Core (fix): improved toUTF8 and getUTF8Char methods / Class (add): i…

…mproved camelCase method, added deCamelCase method / Reader (fix): improved auth and port handling / Resource (fix): nasty typo fixed in setProp method (thx to Stephane Corlosquet) / getFormat (fix): improved SPARQL results detection / TurtleParser (fix): improved WS/non-WS handling around CURIEs (thx to Stephane Corlosquet) / RDFJSONSerializer (fix): improved escaping (thx to Gareth Jones) / RDFXMLSerializer (fix): improved QName generation / RemoteStore (fix): memory leaked (thx to Paul Houle) / DeleteQueryHandler (fix): less greedy triple deletion when datatype or language are declared (thx to Nicholas Rawlings) / LoadQueryHandler (tweak): getOComp now always keeps some trailing chars for improved sorting of URIs / SelectQueryHandler (fix): improved cleanup of temp tables in case of unexpected MySQL errors
contextworks · Feb 23, 2010 · e699756 · e699756
1 parent 4e06cb5
commit e699756
Show file tree

Hide file tree

Showing 12 changed files with 87 additions and 55 deletions.
diff --git a/ARC2.php b/ARC2.php
@@ -6,13 +6,13 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2010-01-15
+ * @version 2010-02-23
 */
 
 class ARC2 {
 
   function getVersion() {
-    return '2010-01-15';
+    return '2010-02-23';
   }
 
   /*  */
@@ -120,14 +120,17 @@ function getPreferredFormat($default = 'plain') {
   /*  */
 
   function toUTF8($v) {
-    if (utf8_decode($v) == $v) return $v;
+    if (urlencode($v) === $v) return $v;
+    //if (utf8_decode($v) == $v) return $v;
 		$v = (strpos(utf8_decode(str_replace('?', '', $v)), '?') === false) ? utf8_decode($v) : $v;
     return preg_replace_callback('/([\x00-\xdf][\x80-\xbf]|[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3}|[\xf8-\xfb][\x80-\xbf]{4}|[\xfc-\xfd][\x80-\xbf]{5}|[^\x00-\x7f])/', array('ARC2', 'getUTF8Char'), $v);
   }
 
   function getUTF8Char($v) {
     $val = $v[1];
-    return (strlen(trim($val)) === 1) ? utf8_encode($val) : $val;
+    if (strlen(trim($val)) === 1) return utf8_encode($val);
+    if (preg_match('/^([\x00-\x7f])(.+)/', $val, $m)) return $m[1] . ARC2::toUTF8($m[2]);
+    return $val;
   }
 
   /*  */

diff --git a/ARC2_Class.php b/ARC2_Class.php
@@ -6,7 +6,7 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2009-12-08
+ * @version 2010-02-17
 */
 
 class ARC2_Class {
@@ -68,10 +68,16 @@ function m($name, $a = false, $default = false, $o = false) {/* call method */
 
   function camelCase($v, $lc_first = 0) {
     $r = ucfirst($v);
-    while (preg_match('/^(.*)[\-\_ ](.*)$/', $r, $m)) {
+    while (preg_match('/^(.*)[^a-z0-9](.*)$/si', $r, $m)) {
       $r = $m[1] . ucfirst($m[2]);
     }
-    return $lc_first ? strtolower(substr($r, 0, 1)) . substr($r, 1) : $r;
+    return $lc_first && !preg_match('/[A-Z]/', $r[1]) ? strtolower($r[0]) . substr($r, 1) : $r;
+  }
+
+  function deCamelCase($v, $uc_first = 0) {
+    $r = str_replace('_', ' ', $v);
+    $r = preg_replace('/([a-z])([A-Z])/e', '"\\1 " . strtolower("\\2")', $r);
+    return $uc_first ? ucfirst($r) : $r;
   }
 
   /*  */

diff --git a/ARC2_Reader.php b/ARC2_Reader.php
@@ -6,7 +6,7 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2009-12-08
+ * @version 2010-01-30
 */
 
 ARC2::inc('Class');
@@ -133,7 +133,7 @@ function setCredentials($url) {
       }
       /* add header */
       if ($auth) {
-        $this->setCustomHeaders('Authorization: ' . $auth);
+        $this->addCustomHeaders('Authorization: ' . $auth);
         break;
       }
     }
@@ -199,8 +199,9 @@ function getHTTPSocket($url, $redirs = 0) {
     else {
       $h_code = $http_mthd . ' ' . $this->v1('path', '/', $parts) . (($v = $this->v1('query', 0, $parts)) ? '?' . $v : '') . (($v = $this->v1('fragment', 0, $parts)) ? '#' . $v : '');
     }
+    $port_code = ($parts['port'] != 80) ? ':' . $parts['port'] : '';
     $h_code .= ' HTTP/1.0' . $nl.
-      'Host: ' . $parts['host'] . ':' . $parts['port'] . $nl .
+      'Host: ' . $parts['host'] . $port_code . $nl .
       (($v = $this->http_accept_header) ? $v . $nl : '') .
       (($v = $this->http_user_agent_header) && !preg_match('/User\-Agent\:/', $this->http_custom_headers) ? $v . $nl : '') .
       (($http_mthd == 'POST') ? 'Content-Length: ' . strlen($this->message_body) . $nl : '') .
@@ -223,7 +224,7 @@ function getHTTPSocket($url, $redirs = 0) {
           stream_context_set_option($context, 'ssl', $m[1], $v);
         }
       }
-      $s = stream_socket_client('ssl://' . $parts['host'] . ':' . $parts['port'], $errno, $errstr, $this->timeout, STREAM_CLIENT_CONNECT, $context);
+      $s = stream_socket_client('ssl://' . $parts['host'] . $port_code, $errno, $errstr, $this->timeout, STREAM_CLIENT_CONNECT, $context);
     }
     elseif ($parts['scheme'] == 'https') {
       $s = @fsockopen('ssl://' . $parts['host'], $parts['port'], $errno, $errstr, $this->timeout);
@@ -285,7 +286,7 @@ function getHTTPSocket($url, $redirs = 0) {
           $this->digest_auth = 1;
           return $this->getHTTPSocket($url);
         }
-        return $this->addError($error . ' "' . (!feof($s) ? trim(strip_tags(fread($s, 64))) . '..."' : ''));
+        return $this->addError($error . ' "' . (!feof($s) ? trim(strip_tags(fread($s, 128))) . '..."' : ''));
       }
       /* redirect */
       if ($this->v('redirect', 0, $h) && ($new_url = $this->v1('location', 0, $h))) {

diff --git a/ARC2_Resource.php b/ARC2_Resource.php
@@ -6,7 +6,7 @@
  * @license http://arc.semsol.org/license
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2009-11-26
+ * @version 2010-02-23
 */
 
 ARC2::inc('Class');
@@ -52,7 +52,7 @@ function setProp($p, $os, $s = '') {
     if (isset($os['value'])) $os = array($os);
     /* list of values */
     foreach ($os as $i => $o) {
-      if (!is_array($o)) $os[i] = array('value' => $o, 'type' => 'literal');
+      if (!is_array($o)) $os[$i] = array('value' => $o, 'type' => 'literal');
     }
     $this->index[$s][$this->expandPName($p)] = $os;
   }

diff --git a/ARC2_getFormat.php b/ARC2_getFormat.php
@@ -5,7 +5,7 @@
  * @author Benjamin Nowack <[email protected]>
  * @license http://arc.semsol.org/license
  * @package ARC2
- * @version 2009-10-16
+ * @version 2010-01-18
 */
 
 function ARC2_getFormat($v, $mtype = '', $ext = '') {
@@ -15,6 +15,7 @@ function ARC2_getFormat($v, $mtype = '', $ext = '') {
   $r = (!$r && preg_match('/\/rdf\+xml/', $mtype)) ? 'rdfxml' : $r;
   $r = (!$r && preg_match('/\/(x\-)?turtle/', $mtype)) ? 'turtle' : $r;
   $r = (!$r && preg_match('/\/rdf\+n3/', $mtype)) ? 'n3' : $r;
+  $r = (!$r && preg_match('/\/sparql-results\+xml/', $mtype)) ? 'sparqlxml' : $r;
   /* xml sniffing */
   if (
     !$r &&
@@ -23,7 +24,7 @@ function ARC2_getFormat($v, $mtype = '', $ext = '') {
     /* has an xmlns:* declaration or a matching pair of tags */
     (preg_match('/\sxmlns\:?/', $v) || preg_match('/\<([^\s]+).+\<\/\\1\>/s', $v)) &&
     /* not a typical ntriples/turtle/n3 file */
-    !preg_match('/\.\s*$/s', $v)
+    !preg_match('/[\>\"\']\s*\.\s*$/s', $v)
   ) {
     while (preg_match('/^\s*\<\?xml[^\r\n]+\?\>\s*/s', $v)) {
       $v = preg_replace('/^\s*\<\?xml[^\r\n]+\?\>\s*/s', '', $v);
@@ -41,7 +42,7 @@ function ARC2_getFormat($v, $mtype = '', $ext = '') {
     $r = (!$r && preg_match('/^\s*\<opml\s/s', $v)) ? 'opml' : $r;
     $r = (!$r && preg_match('/^\s*\<html[\s|\>]/is', $v)) ? 'html' : $r;
     $r = (!$r && preg_match('/^\s*\<sparql\s+[^\>]+http\:\/\/www\.w3\.org\/2005\/sparql\-results\#/s', $v)) ? 'sparqlxml' : $r;
-    $r = (!$r && preg_match('/^\s*\<[^\>]+http\:\/\/www\.w3\.org\/2005\/sparql\-results#/s', $v)) ? 'srx' : $r;    
+    $r = (!$r && preg_match('/^\s*\<[^\>]+http\:\/\/www\.w3\.org\/2005\/sparql\-results#/s', $v)) ? 'srx' : $r;
     $r = (!$r && preg_match('/^\s*\<[^\s]*RDF[\s\>]/s', $v)) ? 'rdfxml' : $r;
     $r = (!$r && preg_match('/^\s*\<[^\>]+http\:\/\/www\.w3\.org\/1999\/02\/22\-rdf/s', $v)) ? 'rdfxml' : $r;
 

diff --git a/parsers/ARC2_TurtleParser.php b/parsers/ARC2_TurtleParser.php
@@ -6,7 +6,7 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2009-11-16
+ * @version 2010-02-16
 */
 
 ARC2::inc('RDFParser');
@@ -831,7 +831,7 @@ function xPN_LOCAL($v) {
         $proceed = 1;
       }
       elseif ($r) {
-        if (($sub_r = $this->x('(\.)', $sub_v)) && !preg_match('/^\s/s', $sub_r[2])) {
+        if (($sub_r = $this->x('(\.)', $sub_v)) && !preg_match('/^[\s\}]/s', $sub_r[2])) {
           $r .= $sub_r[1];
           $sub_v = $sub_r[2];
         }

diff --git a/serializers/ARC2_RDFJSONSerializer.php b/serializers/ARC2_RDFJSONSerializer.php
@@ -1,11 +1,12 @@
 <?php
-/*
-homepage: http://arc.semsol.org/
-license:  http://arc.semsol.org/license
-
-class:    ARC2 RDF/JSON Serializer
-author:   Benjamin Nowack
-version:  2008-07-01 (Fix: Proper jsonEscape method, thx to Keith Alexander)
+/**
+ * ARC2 RDF/JSON Serializer
+ *
+ * @author Benjamin Nowack <[email protected]>
+ * @license http://arc.semsol.org/license
+ * @homepage <http://arc.semsol.org/>
+ * @package ARC2
+ * @version 2010-02-23
 */
 
 ARC2::inc('RDFSerializer');
@@ -39,12 +40,12 @@ function getTerm($v, $term = 's') {
         return $this->getTerm($v['value'], $term);
       }
       if (preg_match('/^\_\:/', $v['value'])) {
-        return '{ "value" : "' . $v['value']. '", "type" : "bnode" }';
+        return '{ "value" : "' . $this->jsonEscape($v['value']) . '", "type" : "bnode" }';
       }
-      return '{ "value" : "' . $v['value']. '", "type" : "uri" }';
+      return '{ "value" : "' . $this->jsonEscape($v['value']) . '", "type" : "uri" }';
     }
     /* literal */
-    $r = '{ "value" : "' . $this->jsonEscape($v['value']). '", "type" : "literal"';
+    $r = '{ "value" : "' . $this->jsonEscape($v['value']) . '", "type" : "literal"';
     $suffix = isset($v['datatype']) ? ', "datatype" : "' . $v['datatype'] . '"' : '';
     $suffix = isset($v['lang']) ? ', "lang" : "' . $v['lang'] . '"' : $suffix;
     $r .= $suffix . ' }';

diff --git a/serializers/ARC2_RDFXMLSerializer.php b/serializers/ARC2_RDFXMLSerializer.php
@@ -6,7 +6,7 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2009-12-03
+ * @version 2010-01-30
  * 
 */
 
@@ -79,7 +79,8 @@ function getTerm($v, $type) {
 
   function getPName($v, $connector = ':') {
     if ($this->default_ns && (strpos($v, $this->default_ns) === 0)) {
-      return substr($v, strlen($this->default_ns));
+      $pname = substr($v, strlen($this->default_ns));
+      if (!preg_match('/\//', $pname)) return $pname;
     }
     return parent::getPName($v, $connector);
   }

diff --git a/store/ARC2_RemoteStore.php b/store/ARC2_RemoteStore.php
@@ -5,7 +5,7 @@
  * @author Benjamin Nowack <[email protected]>
  * @license http://arc.semsol.org/license
  * @package ARC2
- * @version 2009-12-08
+ * @version 2010-02-23
 */
 
 ARC2::inc('Class');
@@ -144,16 +144,26 @@ function runQuery($q, $qt = '', $infos = '') {
     /* ask|load|insert|delete */
     if (in_array($qt, array('ask', 'load', 'insert', 'delete'))) {
       $bid = $parser->getBooleanInsertedDeleted();
-      switch ($qt) {
-        case 'ask': return $bid['boolean'];
-        default: return $bid;
+      if ($qt == 'ask') {
+        $r = $bid['boolean'];
+      }
+      else {
+        $r = $bid;
       }
     }
     /* select */
-    if (($qt == 'select') && !method_exists($parser, 'getRows')) return $resp;
-    if ($qt == 'select') return array('rows' => $parser->getRows(), 'variables' => $parser->getVariables());
+    elseif (($qt == 'select') && !method_exists($parser, 'getRows')) {
+      $r = $resp;
+    }
+    elseif ($qt == 'select') {
+      $r = array('rows' => $parser->getRows(), 'variables' => $parser->getVariables());
+    }
     /* any other */
-    return $parser->getSimpleIndex(0);
+    else {
+      $r = $parser->getSimpleIndex(0);
+    }
+    unset($parser);
+    return $r;
   }
 
   /*  */

diff --git a/store/ARC2_StoreDeleteQueryHandler.php b/store/ARC2_StoreDeleteQueryHandler.php
@@ -1,11 +1,12 @@
 <?php
-/*
-homepage: http://arc.semsol.org/
-license:  http://arc.semsol.org/license
-
-class:    ARC2 RDF Store DELETE Query Handler
-author:   Benjamin Nowack
-version:  2009-06-15 (Addition: cleanValueTables method)
+/**
+ * ARC2 RDF Store DELETE Query Handler
+ *
+ * @author Benjamin Nowack <[email protected]>
+ * @license http://arc.semsol.org/license
+ * @homepage <http://arc.semsol.org/>
+ * @package ARC2
+ * @version 2010-02-23
 */
 
 ARC2::inc('StoreQueryHandler');
@@ -103,8 +104,15 @@ function deleteTriples() {
         }
         else {
           $term_id = $this->getTermID($t[$term], $term);
-          $q .= $q ? ' AND ' : '';
-          $q .= 'T.' . $term . '=' . $term_id;
+          $q .= ($q ? ' AND ' : '') . 'T.' . $term . '=' . $term_id;
+          /* explicit lang/dt restricts the matching */
+          if ($term == 'o') {
+            $o_lang = $this->v1('o_lang', '', $t);
+            $o_lang_dt = $this->v1('o_datatype', $o_lang, $t);
+            if ($o_lang_dt) {
+              $q .= ($q ? ' AND ' : '') . 'T.o_lang_dt=' . $this->getTermID($o_lang_dt, 'lang_dt');
+            }
+          }
         }
       }
       if ($skip) {

diff --git a/store/ARC2_StoreLoadQueryHandler.php b/store/ARC2_StoreLoadQueryHandler.php
@@ -6,7 +6,7 @@
  * @license <http://arc.semsol.org/license>
  * @homepage <http://arc.semsol.org/>
  * @package ARC2
- * @version 2010-01-14
+ * @version 2010-02-17
 */
 
 ARC2::inc('StoreQueryHandler');
@@ -303,7 +303,10 @@ function getOComp($val) {
     //$val = substr(trim(preg_replace('/[\W\s]+/is', '-', strip_tags($val))), 0, 35);
     $re = $this->has_pcre_unicode ? '/[\PL\s]+/isu' : '/[\s\'\"\´\`]+/is';
     $val = trim(preg_replace($re, '-', strip_tags($val)));
-    $val = function_exists("mb_substr") ? mb_substr($val, 0, 35) : substr($val, 0, 35);
+    if (strlen($val) > 35) {
+      $fnc = function_exists("mb_substr") ? 'mb_substr' : 'substr';
+      $val = $fnc($val, 0, 17) . '-' . $fnc($val, -17);
+    }
     if ($this->strip_mb_comp_str) {
       $val = urldecode(preg_replace('/\%[0-9A-F]{2}/', '', urlencode($val)));
     }

diff --git a/store/ARC2_StoreSelectQueryHandler.php b/store/ARC2_StoreSelectQueryHandler.php
@@ -6,7 +6,7 @@
  * @license   http://arc.semsol.org/license
  * @homepage  <http://arc.semsol.org/>
  * @package   ARC2
- * @version   2010-01-04
+ * @version   2010-02-12
  *
 */
 
@@ -108,10 +108,10 @@ function createTempTable($q_sql) {
     $con = $this->store->getDBCon();
     $v = $this->store->getDBVersion();
     if ($this->cache_results) {
-      $tbl = $this->store->getTablePrefix() . '_Q' . md5($q_sql);
+      $tbl = $this->store->getTablePrefix() . 'Q' . md5($q_sql);
     }
     else {
-      $tbl = $this->store->getTablePrefix() . '_Q' . md5($q_sql . time() . uniqid(rand()));
+      $tbl = $this->store->getTablePrefix() . 'Q' . md5($q_sql . time() . uniqid(rand()));
     }
     if (strlen($tbl) > 64) $tbl = 'Q' . md5($tbl);
     $tmp_sql = 'CREATE TEMPORARY TABLE ' . $tbl . ' ( ' . $this->getTempTableDef($tbl, $q_sql) . ') ';
@@ -121,9 +121,7 @@ function createTempTable($q_sql) {
       return $this->addError(mysql_error($con));
     }
     mysql_unbuffered_query('INSERT INTO ' . $tbl . ' ' . "\n" . $q_sql, $con);
-    if ($er = mysql_error($con)) {
-      return $this->addError($er);
-    }
+    if ($er = mysql_error($con)) $this->addError($er);
     return $tbl;
   }