PureBasic Forums - English

Posted: **Fri Feb 14, 2025 9:05 am**

Hi,

this is a PAC file parser using the new WebViewGadget as JavaScript engine. This are the features:

compiles to a single file (needed js functions are included in the compiled binary)
handles the provided PAC file content (you need to load/read the pac file by yourself)
tested with PB 6.20 x64 on Windows 10 and Ubuntu 24.04, very likely also working on MacOS?
the WebViewGadget has some drawback causing crashes in the worst case.
- I solved by giving time to process internal stuff before closing but you might need to increase the waiting time to become more reliable (if you experience crashes).
- See code comment, search for "crash".
the alert() function is not working.
- I encapsulated and forwarded to console.log to prevent any error messages because, as the WebViewGadget is invisible, the alert is never visible to the user.
- In theory, it is possible to implement using a callback, but this likely depends on your environment.
no guarantee that the contained JavaScript functions work in all environments and are error free. I have not tested all of them in detail. If you find errors or made enhancements, please share here below in this thread!

The PB include:

; Parse PAC file using WebViewGadget
;
; PAC reference: 
; https://developer.mozilla.org/en-US/docs/Web/HTTP/Proxy_servers_and_tunneling/Proxy_Auto-Configuration_PAC_file
;
; (c) 2025 / Volker Schmid
;
; PureBasic 6.20

EnableExplicit

DataSection
  ; Include the needed JS code into the generated binary so we do not
  ; create any file dependency for the runtime executable.
  pac_include_js:
  IncludeBinary #PB_Compiler_FilePath + "pac_include.js"
  Data.a 0
EndDataSection

Structure pacInstance
  winId.i       ; the window id
  webViewId.i   ; the webviewgadget it
  call.s        ; the call to execute if DOM has loaded
  result.s      ; catches the result
EndStructure

Global globalPAC.pacInstance

Procedure _createPACEnvironment()
  globalPAC\winId = OpenWindow(#PB_Any, 0, 0, 0, 0, "", #PB_Window_NoActivate)
  If globalPAC\winId = 0
    Debug "PAC: Failed to create dummy window"
    End
  EndIf
  
  globalPAC\webViewId = WebViewGadget(#PB_Any, 0, 0, 0, 0) ; add #PB_WebView_Debug flag if needed
  If globalPAC\webViewId = 0
    Debug "PAC: Failed to create dummy WebViewGadget. Missing libraries/dependencies?"
    End
  EndIf
EndProcedure

Procedure _closePACEnvironment()
  If IsGadget(globalPAC\webViewId)
    FreeGadget(globalPAC\webViewId)
  EndIf
  If IsWindow(globalPAC\winId)
    CloseWindow(globalPAC\winId)
  EndIf
  globalPAC\winId = 0
  globalPAC\webViewId = 0
  globalPAC\result = ""
  globalPAC\call = ""
EndProcedure
 
 ; Triggered upon DOM loaded. Then triggers PAC file execution.
Procedure _pacTrigger(jsonParam.s)
  WebViewExecuteScript(globalPAC\webViewId, globalPAC\call)
  ProcedureReturn #Null
EndProcedure
 
; Receiving the result of PAC file execution
Procedure _pacCallback(jsonParam.s)
  Protected json.i = ParseJSON(#PB_Any, jsonParam.s)
  If JSONArraySize(JSONValue(json.i)) <> 1
    Debug("PAC: _pacCallback had wrong parameter count!")
    FreeJSON(json.i)
    ProcedureReturn #Null
  EndIf
 
  Protected v.i = GetJSONElement(JSONValue(json.i), 0) ; 0 = url
  Protected pacRes.s = GetJSONString(v.i)
  FreeJSON(json.i)
  globalPAC\result = pacRes.s
  ProcedureReturn #Null
EndProcedure
 
Procedure.s ParsePACFile(PACfileContent.s, url.s, host.s)
  _createPACEnvironment()
  Protected includes.s = PeekS(?pac_include_js, -1, #PB_UTF8) ; the needed JS code
  Protected html.s = ~"<!DOCTYPE html>\n<html><body><script>\n" + includes.s +
                     ~"\n\n// ----------- user PAC function ------------\n" + PACfileContent.s + 
                     ~"\n\n// ----------- PB Helper ------------\n" +
                     ~"function runPAC(url, host) {\n" +
                     ~"    var result = FindProxyForURL(url, host);\n" +
                     ~"    window.pacCallback(result); // run PB callback function\n" +
                     ~"}\n" +
                     ~"document.addEventListener(\"DOMContentLoaded\", () => {\n" +
                     ~"    window.pacTrigger();\n" +
                     ~"});\n" +
                     ~"</script></body></html>"
  
  BindWebViewCallback(globalPAC\webViewId, "pacCallback", @_pacCallback())
  BindWebViewCallback(globalPAC\webViewId, "pacTrigger", @_pacTrigger())
  
  SetGadgetItemText(globalPAC\webViewId, #PB_Web_HtmlCode, html.s)
  
  globalPac\call = ~"runPAC(\"" + url.s + ~"\", \"" + host.s + ~"\");"
  
  ; Wait for max 5 seconds for result
  Protected startTime.i = ElapsedMilliseconds()
  Protected event.i
  Repeat
    event = WindowEvent()
    Delay(10)
  Until globalPAC\result <> "" Or ElapsedMilliseconds() > startTime.i + 5000
  Protected result.s = globalPAC\result ; keep because _closePACEnvironment() cleans up
  
  startTime.i = ElapsedMilliseconds()
  Repeat
    event = WindowEvent()
    Delay(1)
  Until ElapsedMilliseconds() > startTime.i + 100 ; increase this time if you encounter a crash
  
  UnbindWebViewCallback(globalPAC\webViewId, "pacCallback")
  UnbindWebViewCallback(globalPAC\webViewId, "pacTrigger")
  
  _closePACEnvironment()
  ProcedureReturn result.s
EndProcedure

; RUNS ONLY IF YOU RUN THIS INCLUDE AS MAIN FILE
CompilerIf #PB_Compiler_IsMainFile = 1

  ; Helpful file loader for testing the PAC library
  Procedure.s _testLoadFile(filename.s)
    Protected fh.i = ReadFile(#PB_Any, filename.s, #PB_File_SharedRead)
    If fh.i = 0
      Debug "PAC: Failed to load [" + filename.s + "]"
      ProcedureReturn ""
    EndIf
    
    Protected fl.i = Lof(fh.i)
    Protected *mem = AllocateMemory(fl.i)
    ReadData(fh.i, *mem, fl.i)
    Protected content.s = PeekS(*mem, fl.i, #PB_UTF8 | #PB_ByteLength)
    FreeMemory(*mem)
    
    CloseFile(fh.i)
    ProcedureReturn content.s
  EndProcedure

  Debug "PAC-Parser Result 1: [" + ParsePACFile(_testLoadFile("pac_testfile.pac"), "https://test.mozilla.org", "test.mozilla.org") + "]"
  Debug "PAC-Parser Result 2: [" + ParsePACFile(_testLoadFile("pac_testfile.pac"), "https://www.google.com", "www.google.com") + "]"
CompilerEndIf

The needed "pac_include.js" bust be saved at the same location than the above include:

Code: Select all

var weekdayMap = {'SUN': 0, 'MON': 1, 'TUE': 2, 'WED': 3, 'THU': 4, 'FRI': 5, 'SAT': 6}; // Map weekday strings to numbers (SUN=0 to SAT=6)
var dnsCache = {}; // Cache DNS lookups
var ipCache = new Map(); // For isInNet caching
var timezoneOffset = new Date().getTimezoneOffset(); // Add time zone offset cache

function isPlainHostName(host) {
    return host.indexOf('.') === -1;
}

function dnsDomainIs(host, domain) {
    if (!domain.startsWith('.')) domain = '.' + domain;
    return host.toLowerCase().endsWith(domain.toLowerCase());
}

function localHostOrDomainIs(host, hostDom) {
    return host === hostDom || (!host.includes('.') && host === hostDom.split('.')[0]);
}

function dnsResolve(host) {
    if (!dnsCache[host]) {
        dnsCache[host] = helper_resolveDNS(host);
    }
    return dnsCache[host];
}

function helper_resolveDNS(host) {
    let xhr = new XMLHttpRequest();
    try {
        xhr.open('GET', 'https://dns.google/resolve?name=' + encodeURIComponent(host), false);
        xhr.send();
        if (xhr.status != 200) {
            console.error(`Error ${xhr.status}: ${xhr.statusText}`);
            return "";
        }
        var response = xhr.response;
        var json = JSON.parse(response);
        console.log(json);
        return json["Answer"][0]["data"];
    } catch(err) { // instead of onerror
        return "";
    }
}

function isResolvable(host) {
    try {
        const ip = dnsResolve(host);
        return ip !== null && ip.length > 0;
    } catch (error) {
        return false;
    }
}

function myIpAddress() {
    if (navigator.connection && navigator.connection.remoteAddresses) {
        return Promise.resolve(
            navigator.connection.remoteAddresses[0]
        );
    }
    resolve(helper_getLocalIP());
}

function helper_getLocalIP() {
    const RTCPeerConnection = window.RTCPeerConnection ||
                             window.mozRTCPeerConnection ||
                             window.webkitRTCPeerConnection;
    if (!RTCPeerConnection) {
        return '127.0.0.1';
    }
    const pc = new RTCPeerConnection();
    const noop = () => {};
    pc.createDataChannel('');
    pc.createOffer(pc.setLocalDescription.bind(pc), noop);
    return new Promise(resolve => {
        pc.onicecandidate = event => {
            if (event.candidate) {
                resolve(event.candidate.candidate.split(':')[3]);
                pc.close();
            }
        };
    });
}

function helper_isValidIP(ip) {
    const ipRegex = /^(\d{1,3}\.){3}\d{1,3}$/;
    if (!ipRegex.test(ip)) return false;

    const octets = ip.split('.');
    return octets.every(octet => {
        const num = parseInt(octet);
        return num >= 0 && num <= 255;
    });
}

function helper_ipToInteger(ip) {
    const octets = ip.split('.');
    let result = 0;
    for (let i = 0; i < 4; i++) {
        result = ((result << 8) | parseInt(octets[i]));
    }
    return result >>> 0;
}

function isInNet(host, pattern, mask) {
    const key = `${host}:${pattern}:${mask}`;
    if (ipCache.has(key)) {
        return ipCache.get(key);
    }
    let ipToCheck = host;
    if (!helper_isValidIP(host)) {
        ipToCheck = dnsResolve(host);
        if (!ipToCheck) return false;
    }
    const ipNum = helper_ipToInteger(ipToCheck);
    const patternNum = helper_ipToInteger(pattern);
    const maskNum = helper_ipToInteger(mask);

    const result = (ipNum & maskNum) === (patternNum & maskNum);
    ipCache.set(key, result);
    return result;
}

function dnsDomainLevels(host) {
    return host.split('.').length - 1;
}

function shExpMatch(str, pattern) {
    const escapeRegex = str.replace(/[-\/\\^$*+?.()|[\]{}]/g, '\\$&');
    const regexPattern = pattern.replace(/\*/g, '.*').replace(/\?/g, '.');
    return new RegExp('^' + regexPattern + '$').test(escapeRegex);
}

function weekdayRange(wd1, wd2, gmt) {
    const weekdays = ['SUN', 'MON', 'TUE', 'WED', 'THU', 'FRI', 'SAT'];
    const now = new Date();
    const currentDay = weekdays[now.getDay()];
    if (!wd2) {
        return currentDay === wd1.toUpperCase();
    }
    const startDay = weekdayMap[wd1.toUpperCase()];
    const endDay = weekdayMap[wd2.toUpperCase()];
    if (startDay > endDay) {
        return currentDay >= wd1.toUpperCase() || 
               currentDay <= wd2.toUpperCase();
    }
    return currentDay >= wd1.toUpperCase() && 
           currentDay <= wd2.toUpperCase();
}

function timeRange(start, end, [gmt]) {
    const now = new Date();
    if (gmt) {
        now.setTime(now.getTime() + (timezoneOffset * 60000));
    }
    const parseTime = (timeStr) => {
        const [hours, minutes] = timeStr.split(':').map(Number);
        return hours * 60 + minutes;
    };
    const currentTime = now.getHours() * 60 + now.getMinutes();
    const startTime = parseTime(start);
    const endTime = parseTime(end);
    if (endTime < startTime) {
        return currentTime >= startTime || currentTime <= endTime;
    }
    return currentTime >= startTime && currentTime <= endTime;
}

function dateRange(date1, date2, gmt = false) {
    const now = new Date();
    if (gmt) {
        now.setTime(now.getTime() + (timezoneOffset * 60000));
    }
    const d1 = typeof date1 === 'string' ? new Date(date1) : date1;
    const d2 = typeof date2 === 'string' ? new Date(date2) : date2;
    return now >= d1 && now <= d2;
}

// Overwrite alert function because this does not
// work for our PB implementation
function alert(message) {
    console.log(message);
}

Some test PAC file:

Code: Select all

function FindProxyForURL(url, host) {
  if (
    (isPlainHostName(host) || dnsDomainIs(host, ".mozilla.org")) &&
    !localHostOrDomainIs(host, "www.mozilla.org") &&
    !localHostOrDomainIs(host, "merchant.mozilla.org")
  ) {
    return "DIRECT";
  } else {
    return "PROXY w3proxy.mozilla.org:8080; DIRECT";
  }
}

See the end of the include for a test with given pac file.

PureBasic Forums - English

PB6.20 | PAC file parser

PB6.20 | PAC file parser