Project

General

Profile

« Previous | Next » 

Revision 5c583de8

Added by koszko almost 2 years ago

start implementing more efficient querying of URL patterns

View differences:

common/patterns.js
17 17 
const proto_regex = /^(\w+):\/\/(.*)$/;
18 18 

  		




  19
  19
  
    
const user_re = "[^/?#@]+@"


  		




  20
  
  
    
const domain_re = "[.a-zA-Z0-9-]+";


  		




  
  20
  
    
const domain_re = "[.*a-zA-Z0-9-]+";


  		




  21
  21
  
    
const path_re = "[^?#]*";


  		




  22
  22
  
    
const query_re = "\\??[^#]*";


  		




  23
  23
  
    

  		




  24
  24
  
    
const http_regex = new RegExp(`^(${domain_re})(${path_re})(${query_re}).*`);


  		




  25
  25
  
    

  		




  26
  
  
    
const file_regex = new RegExp(`^(${path_re}).*`);


  		




  
  26
  
    
const file_regex = new RegExp(`^(/${path_re}).*`);


  		




  27
  27
  
    

  		




  28
  28
  
    
const ftp_regex = new RegExp(`^(${user_re})?(${domain_re})(${path_re}).*`);


  		




  29
  29
  
    

  		




  
  30
  
    
function match_or_throw(regex, string, error_msg)


  		




  
  31
  
    
{


  		




  
  32
  
    
    const match = regex.exec(string);


  		




  
  33
  
    
    if (match === null)


  		




  
  34
  
    
	throw error_msg;


  		




  
  35
  
    

  		




  
  36
  
    
    return match;


  		




  
  37
  
    
}


  		




  
  38
  
    

  		




  30
  39
  
    
function deconstruct_url(url, use_limits=true)


  		




  31
  40
  
    
{


  		




  32
  41
  
    
    const max = MAX;


  		




  ......




  35
  44
  
    
	    max[key] = Infinity;


  		




  36
  45
  
    
    }


  		




  37
  46
  
    

  		




  38
  
  
    
    const proto_match = proto_regex.exec(url);


  		




  39
  
  
    
    if (proto_match === null)


  		




  40
  
  
    
	throw `bad url '${url}'`;


  		




  
  47
  
    
    const matcher = (re, str) => match_or_throw(re, str, `bad url '${url}'`)


  		




  41
  48
  
    

  		




  
  49
  
    
    const proto_match = matcher(proto_regex, url);


  		




  42
  50
  
    
    const deco = {proto: proto_match[1]};


  		




  43
  51
  
    

  		




  44
  52
  
    
    if (deco.proto === "file") {


  		




  45
  
  
    
	deco.path = file_regex.exec(proto_match[2])[1];


  		




  
  53
  
    
	deco.path = matcher(file_regex, proto_match[2])[1];


  		




  46
  54
  
    
    } else if (deco.proto === "ftp") {


  		




  47
  
  
    
	[deco.domain, deco.path] = ftp_regex.exec(proto_match[2]).slice(2, 4);


  		




  
  55
  
    
	[deco.domain, deco.path] =


  		




  
  56
  
    
	    matcher(ftp_regex, proto_match[2]).slice(2, 4);


  		




  48
  57
  
    
    } else if (deco.proto === "http" || deco.proto === "https") {


  		




  49
  
  
    
	const http_match = http_regex.exec(proto_match[2]);


  		




  50
  
  
    
	if (!http_match)


  		




  51
  
  
    
	    return undefined;


  		




  52
  
  
    
	[deco.domain, deco.path, deco.query] = http_match.slice(1, 4);


  		




  
  58
  
    
	[deco.domain, deco.path, deco.query] =


  		




  
  59
  
    
	    matcher(http_regex, proto_match[2]).slice(1, 4);


  		




  53
  60
  
    
	deco.domain = deco.domain.toLowerCase();


  		




  54
  61
  
    
    } else {


  		




  55
  62
  
    
	throw `unsupported protocol in url '${url}'`;


  		












Also available in:  Unified diff