Changeset 31624

Timestamp:

2017-04-21T01:02:28+12:00 (7 years ago)

Author:

davidb

Message:

Combined volume md and full-text page searching

Location:

other-projects/hathitrust/wcsa/extracted-features-solr/trunk/web-portal

Files:

• index.html (modified) (2 diffs)
• index.js (modified) (7 diffs)

other-projects/hathitrust/wcsa/extracted-features-solr/trunk/web-portal/index.html

@@ -173 +173 @@
         <input type="hidden" id="rows" name="rows" value="20"/>
 
+        <!--
         <input type="text"   class="search-in-progress" id="q" name="q" style="width: 600px;"/>
-        <input type="submit" class="search-in-progress" id="search-submit" />
+        <input type="submit" class="search-in-progress" id="search-submit" />       
         <br>
-          <input type="checkbox" id="group-results-by-id" name="group-results-by-vol" value="on">Group by Volume
-          <input type="checkbox" id="search-all-langs"    name="search-all-langs"     value="on">Search all Languages
+        -->
         
-        <div id="pos-fieldsets">
-        </div>      
-
-        <fieldset class="show-hide-lang">
-          <legend style="margin-bottom: 5px; padding-top: 15px;">
-            Other Languages (without modelled parts-of-speech)
-          </legend>
-          <div id="other-langs">
-          </div>
-        </fieldset>
-
+        <table class="query-table">
+          <tr>
+            <td>Volume Metadata:</td>
+            <td><input type="text"   class="search-in-progress" id="vq" name="vq"
+                   placeholder="metaname:metavalue e.g., title_t:farming"
+                   style="width: 600px;"/><br />
+              <input type="checkbox" id="search-all-vfields" name="search-all-vfields" value="on">&nbsp;Search all volume fields
+            </td>
+          </tr>
+          <tr>
+            <td></td>
+            <td style="color: #aaaaaa; width: 400px; font-size: 14px; line-height: 16px; padding-bottom: 10px;">
+              Indexed volume metadata fields: accessProfile_t, genre_t,
+              imprint_t, isbn_t, issn_t, issuance_t,
+              language_t, lccn_t, names_t, oclc_t,
+              pubPlace_t, pubDate_t, rightsAttributes_t,
+              title_t, typeOfResource_t
+            </td>
+          </tr>
+          <tr>
+            <td>Page-level text:</td>
+            <td>
+              <input type="text"   class="search-in-progress" id="q" name="q"
+                 style="width: 600px;"/><br />
+              <input type="checkbox" id="group-results-by-id" name="group-results-by-vol" value="on">&nbsp;<span>Group by Volume</span>
+
+            </td>
+          </tr>
+          <tr>
+            <td></td>
+            <td >
+              <input type="submit" class="search-in-progress" id="search-submit" style="float: right;"/>
+            </td>
+          </tr>
+        </table>
+
+                
+
+        <div>
+          <input type="checkbox" id="search-all-langs"    name="search-all-langs"     value="on">&nbsp;Search all Languages
+        
+          <div id="pos-fieldsets">
+          </div>        
+          
+          <fieldset class="show-hide-lang">
+            <legend style="margin-bottom: 5px; padding-top: 15px;">
+              Other Languages (without modelled parts-of-speech)
+            </legend>
+            <div id="other-langs">
+            </div>
+          </fieldset>
+        </div>
+        
           </form>
 
@@ -215 +257 @@
     <li><a href="admin.html">HTRC-EF Solr Admin UI (log in)</a></li>
     <li><a href="luke_lang_mappings.html">Dynamically Generated Language Map (Part-of-Speech, and Other)</a></li>
-    <!--
-    <li><a href="/solr/htrc-full-ef_shard6_replica1/admin/luke">Luke XML on shard</a></li>
--->
     <li><a href="/solr/htrc-full-ef20_shard6_replica1/admin/luke">Luke XML on shard</a></li>
       </ul>

other-projects/hathitrust/wcsa/extracted-features-solr/trunk/web-portal/index.js

@@ -72 +72 @@
     var page = id_pages[pi];
     
-    var seqnum = page;
+    var seqnum = (page==0) ? 1 : page;
     var babel_url = "https://babel.hathitrust.org/cgi/pt?id="+id+";view=1up;seq="+seqnum;
 
@@ -81 +81 @@
         html_item += '<span style="font-style: italic;" name="'
                      +id+'"><span style="cursor: progress;">Loading ...</span></span><br>';
-
-        html_item += id + ': <a target="_blank" href="' + babel_url + '">seq ' + seqnum + '</a> ';
+        if (page > 0) {
+            html_item += id + ': <a target="_blank" href="' + babel_url + '">seq ' + seqnum + '</a> ';
+        }
+        else {
+            // skip linking to the 'phony' page 0
+            html_item += id;
+        }
         }
         else {
@@ -92 +97 @@
         html_item += ' <span style="font-style: italic;" name="'
                  +id+'"><span style="cursor: progress;">Loading ...</span></span><br>';
-        html_item += '<a target="_blank" href="' + babel_url + '">' + id + ', seq ' + seqnum + '</a>';
+
+        if (page > 0) {
+        html_item += '<a target="_blank" href="' + babel_url + '">' + id + ', seq ' + seqnum + '</a>';
+        }
+        else {
+        html_item += '<a target="_blank" href="' + babel_url + '">' + id + ', all pages</a>';
+        }
+        
         html_item += '</p>';
     }
@@ -141 +153 @@
     var id_and_page = doc.id.split(".page-");
     var id = id_and_page[0];
-    var seqnum = parseInt(id_and_page[1]) +1; // fix up ingest error
+    var seqnum;
+    if (id_and_page.length>1) {
+        seqnum = parseInt(id_and_page[1]) +1; // fix up ingest error
+    }
+    else {
+        seqnum = 0;
+    }
     var page = seqnum;
 
@@ -195 +213 @@
 
 var group_by_vol_checked = 0;
+
+function expand_vfield(q_term,all_vfields)
+{
+    var vfields = [];
+    var metadata_fields = [  "accessProfile_t", "genre_t", "imprint_t", "isbn_t", "issn_t",
+                 "issuance_t", "language_t", "lccn_t", "names_t", "oclc_t",
+                 "pubPlace_t", "pubDate_t", "rightsAttributes_t", "title_t", "typeOfResource_t"  ];
+    
+    if (all_vfields) {
+    for (var fi=0; fi<metadata_fields.length; fi++) {
+        var vfield = metadata_fields[fi];
+        vfields.push(vfield+":"+q_term);
+    }
+    }
+    else {
+    if (q_term.match(/:/)) {
+        vfields.push(q_term);
+    }
+    else {
+        // make searching by title the default
+        vfields.push("title_t:"+q_term);
+    }
+    }
+    
+
+    var vfields_str = vfields.join(" OR ");
+    
+    return vfields_str;
+}
+
+function expand_vquery_field_and_boolean(query,all_vfields)
+{
+    // boolean terms
+    //  => pos and lang field
+    if (query === "") { return "" }
+    
+    var query_terms = query.split(/\s+/);
+    var query_terms_len = query_terms.length;
+
+    var bool_query_term = [];
+
+    var i=0;
+    var prev_bool = "";
+
+    for (var i=0; i<query_terms_len; i++) {
+    var term = query_terms[i];
+    if (term.match(/^(and|or)$/i)) {
+        prev_bool = term.toUpperCase();     
+    }
+    else {
+        if (i>0) {
+        if (prev_bool == "") {
+            prev_bool = "AND";
+        }
+        }
+        
+        var expanded_term = expand_vfield(term,all_vfields); // **** only difference to POS version
+        
+        term = "(" + expanded_term + ")";
+
+        if (prev_bool != "") {
+        bool_query_term.push(prev_bool);
+        prev_bool = "";
+        }
+        bool_query_term.push(term);
+    }
+    }
+
+    var bool_query = bool_query_term.join(" ");
+
+    return bool_query;
+}
 
 
@@ -242 +332 @@
     // boolean terms
     //  => pos and lang field
+    if (query === "") { return "" }
     
     var query_terms = query.split(/\s+/);
@@ -293 +384 @@
     var arg_start  = $('#start').attr('value');
     var arg_rows   = $('#rows').attr('value');
+
     var q_text     = $('#q').val();
-
+    var vq_text     = $('#vq').val();
+    
     group_by_vol_checked = $('#group-results-by-id:checked').length;
-    var search_all_checked = $('#search-all-langs:checked').length;
-    
-    if (q_text === "") {
+
+    var search_all_langs_checked = $('#search-all-langs:checked').length;
+    var search_all_vfields_checked = $('#search-all-vfields:checked').length;
+    
+    if ((q_text === "") && (vq_text === "")) {
     $('.search-in-progress').css("cursor","auto");
-    alert("No query term entered");
+    alert("No query term(s) entered");
     return;
     }
 
-    arg_q = expand_query_field_and_boolean(q_text,langs_with_pos,langs_without_pos,search_all_checked);
-
+    arg_vq = expand_vquery_field_and_boolean(vq_text,search_all_vfields_checked);
+    
+    arg_q = expand_query_field_and_boolean(q_text,langs_with_pos,langs_without_pos,search_all_langs_checked);
+
+    //console.log("*** arg_vq = " + arg_vq);
+    //console.log("*** arg_q = " + arg_q);
+    
     if (arg_q == "") {
-    $('.search-in-progress').css("cursor","auto");
-    alert("No languages selected");
-    return;
-    }
-    
+    if (arg_vq == "") {
+        // arg_vq was empty to start with, but attempt to expand non-empty arg_q
+        //   lead to an empty arg_q being returned
+        $('.search-in-progress').css("cursor","auto");
+        alert("No languages selected");
+        return;
+    }
+    else {
+        arg_q = arg_vq;
+    }
+    }
+    else {
+    if (arg_vq != "") {
+        // join the two with an AND
+        arg_q = "(" + arg_vq + ")" + " OR " + "(" + arg_q + ")";
+
+        // also implies
+        group_by_vol_checked = true;
+    }
+    }
+
+    //console.log("*** NOW arg_q = " + arg_q);
     
     // Example search on one of the htrc-full-ef fields is: