feat: better unzip lib that's browser compatible

libs/reoff/docx-to-vfile/src/index.ts

@@ -1,2 +1,2 @@
 export * from './lib/get-xml-data'
-export * from './lib/docx-to-vfile-yauzl'
+export * from './lib/docx-to-vfile-unzipit'

libs/reoff/docx-to-vfile/src/lib/__snapshots__/docx-to-vfile-unzipit.spec.ts.snap

@@ -0,0 +1,33 @@
+// Jest Snapshot v1, https://goo.gl/fbAQLP
+
+exports[`reoffDocxToVfile should work 1`] = `
+VFile {
+  "cwd": "/Users/thomas/Projects/jote-monorepo",
+  "data": Object {},
+  "history": Array [],
+  "messages": Array [],
+  "value": "[object
+  
+  [
+o
+b
+j
+e
+c
+t
+ 
+A
+r
+r
+a
+y
+B
+u
+f
+f
+e
+r
+]
+  </w:document>",
+}
+`;

libs/reoff/docx-to-vfile/src/lib/docx-to-vfile-unzipit.spec.ts

@@ -0,0 +1,15 @@
+import { docxToVFile } from './docx-to-vfile-unzipit'
+import fs from 'fs'
+import path from 'path'
+
+describe('reoffDocxToVfile', () => {
+  const doc = fs.readFileSync(
+    path.join(__dirname, '../../../reoff-parse/src/test/word-citation.docx')
+  )
+  jest.setTimeout(10000)
+  it('should work', async () => {
+    const vfile = await docxToVFile(new Uint8Array(doc))
+    console.log(vfile)
+    expect(vfile).toMatchSnapshot()
+  })
+})

libs/reoff/docx-to-vfile/src/lib/docx-to-vfile-unzipit.ts

@@ -0,0 +1,39 @@
+import { VFile } from 'vfile'
+import { unzip } from 'unzipit'
+
+const removeHeader = (text: string | undefined) =>
+  text ? text.replace(/<\?xml.*?\?>/, '') : ''
+
+export async function docxToVFile(file: ArrayBuffer) {
+  const { entries } = await unzip(file)
+  console.log(entries['word/document.xml'])
+  const doc = (await entries['word/document.xml'].arrayBuffer()).toString()
+  const foot =
+    (await entries?.['word/footnotes.xml']?.arrayBuffer())?.toString() || ''
+  const bib =
+    (await entries?.['customXml/item1.xml']?.arrayBuffer())?.toString() || ''
+  // const data = await getXMLDatas(file, {
+  //   filenames: [
+  //     /customXml\/item\d+\.xml/,
+  //     'word/document.xml',
+  //     'word/footnotes.xml',
+  //   ],
+  // })
+
+  // const {
+  //   'word/document.xml': document,
+  //   'word/footnotes.xml': footnotes,
+  //   ...bibliography
+  // } = data
+  const total = `${removeHeader(doc).slice(0, -'</w:document>'.length)}
+  ${removeHeader(foot)}
+  ${Object.values(bib)
+    .map((bib) => removeHeader(bib))
+    .join('\n')}
+  </w:document>`
+  const vfile = new VFile(total)
+  // if (footnotes) {
+  //   Object.assign(vfile.data, { footnotes })
+  // }
+  return vfile
+}

package.json

@@ -62,6 +62,7 @@
     "unist-util-remove": "^3.0.0",
     "unist-util-select": "^4.0.1",
     "unist-util-visit": "4.0.0",
+    "unzipit": "^1.4.0",
     "utility-types": "^3.10.0",
     "vfile": "^5.3.0",
     "xast-util-from-xml": "^2.0.1",

yarn.lock

@@ -12542,6 +12542,13 @@ untildify@^4.0.0:
   resolved "https://registry.yarnpkg.com/untildify/-/untildify-4.0.0.tgz#2bc947b953652487e4600949fb091e3ae8cd919b"
   integrity sha512-KK8xQ1mkzZeg9inewmFVDNkg3l5LUhoq9kN6iWYB/CC9YMG8HA+c1Q8HwDe6dEX7kErrEVNVBO3fWsVq5iDgtw==
 
+unzipit@^1.4.0:
+  version "1.4.0"
+  resolved "https://registry.yarnpkg.com/unzipit/-/unzipit-1.4.0.tgz#69fc6eef47730897e39b15e54f972d60a1a92441"
+  integrity sha512-hjoB8j1igXJgmxqaAvqkIW+faKTpG9cPK6QvkBhNCyd8OPWqODXTBVqTEmZbz62K5J/dX4Xa8lTa0NRikQwSjQ==
+  dependencies:
+    uzip-module "^1.0.2"
+
 upath2@^3.1.12:
   version "3.1.12"
   resolved "https://registry.yarnpkg.com/upath2/-/upath2-3.1.12.tgz#441b3dfbadde21731017bd1b7beb169498efd0a9"
@@ -12664,6 +12671,11 @@ uvu@^0.5.0:
     kleur "^4.0.3"
     sade "^1.7.3"
 
+uzip-module@^1.0.2:
+  version "1.0.3"
+  resolved "https://registry.yarnpkg.com/uzip-module/-/uzip-module-1.0.3.tgz#6bbabe2a3efea5d5a4a47479f523a571de3427ce"
+  integrity sha512-AMqwWZaknLM77G+VPYNZLEruMGWGzyigPK3/Whg99B3S6vGHuqsyl5ZrOv1UUF3paGK1U6PM0cnayioaryg/fA==
+
 v8-compile-cache@2.3.0, v8-compile-cache@^2.0.3:
   version "2.3.0"
   resolved "https://registry.yarnpkg.com/v8-compile-cache/-/v8-compile-cache-2.3.0.tgz#2de19618c66dc247dcfb6f99338035d8245a2cee"