Add EasyList filter support

src/asset-inliner.mjs

@@ -144,6 +144,7 @@ export class AssetInliner {
     this.referer = options.referer;
     this.maxAssetBytes = options.maxAssetBytes || 30 * 1024 * 1024;
     this.maxInlineStyleBytes = options.maxInlineStyleBytes || 128 * 1024;
+    this.shouldBlockAsset = options.shouldBlockAsset || null;
     this.cache = new Map();
     this.warnings = [];
   }
@@ -214,7 +215,7 @@ export class AssetInliner {
       if (!absolute || absolute.startsWith("data:")) {
         return "";
       }
-      const css = await this.fetchText(absolute, baseUrl);
+      const css = await this.fetchText(absolute, baseUrl, "stylesheet");
       if (css == null) {
         return "";
       }
@@ -242,7 +243,7 @@ export class AssetInliner {
       }
     }
 
-    const dataUri = await this.toDataUri(href, baseUrl);
+    const dataUri = await this.toDataUri(href, baseUrl, linkResourceType(asValue));
     if (!dataUri) {
       return "";
     }
@@ -251,12 +252,13 @@ export class AssetInliner {
 
   async rewriteMediaAttributes(tag, baseUrl) {
     let output = tag;
+    const tagName = getTagName(output);
     for (const attr of ["src", "poster", "data"]) {
       const value = getAttribute(output, attr);
       if (!value) {
         continue;
       }
-      const dataUri = await this.toDataUri(value, baseUrl);
+      const dataUri = await this.toDataUri(value, baseUrl, mediaResourceType(tagName, attr));
       if (dataUri) {
         output = setAttribute(output, attr, dataUri);
       } else {
@@ -291,7 +293,7 @@ export class AssetInliner {
     if (!absolute || absolute.startsWith("data:")) {
       return tag;
     }
-    const text = await this.fetchText(absolute, baseUrl);
+    const text = await this.fetchText(absolute, baseUrl, "subdocument");
     if (text != null) {
       const inlined = await this.inlineHtml(text, absolute, { depth: depth + 1 });
       let rewritten = removeAttribute(tag, "src");
@@ -308,7 +310,7 @@ export class AssetInliner {
     const rewritten = [];
     for (const candidate of candidates) {
       const [urlPart, ...descriptor] = candidate.split(/\s+/);
-      const dataUri = await this.toDataUri(urlPart, baseUrl);
+      const dataUri = await this.toDataUri(urlPart, baseUrl, "image");
       rewritten.push([dataUri || TRANSPARENT_IMAGE_DATA_URI, ...descriptor].join(" "));
     }
     return rewritten.join(", ");
@@ -323,7 +325,7 @@ export class AssetInliner {
         if (!absolute || absolute.startsWith("data:")) {
           return "";
         }
-        const imported = await this.fetchText(absolute, baseUrl);
+        const imported = await this.fetchText(absolute, baseUrl, "stylesheet");
         if (imported == null) {
           return "";
         }
@@ -336,33 +338,34 @@ export class AssetInliner {
       if (!raw || raw.startsWith("#") || /^%23/i.test(raw) || /^(?:data|blob|about|javascript):/i.test(raw)) {
         return match[0];
       }
-      const dataUri = await this.toDataUri(raw, baseUrl);
+      const dataUri = await this.toDataUri(raw, baseUrl, cssResourceType(raw, baseUrl));
       return dataUri ? `url("${dataUri}")` : "url(about:blank)";
     });
 
     return output;
   }
 
-  async toDataUri(rawUrl, baseUrl) {
+  async toDataUri(rawUrl, baseUrl, resourceType = "other") {
     const absolute = resolveUrl(rawUrl, baseUrl);
     if (!absolute || absolute.startsWith("data:")) {
       return absolute;
     }
-    if (this.cache.has(absolute)) {
-      return this.cache.get(absolute);
+    const cacheKey = `${resourceType}:${absolute}`;
+    if (this.cache.has(cacheKey)) {
+      return this.cache.get(cacheKey);
     }
-    const asset = await this.fetchAsset(absolute, baseUrl);
+    const asset = await this.fetchAsset(absolute, baseUrl, resourceType);
     if (!asset) {
-      this.cache.set(absolute, null);
+      this.cache.set(cacheKey, null);
       return null;
     }
     const dataUri = `data:${asset.contentType};base64,${asset.bytes.toString("base64")}`;
-    this.cache.set(absolute, dataUri);
+    this.cache.set(cacheKey, dataUri);
     return dataUri;
   }
 
-  async fetchText(rawUrl, baseUrl) {
-    const asset = await this.fetchAsset(rawUrl, baseUrl);
+  async fetchText(rawUrl, baseUrl, resourceType = "other") {
+    const asset = await this.fetchAsset(rawUrl, baseUrl, resourceType);
     if (!asset) {
       return null;
     }
@@ -373,12 +376,15 @@ export class AssetInliner {
     return asset.bytes.toString("utf8");
   }
 
-  async fetchAsset(rawUrl, baseUrl) {
+  async fetchAsset(rawUrl, baseUrl, resourceType = "other") {
     const absolute = resolveUrl(rawUrl, baseUrl);
     if (!absolute || absolute.startsWith("data:")) {
       return null;
     }
     try {
+      if (this.shouldBlockAsset?.(absolute, resourceType)) {
+        return null;
+      }
       if (isFileUrl(absolute)) {
         const filePath = fileURLToPath(absolute);
         const bytes = await fs.readFile(filePath);
@@ -512,6 +518,43 @@ function mimeFromUrl(rawUrl) {
   return MIME_BY_EXT.get(path.extname(pathname).toLowerCase()) || "application/octet-stream";
 }
 
+function linkResourceType(asValue) {
+  const normalized = String(asValue || "").toLowerCase();
+  if (normalized === "font") return "font";
+  if (normalized === "script") return "script";
+  if (normalized === "style") return "stylesheet";
+  if (normalized === "document") return "subdocument";
+  if (normalized === "audio" || normalized === "video") return "media";
+  return "image";
+}
+
+function mediaResourceType(tagName, attr) {
+  if (tagName === "iframe") return "subdocument";
+  if (tagName === "object" || tagName === "embed") return "object";
+  if (tagName === "audio" || tagName === "video") return "media";
+  if (attr === "poster") return "image";
+  if (tagName === "track") return "other";
+  return "image";
+}
+
+function cssResourceType(rawUrl, baseUrl) {
+  const absolute = resolveUrl(rawUrl, baseUrl) || rawUrl;
+  let pathname = absolute;
+  try {
+    pathname = new URL(absolute).pathname;
+  } catch {
+    // Keep raw string.
+  }
+  const ext = path.extname(pathname).toLowerCase();
+  if ([".woff", ".woff2", ".ttf", ".otf"].includes(ext)) return "font";
+  if ([".mp4", ".webm", ".mp3", ".m4a"].includes(ext)) return "media";
+  return "image";
+}
+
+function getTagName(markup) {
+  return markup.match(/^<([a-z0-9:-]+)/i)?.[1]?.toLowerCase() || "";
+}
+
 function getAttribute(tag, attr) {
   const openingTag = getOpeningTag(tag);
   if (!openingTag) {