magpie-html 0.1.3 → 0.1.4

package/dist/index.js

@@ -1,7 +1,7 @@
 import { Readability } from '@mozilla/readability';
 import { parseHTML as parseHTML$1 } from 'linkedom';
 import vm4 from 'vm';
-import { setTimeout as setTimeout$1, clearTimeout as clearTimeout$1, setInterval, clearInterval, setImmediate, clearImmediate } from 'timers';
+import { setTimeout, clearTimeout, setInterval, clearInterval, setImmediate, clearImmediate } from 'timers';
 
 // src/content/quality.ts
 function countWords(text) {
@@ -832,6 +832,32 @@ function parseAtomDate(dateString) {
 }
 
 // src/feed/atom/extract-entry.ts
+function extractAtomDate(element) {
+  let dateText = element.querySelector("updated")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  dateText = element.querySelector("modified")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  dateText = element.querySelector("issued")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  const dcDateElements = element.children.filter((child) => child.tagName === "dc:date");
+  if (dcDateElements.length > 0) {
+    dateText = dcDateElements[0].textContent;
+    if (dateText) {
+      const parsed = parseAtomDate(dateText);
+      if (parsed) return parsed;
+    }
+  }
+  return null;
+}
 function extractPerson(element) {
   const name = element.querySelector("name")?.textContent;
   if (!name) {
@@ -974,13 +1000,11 @@ function extractEntry(entryElement) {
   if (!title) {
     throw new Error("Invalid Atom entry: missing required <title> element");
   }
-  const updatedRaw = entryElement.querySelector("updated")?.textContent;
-  if (!updatedRaw) {
-    throw new Error("Invalid Atom entry: missing required <updated> element");
-  }
-  const updated = parseAtomDate(updatedRaw);
+  const updated = extractAtomDate(entryElement);
   if (!updated) {
-    throw new Error("Invalid Atom entry: invalid <updated> date");
+    throw new Error(
+      "Invalid Atom entry: missing or invalid date (tried <updated>, <modified>, <issued>, <dc:date>)"
+    );
   }
   const entry = {
     id: cleanText(id),
@@ -1230,6 +1254,32 @@ function parseXML(xml) {
 }
 
 // src/feed/atom/extract-feed.ts
+function extractAtomDate2(element) {
+  let dateText = element.querySelector("updated")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  dateText = element.querySelector("modified")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  dateText = element.querySelector("issued")?.textContent;
+  if (dateText) {
+    const parsed = parseAtomDate(dateText);
+    if (parsed) return parsed;
+  }
+  const dcDateElements = element.children.filter((child) => child.tagName === "dc:date");
+  if (dcDateElements.length > 0) {
+    dateText = dcDateElements[0].textContent;
+    if (dateText) {
+      const parsed = parseAtomDate(dateText);
+      if (parsed) return parsed;
+    }
+  }
+  return null;
+}
 function extractPerson2(element) {
   const name = element.querySelector("name")?.textContent;
   if (!name) {
@@ -1377,13 +1427,11 @@ function extractFeed(xml) {
   if (!title) {
     throw new Error("Invalid Atom feed: missing required <title> element");
   }
-  const updatedRaw = feed.querySelector("updated")?.textContent;
-  if (!updatedRaw) {
-    throw new Error("Invalid Atom feed: missing required <updated> element");
-  }
-  const updated = parseAtomDate(updatedRaw);
+  const updated = extractAtomDate2(feed);
   if (!updated) {
-    throw new Error("Invalid Atom feed: invalid <updated> date");
+    throw new Error(
+      "Invalid Atom feed: missing or invalid date (tried <updated>, <modified>, <issued>, <dc:date>)"
+    );
   }
   const result = {
     id: cleanText(id),
@@ -2445,13 +2493,12 @@ async function pluck(input, init) {
   const startTime = Date.now();
   const options = normalizeOptions2(init);
   const originalUrl = typeof input === "string" || input instanceof URL ? String(input) : input.url;
-  const abortController = new AbortController();
-  const timeoutId = setTimeout(() => abortController.abort(), options.timeout);
+  const signal = AbortSignal.timeout(options.timeout);
   try {
     const { response, redirectChain, redirectDuration } = await followRedirects(
       input,
       options,
-      abortController.signal
+      signal
     );
     const finalUrl = response.url;
     if (options.throwOnHttpError && !response.ok) {
@@ -2488,15 +2535,13 @@ async function pluck(input, init) {
     if (error instanceof PluckTimeoutError || error instanceof PluckNetworkError) {
       throw error;
     }
-    if (error.name === "AbortError") {
+    if (error.name === "TimeoutError") {
       throw new PluckTimeoutError(`Request timeout after ${options.timeout}ms`, options.timeout);
     }
     if (error instanceof TypeError) {
       throw new PluckNetworkError(`Network error: ${error.message}`, error);
     }
     throw error;
-  } finally {
-    clearTimeout(timeoutId);
   }
 }
 function normalizeOptions2(init) {
@@ -5111,8 +5156,8 @@ ${err.stack}` : ""}`.trim());
 }
 function installAsyncEnv(init) {
   const { globalObj } = init;
-  const hostSetTimeout = setTimeout$1;
-  const hostClearTimeout = clearTimeout$1;
+  const hostSetTimeout = setTimeout;
+  const hostClearTimeout = clearTimeout;
   const hostSetInterval = setInterval;
   const hostClearInterval = clearInterval;
   const hostSetImmediate = setImmediate;
@@ -6211,7 +6256,7 @@ function isNodeRuntime() {
   return typeof process !== "undefined" && typeof process.versions === "object" && typeof process.versions.node === "string";
 }
 function sleep(ms) {
-  return new Promise((resolve) => setTimeout$1(resolve, ms));
+  return new Promise((resolve) => setTimeout(resolve, ms));
 }
 function normalizeInit(init) {
   return {