feat: update extract script for october 2025; misc fixes and robustness improvements

Author: transitive-bullshit

package.json

@@ -33,10 +33,15 @@
     "node-id3": "^0.2.6",
     "openai-fetch": "^3.3.1",
     "p-map": "^7.0.2",
+    "p-race": "^3.1.0",
     "patchright": "^1.56.1",
     "pdfkit": "^0.15.0",
     "playwright": "^1.48.0",
     "playwright-core": "^1.48.0",
+    "sharp": "^0.34.4",
+    "tar": "^7.5.1",
+    "tempy": "^3.1.0",
+    "type-fest": "^5.1.0",
     "unrealspeech-api": "^1.0.2"
   },
   "devDependencies": {

pnpm-lock.yaml

@@ -0,0 +1,2 @@
+onlyBuiltDependencies:
+  - sharp

pnpm-workspace.yaml

@@ -144,7 +144,7 @@ By ${authors.join(', ')}`
       .replaceAll('\n', '\n\n')
 
     // Split the text in this chapter into paragraphs.
-    const t = `${tocItem.title}
+    const t = `${tocItem.label}
 
 ${text}`.split('\n\n')
 
@@ -175,7 +175,7 @@ ${text}`.split('\n\n')
 
     for (const [k, element] of t.entries()) {
       batches.push({
-        title: k === 0 ? tocItem.title : undefined,
+        title: k === 0 ? tocItem.label : undefined,
         text: element!
       })
     }

src/browserbase-experiment.ts

@@ -53,7 +53,7 @@ ${metadata.toc
   )
   .map(
     (tocItem) =>
-      `- [${tocItem.title}](#${tocItem.title.toLowerCase().replaceAll(/[^\da-z]+/g, '-')})`
+      `- [${tocItem.label}](#${tocItem.label.toLowerCase().replaceAll(/[^\da-z]+/g, '-')})`
   )
   .join('\n')}
 
@@ -78,7 +78,7 @@ ${metadata.toc
 
     output += `
 
-## ${tocItem.title}
+## ${tocItem.label}
 
 ${text}`
 

src/export-book-audio.ts

@@ -83,9 +83,9 @@ async function main() {
     const chunks = content.slice(index, nextIndex)
     const text = chunks.map((chunk) => chunk.text).join(' ')
 
-    ;(doc as any).outline.addItem(tocItem.title)
+    ;(doc as any).outline.addItem(tocItem.label)
     doc.fontSize(20)
-    doc.text(tocItem.title, { align: 'center', lineGap: 16 })
+    doc.text(tocItem.label, { align: 'center', lineGap: 16 })
 
     doc.fontSize(fontSize)
     doc.moveDown(1)