From 8614a7f979e3166bff518a92240c4feae0c08ca9 Mon Sep 17 00:00:00 2001
From: gggeek <giunta.gaetano@gmail.com>
Date: Mon, 16 Jan 2023 14:54:34 +0000
Subject: [PATCH] convert data to non-utf8 internalencodings if possible
---
src/Helper/XMLParser.php | 106 ++++++++++++++++++++++++++++++---------
src/Request.php | 14 ++----
src/Server.php | 20 +++-----
3 files changed, 95 insertions(+), 45 deletions(-)
diff --git a/src/Helper/XMLParser.php b/src/Helper/XMLParser.php
index 98b94297..a931e664 100644
--- a/src/Helper/XMLParser.php
+++ b/src/Helper/XMLParser.php
@@ -11,8 +11,7 @@ use PhpXmlRpc\Value;
*
* @todo implement an interface to allow for alternative implementations
* - make access to $_xh protected, return more high-level data structures
- * - move $this->accept, $this->callbacks to an internal-use parsing-options config, along with the private
- * parts of $_xh
+ * - move the private parts of $_xh to the internal-use parsing-options config
* - add parseRequest, parseResponse, parseValue methods
* @todo if iconv() or mb_string() are available, we could allow to convert the received xml to a custom charset encoding
* while parsing, which is faster than doing it later by going over the rebuilt data structure
@@ -91,14 +90,15 @@ class XMLParser
'EX:NIL' => array('VALUE'), // only used when extension activated
);
- /** @var int[] $parsing_options */
+ /** @var array $parsing_options */
protected $parsing_options = array();
+
/** @var int $accept self::ACCEPT_REQUEST | self::ACCEPT_RESPONSE by default */
- protected $accept = 3;
+ //protected $accept = 3;
/** @var int $maxChunkLength 4 MB by default. Any value below 10MB should be good */
protected $maxChunkLength = 4194304;
- /** @var \Callable[] */
- protected $callbacks = array();
+ /** @var array */
+ protected $current_parsing_options = array();
public function getLogger()
{
@@ -158,39 +158,51 @@ class XMLParser
return;
}
- $prevAccept = $this->accept;
- $this->accept = $accept;
+ //$prevAccept = $this->accept;
+ //$this->accept = $accept;
+ $this->current_parsing_options = array('accept' => $accept);
- $this->callbacks = array();
+ $mergedOptions = $this->parsing_options;
foreach ($options as $key => $val) {
+ $mergedOptions[$key] = $val;
+ }
+
+ foreach ($mergedOptions as $key => $val) {
if (is_string($key)) {
switch($key) {
+ case 'target_charset':
+ if (function_exists('mb_convert_encoding')) {
+ $this->current_parsing_options['target_charset'] = $val;
+ } else {
+ $this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ": 'target_charset' option is unsupported without mbstring");
+ }
+ break;
+
case 'methodname_callback':
- if (!is_callable($val)) {
+ if (is_callable($val)) {
+ $this->current_parsing_options['methodname_callback'] = $val;
+ } else {
//$this->_xh['isf'] = 4;
//$this->_xh['isf_reason'] = "Callback passed as 'methodname_callback' is not callable";
//return;
$this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ": Callback passed as 'methodname_callback' is not callable");
- } else {
- $this->callbacks['methodname'] = $val;
}
break;
+
default:
$this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ": unsupported option: $key");
}
- unset($options[$key]);
+ unset($mergedOptions[$key]);
}
}
// NB: we use '' instead of null to force charset detection from the xml declaration
$parser = xml_parser_create('');
- foreach ($this->parsing_options as $key => $val) {
- xml_parser_set_option($parser, $key, $val);
- }
- foreach ($options as $key => $val) {
+ foreach ($mergedOptions as $key => $val) {
xml_parser_set_option($parser, $key, $val);
}
+
// always set this, in case someone tries to disable it via options...
xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 1);
@@ -233,15 +245,15 @@ class XMLParser
}
} catch (\Exception $e) {
xml_parser_free($parser);
- $this->callbacks = array();
- $this->accept = $prevAccept;
+ $this->current_parsing_options = array();
+ //$this->accept = $prevAccept;
/// @todo should we set $this->_xh['isf'] and $this->_xh['isf_reason'] ?
throw $e;
}
xml_parser_free($parser);
- $this->callbacks = array();
- $this->accept = $prevAccept;
+ $this->current_parsing_options = array();
+ //$this->accept = $prevAccept;
}
/**
@@ -270,8 +282,9 @@ class XMLParser
/// there is only a single top level element in xml anyway
// BC
if ($acceptSingleVals === false) {
- $accept = $this->accept;
+ $accept = $this->current_parsing_options['accept'];
} else {
+ //trigger_error('using argument $acceptSingleVals is deprecated', E_USER_DEPRECATED);
$accept = self::ACCEPT_REQUEST | self::ACCEPT_RESPONSE | self::ACCEPT_VALUE;
}
if (($name == 'METHODCALL' && ($accept & self::ACCEPT_REQUEST)) ||
@@ -472,6 +485,11 @@ class XMLParser
$this->_xh['vt'] = Value::$xmlrpcString;
}
+ // in case there is charset conversion required, do it here, to catch both cases of string values
+ if (isset($this->current_parsing_options['target_charset']) && $this->_xh['vt'] === Value::$xmlrpcString) {
+ $this->_xh['vt'] = mb_convert_encoding($this->_xh['vt'], $this->current_parsing_options['target_charset'], 'UTF-8');
+ }
+
if ($rebuildXmlrpcvals > 0) {
// build the xmlrpc val out of the data received, and substitute it
$temp = new Value($this->_xh['value'], $this->_xh['vt']);
@@ -616,8 +634,8 @@ class XMLParser
$methodname = preg_replace('/^[\n\r\t ]+/', '', $this->_xh['ac']);
$this->_xh['method'] = $methodname;
// we allow the callback to f.e. give us back a mangled method name by manipulating $this
- if (isset($this->callbacks['methodname'])) {
- call_user_func($this->callbacks['methodname'], $methodname, $this, $parser);
+ if (isset($this->current_parsing_options['methodname_callback'])) {
+ call_user_func($this->current_parsing_options['methodname_callback'], $methodname, $this, $parser);
}
break;
@@ -840,4 +858,44 @@ class XMLParser
return false;
}
+
+ // BC layer
+
+ public function __set($name, $value)
+ {
+ //trigger_error('setting property Response::' . $name . ' is deprecated', E_USER_DEPRECATED);
+
+ switch ($name) {
+ case 'accept':
+ $this->current_parsing_options['accept'] = $value;
+ break;
+ default:
+ $trace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS);
+ trigger_error('Undefined property via __set(): ' . $name . ' in ' . $trace[0]['file'] . ' on line ' . $trace[0]['line'], E_USER_WARNING);
+ }
+ }
+
+ public function __isset($name)
+ {
+ //trigger_error('checking property Response::' . $name . ' is deprecated', E_USER_DEPRECATED);
+
+ switch ($name) {
+ case 'accept':
+ return isset($this->current_parsing_options['accept']);
+ default:
+ return false;
+ }
+ }
+
+ public function __unset($name)
+ {
+ switch ($name) {
+ case 'accept':
+ unset($this->current_parsing_options['accept']);
+ break;
+ default:
+ $trace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS);
+ trigger_error('Undefined property via __unset(): ' . $name . ' in ' . $trace[0]['file'] . ' on line ' . $trace[0]['line'], E_USER_WARNING);
+ }
+ }
}
diff --git a/src/Request.php b/src/Request.php
index 37cba2be..9f5ad18e 100644
--- a/src/Request.php
+++ b/src/Request.php
@@ -315,8 +315,8 @@ class Request
return new Response($data, 0, '', 'xml', $this->httpResponse);
}
+ /// @todo move this block of code into the XMLParser
if ($respEncoding != '') {
-
// Since parsing will fail if charset is not specified in the xml prologue,
// the encoding is not UTF8 and there are non-ascii chars in the text, we try to work round that...
// The following code might be better for mb_string enabled installs, but makes the lib about 200% slower...
@@ -328,21 +328,17 @@ class Request
if ($respEncoding == 'ISO-8859-1') {
$data = utf8_encode($data);
} else {
- $this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ': invalid charset encoding of received response: ' . $respEncoding);
+ $this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ': unsupported charset encoding of received response: ' . $respEncoding);
}
}
}
}
-
// PHP internally might use ISO-8859-1, so we have to tell the xml parser to give us back data in the expected charset.
// What if internal encoding is not in one of the 3 allowed? We use the broadest one, ie. utf8
- // This allows to send data which is native in various charset, by extending xmlrpc_encode_entities() and
- // setting xmlrpc_internalencoding
- if (!in_array(PhpXmlRpc::$xmlrpc_internalencoding, array('UTF-8', 'ISO-8859-1', 'US-ASCII'))) {
- /// @todo emit a warning
- $options = array(XML_OPTION_TARGET_ENCODING => 'UTF-8');
- } else {
+ if (in_array(PhpXmlRpc::$xmlrpc_internalencoding, array('UTF-8', 'ISO-8859-1', 'US-ASCII'))) {
$options = array(XML_OPTION_TARGET_ENCODING => PhpXmlRpc::$xmlrpc_internalencoding);
+ } else {
+ $options = array(XML_OPTION_TARGET_ENCODING => 'UTF-8', 'target_charset' => PhpXmlRpc::$xmlrpc_internalencoding);
}
$xmlRpcParser = $this->getParser();
diff --git a/src/Server.php b/src/Server.php
index e5871f08..9596c7e4 100644
--- a/src/Server.php
+++ b/src/Server.php
@@ -531,11 +531,11 @@ class Server
if ($this->response_charset_encoding == 'auto') {
$respEncoding = '';
if (isset($_SERVER['HTTP_ACCEPT_CHARSET'])) {
- // here we should check if we can match the client-requested encoding
- // with the encodings we know we can generate.
+ // here we should check if we can match the client-requested encoding with the encodings we know we can generate.
/// @todo we should parse q=0.x preferences instead of getting first charset specified...
$clientAcceptedCharsets = explode(',', strtoupper($_SERVER['HTTP_ACCEPT_CHARSET']));
// Give preference to internal encoding
+/// @todo if mbstring is enabled, we can support other charsets too! Add a method to the Encoder!
$knownCharsets = array(PhpXmlRpc::$xmlrpc_internalencoding, 'UTF-8', 'ISO-8859-1', 'US-ASCII');
foreach ($knownCharsets as $charset) {
foreach ($clientAcceptedCharsets as $accepted) {
@@ -583,14 +583,14 @@ class Server
{
// decompose incoming XML into request structure
+ /// @todo move this block of code into the XMLParser
if ($reqEncoding != '') {
// Since parsing will fail if
// - charset is not specified in the xml prologue,
// - the encoding is not UTF8 and
// - there are non-ascii chars in the text,
// we try to work round that...
- // The following code might be better for mb_string enabled installs, but
- // makes the lib about 200% slower...
+ // The following code might be better for mb_string enabled installs, but it makes the lib about 200% slower...
//if (!is_valid_charset($reqEncoding, array('UTF-8')))
if (!in_array($reqEncoding, array('UTF-8', 'US-ASCII')) && !XMLParser::hasEncoding($data)) {
if (function_exists('mb_convert_encoding')) {
@@ -599,21 +599,17 @@ class Server
if ($reqEncoding == 'ISO-8859-1') {
$data = utf8_encode($data);
} else {
- $this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ': invalid charset encoding of received request: ' . $reqEncoding);
+ $this->getLogger()->errorLog('XML-RPC: ' . __METHOD__ . ': unsupported charset encoding of received request: ' . $reqEncoding);
}
}
}
}
-
// PHP internally might use ISO-8859-1, so we have to tell the xml parser to give us back data in the expected charset.
// What if internal encoding is not in one of the 3 allowed? We use the broadest one, ie. utf8
- // This allows to send data which is native in various charset,
- // by extending xmlrpc_encode_entities() and setting xmlrpc_internalencoding
- if (!in_array(PhpXmlRpc::$xmlrpc_internalencoding, array('UTF-8', 'ISO-8859-1', 'US-ASCII'))) {
- /// @todo emit a warning
- $options = array(XML_OPTION_TARGET_ENCODING => 'UTF-8');
- } else {
+ if (in_array(PhpXmlRpc::$xmlrpc_internalencoding, array('UTF-8', 'ISO-8859-1', 'US-ASCII'))) {
$options = array(XML_OPTION_TARGET_ENCODING => PhpXmlRpc::$xmlrpc_internalencoding);
+ } else {
+ $options = array(XML_OPTION_TARGET_ENCODING => 'UTF-8', 'target_charset' => PhpXmlRpc::$xmlrpc_internalencoding);
}
// register a callback with the xml parser for when it finds the method name
$options['methodname_callback'] = array($this, 'methodNameCallback');
--
2.47.0