- add phrase splitter

- add test harness for phrase splitter - use boolean + and phrases in searchabletext criterion. git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@5170 c91229c3-7414-0410-bfa2-8a42b809f60b

- add phrase splitter
- add test harness for phrase splitter - use boolean + and phrases in searchabletext criterion. git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/trunk@5170 c91229c3-7414-0410-bfa2-8a42b809f60b
Brad Shuttleworth
1 parent bd981125
Showing 3 changed files with 82 additions and 1 deletions
lib/browse/Criteria.inc
lib/util/ktutil.inc
tests/util/ktutil/testPhraseSplit.php
@@ -609,7 +609,20 @@ class SearchableTextCriterion extends BrowseCriterion {
  
         $p = array();
         $p[0] = "MATCH(DST.document_text) AGAINST (? $boolean_mode)";
-        $p[1] = $aRequest[$this->getWidgetBase()];
+        
+        $RESTRICTING_SEARCH = true;
+        
+        if ($RESTRICTING_SEARCH) {
+            // FIXME use a more reasonable approach for splitting this...
+            $q_set = KTUtil::phraseSplit($aRequest[$this->getWidgetBase()]);
+            $temp = $q_set;
+            foreach ($temp as $k => $v) { 
+                $q_set[$k] = sprintf('+"%s"', $v);
+            }
+            $p[1] = implode(' ',$q_set);
+        } else {
+            $p[1] = $aRequest[$this->getWidgetBase()]; 
+        }
         return $p;
     }
  
@@ -540,6 +540,34 @@ class KTUtil {
         return ((float) $microtime_simple[1] + (float) $microtime_simple[0]);
     }
  
+    function phraseSplit($sSearchString) {
+        $a = preg_split('#"#', $sSearchString);
+        $i = 0;
+        $phrases = array();
+        $word_parts = array();
+        foreach ($a as $part) {
+            if ($i%2 == 0) {
+                $word_parts[] = $part;
+            } else {
+                $phrases[] = $part;
+            }
+            $i += 1;
+        }
+        
+        $words = array();
+        foreach ($word_parts as $part) {
+            $w = (array) explode(' ', $part);
+            foreach ($w as $potential) { if (!empty($potential)) { $words[] = $potential; }}
+        }
+        
+        
+        return array(
+            'words' => $words,
+            'phrases' => $phrases,
+        );
+
+    }
+
 }
  
 ?>
+<?php
+
+require_once('../../../config/dmsDefaults.php');
+require_once(KT_LIB_DIR . '/util/ktutil.inc');
+
+$tests = array(
+    // (string, phrases, words)
+    array('a"b c"d', array('b c'), array('a','d')), 
+);
+
+
+foreach ($tests as $t) {
+    print '<pre>';
+    
+    $test = $t[0];
+    $phrases = $t[1];
+    $words = $t[2];
+    
+    
+    $p_expect = implode(', ', $phrases);
+    $w_expect = implode(', ',$words);
+    
+    $res = KTUtil::phraseSplit($test);
+    
+    $p_got = implode(', ', $res['phrases']);
+    $w_got = implode(', ', $res['words']);
+
+    
+    if (($w_got == $w_expect) && ($p_got == $p_expect)) {
+        print sprintf("Passed: %s\n", $test);
+    } else {       
+        print "--------\n";
+        print sprintf("failed: %s\n", $test); 
+        print sprintf("Phrases -  got \"%s\", expected \"%s\"\n", $p_got, $p_expect);
+        print sprintf("Words -  got \"%s\", expected \"%s\"\n", $w_got, $w_expect);
+        print "--------\n";
+    }
+}
+
+?>
 \ No newline at end of file