Project

General

Profile

« Previous | Next » 

Revision d911bf37

Added by koszko over 1 year ago

fix encoding detection in StreamFilter

View differences:

background/stream_filter.js
35 35 
function validate_encoding(charset)
36 36 
{
37 37 
    try {
38 
	new TextDecoder();
38 
	new TextDecoder(charset);
39 39 
	return charset;
40 40 
    } catch(e) {
41 41 
	return undefined;
......
44 44 

  		




  45
  45
  
    
function is_content_type_header(header)


  		




  46
  46
  
    
{


  		




  47
  
  
    
    header.name.toLowerCase().trim() === "content-type";


  		




  
  47
  
    
    return header.name.toLowerCase().trim() === "content-type";


  		




  48
  48
  
    
}


  		




  49
  49
  
    

  		




  50
  50
  
    
const charset_reg = /;\s*charset\s*=\s*([\w-]+)/i;


  		




  ......




  55
  55
  
    

  		




  56
  56
  
    
    for (const header of headers.filter(is_content_type_header)) {


  		




  57
  57
  
    
	const match = charset_reg.exec(header.value);


  		




  58
  
  
    
	if (!properties.detected_charset && validate_encoding(match[1]))


  		




  
  58
  
    
	if (match && !properties.detected_charset &&


  		




  
  59
  
    
	    validate_encoding(match[1]))


  		




  59
  60
  
    
	    properties.detected_charset = match[1];


  		




  60
  61
  
    

  		




  61
  62
  
    
	if (/html/i.test(header.value))


  		




  ......




  105
  106
  
    
function create_decoder(properties, data)


  		




  106
  107
  
    
{


  		




  107
  108
  
    
    let charset = charset_from_BOM(data) || properties.detected_charset;


  		




  108
  
  
    
    if (!charset && data.indexOf(0) !== -1) {


  		




  
  109
  
    

  		




  
  110
  
    
    if (charset)


  		




  
  111
  
    
	return new TextDecoder(charset);


  		




  
  112
  
    

  		




  
  113
  
    
    if (data.indexOf(0) !== -1) {


  		




  109
  114
  
    
        console.warn("Haketilo: zeroes in bytestream, probable cached encoding mismatch. Trying to decode it as UTF-16.",


  		




  110
  115
  
    
		     properties);


  		




  111
  116
  
    
	return new TextDecoder("utf-16be");


  		












Also available in:  Unified diff