Parse HTML tags and implement Quick Find nodes to get node information

Detailed introduction and use, please click Source sources. /** * HTML Tag Parsing package * * @category Tagparse * @package Tagparse * @author Kun * @copyright Kun * @license http://www.php.net/license/3_01.txt PHP license 3.01 * @version 1.0 * @link http://www.blogkun.com * @since 1.0 */ namespace Tagparse; /** * Tagdomroot * * @category Tagparse * @package Tagparse * @author Kun * @copyright Kun * @license http://www.php.net/license/3_01.txt PHP license 3.01 * @version 1.0 * @link http://www.blogkun.com * @since 1.0 */ Class Tagdomroot { Public $tag = ' root '; Public $plaintext; Public $child = Array (); Public $level = 0; public static $TagParseError = false; protected static $TagSet = Array (); protected static $FoundNode = Array (); public static $ErrorTag = Array (); /** * Initproperty * * @access Public * * @return NULL */ Public Function Initproperty () { $TagParseError = false; $TagSet = Array (); $FoundNode = Array (); $DumpScriptCode = Array (); $ErrorTag = Array (); } /** * __construct * * @param string $STR the tag string to be parse. * * @access Public * * @return Tagdomroot */ Public function __construct ($STR) { $this->_removenoise ($STR); if ($str = = = null) { Self:: $TagParseError = true; } else { $l = Strpos ($str, ' < '); if ($l!== false) { $this->plaintext = substr ($str, 0, $l); } $res = Preg_match_all (' ~> (. *?) <~s ', $str, $matches); if ($res!== false && $res > 0) { $this->plaintext. = Implode ($matches [1]); } $r = Strrpos ($str, ' > '); if ($r!== false) { $this->plaintext. = substr ($str, $r + 1); } $tagCollect = Array (); $attrCollect = Array (); $innerContentCollect = Array (); if ($this->parsetag ($str, $tagCollect, $attrCollect, $innerContentCollect) = = = = False) { Self:: $TagParseError = true; } foreach ($tagCollect as $index = = $tag) { $this->child[] = new Tagdomnode ($tag, $this, $attrCollect [$index], $innerContentCollect [$index], $this->level+1 ); } } } /** * Parsetag * * @param mixed $str Description. * @param mixed & $tagCollect Description. * @param mixed & $attrCollect Description. * @param mixed & $innerContentCollect Description. * * @access protected * * @return Boolean Value. */ protected function Parsetag ($STR, array & $tagCollect, array & $attrCollect, array & $innerContentCollect) { $selfClosingTags = Array (' img ' = = 1, ' br ' = + 1, ' input ' = 1, ' meta ' = 1, ' link ' = 1, ' hr ' = 1, ' base ' = 1, ' embed ' = 1, ' spacer ' and 1 '; $end =-2; $close = 0; $error = false; $tag = "; while (true) { $l = Strpos ($str, ' < ', $end +strlen ($tag) +2); if ($l = = = False) {//parse End Break } if (Strpos (substr ($str, $l, 2), '/')!== false) {//surplus closing Tag,discard $error = true; $end = $l +strlen ($tag); Self:: $ErrorTag [] = substr ($str, $l, Strpos ($str, ' > ', $l)-$l + 1); Continue } $r = Strpos ($str, ' > ', $l); $tag = substr ($str, $l +1, $r-$l-1); if (!ctype_alpha ($tag [0]) | | strpos ($tag, ' < ')!== false) { $end = $r + 1; Continue } $tag = Preg_replace ("~\n+~", "', $tag); $space = Strpos ($tag, '); if ($space!== false) { $attrCollect [] = substr ($tag, $space + 1); $tag = substr ($tag, 0, $space); } else { $attrCollect [] = "; } $tagCollect [] = $tag; if (Isset ($selfClosingTags [$tag])) { $innerContentCollect [] = "; $end = $r-strlen ($tag)-2; $close = $r +1; Continue } $countOpen =-1; $open = Strpos ($str, ' < '. $tag, $close); $close = Strpos ($str, " , $open); if ($close = = = False) {//surplus opening tag $innerContentCollect [] = substr ($str, $r + 1); $error = true; Self:: $ErrorTag [] = ' < '. $tag. ' > '; Break } $start = $open; while ($open < $close && $open!== false) { $countOpen + +; $open = Strpos ($str, ' < '. $tag, $open +strlen ($tag)); } while ($countOpen > 0 && $close!== false) { $open = Strpos ($str, ' < '. $tag, $close +strlen ($tag) +3); $close = Strpos ($str, " , $close +strlen ($tag) +3); if ($close = = = False) { Break } $countOpen--; while ($open < $close && $open!== false) { $open = Strpos ($str, ' < '. $tag, $open +strlen ($tag) +3); $countOpen + +; } } if ($close = = = False) {//label closed not paired $innerContentCollect [] = substr ($str, $r + 1); $error = true; Break } $end = $close; $r = Strpos ($str, ' > ', $start); $innerContentCollect [] = substr ($str, $r +1, $end-$r-1); } return! $error; } /** * _removenoise * * @param string & $STR The tag string to is parse. * * @access Private * * @return String */ Private Function _removenoise (& $STR) { $str = Preg_replace (' ~ ~is ', ' ', $str); $str = Preg_replace (' ~ ~is ', ' ', $str); $str = preg_replace (' ~~is ', ' ', $str); } /** * Parseselectors * * @param string $selectors user s Select condition. * @param array & $selectorsTag tags * @param array & $selectorsAttr attributes * * @access protected * * @return NULL */ protected function Parseselectors ($selectors, array & $selectorsTag, array & $selectorsAttr) { Preg_match_all (' ~ ([\w\d]+) (\[[\w\d-= ". _/]+\])? ~ ', $selectors, $matches); $selectorsTag = $matches [1]; foreach ($matches [2] as $key = + $value) { $selectorsAttr [$key] = array (); if ($value!== ") { Preg_match_all (' ~ ([\w\d-]+] = "([\w\d-. _/]+)" ~ ', $value, $matches); foreach ($matches [1] as $index = + $attr) { $selectorsAttr [$key] [$attr] = $matches [2][$index]; } } } } /** * Find * * @param mixed $selectors user ' s select condition. * @param array $selectorsTag tags. * @param array $selectorsAttr attributes. * * @access Public * * @return Array */ Public function Find ($selectors, $selectorsTag = Array (), $selectorsAttr = Array ()) { if ($selectors!== null) { $this->parseselectors ($selectors, $selectorsTag, $selectorsAttr); } Var_dump ($selectorsTag, $selectorsAttr); exit (); if (!empty ($selectorsTag)) { $this->seek ($selectorsTag, $selectorsAttr); foreach ($this->child as $key = + $node) { $node->find (null, $SELECTORSTAG, $selectorsAttr); } } if ($selectors!== null) { $res = self:: $FoundNode; Self:: $FoundNode = Array (); return $res; } } /** * Findglobal * * @param string $selectors user s Select condition. * * @access Public * * @return Array */ Public Function Findglobal ($selectors) { $space = Strpos ($selectors, ' ', Strpos ($selectors, '] ')); if ($space = = = False) { return $this->findoneglobal ($selectors); } else { $selectorsAttr = Array (); $selectorsTag = Array (); $this->findoneglobal (substr ($selectors, 0, $space), false); $this->parseselectors (substr ($selectors, $space + 1), $selectorsTag, $selectorsAttr); if (!empty (self:: $FoundNode) &&!empty ($selectorsTag)) { $nodes = self:: $FoundNode; Self:: $FoundNode = Array (); foreach ($nodes as $key = = $node) { $node->seek ($selectorsTag, $selectorsAttr); } } } $res = self:: $FoundNode; Self:: $FoundNode = Array (); return $res; } /** * Seek * * @param array $selectorsTag tags. * @param array $selectorsAttr attributes. * * @access protected * * @return NULL */ protected function Seek ($selectorsTag, $selectorsAttr) { foreach ($this->child as $key = + $node) { $isFind = true; if ($node->tag = = = = $selectorsTag [0]) { foreach ($selectorsAttr [0] as $attrName = + $value) { if (Isset ($node->attr[$attrName]) && (Preg_match (' ~.*? '. $value. *?~ ', $node->attr[$attrName]) > 0 || Preg_match (' ~^ '. $value. ' $~ ', $node->attr[$attrName]) > 0 || Preg_match (' ~^ '. $value. ' ~ ', $node->attr[$attrName]) > 0 || Preg_match (' ~ '. $value. ' $~ ', $node->attr[$attrName]) > 0) ) { Continue } else { $isFind = false; Break } } } else { $isFind = false; } if ($isFind) { if (count ($selectorsTag) = = = 1) { Self:: $FoundNode [] = $node; } else { $node->seek ( Array_slice ($selectorsTag, 1), Array_slice ($selectorsAttr, 1) ); } } } } /** * Findoneglobal * * @param string $selector user s Select condition. * @param bool $isReturn weather return value. * * @access Public * * @return Array */ Public Function Findoneglobal ($selector, $isReturn = True) { Preg_match (' ~ ([\w\d]+) (\[[\w\d-= ". _/]+\])? ~ ', $selector, $matches); $tag = $matches [1]; $attr = Array (); if (Isset ($matches [2])) { Preg_match_all (' ~ ([\w\d-]+] = "([\w\d-. _/]+)" ~ ', $matches [2], $matches); foreach ($matches [1] as $key = + $value) { $attr [$value] = $matches [2][$key]; } } if (Isset (self:: $TagSet [$tag])) { foreach (self:: $TagSet [$tag] as $attrValue = = $nodeArray) { $isFind = true; foreach ($attr as $attrName = = $value) { if (Preg_match (' ~ '. $attrName. ' =".*? '. $value. '. *? ' ~ ', $attrValue) || Preg_match (' ~ '. $attrName. ' = "'. $value. '. *? ' ~ ', $attrValue) || Preg_match (' ~ '. $attrName. ' =".*? '. $value. ' " ~ ', $attrValue) || Preg_match (' ~ '. $attrName. ' = "'. $value. '" ~ ', $attrValue) ) { Continue } else { $isFind = false; Break } } if ($isFind) { foreach ($nodeArray as $key = = $node) { Self:: $FoundNode [] = $node; } } } } if ($isReturn) { $res = self:: $FoundNode; Self:: $FoundNode = Array (); return $res; } } } /** * Tagdomnode * * @uses Tagdomroot * * @category Tagparse * @package Tagparse * @author Kun * @copyright Kun * @license http://www.php.net/license/3_01.txt PHP license 3.01 * @version 1.0 * @link http://www.blogkun.com * @since 1.0 */ Class Tagdomnode extends Tagdomroot { Public $attr = Array (); public $parent = null; /** * __construct * * @param mixed $tag tag. * @param mixed $parent parent node. * @param mixed $attr attribute. * @param mixed $innerContent tag content. * @param mixed $level node level. * * @access Public * * @return Tagdomnode */ Public function __construct ($tag, $parent, $attr, $innerContent, $level) { $this->tag = $tag; $this->parent = $parent; $this->_parseattr ($attr); $this->level = $level; $l = Strpos ($innerContent, ' < '); if ($l!== false) { $this->plaintext = substr ($innerContent, 0, $l); } $res = Preg_match_all (' ~> (. *?) <~s ', $innerContent, $matches); if ($res!== false && $res > 0) { $this->plaintext. = Implode ($matches [1]); } else { $this->plaintext. = $innerContent; } $r = Strrpos ($innerContent, ' > '); if ($r!== false) { $this->plaintext. = substr ($innerContent, $r + 1); } $tagCollect = Array (); $attrCollect = Array (); $innerContentCollect = Array (); if ($this->parsetag ($innerContent, $tagCollect, $attrCollect, $innerContentCollect) = = = = False) { Self:: $TagParseError = true; } foreach ($tagCollect as $index = = $tag) { $this->child[] = new Tagdomnode ($tag, $this, $attrCollect [$index], $innerContentCollect [$index], $this->level+1 ); } if (!isset (self:: $TagSet [$this->tag])) { Self:: $TagSet [$this->tag] = array (); } if (!isset (self:: $TagSet [$this->tag][$attr])) { Self:: $TagSet [$this->tag][$attr] = array (); } Self:: $TagSet [$this->tag][$attr] = & $this; } /** * _parseattr * * @param string $str attribute string. * * @access Public * * @return NULL */ Private Function _parseattr ($STR) { Preg_match_all (' ~ (? [ \w-]+) = "(?. *?)" ~s ', $str, $matches); foreach ($matches [' attrname '] as $key = + $value) { $this->attr[$value] = $matches [' AttrValue '] [$key]; } } } Copy Code

