Update search : textual search on the columns instead of token based search

twbackends/phpAPI/csv_div.php

@@ -93,14 +93,16 @@ foreach ($sims as $doc => $score) {
 }
 $htmlout .= "</ul>\n";
 
+// $output_mode defined in /twbackends/phpAPI/parameters_details.php
 if ($output_mode == "html") {
   echo '<br/><h4><font color="#0000FF"> Full text of top '.$nb_displayed."/".count($sims).' related publications:</font></h4>';
   echo $htmlout;
 }
 else {
   echo json_encode(array(
-    'hits' => $jsonout,
-    'nhits' => $nb_displayed
+    'hits' => $jsonout, // documents content
+    'nhits' => $nb_displayed, // number of documents displayed
+    'ntotal' => $totalfound // total number of matching documents found
   ));
 }
 

twbackends/phpAPI/info_div.php

@@ -99,14 +99,58 @@ else {
     // DO THE SEARCH
     // -------------
     $searchcols = $my_conf["node".$ntid][$dbtype]['qcols'];
+    //var_dump($searchcols);
 
-    // a - split the query
-    $qtokens = preg_split('/[\p{Z}\p{P}\p{C}]+/u', $_GET["query"]);
-
-    // b - compute freq similarity per doc
     $sims = array();
 
-    // for each token
+    $totalfound=0;
+
+    // ------------   Textual search -------------
+
+    // Format input search string
+    $searchinput = $_GET["query"]; // ex : "[\"medical device\"]"
+    $searchinput=ltrim($searchinput,"[\"");
+    $searchinput=rtrim($searchinput,"\"]");
+    //echodump("Search input :", $searchinput); // ex "medical device"
+    $searchpattern="/".$searchinput."/i"; // ex "/medical device/i"
+    //echodump("Search pattern :", $pattern);
+
+    // Search in $base occurences of $searchtext. Update $sims if found.
+    // $base is a 2D array (array(array).
+    // echodump("base", $base);
+    // The first line, or row is the first element in the $base array:
+    // echodump("base row 0", $base[0]);
+    // The column "abstract" for first row :
+    // echodump("base row 0, 'abstract' col:", $base[0]["abstract"]);
+
+    foreach ( $base  as $idrow => $row) {
+      // test search text with : $row["abstract"]
+
+      // search nb occurences. if > 1, update $sims
+      $res=0;
+      foreach ($searchcols as $idcol => $col){
+        $res=$res+preg_match_all($searchpattern,$row[$col],$out);
+      }
+      if ($res > 0) {
+        $docid = 'd'.$idrow;
+        //echo "doc ". $docid.", ".$res ." occurence(s) found \n";
+        $totalfound++;
+        // build sims array similar to what token search does
+        $sims[$docid] = $res;
+      }
+      //echo "id : ". $id . "\n";
+      //echo "Abstract : ". $row["abstract"]. "\n";
+    }
+    // echo "total doc found :". $totalfound."\n";
+
+    /*
+    // ------- Token based search search ------
+
+    // Split the query
+    $qtokens = preg_split('/[\p{Z}\p{P}\p{C}]+/u', $_GET["query"]);
+    // echodump("tockens :", $qtokens);
+
+    // for each token, compute freq similarity per doc
     for ($k=0 ; $k < count($qtokens) ; $k++) {
 
       $tok = $qtokens[$k];
@@ -148,6 +192,7 @@ else {
 
       }
     }
+    */
 
     // c - sorted score per doc
     //
@@ -159,13 +204,9 @@ else {
   }
 }
 
-
 // just to make sure we use the same conventional name at cache read/write
 function mem_entry_name($graphdbname) {
   return "twbackendmem/".$graphdbname."/csv_search_base";
 }
 
-
-
-
 ?>

twmain/extras_explorerjs.js

@@ -503,10 +503,14 @@ function getTopPapers(qWords, nodetypeId, chosenAPI, tgtDivId) {
         data: urlParams,
         contentType: "application/json",
         success : function(data){
-          cbDisplay(data.hits)
+          // debug
+          // console.log(data);
+          cbDisplay(data.hits);
         },
         error: function(){
           console.log(`Not found: relatedDocs for ${apiurl}`)
+          // debug 
+          // console.log(apiurl + '/info_div.php?'+urlParams)
           cbDisplay([{ "error": stockErrMsg }])
         }
     });