -0,0 +1,59 @@
+<ul id="search-results">
+    {% for r in results %}
+    <li>
+        {% if not site %}
+        <div class="post-logo">
+            <img src="{{ SETTINGS.APP_URL_ROOT }}media/logos/{{ r.site.key }}.png" alt="{{ r.site.name }} logo" />
+        </div>
+        {% endif %}
+        <div class="post-stats-vertical">
+            <div class="post-stat">
+                <p class="post-stat-value {% if r.question.score < 0 %}post-stat-value-poor{% endif %}">
+                    {{ r.question.score }}
+                </p>
+                <p>vote{% if r.question.score != 1 %}s{% endif %}</p>
+            </div>
+            <div class="post-stat">
+                <p class="post-stat-value {% if r.answers|length == 0 %}post-stat-value-poor{% endif %}">
+                    {{ r.answers|length }}
+                </p>
+                <p>answer{% if r.answers|length != 1 %}s{% endif %}</p>
+            </div>
+        </div>
+        <div class="post-summary {% if not site %}post-summary-with-logo{% endif %}">
+            <h3><a href="#">{{ r.question.title }}</a></h3>
+            <p>{{ r.question.body|striptags|truncate(256) }}</p>
+            <p class="post-details">
+                Asked by <strong>{{ r.question.ownerUser.displayName }}</strong> on
+                <strong>{{ r.question.creationDate|format_datetime }}</strong>.
+            </p>
+            <div class="post-tags">
+                {% for t in r.question.tags %}
+                <span class="label">{{ t }}</span>
+                {% endfor %}
+            </div>
+        </div>
+        {# hack to force a clear on all internal elements #}
+        <div class="clearfix"></div>
+    </li>
+    {% endfor %}
+</ul>
+<div class="pagination">
+    <ul>
+        {% if current_page > 1 %}
+        {# the prev page is current_page - 2 because current_page is ones-based, but the p GET parameter is zero-based #}
+        <li class="prev"><a href="{{ REQUEST.url|set_get_parameters('p=' ~ (current_page - 2)) }}">&larr; Previous</a></li>
+        {% else %}
+        <li class="prev disabled"><a href="#">&larr; Previous</a></li>
+        {% endif %}
+        {% for p in range(1, total_pages + 1) %}
+        <li {% if p == current_page %}class="active"{% endif %}><a href="{{ REQUEST.url|set_get_parameters('p=' ~ (p-1)) }}">{{ p }}</a></li>
+    {% endfor %}
+    {% if current_page != total_pages %}
+    {# the next page is just current_page because current_page is ones-based, but the p GET parameter is zero-based #}
+    <li class="next"><a href="{{ REQUEST.url|set_get_parameters('p=' ~ current_page) }}">&rarr; Next</a></li>
+    {% else %}
+    <li class="next disabled"><a href="#">&rarr; Next</a></li>
+    {% endif %}
+    </ul>
+</div>

python/src/stackdump/templates/results.html

@@ -20,60 +20,7 @@
     </div>
     <div class="row">
         <div class="span12">
-            <ul id="search-results">
-                {% for r in results %}
-                <li>
-                    <div class="post-stats-vertical">
-                        <div class="post-stat">
-                            <p class="post-stat-value {% if r.question.score < 0 %}post-stat-value-poor{% endif %}">
-                                {{ r.question.score }}
-                            </p>
-                            <p>vote{% if r.question.score != 1 %}s{% endif %}</p>
-                        </div>
-                        <div class="post-stat">
-                            <p class="post-stat-value {% if r.answers|length == 0 %}post-stat-value-poor{% endif %}">
-                                {{ r.answers|length }}
-                            </p>
-                            <p>answer{% if r.answers|length != 1 %}s{% endif %}</p>
-                        </div>
-                    </div>
-                    <div class="post-summary">
-                        <h3><a href="#">{{ r.question.title }}</a></h3>
-                        <p>{{ r.question.body|striptags|truncate(256) }}</p>
-                        <p class="post-details">
-                            Asked by <strong>{{ r.question.ownerUser.displayName }}</strong> on
-                            <strong>{{ r.question.creationDate|format_datetime }}</strong>.
-                        </p>
-                        <div class="post-tags">
-                            {% for t in r.question.tags %}
-                            <span class="label">{{ t }}</span>
-                            {% endfor %}
-                        </div>
-                    </div>
-                    {# hack to force a clear on all internal elements #}
-                    <div class="clearfix"></div>
-                </li>
-                {% endfor %}
-            </ul>
-            <div class="pagination">
-                <ul>
-                    {% if current_page > 1 %}
-                    {# the prev page is current_page - 2 because current_page is ones-based, but the p GET parameter is zero-based #}
-                    <li class="prev"><a href="{{ REQUEST.url|set_get_parameters('p=' ~ (current_page - 2)) }}">&larr; Previous</a></li>
-                    {% else %}
-                    <li class="prev disabled"><a href="#">&larr; Previous</a></li>
-                    {% endif %}
-                    {% for p in range(1, total_pages + 1) %}
-                    <li {% if p == current_page %}class="active"{% endif %}><a href="{{ REQUEST.url|set_get_parameters('p=' ~ (p-1)) }}">{{ p }}</a></li>
-                {% endfor %}
-                {% if current_page != total_pages %}
-                {# the next page is just current_page because current_page is ones-based, but the p GET parameter is zero-based #}
-                <li class="next"><a href="{{ REQUEST.url|set_get_parameters('p=' ~ current_page) }}">&rarr; Next</a></li>
-                {% else %}
-                <li class="next disabled"><a href="#">&rarr; Next</a></li>
-                {% endif %}
-                </ul>
-            </div>
+        {% include 'includes/results.html.inc' %}
         </div>
         <div class="span4">
             <div class="total-hits">
@@ -92,7 +39,7 @@
                     <li>
                         <img src="{{ SETTINGS.APP_URL_ROOT }}media/logos/{{ s.key }}.png" alt="{{ s.name }} logo" />
                         <h6>
-                            <a href="{{ SETTINGS.APP_URL_ROOT }}{{ s.key }}/search?q={{ query }}">{{ s.name }}</a>
+                            <a href="{{ SETTINGS.APP_URL_ROOT }}{{ s.key }}/search?q={{ query }}&amp;s={{ sort_by }}">{{ s.name }}</a>
                             <small class="tagline">{{ s.dump_date }}</small>
                         </h6>
                     </li>

python/src/stackdump/templates/site_results.html

@@ -0,0 +1,58 @@
+{% extends 'base.html' %}
+
+{% block title %}Stackdump // {{ site.name }} search for "{{ query }}"{% endblock %}
+
+{% block body %}
+    <div class="row">
+        <div class="span16">
+            <form id="search" method="get" action="{{ SETTINGS.APP_URL_ROOT }}search">
+                <input type="text" class="xlarge" name="q" value="{{ query }}" />
+                <input type="hidden" name="s" value="{{ sort_by }}" />
+                <input type="submit" class="btn primary" value="Search{% if site %} {{ site.name }}{% endif %}" />
+            </form>
+            
+            <ul class="tabs">
+                <li {% if sort_by == 'newest' %}class="active"{% endif %}><a href="{{ REQUEST.url|set_get_parameters('s=newest', 'p=0') }}">newest</a></li>
+                <li {% if sort_by == 'votes' %}class="active"{% endif %}><a href="{{ REQUEST.url|set_get_parameters('s=votes', 'p=0') }}">votes</a></li>
+                <li {% if sort_by == 'relevance' %}class="active"{% endif %}><a href="{{ REQUEST.url|set_get_parameters('s=relevance', 'p=0') }}">relevance</a></li>
+            </ul>
+        </div>
+    </div>
+    <div class="row">
+        <div class="span12">
+        {% include 'includes/results.html.inc' %}
+        </div>
+        <div class="span4">
+            <div class="total-hits">
+                <p>
+                    <span class="stat-value">{{ total_hits }}</span>
+                </p>
+                <p>
+                    posts matched your query <em>"{{ query }}"</em>.
+                </p>
+            </div>
+            
+            <div class="well site-list-container">
+                <h3>Search other sites</h3>
+                <ul class="site-list">
+                    {% for s in sites %}
+                    {# don't show the current site #}
+                    {% if s.key != site.key %}
+                    <li>
+                        <img src="{{ SETTINGS.APP_URL_ROOT }}media/logos/{{ s.key }}.png" alt="{{ s.name }} logo" />
+                        <h6>
+                            <a href="{{ SETTINGS.APP_URL_ROOT }}{{ s.key }}/search?q={{ query }}&amp;s={{ sort_by }}">{{ s.name }}</a>
+                            <small class="tagline">{{ s.dump_date }}</small>
+                        </h6>
+                    </li>
+                    {% endif %}
+                    {% endfor %}
+                </ul>
+            </div>
+            
+            <div class="well">
+                <a href="{{ SETTINGS.APP_URL_ROOT }}search?q={{ query }}&amp;s={{ sort_by }}">Search across all sites.</a>
+            </div>
+        </div>
+    </div>
+{% endblock %}