@neurowire/ingest 0.3.0 → 0.4.0

package/dist/index.js

@@ -125,7 +125,10 @@ function attr(node, name) {
 }
 
 // src/util.ts
-import { GENERATOR } from "@neurowire/core";
+import {
+  GENERATOR,
+  stableId
+} from "@neurowire/core";
 function resolveUrl(href, base) {
   try {
     return new URL(href, base).toString();
@@ -153,13 +156,17 @@ function newestEntryDate(entries) {
   }
   return max === Number.NEGATIVE_INFINITY ? void 0 : new Date(max).toISOString();
 }
+function withStableId(entry) {
+  if (entry.id.trim()) return entry;
+  return { ...entry, id: stableId(entry.link, entry.title) };
+}
 function finalizeFeed(draft, ctx) {
   const updated = normDate(draft.updated) ?? newestEntryDate(draft.entries) ?? (/* @__PURE__ */ new Date()).toISOString();
   const feed = {
     id: draft.id?.trim() || ctx.sourceUrl,
     title: draft.title?.trim() || "Untitled",
     updated,
-    entries: draft.entries,
+    entries: draft.entries.map(withStableId),
     generator: { name: GENERATOR.name, version: GENERATOR.version }
   };
   if (draft.home) feed.home = resolveUrl(draft.home, ctx.sourceUrl);
@@ -238,7 +245,7 @@ function jsonLdEntry(node, ctx) {
   const title = str(node.headline) ?? str(node.name);
   if (!url || !title) return null;
   const link = resolveUrl(url, ctx.sourceUrl);
-  const entry = { id: link, title, link };
+  const entry = { id: "", title, link };
   const published = normDate(str(node.datePublished));
   const updated = normDate(str(node.dateModified));
   if (published) entry.published = published;
@@ -287,7 +294,7 @@ function fromSemantic($, ctx) {
     const href = $a.attr("href") ?? $el.find("a[href]").first().attr("href");
     if (!title || !href) return;
     const link = resolveUrl(href, ctx.sourceUrl);
-    const entry = { id: link, title, link };
+    const entry = { id: "", title, link };
     const $time = $el.find("time[datetime]").first();
     const date = normDate($time.attr("datetime") ?? $el.find("time").first().text().trim());
     if (date) entry.published = date;
@@ -351,7 +358,7 @@ function applyTemplate($, template, ctx) {
     const href = $link.attr("href") ?? $link.find("a").first().attr("href");
     if (!title2 || !href) return;
     const link = resolveUrl(href, ctx.sourceUrl);
-    const entry = { id: link, title: title2, link };
+    const entry = { id: "", title: title2, link };
     if (template.date) {
       const $date = $el.find(template.date).first();
       const date = normDate($date.attr("datetime") ?? $date.text().trim());
@@ -432,7 +439,7 @@ function atomEntry(node, ctx) {
   const href = pickLink(links, "alternate") ?? attr(links[0], "href") ?? text(get(node, "id")) ?? "";
   const link = resolveUrl(href, ctx.sourceUrl);
   const entry = {
-    id: text(get(node, "id")) ?? link,
+    id: text(get(node, "id")) ?? "",
     title: text(get(node, "title")) ?? "Untitled",
     link
   };
@@ -468,7 +475,7 @@ function rssItem(node, ctx) {
   const guid = text(get(node, "guid"));
   const link = resolveUrl(text(get(node, "link")) ?? guid ?? "", ctx.sourceUrl);
   const entry = {
-    id: guid ?? link,
+    id: guid ?? "",
     title: text(get(node, "title")) ?? "Untitled",
     link
   };
@@ -526,7 +533,7 @@ function parseJsonFeed(raw, ctx) {
   const entries = toArray(data.items).map((item) => {
     const link = resolveUrl(item.url ?? item.external_url ?? "", ctx.sourceUrl);
     const entry = {
-      id: item.id !== void 0 ? String(item.id) : link,
+      id: item.id !== void 0 ? String(item.id) : "",
       title: item.title ?? "Untitled",
       link
     };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@neurowire/ingest",
-  "version": "0.3.0",
+  "version": "0.4.0",
   "description": "Fetch, detect, and parse RSS/Atom/RDF/JSON feeds and HTML pages into the Neurowire model.",
   "type": "module",
   "main": "./dist/index.js",
@@ -48,7 +48,7 @@
     "cheerio": "^1.0.0",
     "fast-xml-parser": "^4.5.1",
     "zod": "^3.24.1",
-    "@neurowire/core": "0.4.0"
+    "@neurowire/core": "0.5.0"
   },
   "devDependencies": {
     "@types/node": "^22.10.5",