WIP: table of contents search

openlibrary/core/fulltext.py

@@ -8,6 +8,57 @@
 from infogami import config
 from openlibrary.core.lending import get_availability_of_ocaids
 
+logger = logging.getLogger("openlibrary.inside")
+
+def toc_search_api(q, page=1, limit=50):
+    if not hasattr(config, 'plugin_inside'):
+        return {'error': 'Unable to prepare search engine'}
+    offset = (page - 1) * limit
+    params = {
+        'limit': limit,
+        'q': q,
+        'nofacets': 'true',
+        'olonly': 'true',
+        'fields': 'toc',
+        'from': offset,
+        'size': limit
+    }
+
+    toc_search_url = '%s?%s' % (
+        config.toc_search['toc_endpoint'],
+        urllib.urlencode(params, 'utf-8'))
+
+    try:
+        json_data = urllib2.urlopen(toc_search_url, timeout=30).read()
+    except:
+        return {'error': 'Unable to query toc search engine'}
+
+    try:
+        return simplejson.loads(json_data)
+    except:
+        return {
+            'error': 'Error converting toc search engine data to JSON'
+        }
+
+def toc_search(q, page=1, limit=50):
+    ia_results = toc_search_api(q, page=page, limit=limit)
+
+    if 'error' not in ia_results and ia_results['hits']:
+        hits = ia_results['hits'].get('hits', [])
+        ocaids = [hit['_source'].get('meta_identifier', '') for hit in hits]
+        availability = get_availability_of_ocaids(ocaids)                
+        if 'error' in availability:
+            return []
+        editions = web.ctx.site.get_many([
+            '/books/%s' % availability[ocaid].get('openlibrary_edition')
+            for ocaid in availability
+            if availability[ocaid].get('openlibrary_edition')])
+        for ed in editions:
+            idx = ocaids.index(ed.ocaid)
+            ia_results['hits']['hits'][idx]['edition'] = ed
+            ia_results['hits']['hits'][idx]['availability'] = availability[ed.ocaid]
+    return ia_results
+
 
 def fulltext_search_api(params):
     if not hasattr(config, 'plugin_inside'):
@@ -17,7 +68,6 @@ def fulltext_search_api(params):
 
     try:
         json_data = urllib2.urlopen(search_select, timeout=30).read()
-        logger = logging.getLogger("openlibrary.inside")
         logger.debug('URL: ' + search_select)
     except:
         return {'error': 'Unable to query search engine'}

openlibrary/macros/FulltextResults.html

@@ -9,10 +9,10 @@
   $ num_found = results['hits'].get('total', 0)
   <div id="searchResults">
     <ul id="siteSearch">
-      $for doc in hits:                
+      $for doc in hits:
         $if doc.get('edition'):
           $ snippet = macros.FulltextSnippet(query, doc=doc)
           $:macros.SearchResultsWork(doc['edition'], availability=doc['availability'], user=user, extra=snippet)
     </ul>
     $:macros.Pager(page, num_found)
-  </div>          
+  </div>

openlibrary/macros/TocResults.html

@@ -0,0 +1,18 @@
+$def with(query, results, page=1)
+
+$if results and results.get('hits'):
+  $ loans = ctx.user.get_loans() if ctx.user else []
+  $ waiting_loans = ctx.user.get_waitinglist() if ctx.user else []
+  $ user = {'loans': loans, 'waitlists': waiting_loans}
+
+  $ hits = results['hits'].get('hits', [])
+  $ num_found = results['hits'].get('total', 0)
+  <div id="searchResults">
+    <ul id="siteSearch">
+      $for doc in hits:                
+        $if doc.get('edition'):          
+          $ toc = macros.TocSnippet(query, doc=doc)
+          $:macros.SearchResultsWork(doc['edition'], availability=doc['availability'], user=user, extra=toc)
+    </ul>
+    $:macros.Pager(page, num_found)
+  </div>          

openlibrary/macros/TocSnippet.html

@@ -0,0 +1,9 @@
+$def with (q, doc=None)
+
+$ toc = doc['_source']['toc']
+<section class="fulltext-excerpts" aria-label="search results from: $doc['edition']['title']">
+<pre>
+$for sec in toc:
+  $sec.split(':', 1)[1]
+</pre>
+</section>

openlibrary/plugins/inside/code.py

@@ -9,11 +9,28 @@
 import web
 from infogami.utils import delegate
 from infogami.utils.view import render_template
-from openlibrary.core.fulltext import fulltext_search
+from openlibrary.core.fulltext import fulltext_search, toc_search
 
 
 RESULTS_PER_PAGE = 20
 
+class search_toc(delegate.page):
+
+    path = '/search/toc'
+
+    def GET(self):
+        search_start = time()  # should probably use a @timeit decorator
+        i = web.input(q='', page=1)
+        query = i.q
+        page = int(i.page)
+        results = toc_search(query, page=page, limit=RESULTS_PER_PAGE)
+        search_time = time() - search_start
+
+        page = render_template('search/toc.tmpl', query, results, search_time,
+                               page=page, results_per_page=RESULTS_PER_PAGE)
+        page.v2 = True  # page is mobile-first
+        return page
+
 class search_inside(delegate.page):
 
     path = '/search/inside'

openlibrary/plugins/worksearch/code.py

@@ -464,10 +464,6 @@ def GET(self):
 
         if i.get('ftokens') and ',' not in i.ftokens:
             token = i.ftokens
-            #if ftoken_db is None:
-            #    ftoken_db = dbm.open('/olsystem/ftokens', 'r')
-            #if ftoken_db.get(token):
-            #    raise web.seeother('/subjects/' + ftoken_db[token].decode('utf-8').lower().replace(' ', '_'))
 
         if i.get('wisbn'):
             i.isbn = i.wisbn

openlibrary/templates/search/toc.html

@@ -0,0 +1,32 @@
+$def with (q, results, search_time, page=1, results_per_page=20)
+
+$var title: Search Open Library for $q
+
+<div id="contentHead">
+    <h1>$_("Search Tables of Contents")</h1>
+</div>
+
+<div id="contentBody">
+
+  <form class="siteSearch searchInsideForm" action="/search/toc">
+    <input type="text" class="larger" name="q" value="$q"/>
+    <input type="submit" class="generic-button" value="$_('Search')">
+  </form>
+    $ num_found = 0
+    $if q:
+        $if results and 'error' not in results and results['hits']:
+            $ hits = results['hits'].get('hits', [])
+            $ num_found = results['hits'].get('total', 0)
+
+        $if 'error' in results:
+            <div class="searchResultsError">$results['error']</div>
+
+        $if not num_found:
+            <p class="sansserif red collapse">No hits for: <strong>$q</strong></span></p>
+
+        $else:
+            <p class="search-results-stats">About $commify(num_found) result$("s" if num_found != 1 else "") ($("%.2f" % search_time) seconds)</p>
+
+        $:macros.TocResults(q, results, page=page)
+</div>
+