html tag closing detection and repair

html标签闭合检测与修复，说的有点大 ， 并没有考虑的很完整，没有使用正则表达式， 适用于html文件中只有开始标签没有结束标签， 是有结束标签没有开始标签的情况。标签闭合的位置需要根据需求调整

$str = '



        content



        content full

        this is content
        this is content
         This is cont


         This is content

this is content';

$str_len = strlen($str);

//Record starting tag

$pre_data = array();

/ /Record the starting tag position

$pre_pos = array();

$last_data = array();

$error_data = array();

$error_pos = array();

$i = 0;

// Marked as < start

$start_flag = false;

while( $i < $str_len ) {

if($str[$i]=="<" && $str[$i+1]!= '/' && $str[$i+1]!='!') {

$i++;

$_tmp_str = '';

//Mark as < start

$start_flag = true;

//Mark Blank

$space_flag = false;

while($str[$i]!=">" && $str[$i]!="'" && $str[$i]!='"' && $str[ $i] !='/' && $i<$str_len){

if($str[$i]==' ') {

$space_flag = true;

}

if(!$space_flag) {

$_tmp_str .= $str[$i];

}

$i++;

}

$pre_data[] = $_tmp_str;

$pre_pos[] = $i;

} else if ($str[$i ]=="<" && $str[$i+1]=='/') {

$i += 2;

$_tmp_str = '';

while($str[$i]!=" >" && $i<$str_len){

$_tmp_str .= $str[$i];

$i++;

}

$last_data[] = $_tmp_str;

//View the previous value of the start tag

if(count($pre_data)>0) {

$last_pre_node = getLastNode($pre_data, 1);

if($last_pre_node == $_tmp_str) {

//On pairing, delete the value at the corresponding position

array_pop ($pre_data);

array_pop($pre_pos);

array_pop($last_data);

} else {

//No matching, there are two situations

//Case 1: Only closing tag, no start tag

//Case 2: Only opening tag, no closing tag

array_pop($last_data);

$error_data[] = $_tmp_str;

$error_pos[] = $i;

}

} else {

array_pop($last_data );

$error_data[] = $_tmp_str;

$error_pos[] = $i;

}

}else if ($str[$i]=="<" && $str[$i+1] =="!") {

$i++;

while($i<$str_len) {

if($str[$i]=="-" && $str[$i+1]==" -" && $str[$i+2]==">") {

$i++;

break;

} else {

$i++;

}

}

$i++;

}else if($str[$i]=='/' && $str[$i+1]=='>') {

//Skip the automatic single closing tag

if($start_flag) {

array_pop( $pre_data);

array_pop($pre_pos);

$i+=2;

}

}else if($str[$i]=="/" && $str[$i+1]=="* "){

$i++;

while($i<$str_len) {

if($str[$i]=="*" && $str[$i+1]=="/") {

$i++;

break;

} else {

$i++;

}

$i++;

}

}else if($str[$i]=="'"){

$i++ ;

while($str[$i]!="'" && $i<$str_len) {

$i++;

}

$i++;

} else if($str[$i]==' "'){

$i++;

while($str[$i]!='"' && $i<$str_len ) {

$i++;

}

$i++;

} else {

$ i++;

}

}

//Determine the position of the start tag

function confirm_pre_pos($str, $pre_pos){

$str_len = strlen($str);

$j=$pre_pos;

while($j < $str_len) {

if($str[$j] == '"') {

$j++;

while ($j<$str_len) {

if($str[$j]=='" ') {

$j++;

break;

}

$j++;

}

}

else if($str[$j] == "'") {

$j++;

while ($j<$str_len ) {

if($str[$j]=="'") {

$j++;

break;

}

$j++;

}

}

else if($str[$j]==">") {

$j++;

while ($j<$str_len) {

if($str[$j]=="<") {

//Return to the original content position

$j--;

break;

}

$j++;

}

break;

}

else {

$j++;

}

}

return $j;

}

//Determine the position of the starting tag

function confirm_err_pos($str, $err_pos){

$j=$err_pos;

$j--;

while($j > 0) {

if ($str[$j] == '"') {

$j--;

while ($j<$str_len) {

if($str[$j]=='"') {

$j- -;

break;

}

$j--;

}

}

else if($str[$j] == "'") {

$j--;

while ($j<$str_len) {

if($str[$j]=="'") {

$j--;

break;

}

$j--;

}

}

else if($str[$j]= =">") {

$j++;

break;

}

else {

$j--;

}

}

return $j;

}

//Get the last number of the array num values

function getLastNode(array $arr, $num){

$len = count($arr);

if($len > $num) {

return $arr[$len-$num];

} else {

return $arr[0];

}

}

//Organize the data, mainly look backward and conduct further checks

function sort_data(&$pre_data, &$pre_pos, &$error_data, &$error_pos) {

$rem_key_array = array();

$rem_i_array = array();

//Get the value that needs to be deleted

foreach($error_data as $key=>$value){

$count = count($ pre_data);

for($i=($count-1) ; $i>=0; $i--) {

if($pre_data[$i] == $value && !in_array($i, $ rem_i_array)) {

$rem_key_array[] = $key;

$rem_i_array[] = $i;

break;

}

}

}

//Delete the corresponding value of the start tag

foreach($rem_key_array as $_item ) {

unset($error_pos[$_item]);

unset($error_data[$_item]);

}

//Delete the corresponding value of the end tag

foreach($rem_i_array as $_item) {

unset($ pre_data[$_item]);

unset($pre_pos[$_item]);

}

}

//Organize the data and close the tag

function modify_data($str, $pre_data, $pre_pos, $error_data, $error_pos ){

$move_log = array();

//Only data for closed labels

foreach ($error_data as $key => $value) {

# code...

$_tmp_move_count = 0;

foreach ( $move_log as $pos_key => $move_value) {

# code...

if($error_pos[$key]>=$pos_key) {

$_tmp_move_count += $move_value;

}

}

$data = insert_data($str, $value, $error_pos[$key]+$_tmp_move_count, false);

$str = $data['str'];

$move_log[$data['pos']] = $ data['move_count'];

}

//Only data with start tag

foreach ($pre_data as $key => $value) {

# code...

$_tmp_move_count = 0;

foreach ( $move_log as $pos_key => $move_value) {

# code...

if($pre_pos[$key]>=$pos_key) {

$_tmp_move_count += $move_value;

}

}

$data = insert_data($str, $value, $pre_pos[$key]+$_tmp_move_count, true);

$str = $data['str'];

$move_log[$data['pos']] = $ data['move_count'];

}

return $str;

}

//Insert data, $type represents the way to insert data

function insert_data($str, $insert_data, $pos, $type ) {

$len = strlen($str);

//Start tag type

if($type==true) {

$move_count = strlen($insert_data)+3;

$pos = confirm_pre_pos ($str, $pos);

$pre_str = substr($str, 0, $pos);

$end_str = substr($str, $pos);

$mid_str = "";

//Closing tag type

} else {

$pos = confirm_err_pos($str, $pos);

$move_count = strlen($insert_data) + 2;

$ pre_str = substr($str, 0, $pos);

$end_str = substr($str, $pos);

$mid_str = "<" . $insert_data . ">";

}

$str = $pre_str.$mid_str.$end_str;

return array('str'=>$str, 'pos'=>$pos, 'move_count'=>$move_count);

}

sort_data($pre_data, $pre_pos, $error_data, $error_pos);

$new_str = modify_data($str, $pre_data, $pre_pos, $error_data, $error_pos);

echo $new_str;

// print_r($pre_data);

// print_r($pre_pos);

// print_r($error_data);

// print_r($error_pos);

// echo strlen($str);

// foreach($pre_pos as $value){

// $value = confirm_pre_pos($str, $value);

// for($i=$value-5; $i<=$value; $i++) {

// echo $str[$i];

// }

// echo "n";

// }

// foreach($error_pos as $value){

// for($i=$value-5; $i<=$value; $i++) {

// echo $str[$i];

// }

// echo "n";

// }

?>

复制代码