From b442190049cdfda3ddb9cda60effe85b6aef5603 Mon Sep 17 00:00:00 2001
From: Shane Friedman <smoores-gpg@friedmans.us>
Date: Mon, 13 May 2024 20:44:07 +0000
Subject: [PATCH] Properly parse self-closing stop-nodes

Work around a bug in fast-xml-parser: https://github.com/NaturalIntelligence/fast-xml-parser/issues/654

Also use the `parseNodeValue` flag to disable parsing node values to numbers and booleans, removing the need for other workarounds.

Changelog: fixed
---
 .yarn/versions/8340cb85.yml                    |  2 ++
 web/__tests__/epub.test.ts                     | 14 +++++++++++++-
 web/epub.ts                                    | 18 ++++++++++++++----
 .../__tests__/getXhtmlSentences.test.ts        |  4 ++--
 web/synchronize/tagSentences.ts                |  2 +-
 5 files changed, 32 insertions(+), 8 deletions(-)
 create mode 100644 .yarn/versions/8340cb85.yml

diff --git a/.yarn/versions/8340cb85.yml b/.yarn/versions/8340cb85.yml
new file mode 100644
index 0000000..0e2d813
--- /dev/null
+++ b/.yarn/versions/8340cb85.yml
@@ -0,0 +1,2 @@
+releases:
+  "@storyteller/web": patch
diff --git a/web/__tests__/epub.test.ts b/web/__tests__/epub.test.ts
index 96b336a..9921627 100644
--- a/web/__tests__/epub.test.ts
+++ b/web/__tests__/epub.test.ts
@@ -1,9 +1,21 @@
 import { join } from "node:path"
 import { describe, it } from "node:test"
-import { Epub, getBody, textContent } from "../epub"
+import { Epub, ParsedXml, getBody, textContent } from "../epub"
 import assert from "node:assert"
 import { stat } from "node:fs/promises"
 
+void describe("xhtml parsing", () => {
+  void it("can handle self-closing stop nodes", () => {
+    const xmlString = `<script src="script.js"/>`
+    const parsed = Epub.xhtmlParser.parse(xmlString) as ParsedXml
+
+    console.log(parsed)
+    const built = Epub.xhtmlBuilder.build(parsed) as string
+
+    assert.strictEqual(built, xmlString)
+  })
+})
+
 void describe("Epub", () => {
   void it("can read from an archived .epub file", async () => {
     const filepath = join("__fixtures__", "moby-dick.epub")
diff --git a/web/epub.ts b/web/epub.ts
index aa2b685..ef587b5 100644
--- a/web/epub.ts
+++ b/web/epub.ts
@@ -159,10 +159,20 @@ export class Epub {
     htmlEntities: true,
     trimValues: false,
     stopNodes: ["*.pre", "*.script"],
-    numberParseOptions: {
-      hex: false,
-      leadingZeros: false,
-      skipLike: /.*/,
+    parseTagValue: false,
+    updateTag(_tagName, _jPath, attrs) {
+      // There's never an attribute called "/";
+      // this erroneously happens sometimes when parsing
+      // self-closing stop nodes with ignoreAttributes: false
+      // and allowBooleanAttributes: true.
+      //
+      // Also attrs is undefined if there are no attrs;
+      // the types are wrong.
+      // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
+      if (attrs && "@_/" in attrs) {
+        delete attrs["@_/"]
+      }
+      return true
     },
   })
 
diff --git a/web/synchronize/__tests__/getXhtmlSentences.test.ts b/web/synchronize/__tests__/getXhtmlSentences.test.ts
index 16b5712..91fc12e 100644
--- a/web/synchronize/__tests__/getXhtmlSentences.test.ts
+++ b/web/synchronize/__tests__/getXhtmlSentences.test.ts
@@ -3,7 +3,7 @@ import { describe, it } from "node:test"
 import { getXHtmlSentences } from "../getXhtmlSentences"
 import { strict as assert } from "node:assert/strict"
 
-void describe.only("getXhtmlSentences", () => {
+void describe("getXhtmlSentences", () => {
   void it("gets sentences from a text node", () => {
     const input = [
       {
@@ -54,7 +54,7 @@ void describe.only("getXhtmlSentences", () => {
     ])
   })
 
-  void it.only("gets sentences from nested elements", () => {
+  void it("gets sentences from nested elements", () => {
     const input: ParsedXml = [
       {
         p: [
diff --git a/web/synchronize/tagSentences.ts b/web/synchronize/tagSentences.ts
index 689f3f3..44eaac2 100644
--- a/web/synchronize/tagSentences.ts
+++ b/web/synchronize/tagSentences.ts
@@ -92,7 +92,7 @@ function tagSentencesInXml(
     const remainingSentence = sentences[currentSentenceIndex]!.slice(
       currentSentenceProgress,
     )
-    const text = currentNode["#text"].toString()
+    const text = currentNode["#text"]
     const remainingNodeText = text.slice(currentNodeProgress)
     // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
     const index = remainingNodeText.indexOf(remainingSentence[0]!)