@citation-media/html-parser 0.1.0 → 0.1.1

package/dist/clean.js

@@ -19,6 +19,10 @@ const remove = {
 };
 /** The cleaned page as HTML. */
 export const cleanHtml = (input, options) => {
+    // An empty page has nothing to stream; some runtimes never finish transforming an empty body.
+    if (!(input instanceof Response) && input.trim() === "") {
+        return Promise.resolve("");
+    }
     const base = new URL(options.url);
     const removed = options.remove ?? {};
     let rewriter = new HTMLRewriter();

package/dist/parse.js

@@ -58,16 +58,27 @@ const hiddenFromMarkdown = [
 const maxHeadings = 200;
 const maxImages = 500;
 const maxResources = 500;
-// Follows an element to its end tag. Void elements such as <img> have none, and HTMLRewriter throws
-// for them; elements closed implicitly, such as <li> without </li>, end with their parent.
-const enter = (element, onEnd) => {
+/**
+ * Follows an element to its end tag. HTMLRewriter keeps only the last end-tag handler an element
+ * gets, so every handler adds its action to the element's list and registers the whole list; the
+ * last registration then runs them all. Void elements such as <img> have no end tag, and
+ * HTMLRewriter throws for them; elements closed implicitly, such as <li> without </li>, end with
+ * their parent.
+ */
+const enter = (pass, element, onEnd) => {
+    const actions = [...pass.endActions, onEnd];
     try {
-        element.onEndTag(onEnd);
-        return true;
+        element.onEndTag(() => {
+            for (const action of actions) {
+                action();
+            }
+        });
     }
     catch {
         return false;
     }
+    pass.endActions = actions;
+    return true;
 };
 const clean = (text) => collapseWhitespace(removeHyphenation(decodeEntities(text))).trim();
 const addLink = (links, found, base) => {
@@ -96,7 +107,7 @@ const collecting = (pass) => pass.inContext > 0;
 const writing = (pass) => collecting(pass) && pass.hidden === 0 && pass.markdown !== undefined;
 const hide = (pass) => ({
     element(element) {
-        if (enter(element, () => (pass.hidden -= 1))) {
+        if (enter(pass, element, () => (pass.hidden -= 1))) {
             pass.hidden += 1;
         }
     },
@@ -108,7 +119,7 @@ const onStructure = (rewriter, pass) => {
         next = next.on(context, {
             element(element) {
                 pass.contextFound = true;
-                if (enter(element, () => (pass.inContext -= 1))) {
+                if (enter(pass, element, () => (pass.inContext -= 1))) {
                     pass.inContext += 1;
                 }
             },
@@ -138,7 +149,7 @@ const onStructure = (rewriter, pass) => {
                 return;
             }
             pass.markdown?.open(tag);
-            enter(element, () => {
+            enter(pass, element, () => {
                 if (writing(pass)) {
                     pass.markdown?.close(tag);
                 }
@@ -165,7 +176,7 @@ const onLinks = (rewriter, pass) => rewriter.on("a[href]", {
         if (written) {
             pass.markdown?.openLink(url);
         }
-        enter(element, () => {
+        enter(pass, element, () => {
             if (written) {
                 pass.markdown?.closeLink();
             }
@@ -251,7 +262,7 @@ const onMeta = (rewriter, pass) => rewriter
     .on("title", {
     element(element) {
         pass.title = [];
-        enter(element, () => {
+        enter(pass, element, () => {
             pass.meta.title = clean((pass.title ?? []).join(""));
             pass.title = undefined;
         });
@@ -282,7 +293,7 @@ const onHeadings = (rewriter, pass) => rewriter.on("h1, h2, h3, h4, h5, h6", {
         const level = Number(element.tagName.slice(1));
         pass.headingCounts[level - 1] = (pass.headingCounts[level - 1] ?? 0) + 1;
         pass.heading = { level, text: [] };
-        enter(element, () => {
+        enter(pass, element, () => {
             const text = clean(pass.heading?.text.join("") ?? "");
             if (pass.heading && text && pass.headings.length < maxHeadings) {
                 pass.headings.push({ level: pass.heading.level, text });
@@ -338,6 +349,7 @@ export const parseHtml = async (input, options) => {
         base: new URL(options.url),
         classCounts: Object.fromEntries((options.classPrefixes ?? []).map((prefix) => [prefix, 0])),
         contextFound: false,
+        endActions: [],
         heading: undefined,
         headingCounts: [0, 0, 0, 0, 0, 0],
         headings: [],
@@ -356,7 +368,17 @@ export const parseHtml = async (input, options) => {
         resources: { frames: [], links: [], scripts: [] },
         title: undefined,
     };
-    let rewriter = onImages(onLinks(onStructure(new HTMLRewriter(), pass), pass), pass);
+    // An empty page has nothing to stream; some runtimes never finish transforming an empty body.
+    if (!(input instanceof Response) && input.trim() === "") {
+        return resultOf(pass);
+    }
+    // Registered first, so every element starts with an empty list of end-tag actions.
+    const first = new HTMLRewriter().on("*", {
+        element() {
+            pass.endActions = [];
+        },
+    });
+    let rewriter = onImages(onLinks(onStructure(first, pass), pass), pass);
     if (options.resources) {
         rewriter = onResources(rewriter, pass);
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@citation-media/html-parser",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "Streaming HTML parser for Cloudflare Workers: links, images, resources, and Markdown without a DOM, built on HTMLRewriter.",
   "keywords": [
     "cloudflare-workers",