feat(reoff): parse more async

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

@@ -0,0 +1,21 @@
+import { VFile } from 'vfile'
+import { getXMLDatas } from './get-xml-data'
+
+const removeHeader = (text: string | undefined) =>
+  text ? text.replace(/<\?xml.*?\?>/, '') : ''
+
+export async function docxToVFile(file: Buffer | string) {
+  const data = await getXMLDatas(file, {
+    filenames: [/customXml(\/|\\)/, 'word/document.xml', 'word/footnotes.xml'],
+  })
+
+  const total = `${removeHeader(data['word/document.xml'])}
+  ${removeHeader(data['word/footnotes.xml'])}
+
+  </w:document>`
+  const vfile = new VFile(total)
+  // if (footnotes) {
+  //   Object.assign(vfile.data, { footnotes })
+  // }
+  return vfile
+}

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

@@ -1,14 +1,14 @@
 import { VFile } from 'vfile'
 import { getXMLData } from './get-xml-data'
 
+const removeHeader = (text: string) => text.replace(/<\?xml.*?\?>/, '')
+
 export async function docxToVFile(file: Buffer | string) {
   const mainXML = (await getXMLData(file)).slice(0, -'</w:document>'.length)
   // xast-util-from-xml cannot handle two xml headers in one doc
-  const footnotes = (await getXMLData(file, { filename: 'footnotes' })).replace(
-    '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>',
-    ''
-  )
-  // easier to put the footnotes in the same spot
+  const footnotes = removeHeader(
+    await getXMLData(file, { filename: 'footnotes' })
+  ) // easier to put the footnotes in the same spot
   const total = `${mainXML}${footnotes}
   </w:document>`
   console.log(footnotes)

libs/reoff/docx-to-vfile/src/lib/get-xml-data.ts

@@ -1,5 +1,8 @@
-import AdmZip from 'adm-zip'
-import { extname } from 'path'
+import AdmZip, { IZipEntry } from 'adm-zip'
+import { join, extname } from 'path'
+import { promisify } from 'util'
+import yauzl, { Entry, ZipFile } from 'yauzl'
+import { tryCatchPromise } from '@jote/utils'
 
 const tab = '\t',
   cr = '\n\n',
@@ -12,6 +15,85 @@ const paragraphRegex = new RegExp(
   'g'
 )
 
+/**
+ * Get xml data from either a buffer or path,
+ */
+export async function getXMLDatas(
+  file: string | Buffer,
+  {
+    filenames,
+    xml,
+    returnBuffer,
+  }: { filenames: (string | RegExp)[]; xml?: boolean; returnBuffer?: boolean }
+): Promise<{ [key: string]: string }> {
+  return new Promise((resolve, reject) => {
+    if (
+      typeof file === 'string' &&
+      !extensionRegex.test(extname(file).toLowerCase())
+    ) {
+      reject(new Error('The file must be either a .docx, .xlsx or .pptx'))
+    }
+    try {
+      if (typeof file === 'string') {
+        yauzl.open(file, { lazyEntries: true }, (err, zip) =>
+          unzipCallback(err, zip, resolve, reject, filenames)
+        )
+        return
+      }
+      yauzl.fromBuffer(file, { lazyEntries: true }, (err, zip) =>
+        unzipCallback(err, zip, resolve, reject, filenames)
+      )
+    } catch (err) {
+      reject(new Error(`${err} (${file})`))
+    }
+  })
+}
+
+function unzipCallback(
+  err: unknown,
+  zip: ZipFile | undefined,
+  resolve: (
+    value: { [key: string]: string } | PromiseLike<{ [key: string]: string }>
+  ) => void,
+  reject: (reason?: any) => void,
+  filenames: (string | RegExp)[]
+) {
+  if (err) {
+    reject(err)
+    return
+  }
+  if (!zip) {
+    reject(new Error('Empty zip file'))
+    return
+  }
+  let result: { [key: string]: string }
+
+  const openReadStream = promisify(zip.openReadStream.bind(zip))
+  zip.readEntry()
+  zip.on('entry', async (entry: Entry) => {
+    if (
+      /\/$/.test(entry.fileName) ||
+      !filenames.some((filename) => entry.fileName.match(filename))
+    ) {
+      zip.readEntry()
+      return
+    }
+    let stream = await openReadStream(entry)
+    let entryChunks: any[] = []
+    if (!stream) {
+      zip.readEntry()
+      return
+    }
+    stream.on('data', (chunk) => entryChunks.push(chunk))
+    stream.on('end', () => {
+      result[entry.fileName] = Buffer.from(entryChunks).toString()
+    })
+  })
+  zip.on('end', () => {
+    resolve(result)
+  })
+}
+
 export async function getXMLData(
   file: string | Buffer,
   { filename = 'document', xml = true, returnBuffer = false } = {

libs/utils/src/lib/tryCatchPromise.ts

@@ -1,4 +1,4 @@
-export default async function tryCatchPromise(
+export async function tryCatchPromise(
   promise: Promise<any>,
   errorHandler?: (error: any) => void
 ): Promise<[any, any]> {

libs/utils/src/lib/tryCatchVoid.ts

@@ -1,4 +1,4 @@
-export default function tryCatchVoid(
+export function tryCatchVoid(
   func: any,
   errorHandler?: (error: any) => void
 ): unknown | void {