buzzert/sigilbox
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fixes

Browse Source
This commit is contained in:
James Magahern
2026-05-15 09:25:19 -07:00
parent 2e350ce3dc
commit bc6d9893a1
4 changed files with 544 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files

255
src/archiver.mjs
View File

@@ -32,6 +32,7 @@ const PRIVACY_FILTERS_DIR = path.join(__dirname, "..", "privacy-filters");
let privacyFiltersAvailable = false;
let filterRules = { blockRules: [], allowRules: [], cosmeticRules: [] };
let userScriptData = []; // { file, content, matches, excludes }
let userScriptRequireContent = "";
async function loadPrivacyFilters() {
try {
@@ -40,6 +41,7 @@ async function loadPrivacyFilters() {
filterRules = parseFilterRules(filterContent);
const userscriptDir = path.join(PRIVACY_FILTERS_DIR, "userscript");
userScriptRequireContent = await fs.readFile(path.join(userscriptDir, "bpc_func.js"), "utf8");
const userScriptFiles = [
"bpc.en.user.js",
"bpc.de.user.js",
@@ -126,7 +128,7 @@ function parseNetworkRule(line) {
const lastDollar = line.lastIndexOf("$");
if (lastDollar > 0) {
const optsStr = line.slice(lastDollar + 1);
if (/^[a-z,=~\-|0-9]+$/i.test(optsStr)) {
if (/^[a-z,=~_.\-|0-9]+$/i.test(optsStr)) {
options = optsStr.split(",");
pattern = line.slice(0, lastDollar);
}
@@ -134,8 +136,20 @@ function parseNetworkRule(line) {
if (!pattern) return null;
const type = options.find((o) =>
["script", "stylesheet", "image", "media", "xmlhttprequest", "other", "inline-script"].includes(o)
const types = options.filter((o) =>
[
"document",
"font",
"image",
"inline-script",
"media",
"object",
"other",
"script",
"stylesheet",
"subdocument",
"xmlhttprequest"
].includes(o)
);
const isThirdParty = options.includes("third-party");
const isFirstParty = options.includes("~third-party");
@@ -162,7 +176,7 @@ function parseNetworkRule(line) {
kind: "domain",
domain,
path,
type,
types,
isThirdParty,
isFirstParty,
includeDomains,
@@ -171,27 +185,38 @@ function parseNetworkRule(line) {
};
}
if (pattern.startsWith("/")) {
const lastSlash = pattern.lastIndexOf("/");
if (lastSlash > 0) {
const regex = pattern.slice(1, lastSlash);
return {
kind: "regex",
regex,
type,
isThirdParty,
isFirstParty,
includeDomains,
excludeDomains,
important
};
}
if (pattern.startsWith("/") && pattern.endsWith("/") && pattern.length > 1) {
const regex = pattern.slice(1, -1);
return {
kind: "regex",
regex,
types,
isThirdParty,
isFirstParty,
includeDomains,
excludeDomains,
important
};
}
return null;
return {
kind: "pattern",
regex: adblockPatternToRegex(pattern),
types,
isThirdParty,
isFirstParty,
includeDomains,
excludeDomains,
important
};
}
function cosmeticSelectorToCss(selector) {
if (selector.endsWith(":remove()")) {
const baseSelector = selector.slice(0, -":remove()".length);
return baseSelector ? `${baseSelector} { display: none !important; }` : null;
}
const styleMatch = selector.match(/:style\((.+)\)$/);
if (styleMatch) {
const baseSelector = selector.slice(0, selector.lastIndexOf(":style("));
@@ -246,17 +271,8 @@ function matchesNetworkRule(url, urlObj, hostname, resourceType, sourceHostname,
if (blocked) return false;
}
if (rule.type) {
const typeMap = {
script: "script",
stylesheet: "stylesheet",
image: "image",
media: "media",
xmlhttprequest: "xhr",
other: "other",
"inline-script": "script"
};
if (typeMap[rule.type] && resourceType !== typeMap[rule.type]) {
if (rule.types.length > 0) {
if (!rule.types.some((type) => resourceTypeMatches(type, resourceType))) {
return false;
}
}
@@ -271,18 +287,11 @@ function matchesNetworkRule(url, urlObj, hostname, resourceType, sourceHostname,
}
if (rule.kind === "domain") {
const domainRe = new RegExp(
"^" + rule.domain.replace(/\./g, "\\.").replace(/\*/g, "[^.]*") + "$",
"i"
);
if (!domainRe.test(hostname)) return false;
if (!domainPatternMatches(hostname, rule.domain)) return false;
if (rule.path) {
const pathRe = new RegExp(
"^" + rule.path.replace(/\./g, "\\.").replace(/\*/g, ".*").replace(/\?/g, "\\?").replace(/\^/g, ""),
"i"
);
if (!pathRe.test(urlObj.pathname)) return false;
const pathRe = new RegExp("^" + adblockPatternToRegex(rule.path), "i");
if (!pathRe.test(urlObj.pathname + urlObj.search)) return false;
}
return true;
}
@@ -296,9 +305,84 @@ function matchesNetworkRule(url, urlObj, hostname, resourceType, sourceHostname,
}
}
if (rule.kind === "pattern") {
try {
const re = new RegExp(rule.regex, "i");
return re.test(url);
} catch {
return false;
}
}
return false;
}
function resourceTypeMatches(filterType, resourceType) {
const typeMap = {
document: ["document"],
font: ["font"],
image: ["image"],
"inline-script": ["script"],
media: ["media"],
object: ["object"],
other: ["other"],
script: ["script"],
stylesheet: ["stylesheet"],
subdocument: ["document"],
xmlhttprequest: ["fetch", "xhr"]
};
const mapped = typeMap[filterType];
return mapped ? mapped.includes(resourceType) : false;
}
function domainPatternMatches(hostname, pattern) {
const normalized = pattern.replace(/\^$/, "").toLowerCase();
if (!normalized) return false;
if (!normalized.includes("*")) {
return hostname === normalized || hostname.endsWith("." + normalized);
}
const re = new RegExp(
"^" +
normalized
.split("*")
.map((part) => part.replace(/[|\\{}()[\]^$+?.]/g, "\\$&"))
.join("[^.]*") +
"$",
"i"
);
return re.test(hostname);
}
function adblockPatternToRegex(pattern) {
let source = "";
let remaining = pattern;
let anchoredStart = false;
let anchoredEnd = false;
if (remaining.startsWith("|")) {
anchoredStart = true;
remaining = remaining.slice(1);
}
if (remaining.endsWith("|")) {
anchoredEnd = true;
remaining = remaining.slice(0, -1);
}
for (const ch of remaining) {
if (ch === "*") {
source += ".*";
} else if (ch === "^") {
source += "(?:[^A-Za-z0-9_.%-]|$)";
} else {
source += ch.replace(/[|\\{}()[\]^$+?.]/g, "\\$&");
}
}
return `${anchoredStart ? "^" : ""}${source}${anchoredEnd ? "$" : ""}`;
}
function shouldBlockRequest(url, resourceType, sourceHostname) {
if (url === sourceHostname || url.startsWith(sourceHostname + "/")) {
return false;
@@ -512,7 +596,7 @@ async function setupRequestBlocking(page, sourceHostname) {
await page.route("**/*", (route) => {
try {
const request = route.request();
if (request.isNavigationRequest()) {
if (request.isNavigationRequest() && request.frame() === page.mainFrame()) {
route.continue();
return;
}
@@ -584,6 +668,9 @@ async function injectPrivacyUserScripts(page, sourceUrl) {
// Inject GM API mock first.
try {
await page.addScriptTag({ content: GM_MOCK });
if (userScriptRequireContent) {
await page.addScriptTag({ content: userScriptRequireContent });
}
} catch {
return;
}
@@ -731,15 +818,19 @@ function addArchiveComment(html, sourceUrl) {
export function findExternalAssetRefs(html) {
const refs = new Set();
const attrPattern = /\s(?:src|srcset|poster|data)\s*=\s*(["'])([\s\S]*?)\1/gi;
for (const match of html.matchAll(attrPattern)) {
if (isSelfContainedAssetRef(match[2])) {
continue;
}
for (const part of match[2].split(",")) {
const candidate = part.trim().split(/\s+/)[0];
if (candidate && !isSelfContainedAssetRef(candidate)) {
refs.add(candidate);
const assetTagPattern = /<(?:img|source|audio|video|track|embed|object|input|iframe)\b[^>]*>/gi;
for (const match of html.matchAll(assetTagPattern)) {
const tag = match[0];
for (const attr of ["src", "srcset", "poster", "data"]) {
const value = readAttribute(tag, attr);
if (!value || isSelfContainedAssetRef(value)) {
continue;
}
for (const part of value.split(",")) {
const candidate = part.trim().split(/\s+/)[0];
if (candidate && !isSelfContainedAssetRef(candidate)) {
refs.add(candidate);
}
}
}
}
@@ -779,8 +870,8 @@ function isSelfContainedAssetRef(value) {
}
function readAttribute(tag, attr) {
const match = tag.match(new RegExp(`\\b${attr}\\s*=\\s*("([^"]*)"|'([^']*)'|([^\\s>]+))`, "i"));
return match ? match[2] ?? match[3] ?? match[4] ?? "" : "";
const match = findAttribute(tag, attr);
return match ? match.value : "";
}
function cleanCssUrl(value) {
@@ -796,3 +887,61 @@ function cleanCssUrl(value) {
}
return decoded;
}
function findAttribute(openingTag, attr) {
const attrLower = attr.toLowerCase();
const nameMatch = openingTag.match(/^<[^\s/>]+/);
let index = nameMatch ? nameMatch[0].length : 1;
while (index < openingTag.length) {
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
if (index >= openingTag.length || openingTag[index] === ">" || openingTag[index] === "/") {
return null;
}
const start = index;
while (index < openingTag.length && !/[\s=/>]/.test(openingTag[index])) {
index += 1;
}
const name = openingTag.slice(start, index);
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
let value = "";
if (openingTag[index] === "=") {
index += 1;
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
const quote = openingTag[index];
if (quote === '"' || quote === "'") {
index += 1;
const valueStart = index;
while (index < openingTag.length && openingTag[index] !== quote) {
index += 1;
}
value = openingTag.slice(valueStart, index);
if (openingTag[index] === quote) {
index += 1;
}
} else {
const valueStart = index;
while (index < openingTag.length && !/[\s>]/.test(openingTag[index])) {
index += 1;
}
value = openingTag.slice(valueStart, index);
}
}
if (name.toLowerCase() === attrLower) {
return { start, end: index, value };
}
}
return null;
}

151
src/asset-inliner.mjs
View File

@@ -197,11 +197,6 @@ export class AssetInliner {
async (match) => this.rewriteMediaAttributes(match[0], effectiveBase)
);
output = await replaceAsync(output, /srcset=(["'])([\s\S]*?)\1/gi, async (match) => {
const rewritten = await this.inlineSrcset(match[2], effectiveBase);
return `srcset=${match[1]}${htmlEscape(rewritten)}${match[1]}`;
});
return output;
}
@@ -259,12 +254,28 @@ export class AssetInliner {
output = replaceMissingMediaAttribute(output, attr);
}
}
const srcset = getAttribute(output, "srcset");
if (srcset) {
const rewritten = await this.inlineSrcset(srcset, baseUrl);
output = setAttribute(output, "srcset", rewritten);
}
return output;
}
async rewriteIframeTag(tag, baseUrl, depth) {
const srcdoc = getAttribute(tag, "srcdoc");
if (srcdoc) {
let rewritten = removeAttribute(tag, "src");
if (depth >= 2) {
return rewritten;
}
const inlined = await this.inlineHtml(srcdoc, baseUrl, { depth: depth + 1 });
rewritten = setAttribute(rewritten, "srcdoc", inlined);
return rewritten;
}
const src = getAttribute(tag, "src");
if (!src || getAttribute(tag, "srcdoc")) {
if (!src) {
return this.rewriteMediaAttributes(tag, baseUrl);
}
const absolute = resolveUrl(src, baseUrl);
@@ -425,24 +436,42 @@ function mimeFromUrl(rawUrl) {
}
function getAttribute(tag, attr) {
const match = tag.match(new RegExp(`\\b${attr}\\s*=\\s*("([^"]*)"|'([^']*)'|([^\\s>]+))`, "i"));
if (!match) {
const openingTag = getOpeningTag(tag);
if (!openingTag) {
return null;
}
return htmlDecode(match[2] ?? match[3] ?? match[4] ?? "");
const match = findAttribute(openingTag, attr);
return match ? htmlDecode(match.value) : null;
}
function setAttribute(tag, attr, value) {
const escaped = htmlEscape(value);
const attrRegex = new RegExp(`\\b${attr}\\s*=\\s*("([^"]*)"|'([^']*)'|([^\\s>]+))`, "i");
if (attrRegex.test(tag)) {
return tag.replace(attrRegex, `${attr}="${escaped}"`);
}
return tag.replace(/^<[^>]*>/, (openingTag) => openingTag.replace(/\/?>$/, (end) => ` ${attr}="${escaped}"${end}`));
return replaceOpeningTag(tag, (openingTag) => {
const match = findAttribute(openingTag, attr);
if (match) {
return `${openingTag.slice(0, match.start)}${attr}="${escaped}"${openingTag.slice(match.end)}`;
}
const selfClosing = /\/\s*>$/.test(openingTag);
const closeIndex = openingTag.lastIndexOf(">");
const beforeClose = openingTag.slice(0, closeIndex).replace(/\s*\/\s*$/, "");
return `${beforeClose} ${attr}="${escaped}"${selfClosing ? " /" : ""}>`;
});
}
function removeAttribute(tag, attr) {
return tag.replace(new RegExp(`\\s+${attr}\\s*=\\s*("([^"]*)"|'([^']*)'|([^\\s>]+))`, "i"), "");
return replaceOpeningTag(tag, (openingTag) => {
const match = findAttribute(openingTag, attr);
if (!match) {
return openingTag;
}
let start = match.start;
while (start > 0 && /\s/.test(openingTag[start - 1])) {
start -= 1;
}
return `${openingTag.slice(0, start)}${openingTag.slice(match.end)}`;
});
}
function replaceMissingMediaAttribute(tag, attr) {
@@ -474,3 +503,95 @@ function cleanCssUrl(value) {
}
return decoded;
}
function getOpeningTag(markup) {
const end = openingTagEndIndex(markup);
return end >= 0 ? markup.slice(0, end + 1) : null;
}
function replaceOpeningTag(markup, replacer) {
const end = openingTagEndIndex(markup);
if (end < 0) {
return markup;
}
return `${replacer(markup.slice(0, end + 1))}${markup.slice(end + 1)}`;
}
function openingTagEndIndex(markup) {
let quote = "";
for (let i = 0; i < markup.length; i += 1) {
const ch = markup[i];
if (quote) {
if (ch === quote) {
quote = "";
}
continue;
}
if (ch === '"' || ch === "'") {
quote = ch;
continue;
}
if (ch === ">") {
return i;
}
}
return -1;
}
function findAttribute(openingTag, attr) {
const attrLower = attr.toLowerCase();
const nameMatch = openingTag.match(/^<[^\s/>]+/);
let index = nameMatch ? nameMatch[0].length : 1;
while (index < openingTag.length) {
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
if (index >= openingTag.length || openingTag[index] === ">" || openingTag[index] === "/") {
return null;
}
const start = index;
while (index < openingTag.length && !/[\s=/>]/.test(openingTag[index])) {
index += 1;
}
const name = openingTag.slice(start, index);
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
let value = "";
if (openingTag[index] === "=") {
index += 1;
while (index < openingTag.length && /\s/.test(openingTag[index])) {
index += 1;
}
const quote = openingTag[index];
if (quote === '"' || quote === "'") {
index += 1;
const valueStart = index;
while (index < openingTag.length && openingTag[index] !== quote) {
index += 1;
}
value = openingTag.slice(valueStart, index);
if (openingTag[index] === quote) {
index += 1;
}
} else {
const valueStart = index;
while (index < openingTag.length && !/[\s>]/.test(openingTag[index])) {
index += 1;
}
value = openingTag.slice(valueStart, index);
}
}
if (name.toLowerCase() === attrLower) {
return { start, end: index, value };
}
}
return null;
}