Skip to content

Commit f380db2

Browse files
kthelgasonbnoordhuis
authored andcommitted
zlib: add support for concatenated members
According to the spec gzipped archives can contain more than one compressed member. Previously Node's gzip implementation would only unzip the first member and throw away the rest of the compressed data. Issue #4306 is an example of this occurring in daily use. Fixes: #4306 PR-URL: #5120 Reviewed-By: Ben Noordhuis <[email protected]> Reviewed-By: James M Snell <[email protected]>
1 parent 3b20941 commit f380db2

File tree

3 files changed

+86
-0
lines changed

3 files changed

+86
-0
lines changed

src/node_zlib.cc

+18
Original file line numberDiff line numberDiff line change
@@ -41,6 +41,9 @@ enum node_zlib_mode {
4141
UNZIP
4242
};
4343

44+
#define GZIP_HEADER_ID1 0x1f
45+
#define GZIP_HEADER_ID2 0x8b
46+
#define GZIP_MIN_HEADER_SIZE 10
4447

4548
void InitZlib(v8::Local<v8::Object> target);
4649

@@ -254,6 +257,19 @@ class ZCtx : public AsyncWrap {
254257
ctx->err_ = Z_NEED_DICT;
255258
}
256259
}
260+
while (ctx->strm_.avail_in >= GZIP_MIN_HEADER_SIZE &&
261+
ctx->mode_ == GUNZIP) {
262+
// Bytes remain in input buffer. Perhaps this is another compressed
263+
// member in the same archive, or just trailing garbage.
264+
// Check the header to find out.
265+
if (ctx->strm_.next_in[0] != GZIP_HEADER_ID1 ||
266+
ctx->strm_.next_in[1] != GZIP_HEADER_ID2) {
267+
// Not a valid gzip member
268+
break;
269+
}
270+
Reset(ctx);
271+
ctx->err_ = inflate(&ctx->strm_, ctx->flush_);
272+
}
257273
break;
258274
default:
259275
CHECK(0 && "wtf?");
@@ -524,10 +540,12 @@ class ZCtx : public AsyncWrap {
524540
switch (ctx->mode_) {
525541
case DEFLATE:
526542
case DEFLATERAW:
543+
case GZIP:
527544
ctx->err_ = deflateReset(&ctx->strm_);
528545
break;
529546
case INFLATE:
530547
case INFLATERAW:
548+
case GUNZIP:
531549
ctx->err_ = inflateReset(&ctx->strm_);
532550
break;
533551
default:
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
'use strict';
2+
// Test unzipping a gzip file that contains multiple concatenated "members"
3+
4+
const common = require('../common');
5+
const assert = require('assert');
6+
const zlib = require('zlib');
7+
8+
const data = Buffer.concat([
9+
zlib.gzipSync('abc'),
10+
zlib.gzipSync('def')
11+
]);
12+
13+
assert.equal(zlib.gunzipSync(data).toString(), 'abcdef');
14+
15+
zlib.gunzip(data, common.mustCall((err, result) => {
16+
assert.ifError(err);
17+
assert.equal(result, 'abcdef', 'result should match original string');
18+
}));
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,50 @@
1+
'use strict';
2+
// test unzipping a gzip file that has trailing garbage
3+
4+
const common = require('../common');
5+
const assert = require('assert');
6+
const zlib = require('zlib');
7+
8+
// should ignore trailing null-bytes
9+
let data = Buffer.concat([
10+
zlib.gzipSync('abc'),
11+
zlib.gzipSync('def'),
12+
Buffer(10).fill(0)
13+
]);
14+
15+
assert.equal(zlib.gunzipSync(data).toString(), 'abcdef');
16+
17+
zlib.gunzip(data, common.mustCall((err, result) => {
18+
assert.ifError(err);
19+
assert.equal(result, 'abcdef', 'result should match original string');
20+
}));
21+
22+
// if the trailing garbage happens to look like a gzip header, it should
23+
// throw an error.
24+
data = Buffer.concat([
25+
zlib.gzipSync('abc'),
26+
zlib.gzipSync('def'),
27+
Buffer([0x1f, 0x8b, 0xff, 0xff]),
28+
Buffer(10).fill(0)
29+
]);
30+
31+
assert.throws(() => zlib.gunzipSync(data));
32+
33+
zlib.gunzip(data, common.mustCall((err, result) => {
34+
assert(err);
35+
}));
36+
37+
// In this case the trailing junk is too short to be a gzip segment
38+
// So we ignore it and decompression succeeds.
39+
data = Buffer.concat([
40+
zlib.gzipSync('abc'),
41+
zlib.gzipSync('def'),
42+
Buffer([0x1f, 0x8b, 0xff, 0xff])
43+
]);
44+
45+
assert.equal(zlib.gunzipSync(data).toString(), 'abcdef');
46+
47+
zlib.gunzip(data, common.mustCall((err, result) => {
48+
assert.ifError(err);
49+
assert.equal(result, 'abcdef', 'result should match original string');
50+
}));

0 commit comments

Comments
 (0)