Ett 1200 facet search error

services/Search/SearchStructure.php

@@ -85,6 +85,10 @@ static function fromHash($hash)
         $c = __CLASS__;
         $obj = new $c(true);
         $obj->_fillFromHash($hash);
+
+        echo "\n";
+        print_r($obj);
+        echo "\n";
         return $obj;
     }
 
@@ -231,6 +235,8 @@ private function _fillFromHash($hash)
         } elseif (count($ss) == 1) {
 
             $index = $ss[0][0];
+            print_r('^^^^^^^^^^^');
+            print_r($index);
             if (isset($this->force_standard[$index]) && $this->force_standard[$index]) {
                 $this->use_dismax = false;
             }
@@ -909,6 +915,9 @@ static function displayStrip($v)
             }
             return "Between $start and $end";
         } else {
+            // Remove square brackets and double quote from facet display.
+            // Dec 2025 Note: this is just a UI string, so this replacement should be unnecessary.
+            // return $v;
             return preg_replace('/[\[\]\"]/', '', $v);
         }
 

sys/Solr.php

@@ -104,7 +104,11 @@ function simplesearch($ss, $start = 0, $limit = null, $raw = false) {
       $limit = isset($_REQUEST['pagesize']) ? $_REQUEST['pagesize'] : $configArray['Site']['itemsPerPage'];
     }
 
-
+    error_log("SearchStructure: " . print_r($ss) . "\n");
+    echo "\n";
+    print_r($ss);
+    echo "\n";
+
     // The initial query
     if ($ss->use_dismax) {
       $ss->action = 'edismax';
@@ -129,12 +133,12 @@ function simplesearch($ss, $start = 0, $limit = null, $raw = false) {
       $args = array_merge($args, $this->spellcheckComponents($ss));
     }
 
-
     if ($raw) {
       return $this->rawSolrSearch($args, $action);
     }
 
     // Otherwise...
+    print_r('22222222222222222222222222222222 ');
     $rv = $this->solrSearch($args, $action);
     return $rv;
   }
@@ -230,7 +234,8 @@ function facetlist($ss, $fields, $sort = 'index', $skip = 0, $limit = 20) {
       $values = $body['facet_counts']['facet_fields'][$field];
       $rv['values'][$field] = array();
 
-      // skip the hidden ones
+      // Filter out facet values that match the hidden pattern defined on config.ini.
+      // e.g. skip hlbgeneral = "hlb_both:^U\.S\. National and" to hide "U.S. National..." facets
       foreach ($values as $valcnt) {
         if (isset($hide, $hide[$field])) {
           foreach ($hide[$field] as $regexp) {
@@ -275,19 +280,25 @@ function dismaxSearchArguments($ss) {
     // Should just be on "lookfor" and "type"
     $tvb = isset($ss->search[0]) ? $ss->search[0] : array('all', '*:*');
     $type = $tvb[0];
+    print_r('**************type****************');
+    print_r($type);
+    print_r('******************************');
+    // $value is the search string
     $value = $tvb[1];
 
+    // If search is empty/whitespace-only, default to *:* (match-all)
     if (!preg_match('/\S/', $value)) {
       $value = '*:*';
     }
     // Get the yaml file
     $allspecs = yaml_parse_file('conf/dismaxsearchspecs.yaml');
 
     // If the type isn't set, back up to normal arguments
+    // Lianet's notes: $type is extracted from conf/dismaxsearchspecs.yaml so the function always return the args in searchArguments
 
     if (!isset($allspecs[$type])) {
       $args =  $this->searchArguments($ss);
-      // print_r($args);
+      print_r('***************I am here***************');
       return $args;
     }
 
@@ -309,6 +320,12 @@ function dismaxSearchArguments($ss) {
     $rv[] = array('qt', 'edismax');
     $rv[] = array('mm', $this->mm($spec, $ss));
 
+    echo "\n";
+    print_r('******************************');
+    print_r($rv);
+    print_r('******************************');
+    echo "\n";
+
     return array_merge($rv, $this->filterComponents($ss), $this->sortComponents($ss));
   }
 
@@ -334,10 +351,11 @@ function standardSearchComponents($ss) {
 
     $searchComponents = array();
 
+    // Lianet's notes: conf/searchspecs is the config used to build the Solr query.
     $specs = yaml_parse_file('conf/searchspecs.yaml');
     $query = '';
 
-    foreach ($ss->search as $tvb) { // Type, Value (keywords), Boolen AND or OR
+    foreach ($ss->search as $tvb) { // Type, Value (keywords), Boolean AND or OR
       $type = $tvb[0];
       $values = $this->build_and_or_onephrase($tvb[1]);
       $bool = isset($tvb[2]) ? $tvb[2] : false;
@@ -356,8 +374,8 @@ function standardSearchComponents($ss) {
       }
       $query .= "id:(" . implode(' OR ', $ss->extraIDs()) . ')';
     }
-
     $ids = $this->tagIDs($ss);
+    // Check if the query has content, otherwise use *:* to match all
     if (preg_match('/\S/', $query)) {
       $searchComponents[] = array('q', $query);
     }
@@ -370,15 +388,19 @@ function standardSearchComponents($ss) {
 
 
   /** Quote a filter value, skipping it if it starts with a '[' (and hence is assumed
-   * to be a range)
+   * to be a range). Detect date range
    **/
 
   function quoteFilterValue($v) {
     if (preg_match('/^\[/', $v)) {
       return $v;
     }
     else {
-      return '"' . $v . '"';
+      // Escape internal quotes before wrapping
+      // input: He said "hello, the output: He said \"hello
+      $escaped = str_replace('"', '\\"', $v);
+      // String ready to Solr "He said \"hello"
+      return '"' . $escaped . '"';
     }
   }
 
@@ -664,6 +686,9 @@ private function __buildQueryString($structure, $values, $joiner = "OR") {
           }
 
           if ($val == 'stdnum') {
+            // Extract standard number from asis input
+            // Strips leading 0s. Captures digits, dashes, dots: 978-0-123-45678-9
+            // e.g.  0000978-0-12-345678-9 → 978-0-12-345678-9
             if (preg_match('/^\s*0*([\d\-\.]+[xX]?).*$/', $values['asis'], $match)) {
               $stdnum = $match[1];
 //              $stdnum = preg_replace('/[\.\-]/', '', $stdnum);
@@ -675,6 +700,7 @@ private function __buildQueryString($structure, $values, $joiner = "OR") {
         if (!isset($values[$val]) || ($values[$val] == "")) {
           continue;
         }
+        // Lianet's notes: Escape the value for safe embedding in field:value syntax
         $sstring = $field . ':(' . $values[$val] . ')';
         if (isset($weight) && $weight > 0) {
           $sstring .= '^' . $weight;
@@ -688,7 +714,7 @@ private function __buildQueryString($structure, $values, $joiner = "OR") {
 
 
   /**
-   * Turn solr output into a record structure (which shouuld probably be its own class...)
+   * Turn solr output into a record structure (which should probably be its own class...)
    *
    * @param string $result The XML returned by solr
    * @param string $xslfile The path of the XSL file to use to convert the data
@@ -700,6 +726,7 @@ function _process($result, $xslfile = 'xsl/solr-convert.xsl') {
     global $configArray;
 
     if (is_string($result) && preg_match('/^<html/', $result)) {
+      // Detect if Solr returns an error page
       if (preg_match('/ParseException/', $result)) {
         $errorMsg = "Error+in+search+syntax";
       }
@@ -757,6 +784,7 @@ function _process($result, $xslfile = 'xsl/solr-convert.xsl') {
    * @return  array               Tokenized array
    * @access  public
    */
+  // TODO: refactor tokenizer to a single-pass parser instead of regex
   public function tokenizeInput($input) {
     // Tokenize on spaces and quotes
     //preg_match_all('/"[^"]*"|[^ ]+/', $input, $words);
@@ -803,6 +831,8 @@ public function tokenizeInput($input) {
    * @return  string                Fixed input
    * @access  public
    */
+
+  // Lianet's notes: Verify if this function could be used to validate the Solr query
   public function validateInput($input) {
     // Ensure wildcards are not at beginning of input
     if ((substr($input, 0, 1) == '*') ||
@@ -863,18 +893,43 @@ function exactmatcherify($str) {
    *
    * Given a lookfor string, clean it up, tokenize it, and
    * return a structure that includes AND, OR, and Phrase
-   * queries.
+   * queries. lookfor could be single or multi-word
    *
    * @param string $lookfor User's search string
    * @return  array   $values     Includes 'and', 'or', and 'onephrase' elements
    * @access  public
    */
-
+  // Lianet's notes: Check if is necessary to remove illegal characters
+  // TODO: Refactoring this function to avoid the different output
   public function build_and_or_onephrase($lookfor = null) {
     $values = array();
 
+    // Removing this characters if the string only contains any of these characters - len == 1
+    // If there are words and characters, escaped them
+    // Delete all these characters destroy the user intent without noticing the user. I'll block meaningless
+    // single-characters explicitly
     $illegal = array('.', '{', '}', '/', '!', ':', ';', '[', ']', '(', ')', '+ ', '&', '- ');
-    $lookfor = trim(str_replace($illegal, '', $lookfor));
+    // $lookfor = trim(str_replace($illegal, '', $lookfor));
+
+    // Reject the input if is exactly one character and if the character is ~ or /
+    // ~ alone is invalid Lucene syntax causing parsing error
+    // \ escape alone is syntactically incorrect and causes 500 errors because Lucene expects to escape something
+    //if (mb_strlen(trim($lookfor)) === 1 && preg_match('/^[~\\\\]$/', $lookfor)) {
+    //    return false;
+    //}
+
+    // 1. Normalize + trim
+    $lookfor = trim($lookfor);
+
+    // 2. Strip garbage-only input
+    if ($lookfor !== '' && !preg_match('/[\p{L}\p{N}]/u', $lookfor)) {
+        $lookfor = trim(str_replace($illegal, '', $lookfor));
+    }
+
+    // 3. Reject invalid single-char input
+    if (mb_strlen($lookfor) === 1 && preg_match('/^[~\\\\]$/', $lookfor)) {
+        return false;
+    }
 
     // Replace fancy quotes
     $lookfor = str_replace(array('“', '”'), '"', $lookfor);
@@ -897,14 +952,14 @@ public function build_and_or_onephrase($lookfor = null) {
 
     // Tokenize Input
     $tokenized = $this->tokenizeInput($lookfor);
-
     $values['onephrase'] = '"' . preg_replace('/"/', '', implode(' ', $tokenized)) . '"';
     $values['and'] = implode(' AND ', $tokenized);
     $values['or'] = implode(' OR ', $tokenized);
     $values['asis'] = $lookfor;
     $values['compressed'] = preg_replace('/\s/', '', $lookfor);
     $values['exactmatcher'] = $this->exactmatcherify($lookfor);
     $values['emstartswith'] = $values['exactmatcher'] . '*';
+
     return $values;
   }
 
@@ -918,7 +973,7 @@ public function build_and_or_onephrase($lookfor = null) {
    **/
 
   function solrSearch($args, $action = 'standard') {
-    $raw = $this->rawSolrSearch($args, $action);
+    $raw = $this->rawSolrSearch($args, $action); // This is the Solr output
     if (!PEAR::isError($raw)) {
       $processed = $this->_process($raw);
 
@@ -956,6 +1011,8 @@ protected function set_proper_action($action, $args) {
         return $action;
       }
     }
+    // Ensure a non-NULL return from non-edismax cases
+    return $action;
   }
 
   // Do we just want the IDs? Spit 'em out!
@@ -1006,12 +1063,18 @@ function rawSolrSearch($args, $action = 'standard') {
 
     $this->solr_connection->add($args);
 
+    print_r('======================');
+    print_r($args);
+
     # Just want a list of IDs? Produce it and die
     if (isset($_REQUEST['htid_list'])) {
       $this->print_out_list_of_ids($args);
       die();
     }
 
+    error_log("Solr action used: " . $action);
+    echo "Solr action used: " . $action;
+
     # Finally, we can deal with the normal case
     return $this->solr_connection->send();
   }
@@ -1112,13 +1175,48 @@ function mltesc($str) {
     return str_replace(array('(', ')','[', ']', '!', '&', ':', ';', '-', '/', '"'), '', $str);
   }
 
-
+  // TODO: Check this function for correctness
   function lucene_escape($str) {
     $pattern = '/(\+|-|&&|\|\||!|\(|\)|\{|}|\[|]|\^|"|~|\*|\?|:|\\\)/';
     $replace = '\\\$1';
     return preg_replace($pattern, $replace, $str);
   }
 
+  /**
+ * Strict escape for filter query values and explicit field:value fragments.
+ *
+ * Use for all fq values and any time you construct field:value with user data.
+ * This function:
+ * - Normalizes Unicode to NFC form
+ * - Removes control characters
+ * - Escapes backslash FIRST (critical ordering)
+ * - Escapes multi-char tokens (&&, ||)
+ * - Escapes all Lucene special characters: + - ! ( ) { } [ ] ^ " ~ * ? : /
+ *
+ * @param string $s Raw user input or value to escape
+ * @return string Safely escaped value for use in Solr fq or field:value
+ */
+function lucene_escape_fq(string $s): string {
+    // Normalize Unicode to composed form (NFC)
+    if (function_exists('normalizer_normalize')) {
+        $s = normalizer_normalize($s, Normalizer::FORM_C) ?: $s;
+    }
+
+    // Remove control characters (0x00-0x1F, 0x7F)
+    $s = preg_replace('/[\x00-\x1F\x7F]/u', '', $s);
+
+    // CRITICAL: Escape backslash FIRST to avoid double-escaping
+    $s = str_replace('\\', '\\\\', $s);
+
+    // Escape Lucene special characters character-by-character
+    $specials = ['+', '-', '!', '(', ')', '{', '}', '[', ']', '^', '"', '~', '*', '?', ':', '/', '&', '|'];
+    foreach ($specials as $c) {
+        $s = str_replace($c, '\\' . $c, $s);
+    }
+
+    return $s;
+}
+
   function getMoreLikeThis($record, $id, $max = 5) {
     global $configArray;
 
@@ -1168,7 +1266,7 @@ function getMoreLikeThis($record, $id, $max = 5) {
     $query .= ') NOT id:(' . $id . ')';
 
     $ss = new SearchStructure(true); // create a "blank" ss with just the filter queries
-
+    
     $args = array_merge(array(array('q', $query)), $this->filterComponents($ss));
     return $this->solrSearch($args);
   }