pass DOM parser to epub tool

resolves #456
wydengyre · Mar 13, 2024 · 9abfd74 · 9abfd74
1 parent d66016d
commit 9abfd74
Show file tree

Hide file tree

Showing 8 changed files with 18 additions and 11 deletions.
diff --git a/package-lock.json b/package-lock.json
diff --git a/packages/cli/epub-to-text.ts b/packages/cli/epub-to-text.ts
@@ -12,7 +12,7 @@ async function main() {
 
 async function go(r: Readable): Promise<string> {
  const ab = await arrayBuffer(r);
- return epubToText(ab);
+ return epubToText(new DOMParser(), ab);
 }
 
 const currentFile = fileURLToPath(import.meta.url);

diff --git a/packages/cli/package.json b/packages/cli/package.json
@@ -21,6 +21,7 @@
  "dependencies": {
  "@bitextual/core": "*",
  "@bitextual/epub": "*",
+ "@xmldom/xmldom": "^0.8.10",
  "franc-min": "^6.2.0"
  }
 }
diff --git a/packages/core/package.json b/packages/core/package.json
@@ -20,7 +20,6 @@
  },
  "dependencies": {
  "@bitextual/hunalign": "^0.0.1",
- "@xmldom/xmldom": "^0.8.10",
  "html-to-text": "^9.0.5",
  "jszip": "^3.10.1"
  }

diff --git a/packages/epub/epub.test.ts b/packages/epub/epub.test.ts
@@ -2,6 +2,7 @@ import { strict as assert } from "node:assert";
 import { readFile } from "node:fs/promises";
 import { test } from "node:test";
 import { fileURLToPath } from "node:url";
+import { DOMParser } from "@xmldom/xmldom";
 import { epubToText } from "./epub.js";
 
 const EPUB2_PATH_REL = "@bitextual/test/bovary.english.epub";
@@ -22,12 +23,12 @@ test("epub", async (t) => {
 
 const epubToTextEpub2 = (expected: string) => async () => {
  const bytes = await readFile(EPUB2_PATH);
- const text = await epubToText(bytes);
+ const text = await epubToText(new DOMParser(), bytes);
  assert.strictEqual(text, expected);
 };
 
 const epubToTextEpub3 = (expected: string) => async () => {
  const bytes = await readFile(EPUB3_PATH);
- const text = await epubToText(bytes);
+ const text = await epubToText(new DOMParser(), bytes);
  assert.strictEqual(text, expected);
 };
diff --git a/packages/epub/epub.ts b/packages/epub/epub.ts
@@ -1,10 +1,13 @@
-import { DOMParser } from "@xmldom/xmldom";
+import type { DOMParser } from "@xmldom/xmldom";
 import { compile as compileHtmlConvert } from "html-to-text";
 import JSZip from "jszip";
 
 export { epubToText };
 
-async function epubToText(epubBytes: ArrayBuffer): Promise<string> {
+async function epubToText(
+ domParser: DOMParser,
+ epubBytes: ArrayBuffer,
+): Promise<string> {
  const zip = await JSZip.loadAsync(epubBytes);
  const files = zip.files;
  const containerPath = "META-INF/container.xml";
@@ -14,7 +17,7 @@ async function epubToText(epubBytes: ArrayBuffer): Promise<string> {
  }
  const containerTxt = await container.async("text");
 
- const containerDom = new DOMParser().parseFromString(
+ const containerDom = domParser.parseFromString(
  containerTxt,
  "application/xml",
  );
@@ -35,7 +38,7 @@ async function epubToText(epubBytes: ArrayBuffer): Promise<string> {
  throw new Error(`opf file not found at ${rootPath}`);
  }
  const opfTxt = await opf.async("text");
- const opfDom = new DOMParser().parseFromString(opfTxt, "application/xml");
+ const opfDom = domParser.parseFromString(opfTxt, "application/xml");
  if (opfDom === undefined) {
  throw new Error(`failed to parse XML DOM from ${opfTxt}`);
  }

diff --git a/packages/epub/package.json b/packages/epub/package.json
@@ -15,6 +15,7 @@
  "@tsconfig/strictest": "2.0.3",
  "@types/html-to-text": "9.0.4",
  "@types/node": "20.11.26",
+ "@xmldom/xmldom": "0.8.10",
  "tsx": "4.7.1",
  "typescript": "5.4.2"
  },

diff --git a/packages/web/src/worker.ts b/packages/web/src/worker.ts
@@ -20,9 +20,10 @@ async function renderAlignment(
  source.arrayBuffer(),
  target.arrayBuffer(),
  ]);
+ const domParser = new DOMParser();
  const [sourceText, targetText] = await Promise.all([
- epubToText(sourceData),
- epubToText(targetData),
+ epubToText(domParser, sourceData),
+ epubToText(domParser, targetData),
  ]);
  const sourceLang = franc(sourceText);
  const targetLang = franc(targetText);