From f6cf4339d9699a779d4f6700c3f34e1c21136cde Mon Sep 17 00:00:00 2001
From: Robert Sesek <rsesek@bluestatic.org>
Date: Sat, 22 Jan 2005 02:29:38 +0000
Subject: [PATCH] Broken initial XML parser. Just getting it in here so we can
 preserve debug stuff.

---
 xml.php | 224 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 224 insertions(+)
 create mode 100644 xml.php

diff --git a/xml.php b/xml.php
new file mode 100644
index 0000000..42e5aa2
--- /dev/null
+++ b/xml.php
@@ -0,0 +1,224 @@
+<?php
+/*=====================================================================*\
+|| ################################################################### ||
+|| # Iris Studios Shared Object Framework [#]version[#]
+|| # --------------------------------------------------------------- # ||
+|| # All parts of this file are ©2003-[#]year[#] Iris Studios, Inc. No     # || 
+|| # part of this file may be reproduced in any way: part or whole.  # ||
+|| # --------------------------------------------------------------- # ||
+|| # ©2003 - [#]year[#] Iris Studios, Inc. | http://www.iris-studios.com   # ||
+|| ################################################################### ||
+\*=====================================================================*/
+
+$OBJECT = 'XML Parser';
+$CLASS = 'XML_Parser';
+$OBJ = 'xml';
+
+/**
+* XML Parser
+*
+* This framework is a wrapper for a robust XML parser.
+*
+* @author		Iris Studios, Inc.
+* @copyright	Copyright ©2003 - [#]year[#], Iris Studios, Inc.
+* @version		$Revision$
+* 
+*/
+class XML_Parser
+{
+	/**
+	* Global environment variables
+	*
+	* @var	parser		Current XML parser
+	* @var	taginfo		Array of all the processed tag data
+	* @var	taghandler	What to do for each tag encountered
+	* @var	trimdata	Trim CDATA elements?
+	* @var	attrdata	Data parsed from global attributes
+	* @var	tagid		Current tag ID in the document
+	* @var	tagname		Current tag name in the document
+	* @var	tree		Tag stack to use for proper nesting
+	* @var	parent		The tag's parent
+	*/
+	var $parser = null;
+	var $taginfo = array();
+	var $taghandler = array();
+	var $trimdata = true;
+	var $attrdata = array();
+	var $tagid = -1;
+	var $tagname = '';
+	var $tree = array();
+	var $parent = -1;
+	
+	/**
+	* Parse an XML file
+	*
+	* @param	str		XML file data
+	*
+	* @return	array	Array with all the XML data parsed
+	*/
+	function parse($data)
+	{
+		$this->parser = xml_parser_create();
+		xml_set_object($this->parser, $this);
+		xml_parser_set_option($this->parser, XML_OPTION_CASE_FOLDING, 0);
+		xml_set_element_handler($this->parser, '_tag_start', '_tag_end');
+		xml_set_character_data_handler($this->parser, '_cdata');
+		
+		if (!xml_parse($this->parser, $data))
+		{
+			$error['code'] = xml_get_error_code($this->parser);
+			$error['string'] = xml_error_string($error['code']);
+			$error['line'] = xml_get_current_line_number($this->parser);
+			$error['column'] = xml_get_current_column_number($this->parser);
+			print(sprintf("XML Error: %s (%d) at line %d colunn %d", $error['string'], $error['code'], $error['line'], $error['column']));
+			exit;
+		}
+		
+		xml_parser_free($this->parser);
+		
+		return array();
+	}
+	
+	/**
+	* Process the opening location of an XML tag
+	*
+	* @param	res		XML parser
+	* @param	str		Tag name
+	* @param	array	Tag attributes
+	*/
+	function _tag_start($parser, $name, $attributes)
+	{
+		// we're opening a new tag
+		$this->tagid++;
+		$this->tagname = $name;
+		
+		// copy all tag attributes
+		array_walk($attributes, 'trim');
+		$this->attrdata[ $this->tagid ] = $attributes;
+		
+		$this->parent = current($this->tree);
+		if ($this->parent === false)
+		{
+			$this->parent = -1;
+		}
+		// advance the parent counter because the tree is yet to be updated
+		else
+		{
+			$this->parent++;
+		}
+		
+		// initialize the data set
+		$this->taginfo["$name"][ $this->parent ][ $this->tagid ] = '';
+		
+		// add the current tag into the tree
+		$this->tree[] = $this->tagid;
+	}
+	
+	/**
+	* Process XML CDATA
+	*
+	* @param	res		XML parser
+	* @param	str		CDATA from tag
+	*/
+	function _cdata($parser, $data)
+	{
+		global $_isso;
+		static $count;
+		
+		/*if (preg_replace('#(^[[:space:]]+|[[:space:]]+$)#', '', trim($data)) == '')
+		{
+			$count++;
+			$datanew = str_replace(array("\t", "\n", "\r", " "), array("{t}", "{n}", "{r}", "{s}"), $data);
+			$_isso->debug("strangechars[" . $this->tagid . "] = '$datanew'");
+			$data = $datanew . "||$count||";
+			//return;
+		}*/
+		
+		$_isso->debug("cdata[] = '$data'");
+		// read in the CDATA
+		if ($this->parent == $this->tagid)
+		{
+			if (!isset($this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ]))
+			{
+				$this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ] = '';
+			}
+			
+			$this->taginfo[ $this->tagname ][ $this->parent - 1 ][ $this->tagid ] .= $data;
+		}
+		else
+		{
+			$this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ] .= $data;
+		}
+	}
+	
+	/**
+	* Process the closing of an XML tag
+	*
+	* @param	res		XML parser
+	* @param	str		Tag name
+	*/
+	function _tag_end($parser, $name)
+	{
+		global $_isso;
+		
+		// fetch the data
+		if (isset($this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ]))
+		{
+			$data = $this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ];
+		}
+		else
+		{
+			return;
+		}
+		
+		if (trim($data) == '')
+		{
+			$_isso->debug("not going to bother with '" . $this->tagid . "'");
+		}
+		
+		// if we have a data handler, operate it now
+		if (isset($this->taghandler[ $this->tagname ]))
+		{
+			$_isso->debug("handler: {$this->taghandler[ " . $this->tagname . " ]}");
+			if (function_exists($this->taghandler[ $this->tagname ]))
+			{
+				$data = $this->taghandler[ $this->tagname ]($data);
+			}
+		}
+
+		// check for necessary trims
+		if ($this->trimdata)
+		{
+			// trim() doesn't take off the edge for more than a few white characters, preg_replace does; don't use [:space:] either...
+			//$data = preg_replace('#(^\s+|\s+$)#', '', $data);
+			//$data = preg_replace('#(^[[:space:]]+|[[:space:]]+$)#', '', trim($data));	
+			$data = preg_replace('#^\s+#', '', $data);
+			$data = preg_replace('#\s+$#', '', $data);
+			$_isso->debug("trimming data ('" . $this->tagid . "'): '" . $data . "'");
+			//$data = trim($data);
+		}
+		
+		// don't want blank fields
+		if (trim($data) != '')
+		{
+			$this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ] = $data;
+		}
+		else
+		{
+			unset($this->taginfo[ $this->tagname ][ $this->parent ][ $this->tagid ]);
+		}
+		
+		$_isso->debug("tree[] = " . current($this->tree));
+		
+		// make sure that we're always at the end of the tree
+		end($this->tree);
+	}
+}
+
+/*=====================================================================*\
+|| ###################################################################
+|| # $HeadURL$
+|| # $Id$
+|| ###################################################################
+\*=====================================================================*/
+?>
\ No newline at end of file
-- 
2.43.5