1<?php 2 /* 3 * dumpit5.php 4 * 5 * a command-line script which dumps the given HTML, PHP, ASP, XHTML, etc. 6 * file as it is represented in the document model. 7 * 8 * NOTE: Only works with tidy for PHP 5+, for tidy in 4.3.x, see dumpit.php 9 * 10 * By: John Coggeshall <john@php.net> 11 * 12 * Usage; php dumpit5.php <filename> 13 */ 14 15 $tidy = tidy_parse_file($_SERVER['argv'][1]); 16 17 /* Optionally you can do this here if you want to fix up the document */ 18 19 /* $tidy->clean_repair() */ 20 21 $tree = $tidy->root(); 22 dump_tree($tree); 23 echo "\n"; 24 25 function node_type($type) { 26 27 switch($type) { 28 29 case TIDY_NODETYPE_ROOT: return "Root Node"; 30 case TIDY_NODETYPE_DOCTYPE: return "DocType Node"; 31 case TIDY_NODETYPE_COMMENT: return "Comment Node"; 32 case TIDY_NODETYPE_PROCINS: return "ProcIns Node"; 33 case TIDY_NODETYPE_TEXT: return "Text Node"; 34 case TIDY_NODETYPE_START: return "Start Node"; 35 case TIDY_NODETYPE_END: return "End Node"; 36 case TIDY_NODETYPE_STARTEND: return "Start/End Node"; 37 case TIDY_NODETYPE_CDATA: return "CDATA Node"; 38 case TIDY_NODETYPE_SECTION: return "Section Node"; 39 case TIDY_NODETYPE_ASP: return "ASP Source Code Node"; 40 case TIDY_NODETYPE_PHP: return "PHP Source Code Node"; 41 case TIDY_NODETYPE_JSTE: return "JSTE Source Code"; 42 case TIDY_NODETYPE_XMLDECL: return "XML Declaration Node"; 43 default: return "Unknown Node"; 44 } 45 } 46 47 function do_leaf($string, $indent) { 48 for($i = 0; $i < $indent; $i++) { 49 echo " "; 50 } 51 echo $string; 52 } 53 54 function dump_tree(tidyNode $node, $indent = 0) { 55 56 /* Put something there if the node name is empty */ 57 $nodename = trim(strtoupper($node->name)); 58 $nodename = (empty($nodename)) ? "[EMPTY]" : $nodename; 59 60 /* Generate the Node, and a pretty name for it */ 61 do_leaf(" + $nodename (".node_type($node->type).")\n", $indent); 62 63 /* Check to see if this node is a text node. Text nodes are 64 generated by start/end tags and contain the text in between. 65 i.e. <B>foo</B> will create a text node with $node->value 66 equal to 'foo' */ 67 if($node->type == TIDY_NODETYPE_TEXT) { 68 do_leaf(" |\n", $indent); 69 do_leaf(" +---- Value: '{$node->value}'\n", $indent); 70 } 71 72 if(count($node->attribute)) { 73 do_leaf(" |\n", $indent); 74 do_leaf(" +---- Attributes\n", $indent); 75 76 foreach($node->attribute as $name=>$value) { 77 @do_leaf(" +-- $name\n", $indent); 78 do_leaf(" | +-- Value: $value\n", $indent); 79 } 80 } 81 82 /* Recurse along the children to generate the remaining nodes */ 83 if($node->hasChildren()) { 84 foreach($node->child as $child) { 85 dump_tree($child, $indent + 3); 86 } 87 } 88 89 } 90 91 92?> 93
