xpressengine /
xe-core
@@ -36,58 +36,58 @@ |
||
| 36 | 36 | */ |
| 37 | 37 | class HTTP_Request2_Exception extends PEAR_Exception |
| 38 | 38 | {
|
| 39 | - /** An invalid argument was passed to a method */ |
|
| 40 | - const INVALID_ARGUMENT = 1; |
|
| 41 | - /** Some required value was not available */ |
|
| 42 | - const MISSING_VALUE = 2; |
|
| 43 | - /** Request cannot be processed due to errors in PHP configuration */ |
|
| 44 | - const MISCONFIGURATION = 3; |
|
| 45 | - /** Error reading the local file */ |
|
| 46 | - const READ_ERROR = 4; |
|
| 39 | + /** An invalid argument was passed to a method */ |
|
| 40 | + const INVALID_ARGUMENT = 1; |
|
| 41 | + /** Some required value was not available */ |
|
| 42 | + const MISSING_VALUE = 2; |
|
| 43 | + /** Request cannot be processed due to errors in PHP configuration */ |
|
| 44 | + const MISCONFIGURATION = 3; |
|
| 45 | + /** Error reading the local file */ |
|
| 46 | + const READ_ERROR = 4; |
|
| 47 | 47 | |
| 48 | - /** Server returned a response that does not conform to HTTP protocol */ |
|
| 49 | - const MALFORMED_RESPONSE = 10; |
|
| 50 | - /** Failure decoding Content-Encoding or Transfer-Encoding of response */ |
|
| 51 | - const DECODE_ERROR = 20; |
|
| 52 | - /** Operation timed out */ |
|
| 53 | - const TIMEOUT = 30; |
|
| 54 | - /** Number of redirects exceeded 'max_redirects' configuration parameter */ |
|
| 55 | - const TOO_MANY_REDIRECTS = 40; |
|
| 56 | - /** Redirect to a protocol other than http(s):// */ |
|
| 57 | - const NON_HTTP_REDIRECT = 50; |
|
| 48 | + /** Server returned a response that does not conform to HTTP protocol */ |
|
| 49 | + const MALFORMED_RESPONSE = 10; |
|
| 50 | + /** Failure decoding Content-Encoding or Transfer-Encoding of response */ |
|
| 51 | + const DECODE_ERROR = 20; |
|
| 52 | + /** Operation timed out */ |
|
| 53 | + const TIMEOUT = 30; |
|
| 54 | + /** Number of redirects exceeded 'max_redirects' configuration parameter */ |
|
| 55 | + const TOO_MANY_REDIRECTS = 40; |
|
| 56 | + /** Redirect to a protocol other than http(s):// */ |
|
| 57 | + const NON_HTTP_REDIRECT = 50; |
|
| 58 | 58 | |
| 59 | - /** |
|
| 60 | - * Native error code |
|
| 61 | - * @var int |
|
| 62 | - */ |
|
| 63 | - private $_nativeCode; |
|
| 59 | + /** |
|
| 60 | + * Native error code |
|
| 61 | + * @var int |
|
| 62 | + */ |
|
| 63 | + private $_nativeCode; |
|
| 64 | 64 | |
| 65 | - /** |
|
| 66 | - * Constructor, can set package error code and native error code |
|
| 67 | - * |
|
| 68 | - * @param string $message exception message |
|
| 69 | - * @param int $code package error code, one of class constants |
|
| 70 | - * @param int $nativeCode error code from underlying PHP extension |
|
| 71 | - */ |
|
| 72 | - public function __construct($message = null, $code = null, $nativeCode = null) |
|
| 73 | - {
|
|
| 74 | - parent::__construct($message, $code); |
|
| 75 | - $this->_nativeCode = $nativeCode; |
|
| 76 | - } |
|
| 65 | + /** |
|
| 66 | + * Constructor, can set package error code and native error code |
|
| 67 | + * |
|
| 68 | + * @param string $message exception message |
|
| 69 | + * @param int $code package error code, one of class constants |
|
| 70 | + * @param int $nativeCode error code from underlying PHP extension |
|
| 71 | + */ |
|
| 72 | + public function __construct($message = null, $code = null, $nativeCode = null) |
|
| 73 | + {
|
|
| 74 | + parent::__construct($message, $code); |
|
| 75 | + $this->_nativeCode = $nativeCode; |
|
| 76 | + } |
|
| 77 | 77 | |
| 78 | - /** |
|
| 79 | - * Returns error code produced by underlying PHP extension |
|
| 80 | - * |
|
| 81 | - * For Socket Adapter this may contain error number returned by |
|
| 82 | - * stream_socket_client(), for Curl Adapter this will contain error number |
|
| 83 | - * returned by curl_errno() |
|
| 84 | - * |
|
| 85 | - * @return integer |
|
| 86 | - */ |
|
| 87 | - public function getNativeCode() |
|
| 88 | - {
|
|
| 89 | - return $this->_nativeCode; |
|
| 90 | - } |
|
| 78 | + /** |
|
| 79 | + * Returns error code produced by underlying PHP extension |
|
| 80 | + * |
|
| 81 | + * For Socket Adapter this may contain error number returned by |
|
| 82 | + * stream_socket_client(), for Curl Adapter this will contain error number |
|
| 83 | + * returned by curl_errno() |
|
| 84 | + * |
|
| 85 | + * @return integer |
|
| 86 | + */ |
|
| 87 | + public function getNativeCode() |
|
| 88 | + {
|
|
| 89 | + return $this->_nativeCode; |
|
| 90 | + } |
|
| 91 | 91 | } |
| 92 | 92 | |
| 93 | 93 | /** |
@@ -1,13 +1,13 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /* Copyright (C) NAVER <http://www.navercorp.com> */ |
| 3 | 3 | /** |
| 4 | - * importer |
|
| 5 | - * high class of importer module |
|
| 6 | - * |
|
| 7 | - * @author NAVER ([email protected]) |
|
| 8 | - * @package /modules/importer |
|
| 9 | - * @version 0.1 |
|
| 10 | - */ |
|
| 4 | + * importer |
|
| 5 | + * high class of importer module |
|
| 6 | + * |
|
| 7 | + * @author NAVER ([email protected]) |
|
| 8 | + * @package /modules/importer |
|
| 9 | + * @version 0.1 |
|
| 10 | + */ |
|
| 11 | 11 | class importer extends ModuleObject |
| 12 | 12 | { |
| 13 | 13 | /** |
@@ -1,10 +1,10 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /* Copyright (C) NAVER <http://www.navercorp.com> */ |
| 3 | 3 | /** |
| 4 | - * @class install |
|
| 5 | - * @author NAVER ([email protected]) |
|
| 6 | - * @brief install module of the high class |
|
| 7 | - */ |
|
| 4 | + * @class install |
|
| 5 | + * @author NAVER ([email protected]) |
|
| 6 | + * @brief install module of the high class |
|
| 7 | + */ |
|
| 8 | 8 | class install extends ModuleObject |
| 9 | 9 | { |
| 10 | 10 | /** |
@@ -5,29 +5,29 @@ |
||
| 5 | 5 | */ |
| 6 | 6 | class PEAR5 |
| 7 | 7 | { |
| 8 | - /** |
|
| 9 | - * If you have a class that's mostly/entirely static, and you need static |
|
| 10 | - * properties, you can use this method to simulate them. Eg. in your method(s) |
|
| 11 | - * do this: $myVar = &PEAR5::getStaticProperty('myclass', 'myVar'); |
|
| 12 | - * You MUST use a reference, or they will not persist! |
|
| 13 | - * |
|
| 14 | - * @access public |
|
| 15 | - * @param string $class The calling classname, to prevent clashes |
|
| 16 | - * @param string $var The variable to retrieve. |
|
| 17 | - * @return mixed A reference to the variable. If not set it will be |
|
| 18 | - * auto initialised to NULL. |
|
| 19 | - */ |
|
| 20 | - static function &getStaticProperty($class, $var) |
|
| 21 | - { |
|
| 22 | - static $properties; |
|
| 23 | - if (!isset($properties[$class])) { |
|
| 24 | - $properties[$class] = array(); |
|
| 25 | - } |
|
| 8 | + /** |
|
| 9 | + * If you have a class that's mostly/entirely static, and you need static |
|
| 10 | + * properties, you can use this method to simulate them. Eg. in your method(s) |
|
| 11 | + * do this: $myVar = &PEAR5::getStaticProperty('myclass', 'myVar'); |
|
| 12 | + * You MUST use a reference, or they will not persist! |
|
| 13 | + * |
|
| 14 | + * @access public |
|
| 15 | + * @param string $class The calling classname, to prevent clashes |
|
| 16 | + * @param string $var The variable to retrieve. |
|
| 17 | + * @return mixed A reference to the variable. If not set it will be |
|
| 18 | + * auto initialised to NULL. |
|
| 19 | + */ |
|
| 20 | + static function &getStaticProperty($class, $var) |
|
| 21 | + { |
|
| 22 | + static $properties; |
|
| 23 | + if (!isset($properties[$class])) { |
|
| 24 | + $properties[$class] = array(); |
|
| 25 | + } |
|
| 26 | 26 | |
| 27 | - if (!array_key_exists($var, $properties[$class])) { |
|
| 28 | - $properties[$class][$var] = null; |
|
| 29 | - } |
|
| 27 | + if (!array_key_exists($var, $properties[$class])) { |
|
| 28 | + $properties[$class][$var] = null; |
|
| 29 | + } |
|
| 30 | 30 | |
| 31 | - return $properties[$class][$var]; |
|
| 32 | - } |
|
| 31 | + return $properties[$class][$var]; |
|
| 32 | + } |
|
| 33 | 33 | } |
| 34 | 34 | \ No newline at end of file |
@@ -22,7 +22,7 @@ discard block |
||
| 22 | 22 | |
| 23 | 23 | public function getInstance() |
| 24 | 24 | { |
| 25 | - if(!isset($GLOBALS['__PURIFIER_INSTANCE__'])) |
|
| 25 | + if (!isset($GLOBALS['__PURIFIER_INSTANCE__'])) |
|
| 26 | 26 | { |
| 27 | 27 | $GLOBALS['__PURIFIER_INSTANCE__'] = new Purifier(); |
| 28 | 28 | } |
@@ -52,9 +52,9 @@ discard block |
||
| 52 | 52 | { |
| 53 | 53 | // add attribute for edit component |
| 54 | 54 | $editComponentAttrs = $this->_searchEditComponent($content); |
| 55 | - if(is_array($editComponentAttrs)) |
|
| 55 | + if (is_array($editComponentAttrs)) |
|
| 56 | 56 | { |
| 57 | - foreach($editComponentAttrs AS $k => $v) |
|
| 57 | + foreach ($editComponentAttrs AS $k => $v) |
|
| 58 | 58 | { |
| 59 | 59 | $this->_def->addAttribute('img', $v, 'CDATA'); |
| 60 | 60 | $this->_def->addAttribute('div', $v, 'CDATA'); |
@@ -63,9 +63,9 @@ discard block |
||
| 63 | 63 | |
| 64 | 64 | // add attribute for widget component |
| 65 | 65 | $widgetAttrs = $this->_searchWidget($content); |
| 66 | - if(is_array($widgetAttrs)) |
|
| 66 | + if (is_array($widgetAttrs)) |
|
| 67 | 67 | { |
| 68 | - foreach($widgetAttrs AS $k => $v) |
|
| 68 | + foreach ($widgetAttrs AS $k => $v) |
|
| 69 | 69 | { |
| 70 | 70 | $this->_def->addAttribute('img', $v, 'CDATA'); |
| 71 | 71 | } |
@@ -82,19 +82,19 @@ discard block |
||
| 82 | 82 | preg_match_all('!<(?:(div)|img)([^>]*)editor_component=([^>]*)>(?(1)(.*?)</div>)!is', $content, $m); |
| 83 | 83 | |
| 84 | 84 | $attributeList = array(); |
| 85 | - if(is_array($m[2])) |
|
| 85 | + if (is_array($m[2])) |
|
| 86 | 86 | { |
| 87 | - foreach($m[2] as $key => $value) |
|
| 87 | + foreach ($m[2] as $key => $value) |
|
| 88 | 88 | { |
| 89 | 89 | unset($script, $m2); |
| 90 | 90 | $script = " {$m[2][$key]} editor_component={$m[3][$key]}"; |
| 91 | 91 | |
| 92 | - if(preg_match_all('/([a-z0-9_-]+)="([^"]+)"/is', $script, $m2)) |
|
| 92 | + if (preg_match_all('/([a-z0-9_-]+)="([^"]+)"/is', $script, $m2)) |
|
| 93 | 93 | { |
| 94 | - foreach($m2[1] as $value2) |
|
| 94 | + foreach ($m2[1] as $value2) |
|
| 95 | 95 | { |
| 96 | 96 | //SECISSUE check style attr |
| 97 | - if($value2 == 'style') |
|
| 97 | + if ($value2 == 'style') |
|
| 98 | 98 | { |
| 99 | 99 | continue; |
| 100 | 100 | } |
@@ -117,18 +117,18 @@ discard block |
||
| 117 | 117 | preg_match_all('!<(?:(div)|img)([^>]*)class="zbxe_widget_output"([^>]*)>(?(1)(.*?)</div>)!is', $content, $m); |
| 118 | 118 | |
| 119 | 119 | $attributeList = array(); |
| 120 | - if(is_array($m[3])) |
|
| 120 | + if (is_array($m[3])) |
|
| 121 | 121 | { |
| 122 | 122 | $content = str_replace('<img class="zbxe_widget_output"', '<img src="" class="zbxe_widget_output"', $content); |
| 123 | 123 | |
| 124 | - foreach($m[3] as $key => $value) |
|
| 124 | + foreach ($m[3] as $key => $value) |
|
| 125 | 125 | { |
| 126 | 126 | if (preg_match_all('/([a-z0-9_-]+)="([^"]+)"/is', $m[3][$key], $m2)) |
| 127 | 127 | { |
| 128 | - foreach($m2[1] as $value2) |
|
| 128 | + foreach ($m2[1] as $value2) |
|
| 129 | 129 | { |
| 130 | 130 | //SECISSUE check style attr |
| 131 | - if($value2 == 'style') |
|
| 131 | + if ($value2 == 'style') |
|
| 132 | 132 | { |
| 133 | 133 | continue; |
| 134 | 134 | } |
@@ -146,7 +146,7 @@ discard block |
||
| 146 | 146 | $whiteIframeUrlList = $oEmbedFilter->getWhiteIframeUrlList(); |
| 147 | 147 | |
| 148 | 148 | $whiteDomain = array(); |
| 149 | - foreach($whiteIframeUrlList as $value) |
|
| 149 | + foreach ($whiteIframeUrlList as $value) |
|
| 150 | 150 | { |
| 151 | 151 | $whiteDomain[] = preg_quote($value, '%'); |
| 152 | 152 | } |
@@ -18,197 +18,197 @@ |
||
| 18 | 18 | */ |
| 19 | 19 | class Html2Text {
|
| 20 | 20 | |
| 21 | - // Private fields |
|
| 21 | + // Private fields |
|
| 22 | 22 | |
| 23 | - var $iCurrentLine = ""; |
|
| 24 | - var $iCurrentWord = ""; |
|
| 25 | - var $iCurrentWordArray; |
|
| 26 | - var $iCurrentWordIndex; |
|
| 27 | - var $iInScript; |
|
| 28 | - var $iListLevel = 0; |
|
| 29 | - var $iHtmlText; |
|
| 30 | - var $iMaxColumns; |
|
| 31 | - var $iHtmlParser; |
|
| 23 | + var $iCurrentLine = ""; |
|
| 24 | + var $iCurrentWord = ""; |
|
| 25 | + var $iCurrentWordArray; |
|
| 26 | + var $iCurrentWordIndex; |
|
| 27 | + var $iInScript; |
|
| 28 | + var $iListLevel = 0; |
|
| 29 | + var $iHtmlText; |
|
| 30 | + var $iMaxColumns; |
|
| 31 | + var $iHtmlParser; |
|
| 32 | 32 | |
| 33 | - // Constants |
|
| 33 | + // Constants |
|
| 34 | 34 | |
| 35 | - var $TOKEN_BR = 0; |
|
| 36 | - var $TOKEN_P = 1; |
|
| 37 | - var $TOKEN_LI = 2; |
|
| 38 | - var $TOKEN_AFTERLI = 3; |
|
| 39 | - var $TOKEN_UL = 4; |
|
| 40 | - var $TOKEN_ENDUL = 5; |
|
| 35 | + var $TOKEN_BR = 0; |
|
| 36 | + var $TOKEN_P = 1; |
|
| 37 | + var $TOKEN_LI = 2; |
|
| 38 | + var $TOKEN_AFTERLI = 3; |
|
| 39 | + var $TOKEN_UL = 4; |
|
| 40 | + var $TOKEN_ENDUL = 5; |
|
| 41 | 41 | |
| 42 | - function Html2Text ($aHtmlText, $aMaxColumns) {
|
|
| 43 | - $this->iHtmlText = $aHtmlText; |
|
| 44 | - $this->iMaxColumns = $aMaxColumns; |
|
| 45 | - } |
|
| 42 | + function Html2Text ($aHtmlText, $aMaxColumns) {
|
|
| 43 | + $this->iHtmlText = $aHtmlText; |
|
| 44 | + $this->iMaxColumns = $aMaxColumns; |
|
| 45 | + } |
|
| 46 | 46 | |
| 47 | - function convert() {
|
|
| 48 | - $this->iHtmlParser = new HtmlParser($this->iHtmlText); |
|
| 49 | - $wholeText = ""; |
|
| 50 | - while (($line = $this->getLine()) !== false) {
|
|
| 51 | - $wholeText .= ($line . "\r\n"); |
|
| 52 | - } |
|
| 53 | - return $wholeText; |
|
| 54 | - } |
|
| 47 | + function convert() {
|
|
| 48 | + $this->iHtmlParser = new HtmlParser($this->iHtmlText); |
|
| 49 | + $wholeText = ""; |
|
| 50 | + while (($line = $this->getLine()) !== false) {
|
|
| 51 | + $wholeText .= ($line . "\r\n"); |
|
| 52 | + } |
|
| 53 | + return $wholeText; |
|
| 54 | + } |
|
| 55 | 55 | |
| 56 | - function getLine() {
|
|
| 57 | - while (true) {
|
|
| 58 | - if (!$this->addWordToLine($this->iCurrentWord)) {
|
|
| 59 | - $retvalue = $this->iCurrentLine; |
|
| 60 | - $this->iCurrentLine = ""; |
|
| 61 | - return $retvalue; |
|
| 62 | - } |
|
| 63 | - $word = $this->getWord(); |
|
| 64 | - if ($word === false) {
|
|
| 65 | - if ($this->iCurrentLine == "") {
|
|
| 66 | - break; |
|
| 67 | - } |
|
| 68 | - $retvalue = $this->iCurrentLine; |
|
| 69 | - $this->iCurrentLine = ""; |
|
| 70 | - $this->iInText = false; |
|
| 71 | - $this->iCurrentWord = ""; |
|
| 72 | - return $retvalue; |
|
| 73 | - } |
|
| 74 | - } |
|
| 75 | - return false; |
|
| 76 | - } |
|
| 56 | + function getLine() {
|
|
| 57 | + while (true) {
|
|
| 58 | + if (!$this->addWordToLine($this->iCurrentWord)) {
|
|
| 59 | + $retvalue = $this->iCurrentLine; |
|
| 60 | + $this->iCurrentLine = ""; |
|
| 61 | + return $retvalue; |
|
| 62 | + } |
|
| 63 | + $word = $this->getWord(); |
|
| 64 | + if ($word === false) {
|
|
| 65 | + if ($this->iCurrentLine == "") {
|
|
| 66 | + break; |
|
| 67 | + } |
|
| 68 | + $retvalue = $this->iCurrentLine; |
|
| 69 | + $this->iCurrentLine = ""; |
|
| 70 | + $this->iInText = false; |
|
| 71 | + $this->iCurrentWord = ""; |
|
| 72 | + return $retvalue; |
|
| 73 | + } |
|
| 74 | + } |
|
| 75 | + return false; |
|
| 76 | + } |
|
| 77 | 77 | |
| 78 | - function addWordToLine ($word) {
|
|
| 79 | - if ($this->iInScript) {
|
|
| 80 | - return true; |
|
| 81 | - } |
|
| 82 | - $prevLine = $this->iCurrentLine; |
|
| 83 | - if ($word === $this->TOKEN_BR) {
|
|
| 84 | - $this->iCurrentWord = ""; |
|
| 85 | - return false; |
|
| 86 | - } |
|
| 87 | - if ($word === $this->TOKEN_P) {
|
|
| 88 | - $this->iCurrentWord = $this->TOKEN_BR; |
|
| 89 | - return false; |
|
| 90 | - } |
|
| 91 | - if ($word === $this->TOKEN_UL) {
|
|
| 92 | - $this->iCurrentWord = $this->TOKEN_BR; |
|
| 93 | - return false; |
|
| 94 | - } |
|
| 95 | - if ($word === $this->TOKEN_ENDUL) {
|
|
| 96 | - $this->iCurrentWord = $this->TOKEN_BR; |
|
| 97 | - return false; |
|
| 98 | - } |
|
| 99 | - if ($word === $this->TOKEN_LI) {
|
|
| 100 | - $this->iCurrentWord = $this->TOKEN_AFTERLI; |
|
| 101 | - return false; |
|
| 102 | - } |
|
| 103 | - $toAdd = $word; |
|
| 104 | - if ($word === $this->TOKEN_AFTERLI) {
|
|
| 105 | - $toAdd = ""; |
|
| 106 | - } |
|
| 107 | - if ($prevLine != "") {
|
|
| 108 | - $prevLine .= " "; |
|
| 109 | - } |
|
| 110 | - else {
|
|
| 111 | - $prevLine = $this->getIndentation($word === $this->TOKEN_AFTERLI); |
|
| 112 | - } |
|
| 113 | - $candidateLine = $prevLine . $toAdd; |
|
| 114 | - if (strlen ($candidateLine) > $this->iMaxColumns && $prevLine != "") {
|
|
| 115 | - return false; |
|
| 116 | - } |
|
| 117 | - $this->iCurrentLine = $candidateLine; |
|
| 118 | - return true; |
|
| 119 | - } |
|
| 78 | + function addWordToLine ($word) {
|
|
| 79 | + if ($this->iInScript) {
|
|
| 80 | + return true; |
|
| 81 | + } |
|
| 82 | + $prevLine = $this->iCurrentLine; |
|
| 83 | + if ($word === $this->TOKEN_BR) {
|
|
| 84 | + $this->iCurrentWord = ""; |
|
| 85 | + return false; |
|
| 86 | + } |
|
| 87 | + if ($word === $this->TOKEN_P) {
|
|
| 88 | + $this->iCurrentWord = $this->TOKEN_BR; |
|
| 89 | + return false; |
|
| 90 | + } |
|
| 91 | + if ($word === $this->TOKEN_UL) {
|
|
| 92 | + $this->iCurrentWord = $this->TOKEN_BR; |
|
| 93 | + return false; |
|
| 94 | + } |
|
| 95 | + if ($word === $this->TOKEN_ENDUL) {
|
|
| 96 | + $this->iCurrentWord = $this->TOKEN_BR; |
|
| 97 | + return false; |
|
| 98 | + } |
|
| 99 | + if ($word === $this->TOKEN_LI) {
|
|
| 100 | + $this->iCurrentWord = $this->TOKEN_AFTERLI; |
|
| 101 | + return false; |
|
| 102 | + } |
|
| 103 | + $toAdd = $word; |
|
| 104 | + if ($word === $this->TOKEN_AFTERLI) {
|
|
| 105 | + $toAdd = ""; |
|
| 106 | + } |
|
| 107 | + if ($prevLine != "") {
|
|
| 108 | + $prevLine .= " "; |
|
| 109 | + } |
|
| 110 | + else {
|
|
| 111 | + $prevLine = $this->getIndentation($word === $this->TOKEN_AFTERLI); |
|
| 112 | + } |
|
| 113 | + $candidateLine = $prevLine . $toAdd; |
|
| 114 | + if (strlen ($candidateLine) > $this->iMaxColumns && $prevLine != "") {
|
|
| 115 | + return false; |
|
| 116 | + } |
|
| 117 | + $this->iCurrentLine = $candidateLine; |
|
| 118 | + return true; |
|
| 119 | + } |
|
| 120 | 120 | |
| 121 | - function getWord() {
|
|
| 122 | - while (true) {
|
|
| 123 | - if ($this->iHtmlParser->iNodeType == NODE_TYPE_TEXT) {
|
|
| 124 | - if (!$this->iInText) {
|
|
| 125 | - $words = $this->splitWords($this->iHtmlParser->iNodeValue); |
|
| 126 | - $this->iCurrentWordArray = $words; |
|
| 127 | - $this->iCurrentWordIndex = 0; |
|
| 128 | - $this->iInText = true; |
|
| 129 | - } |
|
| 130 | - if ($this->iCurrentWordIndex < count($this->iCurrentWordArray)) {
|
|
| 131 | - $this->iCurrentWord = $this->iCurrentWordArray[$this->iCurrentWordIndex++]; |
|
| 132 | - return $this->iCurrentWord; |
|
| 133 | - } |
|
| 134 | - else {
|
|
| 135 | - $this->iInText = false; |
|
| 136 | - } |
|
| 137 | - } |
|
| 138 | - else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ELEMENT) {
|
|
| 139 | - if (strcasecmp ($this->iHtmlParser->iNodeName, "br") == 0) {
|
|
| 140 | - $this->iHtmlParser->parse(); |
|
| 141 | - $this->iCurrentWord = $this->TOKEN_BR; |
|
| 142 | - return $this->iCurrentWord; |
|
| 143 | - } |
|
| 144 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 145 | - $this->iHtmlParser->parse(); |
|
| 146 | - $this->iCurrentWord = $this->TOKEN_P; |
|
| 147 | - return $this->iCurrentWord; |
|
| 148 | - } |
|
| 149 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 150 | - $this->iHtmlParser->parse(); |
|
| 151 | - $this->iCurrentWord = ""; |
|
| 152 | - $this->iInScript = true; |
|
| 153 | - return $this->iCurrentWord; |
|
| 154 | - } |
|
| 155 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 156 | - $this->iHtmlParser->parse(); |
|
| 157 | - $this->iCurrentWord = $this->TOKEN_UL; |
|
| 158 | - $this->iListLevel++; |
|
| 159 | - return $this->iCurrentWord; |
|
| 160 | - } |
|
| 161 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 162 | - $this->iHtmlParser->parse(); |
|
| 163 | - $this->iCurrentWord = $this->TOKEN_LI; |
|
| 164 | - return $this->iCurrentWord; |
|
| 165 | - } |
|
| 166 | - } |
|
| 167 | - else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ENDELEMENT) {
|
|
| 168 | - if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 169 | - $this->iHtmlParser->parse(); |
|
| 170 | - $this->iCurrentWord = ""; |
|
| 171 | - $this->iInScript = false; |
|
| 172 | - return $this->iCurrentWord; |
|
| 173 | - } |
|
| 174 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 175 | - $this->iHtmlParser->parse(); |
|
| 176 | - $this->iCurrentWord = $this->TOKEN_ENDUL; |
|
| 177 | - if ($this->iListLevel > 0) {
|
|
| 178 | - $this->iListLevel--; |
|
| 179 | - } |
|
| 180 | - return $this->iCurrentWord; |
|
| 181 | - } |
|
| 182 | - } |
|
| 183 | - if (!$this->iHtmlParser->parse()) {
|
|
| 184 | - break; |
|
| 185 | - } |
|
| 186 | - } |
|
| 187 | - return false; |
|
| 188 | - } |
|
| 121 | + function getWord() {
|
|
| 122 | + while (true) {
|
|
| 123 | + if ($this->iHtmlParser->iNodeType == NODE_TYPE_TEXT) {
|
|
| 124 | + if (!$this->iInText) {
|
|
| 125 | + $words = $this->splitWords($this->iHtmlParser->iNodeValue); |
|
| 126 | + $this->iCurrentWordArray = $words; |
|
| 127 | + $this->iCurrentWordIndex = 0; |
|
| 128 | + $this->iInText = true; |
|
| 129 | + } |
|
| 130 | + if ($this->iCurrentWordIndex < count($this->iCurrentWordArray)) {
|
|
| 131 | + $this->iCurrentWord = $this->iCurrentWordArray[$this->iCurrentWordIndex++]; |
|
| 132 | + return $this->iCurrentWord; |
|
| 133 | + } |
|
| 134 | + else {
|
|
| 135 | + $this->iInText = false; |
|
| 136 | + } |
|
| 137 | + } |
|
| 138 | + else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ELEMENT) {
|
|
| 139 | + if (strcasecmp ($this->iHtmlParser->iNodeName, "br") == 0) {
|
|
| 140 | + $this->iHtmlParser->parse(); |
|
| 141 | + $this->iCurrentWord = $this->TOKEN_BR; |
|
| 142 | + return $this->iCurrentWord; |
|
| 143 | + } |
|
| 144 | + else if (strcasecmp ($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 145 | + $this->iHtmlParser->parse(); |
|
| 146 | + $this->iCurrentWord = $this->TOKEN_P; |
|
| 147 | + return $this->iCurrentWord; |
|
| 148 | + } |
|
| 149 | + else if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 150 | + $this->iHtmlParser->parse(); |
|
| 151 | + $this->iCurrentWord = ""; |
|
| 152 | + $this->iInScript = true; |
|
| 153 | + return $this->iCurrentWord; |
|
| 154 | + } |
|
| 155 | + else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 156 | + $this->iHtmlParser->parse(); |
|
| 157 | + $this->iCurrentWord = $this->TOKEN_UL; |
|
| 158 | + $this->iListLevel++; |
|
| 159 | + return $this->iCurrentWord; |
|
| 160 | + } |
|
| 161 | + else if (strcasecmp ($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 162 | + $this->iHtmlParser->parse(); |
|
| 163 | + $this->iCurrentWord = $this->TOKEN_LI; |
|
| 164 | + return $this->iCurrentWord; |
|
| 165 | + } |
|
| 166 | + } |
|
| 167 | + else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ENDELEMENT) {
|
|
| 168 | + if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 169 | + $this->iHtmlParser->parse(); |
|
| 170 | + $this->iCurrentWord = ""; |
|
| 171 | + $this->iInScript = false; |
|
| 172 | + return $this->iCurrentWord; |
|
| 173 | + } |
|
| 174 | + else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 175 | + $this->iHtmlParser->parse(); |
|
| 176 | + $this->iCurrentWord = $this->TOKEN_ENDUL; |
|
| 177 | + if ($this->iListLevel > 0) {
|
|
| 178 | + $this->iListLevel--; |
|
| 179 | + } |
|
| 180 | + return $this->iCurrentWord; |
|
| 181 | + } |
|
| 182 | + } |
|
| 183 | + if (!$this->iHtmlParser->parse()) {
|
|
| 184 | + break; |
|
| 185 | + } |
|
| 186 | + } |
|
| 187 | + return false; |
|
| 188 | + } |
|
| 189 | 189 | |
| 190 | - function splitWords ($text) {
|
|
| 191 | - $words = split ("[ \t\r\n]+", $text);
|
|
| 192 | - for ($idx = 0; $idx < count($words); $idx++) {
|
|
| 193 | - $words[$idx] = $this->htmlDecode($words[$idx]); |
|
| 194 | - } |
|
| 195 | - return $words; |
|
| 196 | - } |
|
| 190 | + function splitWords ($text) {
|
|
| 191 | + $words = split ("[ \t\r\n]+", $text);
|
|
| 192 | + for ($idx = 0; $idx < count($words); $idx++) {
|
|
| 193 | + $words[$idx] = $this->htmlDecode($words[$idx]); |
|
| 194 | + } |
|
| 195 | + return $words; |
|
| 196 | + } |
|
| 197 | 197 | |
| 198 | - function htmlDecode ($text) {
|
|
| 199 | - // TBD |
|
| 200 | - return $text; |
|
| 201 | - } |
|
| 198 | + function htmlDecode ($text) {
|
|
| 199 | + // TBD |
|
| 200 | + return $text; |
|
| 201 | + } |
|
| 202 | 202 | |
| 203 | - function getIndentation ($hasLI) {
|
|
| 204 | - $indent = ""; |
|
| 205 | - $idx = 0; |
|
| 206 | - for ($idx = 0; $idx < ($this->iListLevel - 1); $idx++) {
|
|
| 207 | - $indent .= " "; |
|
| 208 | - } |
|
| 209 | - if ($this->iListLevel > 0) {
|
|
| 210 | - $indent = $hasLI ? ($indent . "- ") : ($indent . " "); |
|
| 211 | - } |
|
| 212 | - return $indent; |
|
| 213 | - } |
|
| 203 | + function getIndentation ($hasLI) {
|
|
| 204 | + $indent = ""; |
|
| 205 | + $idx = 0; |
|
| 206 | + for ($idx = 0; $idx < ($this->iListLevel - 1); $idx++) {
|
|
| 207 | + $indent .= " "; |
|
| 208 | + } |
|
| 209 | + if ($this->iListLevel > 0) {
|
|
| 210 | + $indent = $hasLI ? ($indent . "- ") : ($indent . " "); |
|
| 211 | + } |
|
| 212 | + return $indent; |
|
| 213 | + } |
|
| 214 | 214 | } |
@@ -39,7 +39,7 @@ discard block |
||
| 39 | 39 | var $TOKEN_UL = 4; |
| 40 | 40 | var $TOKEN_ENDUL = 5; |
| 41 | 41 | |
| 42 | - function Html2Text ($aHtmlText, $aMaxColumns) {
|
|
| 42 | + function Html2Text($aHtmlText, $aMaxColumns) {
|
|
| 43 | 43 | $this->iHtmlText = $aHtmlText; |
| 44 | 44 | $this->iMaxColumns = $aMaxColumns; |
| 45 | 45 | } |
@@ -75,7 +75,7 @@ discard block |
||
| 75 | 75 | return false; |
| 76 | 76 | } |
| 77 | 77 | |
| 78 | - function addWordToLine ($word) {
|
|
| 78 | + function addWordToLine($word) {
|
|
| 79 | 79 | if ($this->iInScript) {
|
| 80 | 80 | return true; |
| 81 | 81 | } |
@@ -111,7 +111,7 @@ discard block |
||
| 111 | 111 | $prevLine = $this->getIndentation($word === $this->TOKEN_AFTERLI); |
| 112 | 112 | } |
| 113 | 113 | $candidateLine = $prevLine . $toAdd; |
| 114 | - if (strlen ($candidateLine) > $this->iMaxColumns && $prevLine != "") {
|
|
| 114 | + if (strlen($candidateLine) > $this->iMaxColumns && $prevLine != "") {
|
|
| 115 | 115 | return false; |
| 116 | 116 | } |
| 117 | 117 | $this->iCurrentLine = $candidateLine; |
@@ -136,42 +136,42 @@ discard block |
||
| 136 | 136 | } |
| 137 | 137 | } |
| 138 | 138 | else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ELEMENT) {
|
| 139 | - if (strcasecmp ($this->iHtmlParser->iNodeName, "br") == 0) {
|
|
| 139 | + if (strcasecmp($this->iHtmlParser->iNodeName, "br") == 0) {
|
|
| 140 | 140 | $this->iHtmlParser->parse(); |
| 141 | 141 | $this->iCurrentWord = $this->TOKEN_BR; |
| 142 | 142 | return $this->iCurrentWord; |
| 143 | 143 | } |
| 144 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 144 | + else if (strcasecmp($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 145 | 145 | $this->iHtmlParser->parse(); |
| 146 | 146 | $this->iCurrentWord = $this->TOKEN_P; |
| 147 | 147 | return $this->iCurrentWord; |
| 148 | 148 | } |
| 149 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 149 | + else if (strcasecmp($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 150 | 150 | $this->iHtmlParser->parse(); |
| 151 | 151 | $this->iCurrentWord = ""; |
| 152 | 152 | $this->iInScript = true; |
| 153 | 153 | return $this->iCurrentWord; |
| 154 | 154 | } |
| 155 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 155 | + else if (strcasecmp($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 156 | 156 | $this->iHtmlParser->parse(); |
| 157 | 157 | $this->iCurrentWord = $this->TOKEN_UL; |
| 158 | 158 | $this->iListLevel++; |
| 159 | 159 | return $this->iCurrentWord; |
| 160 | 160 | } |
| 161 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 161 | + else if (strcasecmp($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 162 | 162 | $this->iHtmlParser->parse(); |
| 163 | 163 | $this->iCurrentWord = $this->TOKEN_LI; |
| 164 | 164 | return $this->iCurrentWord; |
| 165 | 165 | } |
| 166 | 166 | } |
| 167 | 167 | else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ENDELEMENT) {
|
| 168 | - if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 168 | + if (strcasecmp($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 169 | 169 | $this->iHtmlParser->parse(); |
| 170 | 170 | $this->iCurrentWord = ""; |
| 171 | 171 | $this->iInScript = false; |
| 172 | 172 | return $this->iCurrentWord; |
| 173 | 173 | } |
| 174 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 174 | + else if (strcasecmp($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 175 | 175 | $this->iHtmlParser->parse(); |
| 176 | 176 | $this->iCurrentWord = $this->TOKEN_ENDUL; |
| 177 | 177 | if ($this->iListLevel > 0) {
|
@@ -187,20 +187,20 @@ discard block |
||
| 187 | 187 | return false; |
| 188 | 188 | } |
| 189 | 189 | |
| 190 | - function splitWords ($text) {
|
|
| 191 | - $words = split ("[ \t\r\n]+", $text);
|
|
| 190 | + function splitWords($text) {
|
|
| 191 | + $words = split("[ \t\r\n]+", $text);
|
|
| 192 | 192 | for ($idx = 0; $idx < count($words); $idx++) {
|
| 193 | 193 | $words[$idx] = $this->htmlDecode($words[$idx]); |
| 194 | 194 | } |
| 195 | 195 | return $words; |
| 196 | 196 | } |
| 197 | 197 | |
| 198 | - function htmlDecode ($text) {
|
|
| 198 | + function htmlDecode($text) {
|
|
| 199 | 199 | // TBD |
| 200 | 200 | return $text; |
| 201 | 201 | } |
| 202 | 202 | |
| 203 | - function getIndentation ($hasLI) {
|
|
| 203 | + function getIndentation($hasLI) {
|
|
| 204 | 204 | $indent = ""; |
| 205 | 205 | $idx = 0; |
| 206 | 206 | for ($idx = 0; $idx < ($this->iListLevel - 1); $idx++) {
|
@@ -106,8 +106,7 @@ discard block |
||
| 106 | 106 | } |
| 107 | 107 | if ($prevLine != "") {
|
| 108 | 108 | $prevLine .= " "; |
| 109 | - } |
|
| 110 | - else {
|
|
| 109 | + } else {
|
|
| 111 | 110 | $prevLine = $this->getIndentation($word === $this->TOKEN_AFTERLI); |
| 112 | 111 | } |
| 113 | 112 | $candidateLine = $prevLine . $toAdd; |
@@ -130,48 +129,40 @@ discard block |
||
| 130 | 129 | if ($this->iCurrentWordIndex < count($this->iCurrentWordArray)) {
|
| 131 | 130 | $this->iCurrentWord = $this->iCurrentWordArray[$this->iCurrentWordIndex++]; |
| 132 | 131 | return $this->iCurrentWord; |
| 133 | - } |
|
| 134 | - else {
|
|
| 132 | + } else {
|
|
| 135 | 133 | $this->iInText = false; |
| 136 | 134 | } |
| 137 | - } |
|
| 138 | - else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ELEMENT) {
|
|
| 135 | + } else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ELEMENT) {
|
|
| 139 | 136 | if (strcasecmp ($this->iHtmlParser->iNodeName, "br") == 0) {
|
| 140 | 137 | $this->iHtmlParser->parse(); |
| 141 | 138 | $this->iCurrentWord = $this->TOKEN_BR; |
| 142 | 139 | return $this->iCurrentWord; |
| 143 | - } |
|
| 144 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 140 | + } else if (strcasecmp ($this->iHtmlParser->iNodeName, "p") == 0) {
|
|
| 145 | 141 | $this->iHtmlParser->parse(); |
| 146 | 142 | $this->iCurrentWord = $this->TOKEN_P; |
| 147 | 143 | return $this->iCurrentWord; |
| 148 | - } |
|
| 149 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 144 | + } else if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
|
| 150 | 145 | $this->iHtmlParser->parse(); |
| 151 | 146 | $this->iCurrentWord = ""; |
| 152 | 147 | $this->iInScript = true; |
| 153 | 148 | return $this->iCurrentWord; |
| 154 | - } |
|
| 155 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 149 | + } else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 156 | 150 | $this->iHtmlParser->parse(); |
| 157 | 151 | $this->iCurrentWord = $this->TOKEN_UL; |
| 158 | 152 | $this->iListLevel++; |
| 159 | 153 | return $this->iCurrentWord; |
| 160 | - } |
|
| 161 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 154 | + } else if (strcasecmp ($this->iHtmlParser->iNodeName, "li") == 0) {
|
|
| 162 | 155 | $this->iHtmlParser->parse(); |
| 163 | 156 | $this->iCurrentWord = $this->TOKEN_LI; |
| 164 | 157 | return $this->iCurrentWord; |
| 165 | 158 | } |
| 166 | - } |
|
| 167 | - else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ENDELEMENT) {
|
|
| 159 | + } else if ($this->iHtmlParser->iNodeType == NODE_TYPE_ENDELEMENT) {
|
|
| 168 | 160 | if (strcasecmp ($this->iHtmlParser->iNodeName, "script") == 0) {
|
| 169 | 161 | $this->iHtmlParser->parse(); |
| 170 | 162 | $this->iCurrentWord = ""; |
| 171 | 163 | $this->iInScript = false; |
| 172 | 164 | return $this->iCurrentWord; |
| 173 | - } |
|
| 174 | - else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 165 | + } else if (strcasecmp ($this->iHtmlParser->iNodeName, "ul") == 0 || strcasecmp ($this->iHtmlParser->iNodeName, "ol") == 0) {
|
|
| 175 | 166 | $this->iHtmlParser->parse(); |
| 176 | 167 | $this->iCurrentWord = $this->TOKEN_ENDUL; |
| 177 | 168 | if ($this->iListLevel > 0) {
|
@@ -30,336 +30,336 @@ |
||
| 30 | 30 | */ |
| 31 | 31 | class HtmlParser {
|
| 32 | 32 | |
| 33 | - /** |
|
| 34 | - * Field iNodeType. |
|
| 35 | - * May be one of the NODE_TYPE_* constants above. |
|
| 36 | - */ |
|
| 37 | - var $iNodeType; |
|
| 33 | + /** |
|
| 34 | + * Field iNodeType. |
|
| 35 | + * May be one of the NODE_TYPE_* constants above. |
|
| 36 | + */ |
|
| 37 | + var $iNodeType; |
|
| 38 | 38 | |
| 39 | - /** |
|
| 40 | - * Field iNodeName. |
|
| 41 | - * For elements, it's the name of the element. |
|
| 42 | - */ |
|
| 43 | - var $iNodeName = ""; |
|
| 39 | + /** |
|
| 40 | + * Field iNodeName. |
|
| 41 | + * For elements, it's the name of the element. |
|
| 42 | + */ |
|
| 43 | + var $iNodeName = ""; |
|
| 44 | 44 | |
| 45 | - /** |
|
| 46 | - * Field iNodeValue. |
|
| 47 | - * For text nodes, it's the text. |
|
| 48 | - */ |
|
| 49 | - var $iNodeValue = ""; |
|
| 45 | + /** |
|
| 46 | + * Field iNodeValue. |
|
| 47 | + * For text nodes, it's the text. |
|
| 48 | + */ |
|
| 49 | + var $iNodeValue = ""; |
|
| 50 | 50 | |
| 51 | - /** |
|
| 52 | - * Field iNodeAttributes. |
|
| 53 | - * A string-indexed array containing attribute values |
|
| 54 | - * of the current node. Indexes are always lowercase. |
|
| 55 | - */ |
|
| 56 | - var $iNodeAttributes; |
|
| 51 | + /** |
|
| 52 | + * Field iNodeAttributes. |
|
| 53 | + * A string-indexed array containing attribute values |
|
| 54 | + * of the current node. Indexes are always lowercase. |
|
| 55 | + */ |
|
| 56 | + var $iNodeAttributes; |
|
| 57 | 57 | |
| 58 | - // The following fields should be |
|
| 59 | - // considered private: |
|
| 58 | + // The following fields should be |
|
| 59 | + // considered private: |
|
| 60 | 60 | |
| 61 | - var $iHtmlText; |
|
| 62 | - var $iHtmlTextLength; |
|
| 63 | - var $iHtmlTextIndex = 0; |
|
| 64 | - var $iHtmlCurrentChar; |
|
| 65 | - var $BOE_ARRAY; |
|
| 66 | - var $B_ARRAY; |
|
| 67 | - var $BOS_ARRAY; |
|
| 61 | + var $iHtmlText; |
|
| 62 | + var $iHtmlTextLength; |
|
| 63 | + var $iHtmlTextIndex = 0; |
|
| 64 | + var $iHtmlCurrentChar; |
|
| 65 | + var $BOE_ARRAY; |
|
| 66 | + var $B_ARRAY; |
|
| 67 | + var $BOS_ARRAY; |
|
| 68 | 68 | |
| 69 | - /** |
|
| 70 | - * Constructor. |
|
| 71 | - * Constructs an HtmlParser instance with |
|
| 72 | - * the HTML text given. |
|
| 73 | - */ |
|
| 74 | - function HtmlParser ($aHtmlText) {
|
|
| 75 | - $this->iHtmlText = $aHtmlText; |
|
| 76 | - $this->iHtmlTextLength = strlen($aHtmlText); |
|
| 77 | - $this->iNodeAttributes = array(); |
|
| 78 | - $this->setTextIndex (0); |
|
| 69 | + /** |
|
| 70 | + * Constructor. |
|
| 71 | + * Constructs an HtmlParser instance with |
|
| 72 | + * the HTML text given. |
|
| 73 | + */ |
|
| 74 | + function HtmlParser ($aHtmlText) {
|
|
| 75 | + $this->iHtmlText = $aHtmlText; |
|
| 76 | + $this->iHtmlTextLength = strlen($aHtmlText); |
|
| 77 | + $this->iNodeAttributes = array(); |
|
| 78 | + $this->setTextIndex (0); |
|
| 79 | 79 | |
| 80 | - $this->BOE_ARRAY = array (" ", "\t", "\r", "\n", "=" );
|
|
| 81 | - $this->B_ARRAY = array (" ", "\t", "\r", "\n" );
|
|
| 82 | - $this->BOS_ARRAY = array (" ", "\t", "\r", "\n", "/" );
|
|
| 83 | - } |
|
| 80 | + $this->BOE_ARRAY = array (" ", "\t", "\r", "\n", "=" );
|
|
| 81 | + $this->B_ARRAY = array (" ", "\t", "\r", "\n" );
|
|
| 82 | + $this->BOS_ARRAY = array (" ", "\t", "\r", "\n", "/" );
|
|
| 83 | + } |
|
| 84 | 84 | |
| 85 | - /** |
|
| 86 | - * Method parse. |
|
| 87 | - * Parses the next node. Returns false only if |
|
| 88 | - * the end of the HTML text has been reached. |
|
| 89 | - * Updates values of iNode* fields. |
|
| 90 | - */ |
|
| 91 | - function parse() {
|
|
| 92 | - $text = $this->skipToElement(); |
|
| 93 | - if ($text != "") {
|
|
| 94 | - $this->iNodeType = NODE_TYPE_TEXT; |
|
| 95 | - $this->iNodeName = "Text"; |
|
| 96 | - $this->iNodeValue = $text; |
|
| 97 | - return true; |
|
| 98 | - } |
|
| 99 | - return $this->readTag(); |
|
| 100 | - } |
|
| 85 | + /** |
|
| 86 | + * Method parse. |
|
| 87 | + * Parses the next node. Returns false only if |
|
| 88 | + * the end of the HTML text has been reached. |
|
| 89 | + * Updates values of iNode* fields. |
|
| 90 | + */ |
|
| 91 | + function parse() {
|
|
| 92 | + $text = $this->skipToElement(); |
|
| 93 | + if ($text != "") {
|
|
| 94 | + $this->iNodeType = NODE_TYPE_TEXT; |
|
| 95 | + $this->iNodeName = "Text"; |
|
| 96 | + $this->iNodeValue = $text; |
|
| 97 | + return true; |
|
| 98 | + } |
|
| 99 | + return $this->readTag(); |
|
| 100 | + } |
|
| 101 | 101 | |
| 102 | - function clearAttributes() {
|
|
| 103 | - $this->iNodeAttributes = array(); |
|
| 104 | - } |
|
| 102 | + function clearAttributes() {
|
|
| 103 | + $this->iNodeAttributes = array(); |
|
| 104 | + } |
|
| 105 | 105 | |
| 106 | - function readTag() {
|
|
| 107 | - if ($this->iCurrentChar != "<") {
|
|
| 108 | - $this->iNodeType = NODE_TYPE_DONE; |
|
| 109 | - return false; |
|
| 110 | - } |
|
| 111 | - $this->clearAttributes(); |
|
| 112 | - $this->skipMaxInTag ("<", 1);
|
|
| 113 | - if ($this->iCurrentChar == '/') {
|
|
| 114 | - $this->moveNext(); |
|
| 115 | - $name = $this->skipToBlanksInTag(); |
|
| 116 | - $this->iNodeType = NODE_TYPE_ENDELEMENT; |
|
| 117 | - $this->iNodeName = $name; |
|
| 118 | - $this->iNodeValue = ""; |
|
| 119 | - $this->skipEndOfTag(); |
|
| 120 | - return true; |
|
| 121 | - } |
|
| 122 | - $name = $this->skipToBlanksOrSlashInTag(); |
|
| 123 | - if (!$this->isValidTagIdentifier ($name)) {
|
|
| 124 | - $comment = false; |
|
| 125 | - if (strpos($name, "!--") === 0) {
|
|
| 126 | - $ppos = strpos($name, "--", 3); |
|
| 127 | - if (strpos($name, "--", 3) === (strlen($name) - 2)) {
|
|
| 128 | - $this->iNodeType = NODE_TYPE_COMMENT; |
|
| 129 | - $this->iNodeName = "Comment"; |
|
| 130 | - $this->iNodeValue = "<" . $name . ">"; |
|
| 131 | - $comment = true; |
|
| 132 | - } |
|
| 133 | - else {
|
|
| 134 | - $rest = $this->skipToStringInTag ("-->");
|
|
| 135 | - if ($rest != "") {
|
|
| 136 | - $this->iNodeType = NODE_TYPE_COMMENT; |
|
| 137 | - $this->iNodeName = "Comment"; |
|
| 138 | - $this->iNodeValue = "<" . $name . $rest; |
|
| 139 | - $comment = true; |
|
| 140 | - // Already skipped end of tag |
|
| 141 | - return true; |
|
| 142 | - } |
|
| 143 | - } |
|
| 144 | - } |
|
| 145 | - if (!$comment) {
|
|
| 146 | - $this->iNodeType = NODE_TYPE_TEXT; |
|
| 147 | - $this->iNodeName = "Text"; |
|
| 148 | - $this->iNodeValue = "<" . $name; |
|
| 149 | - return true; |
|
| 150 | - } |
|
| 151 | - } |
|
| 152 | - else {
|
|
| 153 | - $this->iNodeType = NODE_TYPE_ELEMENT; |
|
| 154 | - $this->iNodeValue = ""; |
|
| 155 | - $this->iNodeName = $name; |
|
| 156 | - while ($this->skipBlanksInTag()) {
|
|
| 157 | - $attrName = $this->skipToBlanksOrEqualsInTag(); |
|
| 158 | - if ($attrName != "" && $attrName != "/") {
|
|
| 159 | - $this->skipBlanksInTag(); |
|
| 160 | - if ($this->iCurrentChar == "=") {
|
|
| 161 | - $this->skipEqualsInTag(); |
|
| 162 | - $this->skipBlanksInTag(); |
|
| 163 | - $value = $this->readValueInTag(); |
|
| 164 | - $this->iNodeAttributes[strtolower($attrName)] = $value; |
|
| 165 | - } |
|
| 166 | - else {
|
|
| 167 | - $this->iNodeAttributes[strtolower($attrName)] = ""; |
|
| 168 | - } |
|
| 169 | - } |
|
| 170 | - } |
|
| 171 | - } |
|
| 172 | - $this->skipEndOfTag(); |
|
| 173 | - return true; |
|
| 174 | - } |
|
| 106 | + function readTag() {
|
|
| 107 | + if ($this->iCurrentChar != "<") {
|
|
| 108 | + $this->iNodeType = NODE_TYPE_DONE; |
|
| 109 | + return false; |
|
| 110 | + } |
|
| 111 | + $this->clearAttributes(); |
|
| 112 | + $this->skipMaxInTag ("<", 1);
|
|
| 113 | + if ($this->iCurrentChar == '/') {
|
|
| 114 | + $this->moveNext(); |
|
| 115 | + $name = $this->skipToBlanksInTag(); |
|
| 116 | + $this->iNodeType = NODE_TYPE_ENDELEMENT; |
|
| 117 | + $this->iNodeName = $name; |
|
| 118 | + $this->iNodeValue = ""; |
|
| 119 | + $this->skipEndOfTag(); |
|
| 120 | + return true; |
|
| 121 | + } |
|
| 122 | + $name = $this->skipToBlanksOrSlashInTag(); |
|
| 123 | + if (!$this->isValidTagIdentifier ($name)) {
|
|
| 124 | + $comment = false; |
|
| 125 | + if (strpos($name, "!--") === 0) {
|
|
| 126 | + $ppos = strpos($name, "--", 3); |
|
| 127 | + if (strpos($name, "--", 3) === (strlen($name) - 2)) {
|
|
| 128 | + $this->iNodeType = NODE_TYPE_COMMENT; |
|
| 129 | + $this->iNodeName = "Comment"; |
|
| 130 | + $this->iNodeValue = "<" . $name . ">"; |
|
| 131 | + $comment = true; |
|
| 132 | + } |
|
| 133 | + else {
|
|
| 134 | + $rest = $this->skipToStringInTag ("-->");
|
|
| 135 | + if ($rest != "") {
|
|
| 136 | + $this->iNodeType = NODE_TYPE_COMMENT; |
|
| 137 | + $this->iNodeName = "Comment"; |
|
| 138 | + $this->iNodeValue = "<" . $name . $rest; |
|
| 139 | + $comment = true; |
|
| 140 | + // Already skipped end of tag |
|
| 141 | + return true; |
|
| 142 | + } |
|
| 143 | + } |
|
| 144 | + } |
|
| 145 | + if (!$comment) {
|
|
| 146 | + $this->iNodeType = NODE_TYPE_TEXT; |
|
| 147 | + $this->iNodeName = "Text"; |
|
| 148 | + $this->iNodeValue = "<" . $name; |
|
| 149 | + return true; |
|
| 150 | + } |
|
| 151 | + } |
|
| 152 | + else {
|
|
| 153 | + $this->iNodeType = NODE_TYPE_ELEMENT; |
|
| 154 | + $this->iNodeValue = ""; |
|
| 155 | + $this->iNodeName = $name; |
|
| 156 | + while ($this->skipBlanksInTag()) {
|
|
| 157 | + $attrName = $this->skipToBlanksOrEqualsInTag(); |
|
| 158 | + if ($attrName != "" && $attrName != "/") {
|
|
| 159 | + $this->skipBlanksInTag(); |
|
| 160 | + if ($this->iCurrentChar == "=") {
|
|
| 161 | + $this->skipEqualsInTag(); |
|
| 162 | + $this->skipBlanksInTag(); |
|
| 163 | + $value = $this->readValueInTag(); |
|
| 164 | + $this->iNodeAttributes[strtolower($attrName)] = $value; |
|
| 165 | + } |
|
| 166 | + else {
|
|
| 167 | + $this->iNodeAttributes[strtolower($attrName)] = ""; |
|
| 168 | + } |
|
| 169 | + } |
|
| 170 | + } |
|
| 171 | + } |
|
| 172 | + $this->skipEndOfTag(); |
|
| 173 | + return true; |
|
| 174 | + } |
|
| 175 | 175 | |
| 176 | - function isValidTagIdentifier ($name) {
|
|
| 177 | - return ereg ("^[A-Za-z0-9_\\-]+$", $name);
|
|
| 178 | - } |
|
| 176 | + function isValidTagIdentifier ($name) {
|
|
| 177 | + return ereg ("^[A-Za-z0-9_\\-]+$", $name);
|
|
| 178 | + } |
|
| 179 | 179 | |
| 180 | - function skipBlanksInTag() {
|
|
| 181 | - return "" != ($this->skipInTag ($this->B_ARRAY)); |
|
| 182 | - } |
|
| 180 | + function skipBlanksInTag() {
|
|
| 181 | + return "" != ($this->skipInTag ($this->B_ARRAY)); |
|
| 182 | + } |
|
| 183 | 183 | |
| 184 | - function skipToBlanksOrEqualsInTag() {
|
|
| 185 | - return $this->skipToInTag ($this->BOE_ARRAY); |
|
| 186 | - } |
|
| 184 | + function skipToBlanksOrEqualsInTag() {
|
|
| 185 | + return $this->skipToInTag ($this->BOE_ARRAY); |
|
| 186 | + } |
|
| 187 | 187 | |
| 188 | - function skipToBlanksInTag() {
|
|
| 189 | - return $this->skipToInTag ($this->B_ARRAY); |
|
| 190 | - } |
|
| 188 | + function skipToBlanksInTag() {
|
|
| 189 | + return $this->skipToInTag ($this->B_ARRAY); |
|
| 190 | + } |
|
| 191 | 191 | |
| 192 | - function skipToBlanksOrSlashInTag() {
|
|
| 193 | - return $this->skipToInTag ($this->BOS_ARRAY); |
|
| 194 | - } |
|
| 192 | + function skipToBlanksOrSlashInTag() {
|
|
| 193 | + return $this->skipToInTag ($this->BOS_ARRAY); |
|
| 194 | + } |
|
| 195 | 195 | |
| 196 | - function skipEqualsInTag() {
|
|
| 197 | - return $this->skipMaxInTag ("=", 1);
|
|
| 198 | - } |
|
| 196 | + function skipEqualsInTag() {
|
|
| 197 | + return $this->skipMaxInTag ("=", 1);
|
|
| 198 | + } |
|
| 199 | 199 | |
| 200 | - function readValueInTag() {
|
|
| 201 | - $ch = $this->iCurrentChar; |
|
| 202 | - $value = ""; |
|
| 203 | - if ($ch == "\"") {
|
|
| 204 | - $this->skipMaxInTag ("\"", 1);
|
|
| 205 | - $value = $this->skipToInTag ("\"");
|
|
| 206 | - $this->skipMaxInTag ("\"", 1);
|
|
| 207 | - } |
|
| 208 | - else if ($ch == "'") {
|
|
| 209 | - $this->skipMaxInTag ("'", 1);
|
|
| 210 | - $value = $this->skipToInTag ("'");
|
|
| 211 | - $this->skipMaxInTag ("'", 1);
|
|
| 212 | - } |
|
| 213 | - else {
|
|
| 214 | - $value = $this->skipToBlanksInTag(); |
|
| 215 | - } |
|
| 216 | - return $value; |
|
| 217 | - } |
|
| 200 | + function readValueInTag() {
|
|
| 201 | + $ch = $this->iCurrentChar; |
|
| 202 | + $value = ""; |
|
| 203 | + if ($ch == "\"") {
|
|
| 204 | + $this->skipMaxInTag ("\"", 1);
|
|
| 205 | + $value = $this->skipToInTag ("\"");
|
|
| 206 | + $this->skipMaxInTag ("\"", 1);
|
|
| 207 | + } |
|
| 208 | + else if ($ch == "'") {
|
|
| 209 | + $this->skipMaxInTag ("'", 1);
|
|
| 210 | + $value = $this->skipToInTag ("'");
|
|
| 211 | + $this->skipMaxInTag ("'", 1);
|
|
| 212 | + } |
|
| 213 | + else {
|
|
| 214 | + $value = $this->skipToBlanksInTag(); |
|
| 215 | + } |
|
| 216 | + return $value; |
|
| 217 | + } |
|
| 218 | 218 | |
| 219 | - function setTextIndex ($index) {
|
|
| 220 | - $this->iHtmlTextIndex = $index; |
|
| 221 | - if ($index >= $this->iHtmlTextLength) {
|
|
| 222 | - $this->iCurrentChar = -1; |
|
| 223 | - } |
|
| 224 | - else {
|
|
| 225 | - $this->iCurrentChar = $this->iHtmlText{$index};
|
|
| 226 | - } |
|
| 227 | - } |
|
| 219 | + function setTextIndex ($index) {
|
|
| 220 | + $this->iHtmlTextIndex = $index; |
|
| 221 | + if ($index >= $this->iHtmlTextLength) {
|
|
| 222 | + $this->iCurrentChar = -1; |
|
| 223 | + } |
|
| 224 | + else {
|
|
| 225 | + $this->iCurrentChar = $this->iHtmlText{$index};
|
|
| 226 | + } |
|
| 227 | + } |
|
| 228 | 228 | |
| 229 | - function moveNext() {
|
|
| 230 | - if ($this->iHtmlTextIndex < $this->iHtmlTextLength) {
|
|
| 231 | - $this->setTextIndex ($this->iHtmlTextIndex + 1); |
|
| 232 | - return true; |
|
| 233 | - } |
|
| 234 | - else {
|
|
| 235 | - return false; |
|
| 236 | - } |
|
| 237 | - } |
|
| 229 | + function moveNext() {
|
|
| 230 | + if ($this->iHtmlTextIndex < $this->iHtmlTextLength) {
|
|
| 231 | + $this->setTextIndex ($this->iHtmlTextIndex + 1); |
|
| 232 | + return true; |
|
| 233 | + } |
|
| 234 | + else {
|
|
| 235 | + return false; |
|
| 236 | + } |
|
| 237 | + } |
|
| 238 | 238 | |
| 239 | - function skipEndOfTag() {
|
|
| 240 | - while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 241 | - if ($ch == ">") {
|
|
| 242 | - $this->moveNext(); |
|
| 243 | - return; |
|
| 244 | - } |
|
| 245 | - $this->moveNext(); |
|
| 246 | - } |
|
| 247 | - } |
|
| 239 | + function skipEndOfTag() {
|
|
| 240 | + while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 241 | + if ($ch == ">") {
|
|
| 242 | + $this->moveNext(); |
|
| 243 | + return; |
|
| 244 | + } |
|
| 245 | + $this->moveNext(); |
|
| 246 | + } |
|
| 247 | + } |
|
| 248 | 248 | |
| 249 | - function skipInTag ($chars) {
|
|
| 250 | - $sb = ""; |
|
| 251 | - while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 252 | - if ($ch == ">") {
|
|
| 253 | - return $sb; |
|
| 254 | - } else {
|
|
| 255 | - $match = false; |
|
| 256 | - for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 257 | - if ($ch == $chars[$idx]) {
|
|
| 258 | - $match = true; |
|
| 259 | - break; |
|
| 260 | - } |
|
| 261 | - } |
|
| 262 | - if (!$match) {
|
|
| 263 | - return $sb; |
|
| 264 | - } |
|
| 265 | - $sb .= $ch; |
|
| 266 | - $this->moveNext(); |
|
| 267 | - } |
|
| 268 | - } |
|
| 269 | - return $sb; |
|
| 270 | - } |
|
| 249 | + function skipInTag ($chars) {
|
|
| 250 | + $sb = ""; |
|
| 251 | + while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 252 | + if ($ch == ">") {
|
|
| 253 | + return $sb; |
|
| 254 | + } else {
|
|
| 255 | + $match = false; |
|
| 256 | + for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 257 | + if ($ch == $chars[$idx]) {
|
|
| 258 | + $match = true; |
|
| 259 | + break; |
|
| 260 | + } |
|
| 261 | + } |
|
| 262 | + if (!$match) {
|
|
| 263 | + return $sb; |
|
| 264 | + } |
|
| 265 | + $sb .= $ch; |
|
| 266 | + $this->moveNext(); |
|
| 267 | + } |
|
| 268 | + } |
|
| 269 | + return $sb; |
|
| 270 | + } |
|
| 271 | 271 | |
| 272 | - function skipMaxInTag ($chars, $maxChars) {
|
|
| 273 | - $sb = ""; |
|
| 274 | - $count = 0; |
|
| 275 | - while (($ch = $this->iCurrentChar) !== -1 && $count++ < $maxChars) {
|
|
| 276 | - if ($ch == ">") {
|
|
| 277 | - return $sb; |
|
| 278 | - } else {
|
|
| 279 | - $match = false; |
|
| 280 | - for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 281 | - if ($ch == $chars[$idx]) {
|
|
| 282 | - $match = true; |
|
| 283 | - break; |
|
| 284 | - } |
|
| 285 | - } |
|
| 286 | - if (!$match) {
|
|
| 287 | - return $sb; |
|
| 288 | - } |
|
| 289 | - $sb .= $ch; |
|
| 290 | - $this->moveNext(); |
|
| 291 | - } |
|
| 292 | - } |
|
| 293 | - return $sb; |
|
| 294 | - } |
|
| 272 | + function skipMaxInTag ($chars, $maxChars) {
|
|
| 273 | + $sb = ""; |
|
| 274 | + $count = 0; |
|
| 275 | + while (($ch = $this->iCurrentChar) !== -1 && $count++ < $maxChars) {
|
|
| 276 | + if ($ch == ">") {
|
|
| 277 | + return $sb; |
|
| 278 | + } else {
|
|
| 279 | + $match = false; |
|
| 280 | + for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 281 | + if ($ch == $chars[$idx]) {
|
|
| 282 | + $match = true; |
|
| 283 | + break; |
|
| 284 | + } |
|
| 285 | + } |
|
| 286 | + if (!$match) {
|
|
| 287 | + return $sb; |
|
| 288 | + } |
|
| 289 | + $sb .= $ch; |
|
| 290 | + $this->moveNext(); |
|
| 291 | + } |
|
| 292 | + } |
|
| 293 | + return $sb; |
|
| 294 | + } |
|
| 295 | 295 | |
| 296 | - function skipToInTag ($chars) {
|
|
| 297 | - $sb = ""; |
|
| 298 | - while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 299 | - $match = $ch == ">"; |
|
| 300 | - if (!$match) {
|
|
| 301 | - for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 302 | - if ($ch == $chars[$idx]) {
|
|
| 303 | - $match = true; |
|
| 304 | - break; |
|
| 305 | - } |
|
| 306 | - } |
|
| 307 | - } |
|
| 308 | - if ($match) {
|
|
| 309 | - return $sb; |
|
| 310 | - } |
|
| 311 | - $sb .= $ch; |
|
| 312 | - $this->moveNext(); |
|
| 313 | - } |
|
| 314 | - return $sb; |
|
| 315 | - } |
|
| 296 | + function skipToInTag ($chars) {
|
|
| 297 | + $sb = ""; |
|
| 298 | + while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 299 | + $match = $ch == ">"; |
|
| 300 | + if (!$match) {
|
|
| 301 | + for ($idx = 0; $idx < count($chars); $idx++) {
|
|
| 302 | + if ($ch == $chars[$idx]) {
|
|
| 303 | + $match = true; |
|
| 304 | + break; |
|
| 305 | + } |
|
| 306 | + } |
|
| 307 | + } |
|
| 308 | + if ($match) {
|
|
| 309 | + return $sb; |
|
| 310 | + } |
|
| 311 | + $sb .= $ch; |
|
| 312 | + $this->moveNext(); |
|
| 313 | + } |
|
| 314 | + return $sb; |
|
| 315 | + } |
|
| 316 | 316 | |
| 317 | - function skipToElement() {
|
|
| 318 | - $sb = ""; |
|
| 319 | - while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 320 | - if ($ch == "<") {
|
|
| 321 | - return $sb; |
|
| 322 | - } |
|
| 323 | - $sb .= $ch; |
|
| 324 | - $this->moveNext(); |
|
| 325 | - } |
|
| 326 | - return $sb; |
|
| 327 | - } |
|
| 317 | + function skipToElement() {
|
|
| 318 | + $sb = ""; |
|
| 319 | + while (($ch = $this->iCurrentChar) !== -1) {
|
|
| 320 | + if ($ch == "<") {
|
|
| 321 | + return $sb; |
|
| 322 | + } |
|
| 323 | + $sb .= $ch; |
|
| 324 | + $this->moveNext(); |
|
| 325 | + } |
|
| 326 | + return $sb; |
|
| 327 | + } |
|
| 328 | 328 | |
| 329 | - /** |
|
| 330 | - * Returns text between current position and $needle, |
|
| 331 | - * inclusive, or "" if not found. The current index is moved to a point |
|
| 332 | - * after the location of $needle, or not moved at all |
|
| 333 | - * if nothing is found. |
|
| 334 | - */ |
|
| 335 | - function skipToStringInTag ($needle) {
|
|
| 336 | - $pos = strpos ($this->iHtmlText, $needle, $this->iHtmlTextIndex); |
|
| 337 | - if ($pos === false) {
|
|
| 338 | - return ""; |
|
| 339 | - } |
|
| 340 | - $top = $pos + strlen($needle); |
|
| 341 | - $retvalue = substr ($this->iHtmlText, $this->iHtmlTextIndex, $top - $this->iHtmlTextIndex); |
|
| 342 | - $this->setTextIndex ($top); |
|
| 343 | - return $retvalue; |
|
| 344 | - } |
|
| 329 | + /** |
|
| 330 | + * Returns text between current position and $needle, |
|
| 331 | + * inclusive, or "" if not found. The current index is moved to a point |
|
| 332 | + * after the location of $needle, or not moved at all |
|
| 333 | + * if nothing is found. |
|
| 334 | + */ |
|
| 335 | + function skipToStringInTag ($needle) {
|
|
| 336 | + $pos = strpos ($this->iHtmlText, $needle, $this->iHtmlTextIndex); |
|
| 337 | + if ($pos === false) {
|
|
| 338 | + return ""; |
|
| 339 | + } |
|
| 340 | + $top = $pos + strlen($needle); |
|
| 341 | + $retvalue = substr ($this->iHtmlText, $this->iHtmlTextIndex, $top - $this->iHtmlTextIndex); |
|
| 342 | + $this->setTextIndex ($top); |
|
| 343 | + return $retvalue; |
|
| 344 | + } |
|
| 345 | 345 | } |
| 346 | 346 | |
| 347 | 347 | function HtmlParser_ForFile ($fileName) {
|
| 348 | - return HtmlParser_ForURL($fileName); |
|
| 348 | + return HtmlParser_ForURL($fileName); |
|
| 349 | 349 | } |
| 350 | 350 | |
| 351 | 351 | function HtmlParser_ForURL ($url) {
|
| 352 | - $fp = fopen ($url, "r"); |
|
| 353 | - $content = ""; |
|
| 354 | - while (true) {
|
|
| 355 | - $data = fread ($fp, 8192); |
|
| 356 | - if (strlen($data) == 0) {
|
|
| 357 | - break; |
|
| 358 | - } |
|
| 359 | - $content .= $data; |
|
| 360 | - } |
|
| 361 | - fclose ($fp); |
|
| 362 | - return new HtmlParser ($content); |
|
| 352 | + $fp = fopen ($url, "r"); |
|
| 353 | + $content = ""; |
|
| 354 | + while (true) {
|
|
| 355 | + $data = fread ($fp, 8192); |
|
| 356 | + if (strlen($data) == 0) {
|
|
| 357 | + break; |
|
| 358 | + } |
|
| 359 | + $content .= $data; |
|
| 360 | + } |
|
| 361 | + fclose ($fp); |
|
| 362 | + return new HtmlParser ($content); |
|
| 363 | 363 | } |
| 364 | 364 | |
| 365 | 365 | php?> |
@@ -10,12 +10,12 @@ discard block |
||
| 10 | 10 | * - Leo West (performance improvements) |
| 11 | 11 | */ |
| 12 | 12 | |
| 13 | -define ("NODE_TYPE_START",0);
|
|
| 14 | -define ("NODE_TYPE_ELEMENT",1);
|
|
| 15 | -define ("NODE_TYPE_ENDELEMENT",2);
|
|
| 16 | -define ("NODE_TYPE_TEXT",3);
|
|
| 17 | -define ("NODE_TYPE_COMMENT",4);
|
|
| 18 | -define ("NODE_TYPE_DONE",5);
|
|
| 13 | +define("NODE_TYPE_START", 0);
|
|
| 14 | +define("NODE_TYPE_ELEMENT", 1);
|
|
| 15 | +define("NODE_TYPE_ENDELEMENT", 2);
|
|
| 16 | +define("NODE_TYPE_TEXT", 3);
|
|
| 17 | +define("NODE_TYPE_COMMENT", 4);
|
|
| 18 | +define("NODE_TYPE_DONE", 5);
|
|
| 19 | 19 | |
| 20 | 20 | /** |
| 21 | 21 | * Class HtmlParser. |
@@ -71,15 +71,15 @@ discard block |
||
| 71 | 71 | * Constructs an HtmlParser instance with |
| 72 | 72 | * the HTML text given. |
| 73 | 73 | */ |
| 74 | - function HtmlParser ($aHtmlText) {
|
|
| 74 | + function HtmlParser($aHtmlText) {
|
|
| 75 | 75 | $this->iHtmlText = $aHtmlText; |
| 76 | 76 | $this->iHtmlTextLength = strlen($aHtmlText); |
| 77 | 77 | $this->iNodeAttributes = array(); |
| 78 | - $this->setTextIndex (0); |
|
| 78 | + $this->setTextIndex(0); |
|
| 79 | 79 | |
| 80 | - $this->BOE_ARRAY = array (" ", "\t", "\r", "\n", "=" );
|
|
| 81 | - $this->B_ARRAY = array (" ", "\t", "\r", "\n" );
|
|
| 82 | - $this->BOS_ARRAY = array (" ", "\t", "\r", "\n", "/" );
|
|
| 80 | + $this->BOE_ARRAY = array(" ", "\t", "\r", "\n", "=");
|
|
| 81 | + $this->B_ARRAY = array(" ", "\t", "\r", "\n");
|
|
| 82 | + $this->BOS_ARRAY = array(" ", "\t", "\r", "\n", "/");
|
|
| 83 | 83 | } |
| 84 | 84 | |
| 85 | 85 | /** |
@@ -109,7 +109,7 @@ discard block |
||
| 109 | 109 | return false; |
| 110 | 110 | } |
| 111 | 111 | $this->clearAttributes(); |
| 112 | - $this->skipMaxInTag ("<", 1);
|
|
| 112 | + $this->skipMaxInTag("<", 1);
|
|
| 113 | 113 | if ($this->iCurrentChar == '/') {
|
| 114 | 114 | $this->moveNext(); |
| 115 | 115 | $name = $this->skipToBlanksInTag(); |
@@ -120,7 +120,7 @@ discard block |
||
| 120 | 120 | return true; |
| 121 | 121 | } |
| 122 | 122 | $name = $this->skipToBlanksOrSlashInTag(); |
| 123 | - if (!$this->isValidTagIdentifier ($name)) {
|
|
| 123 | + if (!$this->isValidTagIdentifier($name)) {
|
|
| 124 | 124 | $comment = false; |
| 125 | 125 | if (strpos($name, "!--") === 0) {
|
| 126 | 126 | $ppos = strpos($name, "--", 3); |
@@ -131,7 +131,7 @@ discard block |
||
| 131 | 131 | $comment = true; |
| 132 | 132 | } |
| 133 | 133 | else {
|
| 134 | - $rest = $this->skipToStringInTag ("-->");
|
|
| 134 | + $rest = $this->skipToStringInTag("-->");
|
|
| 135 | 135 | if ($rest != "") {
|
| 136 | 136 | $this->iNodeType = NODE_TYPE_COMMENT; |
| 137 | 137 | $this->iNodeName = "Comment"; |
@@ -173,42 +173,42 @@ discard block |
||
| 173 | 173 | return true; |
| 174 | 174 | } |
| 175 | 175 | |
| 176 | - function isValidTagIdentifier ($name) {
|
|
| 177 | - return ereg ("^[A-Za-z0-9_\\-]+$", $name);
|
|
| 176 | + function isValidTagIdentifier($name) {
|
|
| 177 | + return ereg("^[A-Za-z0-9_\\-]+$", $name);
|
|
| 178 | 178 | } |
| 179 | 179 | |
| 180 | 180 | function skipBlanksInTag() {
|
| 181 | - return "" != ($this->skipInTag ($this->B_ARRAY)); |
|
| 181 | + return "" != ($this->skipInTag($this->B_ARRAY)); |
|
| 182 | 182 | } |
| 183 | 183 | |
| 184 | 184 | function skipToBlanksOrEqualsInTag() {
|
| 185 | - return $this->skipToInTag ($this->BOE_ARRAY); |
|
| 185 | + return $this->skipToInTag($this->BOE_ARRAY); |
|
| 186 | 186 | } |
| 187 | 187 | |
| 188 | 188 | function skipToBlanksInTag() {
|
| 189 | - return $this->skipToInTag ($this->B_ARRAY); |
|
| 189 | + return $this->skipToInTag($this->B_ARRAY); |
|
| 190 | 190 | } |
| 191 | 191 | |
| 192 | 192 | function skipToBlanksOrSlashInTag() {
|
| 193 | - return $this->skipToInTag ($this->BOS_ARRAY); |
|
| 193 | + return $this->skipToInTag($this->BOS_ARRAY); |
|
| 194 | 194 | } |
| 195 | 195 | |
| 196 | 196 | function skipEqualsInTag() {
|
| 197 | - return $this->skipMaxInTag ("=", 1);
|
|
| 197 | + return $this->skipMaxInTag("=", 1);
|
|
| 198 | 198 | } |
| 199 | 199 | |
| 200 | 200 | function readValueInTag() {
|
| 201 | 201 | $ch = $this->iCurrentChar; |
| 202 | 202 | $value = ""; |
| 203 | 203 | if ($ch == "\"") {
|
| 204 | - $this->skipMaxInTag ("\"", 1);
|
|
| 205 | - $value = $this->skipToInTag ("\"");
|
|
| 206 | - $this->skipMaxInTag ("\"", 1);
|
|
| 204 | + $this->skipMaxInTag("\"", 1);
|
|
| 205 | + $value = $this->skipToInTag("\"");
|
|
| 206 | + $this->skipMaxInTag("\"", 1);
|
|
| 207 | 207 | } |
| 208 | 208 | else if ($ch == "'") {
|
| 209 | - $this->skipMaxInTag ("'", 1);
|
|
| 210 | - $value = $this->skipToInTag ("'");
|
|
| 211 | - $this->skipMaxInTag ("'", 1);
|
|
| 209 | + $this->skipMaxInTag("'", 1);
|
|
| 210 | + $value = $this->skipToInTag("'");
|
|
| 211 | + $this->skipMaxInTag("'", 1);
|
|
| 212 | 212 | } |
| 213 | 213 | else {
|
| 214 | 214 | $value = $this->skipToBlanksInTag(); |
@@ -216,7 +216,7 @@ discard block |
||
| 216 | 216 | return $value; |
| 217 | 217 | } |
| 218 | 218 | |
| 219 | - function setTextIndex ($index) {
|
|
| 219 | + function setTextIndex($index) {
|
|
| 220 | 220 | $this->iHtmlTextIndex = $index; |
| 221 | 221 | if ($index >= $this->iHtmlTextLength) {
|
| 222 | 222 | $this->iCurrentChar = -1; |
@@ -228,7 +228,7 @@ discard block |
||
| 228 | 228 | |
| 229 | 229 | function moveNext() {
|
| 230 | 230 | if ($this->iHtmlTextIndex < $this->iHtmlTextLength) {
|
| 231 | - $this->setTextIndex ($this->iHtmlTextIndex + 1); |
|
| 231 | + $this->setTextIndex($this->iHtmlTextIndex + 1); |
|
| 232 | 232 | return true; |
| 233 | 233 | } |
| 234 | 234 | else {
|
@@ -246,7 +246,7 @@ discard block |
||
| 246 | 246 | } |
| 247 | 247 | } |
| 248 | 248 | |
| 249 | - function skipInTag ($chars) {
|
|
| 249 | + function skipInTag($chars) {
|
|
| 250 | 250 | $sb = ""; |
| 251 | 251 | while (($ch = $this->iCurrentChar) !== -1) {
|
| 252 | 252 | if ($ch == ">") {
|
@@ -269,7 +269,7 @@ discard block |
||
| 269 | 269 | return $sb; |
| 270 | 270 | } |
| 271 | 271 | |
| 272 | - function skipMaxInTag ($chars, $maxChars) {
|
|
| 272 | + function skipMaxInTag($chars, $maxChars) {
|
|
| 273 | 273 | $sb = ""; |
| 274 | 274 | $count = 0; |
| 275 | 275 | while (($ch = $this->iCurrentChar) !== -1 && $count++ < $maxChars) {
|
@@ -293,7 +293,7 @@ discard block |
||
| 293 | 293 | return $sb; |
| 294 | 294 | } |
| 295 | 295 | |
| 296 | - function skipToInTag ($chars) {
|
|
| 296 | + function skipToInTag($chars) {
|
|
| 297 | 297 | $sb = ""; |
| 298 | 298 | while (($ch = $this->iCurrentChar) !== -1) {
|
| 299 | 299 | $match = $ch == ">"; |
@@ -332,34 +332,34 @@ discard block |
||
| 332 | 332 | * after the location of $needle, or not moved at all |
| 333 | 333 | * if nothing is found. |
| 334 | 334 | */ |
| 335 | - function skipToStringInTag ($needle) {
|
|
| 336 | - $pos = strpos ($this->iHtmlText, $needle, $this->iHtmlTextIndex); |
|
| 335 | + function skipToStringInTag($needle) {
|
|
| 336 | + $pos = strpos($this->iHtmlText, $needle, $this->iHtmlTextIndex); |
|
| 337 | 337 | if ($pos === false) {
|
| 338 | 338 | return ""; |
| 339 | 339 | } |
| 340 | 340 | $top = $pos + strlen($needle); |
| 341 | - $retvalue = substr ($this->iHtmlText, $this->iHtmlTextIndex, $top - $this->iHtmlTextIndex); |
|
| 342 | - $this->setTextIndex ($top); |
|
| 341 | + $retvalue = substr($this->iHtmlText, $this->iHtmlTextIndex, $top - $this->iHtmlTextIndex); |
|
| 342 | + $this->setTextIndex($top); |
|
| 343 | 343 | return $retvalue; |
| 344 | 344 | } |
| 345 | 345 | } |
| 346 | 346 | |
| 347 | -function HtmlParser_ForFile ($fileName) {
|
|
| 347 | +function HtmlParser_ForFile($fileName) {
|
|
| 348 | 348 | return HtmlParser_ForURL($fileName); |
| 349 | 349 | } |
| 350 | 350 | |
| 351 | -function HtmlParser_ForURL ($url) {
|
|
| 352 | - $fp = fopen ($url, "r"); |
|
| 351 | +function HtmlParser_ForURL($url) {
|
|
| 352 | + $fp = fopen($url, "r"); |
|
| 353 | 353 | $content = ""; |
| 354 | 354 | while (true) {
|
| 355 | - $data = fread ($fp, 8192); |
|
| 355 | + $data = fread($fp, 8192); |
|
| 356 | 356 | if (strlen($data) == 0) {
|
| 357 | 357 | break; |
| 358 | 358 | } |
| 359 | 359 | $content .= $data; |
| 360 | 360 | } |
| 361 | - fclose ($fp); |
|
| 362 | - return new HtmlParser ($content); |
|
| 361 | + fclose($fp); |
|
| 362 | + return new HtmlParser($content); |
|
| 363 | 363 | } |
| 364 | 364 | |
| 365 | 365 | php?> |
@@ -129,8 +129,7 @@ discard block |
||
| 129 | 129 | $this->iNodeName = "Comment"; |
| 130 | 130 | $this->iNodeValue = "<" . $name . ">"; |
| 131 | 131 | $comment = true; |
| 132 | - } |
|
| 133 | - else {
|
|
| 132 | + } else {
|
|
| 134 | 133 | $rest = $this->skipToStringInTag ("-->");
|
| 135 | 134 | if ($rest != "") {
|
| 136 | 135 | $this->iNodeType = NODE_TYPE_COMMENT; |
@@ -148,8 +147,7 @@ discard block |
||
| 148 | 147 | $this->iNodeValue = "<" . $name; |
| 149 | 148 | return true; |
| 150 | 149 | } |
| 151 | - } |
|
| 152 | - else {
|
|
| 150 | + } else {
|
|
| 153 | 151 | $this->iNodeType = NODE_TYPE_ELEMENT; |
| 154 | 152 | $this->iNodeValue = ""; |
| 155 | 153 | $this->iNodeName = $name; |
@@ -162,8 +160,7 @@ discard block |
||
| 162 | 160 | $this->skipBlanksInTag(); |
| 163 | 161 | $value = $this->readValueInTag(); |
| 164 | 162 | $this->iNodeAttributes[strtolower($attrName)] = $value; |
| 165 | - } |
|
| 166 | - else {
|
|
| 163 | + } else {
|
|
| 167 | 164 | $this->iNodeAttributes[strtolower($attrName)] = ""; |
| 168 | 165 | } |
| 169 | 166 | } |
@@ -204,13 +201,11 @@ discard block |
||
| 204 | 201 | $this->skipMaxInTag ("\"", 1);
|
| 205 | 202 | $value = $this->skipToInTag ("\"");
|
| 206 | 203 | $this->skipMaxInTag ("\"", 1);
|
| 207 | - } |
|
| 208 | - else if ($ch == "'") {
|
|
| 204 | + } else if ($ch == "'") {
|
|
| 209 | 205 | $this->skipMaxInTag ("'", 1);
|
| 210 | 206 | $value = $this->skipToInTag ("'");
|
| 211 | 207 | $this->skipMaxInTag ("'", 1);
|
| 212 | - } |
|
| 213 | - else {
|
|
| 208 | + } else {
|
|
| 214 | 209 | $value = $this->skipToBlanksInTag(); |
| 215 | 210 | } |
| 216 | 211 | return $value; |
@@ -220,8 +215,7 @@ discard block |
||
| 220 | 215 | $this->iHtmlTextIndex = $index; |
| 221 | 216 | if ($index >= $this->iHtmlTextLength) {
|
| 222 | 217 | $this->iCurrentChar = -1; |
| 223 | - } |
|
| 224 | - else {
|
|
| 218 | + } else {
|
|
| 225 | 219 | $this->iCurrentChar = $this->iHtmlText{$index};
|
| 226 | 220 | } |
| 227 | 221 | } |
@@ -230,8 +224,7 @@ discard block |
||
| 230 | 224 | if ($this->iHtmlTextIndex < $this->iHtmlTextLength) {
|
| 231 | 225 | $this->setTextIndex ($this->iHtmlTextIndex + 1); |
| 232 | 226 | return true; |
| 233 | - } |
|
| 234 | - else {
|
|
| 227 | + } else {
|
|
| 235 | 228 | return false; |
| 236 | 229 | } |
| 237 | 230 | } |
