Improve Listing Description Generation

-> generalize the rendered content stripping behavior

Author: dragonstyle

src/project/types/website/listing/website-listing-feed.ts

@@ -28,11 +28,14 @@ import {
   kFieldAuthor,
   kFieldCategories,
   kFieldImage,
+  kInlineCodeStyle,
   kItems,
   kListing,
+  kUrlsToAbsolute,
   ListingDescriptor,
   ListingFeedOptions,
   ListingItem,
+  RenderedContentOptions,
   renderedContentReader,
   RenderedContents,
 } from "./website-listing-shared.ts";
@@ -232,6 +235,17 @@ export async function createFeed(
   return feedFiles;
 }
 
+const kFeedOptions: RenderedContentOptions = {
+  remove: {
+    anchors: true,
+  },
+  transform: {
+    [kUrlsToAbsolute]: true,
+    [kInlineCodeStyle]: true,
+    math: true,
+  },
+};
+
 export function completeStagedFeeds(
   context: ProjectContext,
   outputFiles: ProjectOutputFile[],
@@ -251,7 +265,11 @@ export function completeStagedFeeds(
         // Any feed files for this output file
         const files = resolvePathGlobs(dir, [`${stem}${kStagedFileGlob}`], []);
 
-        const contentReader = renderedContentReader(context, true, siteUrl);
+        const contentReader = renderedContentReader(
+          context,
+          kFeedOptions,
+          siteUrl,
+        );
 
         for (const feedFile of files.include) {
           // Info about this feed file

src/project/types/website/listing/website-listing-read.ts

@@ -272,7 +272,9 @@ export function completeListingDescriptions(
   outputFiles: ProjectOutputFile[],
   _incremental: boolean,
 ) {
-  const contentReader = renderedContentReader(context, false);
+  const contentReader = renderedContentReader(context, {
+    remove: { links: true, images: true },
+  });
 
   // Go through any output files and fix up any feeds associated with them
   outputFiles.forEach((outputFile) => {

src/project/types/website/listing/website-listing-shared.ts

@@ -201,9 +201,25 @@ export interface RenderedContents {
   fullContents: string | undefined;
 }
 
+export const kInlineCodeStyle = "inline-code-style";
+export const kUrlsToAbsolute = "urls-to-absolute";
+
+export interface RenderedContentOptions {
+  remove?: {
+    anchors?: boolean;
+    links?: boolean;
+    images?: boolean;
+  };
+  transform?: {
+    [kInlineCodeStyle]?: boolean;
+    math?: boolean;
+    [kUrlsToAbsolute]?: boolean;
+  };
+}
+
 export const renderedContentReader = (
   project: ProjectContext,
-  forFeed: boolean,
+  options: RenderedContentOptions,
   siteUrl?: string,
 ) => {
   const renderedContent: Record<string, RenderedContents> = {};
@@ -212,7 +228,7 @@ export const renderedContentReader = (
       renderedContent[filePath] = readRenderedContents(
         filePath,
         project,
-        forFeed,
+        options,
         siteUrl,
       );
     }
@@ -238,7 +254,7 @@ export const absoluteUrl = (siteUrl: string, url: string) => {
 export function readRenderedContents(
   filePath: string,
   project: ProjectContext,
-  forFeed: boolean,
+  options: RenderedContentOptions,
   siteUrl?: string,
 ): RenderedContents {
   const htmlInput = Deno.readTextFileSync(filePath);
@@ -265,7 +281,7 @@ export function readRenderedContents(
   }
 
   // Convert any images to have absolute paths
-  if (forFeed && siteUrl) {
+  if (options.transform?.[kUrlsToAbsolute] && siteUrl) {
     const imgNodes = doc.querySelectorAll("img");
     if (imgNodes) {
       for (const imgNode of imgNodes) {
@@ -305,16 +321,31 @@ export function readRenderedContents(
     });
   });
 
-  if (forFeed) {
-    // String unacceptable links
+  // Strip unacceptable tags
+  const stripTags = [];
+  if (options.remove?.images) {
+    stripTags.push("img");
+  }
+  stripTags.forEach((tag) => {
+    const nodes = doc.querySelectorAll(`${tag}`);
+    nodes?.forEach((node) => {
+      const el = node as Element;
+      el.remove();
+    });
+  });
+
+  if (options.remove?.anchors) {
+    // Strip unacceptable links
     const relativeLinkSel = 'a[href^="#"]';
     const linkNodes = doc.querySelectorAll(relativeLinkSel);
     linkNodes.forEach((linkNode) => {
       const nodesToMove = linkNode.childNodes;
       linkNode.after(...nodesToMove);
       linkNode.remove();
     });
+  }
 
+  if (options.transform?.[kInlineCodeStyle]) {
     // Process code to apply styles for syntax highlighting
     const highlightingMap = defaultSyntaxHighlightingClassMap();
     const spanNodes = doc.querySelectorAll("code span");
@@ -337,7 +368,9 @@ export function readRenderedContents(
       const codeBlockEl = codeBlockNode as Element;
       codeBlockEl.setAttribute("style", codeStyle);
     }
+  }
 
+  if (options.transform?.math) {
     // Process math using webtex
     const trimMath = (str: string) => {
       // Text of math is prefixed by the below
@@ -359,7 +392,9 @@ export function readRenderedContents(
       );
       mathNode.parentElement?.replaceChild(imgEl, mathNode);
     }
-  } else {
+  }
+
+  if (options.remove?.links) {
     // String all links
     const relativeLinkSel = "a";
     const linkNodes = doc.querySelectorAll(relativeLinkSel);