From 1c4ba0706a460a6ba7eabe459ea08babef187bd0 Mon Sep 17 00:00:00 2001
From: Ziggy Jonsson <ziggy.jonsson.nyc@gmail.com>
Date: Sat, 11 May 2024 14:38:51 -0400
Subject: [PATCH] Retry stream if not big enough

---
 lib/Open/unzip.js    | 17 ++++++++++++++---
 test/office-files.js | 26 ++++++++++++++++++++++++--
 2 files changed, 38 insertions(+), 5 deletions(-)

diff --git a/lib/Open/unzip.js b/lib/Open/unzip.js
index 5cb0ef1..079ca7d 100644
--- a/lib/Open/unzip.js
+++ b/lib/Open/unzip.js
@@ -6,9 +6,9 @@ const parseExtraField = require('../parseExtraField');
 const parseDateTime = require('../parseDateTime');
 const parseBuffer = require('../parseBuffer');
 
-module.exports = function unzip(source, offset, _password, directoryVars, length) {
-  const file = PullStream(),
-    entry = Stream.PassThrough();
+module.exports = function unzip(source, offset, _password, directoryVars, length, _entry) {
+  var file = PullStream(),
+      entry = _entry || Stream.PassThrough();
 
   const req = source.stream(offset, length);
   req.pipe(file).on('error', function(e) {
@@ -34,6 +34,17 @@ module.exports = function unzip(source, offset, _password, directoryVars, length
         ['extraFieldLength', 2],
       ]);
 
+      var localSize = 30
+        + 100 // add extra padding
+        + (vars.extraFieldLength || 0) 
+        + (vars.fileNameLength || 0)
+        + vars.compressedSize;
+      
+      if (localSize > length) {
+        entry.emit('streamRetry', localSize);
+        return unzip(source, offset, _password, directoryVars, localSize, entry);
+      }
+
       vars.lastModifiedDateTime = parseDateTime(vars.lastModifiedDate, vars.lastModifiedTime);
 
       const fileName = await file.pull(vars.fileNameLength);
diff --git a/test/office-files.js b/test/office-files.js
index b7baabe..00068e2 100644
--- a/test/office-files.js
+++ b/test/office-files.js
@@ -2,8 +2,16 @@ const test = require('tap').test;
 const path = require('path');
 const unzip = require('../');
 
-test("get content a docx file without errors", async function () {
-  const archive = path.join(__dirname, '../testData/office/testfile.docx');
+var test = require('tap').test;
+var fs = require('fs');
+var path = require('path');
+var unzip = require('../');
+var il = require('iconv-lite');
+var Promise = require('bluebird');
+var NoopStream = require('../lib/NoopStream');
+
+test("get content a docx file without errors", async function (t) {
+  var archive = path.join(__dirname, '../testData/office/testfile.docx');
 
   const directory = await unzip.Open.file(archive);
   await Promise.all(directory.files.map(file => file.buffer()));
@@ -14,4 +22,18 @@ test("get content a xlsx file without errors", async function () {
 
   const directory = await unzip.Open.file(archive);
   await Promise.all(directory.files.map(file => file.buffer()));
+});
+
+test("stream retries when the local file header indicates bigger size than central directory", async function (t) {
+  var archive = path.join(__dirname, '../testData/office/testfile.xlsx');
+  let retries = 0, size;
+  const directory = await unzip.Open.file(archive, {padding: 10});
+  const stream = directory.files[0].stream();
+  stream.on('streamRetry', _size => {
+    retries += 1;
+    size = _size;
+  });
+  await new Promise(resolve => stream.pipe(NoopStream()).on('finish', resolve));
+  t.ok(retries === 1, 'retries once');
+  t.ok(size > 0, 'size is set');
 });
\ No newline at end of file