Merge pull request #105 from multnomah-county-it/dev

Added prepare_search function to clean up text for use in an ILSWS se…

src/Libilsws.php

@@ -996,6 +996,44 @@ public function describe_bib ($token = null)
         return $this->send_get("$this->base_url/catalog/bib/describe", $token, []);
     }
 
+    /**
+     * Removes accents, punctuation, and non-ascii characters to
+     * create search string acceptable to ILSWS
+     *
+     * @param  string $terms
+     * @return string $terms
+     */
+
+    public function prepare_search ($terms = null)
+    {
+        // Trim leading and trailing whitespace
+        $terms = trim($terms);
+
+        // Validate
+        $this->validate('terms', $terms, 's:256');
+
+        // Change utf8 letters with accents to ascii characters
+        setlocale(LC_ALL, "en_US.utf8");
+        $terms = iconv("utf-8", "ASCII//TRANSLIT", $terms);
+
+        // Remove boolean operators
+        $terms = preg_replace("/(\s+)(and|or|not)(\s+)/", ' ', $terms);
+
+        // Replace certain characters with a space
+        $terms = preg_replace("/[.:;?!,\/]/", ' ', $terms);
+
+        // Remove most punctuation and other unwanted characters
+        $terms = preg_replace("/[&+=><%\'\"\|\{\}\(\)]/", '', $terms);
+
+        // Remove internal non-printing characters
+        $terms = preg_replace('/[^\x20-\x7E]/','', $terms);
+
+        // Replace multiple spaces with a single space
+        $terms = preg_replace('/\s+/', ' ', $terms);
+
+        return $terms;
+    }
+
     /**
      * Search the catalog for bib records
      * 

test/search_bib.php

@@ -18,6 +18,11 @@
 // Connect and get token
 $token = $ilsws->connect();
 
+// Remove unwanted characters from search string
+$search = $ilsws->prepare_search($search);
+
+print "$search\n";
+
 /**
  * Search for a patron. If the $params array is empty or any item is omitted,
  * default values will be supplied as shown, with the exception of the