[codex] Lazy parse deprecated iTunes summaries

src/parser/__test__/entity-expansion.test.ts

@@ -49,4 +49,42 @@ describe("entity expansion limit", () => {
     expect(result?.items).toHaveLength(1);
     expect(result?.items?.[0]).toHaveProperty("title", "Item");
   });
+
+  it("does not count deprecated itunes:summary entities when description exists", () => {
+    const entityRef = "<p>"hello"</p>";
+    const count = 100001;
+    const xml = `<?xml version="1.0" encoding="UTF-8"?>
+<rss version="2.0" xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd">
+  <channel>
+    <title>Entity expansion test</title>
+    <description>Channel description</description>
+    <itunes:summary>${entityRef.repeat(count)}</itunes:summary>
+    <item>
+      <title>Item</title>
+      <description>Item description</description>
+      <itunes:summary>${entityRef.repeat(count)}</itunes:summary>
+      <enclosure url="https://example.com/ep.mp3" length="1000" type="audio/mpeg"/>
+    </item>
+  </channel>
+</rss>`;
+
+    const result = parseFeed(xml, { allowMissingGuid: true });
+
+    expect(result).not.toBeNull();
+    expect(result).toHaveProperty("description", "Channel description");
+    expect(result).not.toHaveProperty("summary");
+    expect(result?.items?.[0]).toHaveProperty("description", "Item description");
+    expect(result?.items?.[0]).not.toHaveProperty("summary");
+  });
+
+  it("still rejects too many entities outside deprecated itunes:summary", () => {
+    const entityRef = "&amp; ";
+    const count = 100001;
+    const descriptionContent = entityRef.repeat(count);
+    const xml = minimalRssPrefix + descriptionContent + minimalRssSuffix;
+
+    expect(() => parseFeed(xml, { allowMissingGuid: true })).toThrow(
+      "Entity expansion count limit exceeded"
+    );
+  });
 });

src/parser/__test__/feed.test.ts

@@ -142,6 +142,20 @@ describe("feed handling", () => {
       expect(result).toHaveProperty("description", "<p>bye</p>");
     });
 
+    it("decodes itunes:summary only when it is used as a fallback", () => {
+      const xml = helpers.spliceFeed(
+        feed,
+        `
+          <itunes:summary>&lt;p&gt;bye&lt;/p&gt;</itunes:summary>
+        `
+      );
+
+      const result = parseFeed(xml);
+
+      expect(result).toHaveProperty("description", "<p>bye</p>");
+      expect(result).not.toHaveProperty("summary");
+    });
+
     it("prefers description value when the fall back exists", () => {
       const xml = helpers.spliceFeed(
         feed,
@@ -173,7 +187,7 @@ describe("feed handling", () => {
   });
 
   describe("summary", () => {
-    it("extracts the value", () => {
+    it("ignores the deprecated value by default", () => {
       const xml = helpers.spliceFeed(
         feed,
         `
@@ -183,6 +197,19 @@ describe("feed handling", () => {
 
       const result = parseFeed(xml);
 
+      expect(result).not.toHaveProperty("summary");
+    });
+
+    it("extracts the value when requested", () => {
+      const xml = helpers.spliceFeed(
+        feed,
+        `
+          <itunes:summary>hi</itunes:summary>
+        `
+      );
+
+      const result = parseFeed(xml, { includeItunesSummary: true });
+
       expect(result).toHaveProperty("summary", "hi");
     });
   });

src/parser/__test__/item.test.ts

@@ -1415,6 +1415,25 @@ describe("item handling", () => {
       expect(first).toHaveProperty("description", "<p>bye</p>");
     });
 
+    it("decodes itunes:summary only when it is used as a fallback", () => {
+      const xml = helpers.spliceFeed(
+        feed,
+        `
+        <item>
+          <itunes:summary>&lt;p&gt;bye&lt;/p&gt;</itunes:summary>
+          <guid isPermaLink="true">https://example.com/ep0003</guid>
+          <enclosure url="https://aphid.fireside.fm/d/1437767933/65632ad5-59b2-4e30-82d1-13845dce07dd/d11384ea-69b5-4e33-bd0e-5d33fdba8a0d.mp3" length="78034115" type="audio/mpeg"/>
+        </item>
+        `
+      );
+
+      const result = parseFeed(xml);
+      const [first] = result.items;
+
+      expect(first).toHaveProperty("description", "<p>bye</p>");
+      expect(first).not.toHaveProperty("summary");
+    });
+
     it("prefers content:encoded value when falling back", () => {
       const xml = helpers.spliceFeed(
         feed,
@@ -1441,7 +1460,7 @@ describe("item handling", () => {
   });
 
   describe("summary", () => {
-    it("extracts the value", () => {
+    it("ignores the deprecated value by default", () => {
       const xml = helpers.spliceFeed(
         feed,
         `
@@ -1457,6 +1476,25 @@ describe("item handling", () => {
       const result = parseFeed(xml);
       const [first] = result.items;
 
+      expect(first).not.toHaveProperty("summary");
+    });
+
+    it("extracts the value when requested", () => {
+      const xml = helpers.spliceFeed(
+        feed,
+        `
+        <item>
+          <itunes:summary>
+          <![CDATA[ <p>bye</p> ]]>
+          </itunes:summary>          <guid isPermaLink="true">https://example.com/ep0003</guid>
+          <enclosure url="https://aphid.fireside.fm/d/1437767933/65632ad5-59b2-4e30-82d1-13845dce07dd/d11384ea-69b5-4e33-bd0e-5d33fdba8a0d.mp3" length="78034115" type="audio/mpeg"/>
+        </item>
+        `
+      );
+
+      const result = parseFeed(xml, { includeItunesSummary: true });
+      const [first] = result.items;
+
       expect(first).toHaveProperty("summary", "<p>bye</p>");
     });
 
@@ -1473,7 +1511,7 @@ describe("item handling", () => {
         `
       );
 
-      const result = parseFeed(xml);
+      const result = parseFeed(xml, { includeItunesSummary: true });
       const [first] = result.items;
 
       expect(first).toHaveProperty("summary", "bye");

src/parser/feed.ts

@@ -1,5 +1,7 @@
 /* eslint-disable @typescript-eslint/no-unsafe-member-access */
 /* eslint-disable @typescript-eslint/no-unsafe-assignment */
+import he from "he";
+
 import { logger } from "../logger";
 
 import {
@@ -18,6 +20,7 @@ import {
 import { BasicFeed, ItunesFeedType } from "./types";
 import type { FeedType, XmlNode } from "./types";
 import { categoryLookup } from "./itunes-categories";
+import type { ParserOptions } from "./unified";
 
 function getTitle(feed: XmlNode): string {
   const node = firstWithValue(feed.title);
@@ -85,7 +88,7 @@ function getSummary(feed: XmlNode): undefined | { summary: string } {
   if (node) {
     const nodeValue = getText(node);
     if (nodeValue) {
-      return { summary: sanitizeMultipleSpaces(sanitizeNewLines(nodeValue)) };
+      return { summary: sanitizeMultipleSpaces(sanitizeNewLines(he.decode(nodeValue))) };
     }
   }
   return undefined;
@@ -627,7 +630,7 @@ function getTimeToLive(feed: XmlNode): undefined | { ttl: number } {
   return undefined;
 }
 
-export function handleFeed(feed: XmlNode, feedType: FeedType): BasicFeed {
+export function handleFeed(feed: XmlNode, feedType: FeedType, options?: ParserOptions): BasicFeed {
   return {
     lastUpdate: new Date(),
     type: feedType,
@@ -650,7 +653,7 @@ export function handleFeed(feed: XmlNode, feedType: FeedType): BasicFeed {
     ...getAuthor(feed),
     ...getOwner(feed),
     ...getImage(feed),
-    ...getSummary(feed),
+    ...(options?.includeItunesSummary ? getSummary(feed) : undefined),
     ...getSubtitle(feed),
     ...getItunesTitle(feed),
     ...getCopyright(feed),

src/parser/item.ts

@@ -3,6 +3,8 @@
 /* eslint-disable @typescript-eslint/no-unsafe-member-access */
 /* eslint-disable @typescript-eslint/no-unsafe-call */
 
+import he from "he";
+
 import { logger } from "../logger";
 
 import {
@@ -23,6 +25,7 @@ import {
 } from "./shared";
 import type { BasicFeed, Enclosure, Episode, XmlNode } from "./types";
 import { unescape } from "./unescape";
+import type { ParserOptions } from "./unified";
 
 export enum ItunesEpisodeType {
   Full = "full",
@@ -292,7 +295,7 @@ function getSummary(item: XmlNode): undefined | { summary: string } {
   if (node) {
     const summaryValue = getText(node);
     if (summaryValue) {
-      return { summary: summaryValue };
+      return { summary: he.decode(summaryValue) };
     }
   }
 
@@ -311,7 +314,7 @@ function getSubtitle(item: XmlNode): undefined | { subtitle: string } {
   return undefined;
 }
 
-export function handleItem(item: XmlNode, _feed: BasicFeed): Episode {
+export function handleItem(item: XmlNode, _feed: BasicFeed, options?: ParserOptions): Episode {
   return {
     guid: getGuid(item),
     // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
@@ -329,7 +332,7 @@ export function handleItem(item: XmlNode, _feed: BasicFeed): Episode {
     ...getKeywords(item),
     ...getPubDate(item),
     ...getImage(item),
-    ...getSummary(item),
+    ...(options?.includeItunesSummary ? getSummary(item) : undefined),
     ...getDescription(item),
     ...getSubtitle(item),
     ...getContent(item),

src/parser/unified.ts

@@ -31,6 +31,8 @@ export type ParserOptions = {
   allowInsecureHTTPMetaboost?: boolean;
   /** Max UTF-8 bytes for the XML string. When set, overrides PARSER_MAX_FEED_BODY_BYTES and the 10 MiB default. */
   maxFeedBodyBytes?: number;
+  /** Include deprecated itunes:summary fields as summary properties. Description fallback still uses itunes:summary when needed. */
+  includeItunesSummary?: boolean;
 };
 
 function handlePodcastSeasons(feedObj: BasicFeed) {
@@ -63,7 +65,7 @@ export function unifiedParser(theFeed: XmlNode, type: FeedType, options?: Parser
     return null;
   }
 
-  let feedObj = handleFeed(theFeed.rss.channel, type);
+  let feedObj = handleFeed(theFeed.rss.channel, type, options);
 
   let phaseSupport: PhaseUpdate = {};
 
@@ -84,7 +86,7 @@ export function unifiedParser(theFeed: XmlNode, type: FeedType, options?: Parser
           return undefined;
         }
 
-        let newFeedItem: Episode = handleItem(item, feedObj);
+        let newFeedItem: Episode = handleItem(item, feedObj, options);
 
         // Item Phase Support
         const itemResult = updateItem(item, theFeed, undefined, options);

src/parser/xml-parser.ts

@@ -18,13 +18,19 @@ const parserOptions = {
   parseAttributeValue: false,
   trimValues: true,
   parseTrueNumberOnly: false,
-  tagValueProcessor: (_tagName: string, tagValue: string) => he.decode(tagValue),
+  tagValueProcessor: (_tagName: string, tagValue: string) => {
+    if (_tagName === "itunes:summary") {
+      return tagValue;
+    }
+    return he.decode(tagValue);
+  },
   attributeValueProcessor: (_tagName: string, tagValue: string) => he.decode(tagValue),
   stopNodes: ["parse-me-as-string"],
   processEntities: {
     enabled: true,
     maxEntityCount: ENTITY_EXPANSION_LIMIT,
     maxTotalExpansions: ENTITY_EXPANSION_LIMIT,
+    tagFilter: (tagName: string) => tagName !== "itunes:summary",
   },
 };