compress: add a new module compress.gzip too (#14686)

vlib/compress/README.md

@@ -1,4 +1,8 @@
 ## Description:
 
 `compress` is a namespace for (multiple) compression algorithms supported by V.
-At the moment, only `compress.zlib` and `compress.deflate` are implemented.
+
+At the moment, the following compression algorithms are implemented:
+- `compress.deflate`
+- `compress.gzip`
+- `compress.zlib`

vlib/compress/compress.v

@@ -9,7 +9,7 @@ fn C.tdefl_compress_mem_to_heap(source_buf voidptr, source_buf_len usize, out_le
 fn C.tinfl_decompress_mem_to_heap(source_buf voidptr, source_buf_len usize, out_len &usize, flags int) voidptr
 
 // compresses an array of bytes based on providing flags and returns the compressed bytes in a new array
-// see `gzip.compress([]u8)` and `zlib.compress([]u8)` for default implementations.
+// NB: this is a low level api, a high level implementation like zlib/gzip should be preferred
 [manualfree]
 pub fn compress(data []u8, flags int) ?[]u8 {
 	if u64(data.len) > compress.max_size {
@@ -28,7 +28,7 @@ pub fn compress(data []u8, flags int) ?[]u8 {
 }
 
 // decompresses an array of bytes based on providing flags and returns the decompressed bytes in a new array
-// see `gzip.decompress([]u8)` and `zlib.decompress([]u8)` for default implementations.
+// NB: this is a low level api, a high level implementation like zlib/gzip should be preferred
 [manualfree]
 pub fn decompress(data []u8, flags int) ?[]u8 {
 	mut out_len := usize(0)

vlib/compress/deflate/README.md

@@ -3,9 +3,6 @@
 `compress.deflate` is a module that assists in the compression and
 decompression of binary data using `deflate` compression
 
-NOTE: To decompress gzip, discard first 10 bytes of compressed bytes
-then use `compress.deflate.decompress`. (Header validation won't be
-performed in this case)
 
 ## Examples:
 

vlib/compress/deflate/deflate.v

@@ -2,15 +2,14 @@ module deflate
 
 import compress
 
-// compresses an array of bytes using gzip and returns the compressed bytes in a new array
-// Example: compressed := gzip.compress(b)?
+// compresses an array of bytes using deflate and returns the compressed bytes in a new array
+// Example: compressed := deflate.compress(b)?
 pub fn compress(data []u8) ?[]u8 {
 	return compress.compress(data, 0)
 }
 
-// decompresses an array of bytes using zlib and returns the decompressed bytes in a new array
-// Example: decompressed := zlib.decompress(b)?
-[manualfree]
+// decompresses an array of bytes using deflate and returns the decompressed bytes in a new array
+// Example: decompressed := deflate.decompress(b)?
 pub fn decompress(data []u8) ?[]u8 {
 	return compress.decompress(data, 0)
 }

vlib/compress/gzip/README.md

@@ -0,0 +1,18 @@
+## Description:
+
+`compress.gzip` is a module that assists in the compression and
+decompression of binary data using `gzip` compression
+
+
+## Examples:
+
+```v
+import compress.gzip
+
+fn main() {
+	uncompressed := 'Hello world!'
+	compressed := gzip.compress(uncompressed.bytes())?
+	decompressed := gzip.decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+```

vlib/compress/gzip/gzip.v

@@ -0,0 +1,114 @@
+// [rfc1952](https://datatracker.ietf.org/doc/html/rfc1952) compliant
+// gzip compression/decompression
+
+module gzip
+
+import compress
+import hash.crc32
+
+// compresses an array of bytes using gzip and returns the compressed bytes in a new array
+// Example: compressed := gzip.compress(b)?
+pub fn compress(data []u8) ?[]u8 {
+	compressed := compress.compress(data, 0)?
+	// header
+	mut result := [
+		u8(0x1f), // magic numbers (1F 8B)
+		0x8b,
+		0x08, // deflate
+		0x00, // header flags
+		0x00, // 4-byte timestamp, 0 = no timestamp (00 00 00 00)
+		0x00,
+		0x00,
+		0x00,
+		0x00, // extra flags
+		0xff, // operating system id (0xff = unknown)
+	] // 10 bytes
+	result << compressed
+	// trailer
+	checksum := crc32.sum(data)
+	length := data.len
+	result << [
+		u8(checksum >> 24),
+		u8(checksum >> 16),
+		u8(checksum >> 8),
+		u8(checksum),
+		u8(length >> 24),
+		u8(length >> 16),
+		u8(length >> 8),
+		u8(length),
+	] // 8 bytes
+	return result
+}
+
+// decompresses an array of bytes using zlib and returns the decompressed bytes in a new array
+// Example: decompressed := gzip.decompress(b)?
+pub fn decompress(data []u8) ?[]u8 {
+	if data.len < 18 {
+		return error('data is too short, not gzip compressed?')
+	} else if data[0] != 0x1f || data[1] != 0x8b {
+		return error('wrong magic numbers, not gzip compressed?')
+	} else if data[2] != 0x08 {
+		return error('gzip data is not compressed with DEFLATE')
+	}
+	mut header_length := 10
+
+	// parse flags, we ignore most of them, but we still need to parse them
+	// correctly, so we dont accidently decompress something that belongs
+	// to the header
+
+	if data[4] & 0b1110_0000 > 0 { // reserved bits
+		// rfc 1952 2.3.1.2 Compliance
+		// A compliant decompressor must give an error indication if any
+		// reserved bit is non-zero, since such a bit could indicate the
+		// presence of a new field that would cause subsequent data to be
+		// interpreted incorrectly.
+		return error('reserved flags are set, unsupported field detected')
+	}
+
+	// if data[4] & 0b0000_0001 {} // FTEXT
+	if data[4] & 0b0000_0100 > 0 { // FEXTRA, extra data
+		xlen := data[header_length]
+		header_length += xlen + 1
+	}
+	if data[4] & 0b0000_1000 > 0 { // FNAME, file name
+		// filename is zero-terminated, so skip until we hit a zero byte
+		for header_length < data.len && data[header_length] != 0x00 {
+			header_length++
+		}
+		header_length++
+	}
+	if data[4] & 0b0001_0000 > 0 { // FCOMMENT
+		// comment is zero-terminated, so skip until we hit a zero byte
+		for header_length < data.len && data[header_length] != 0x00 {
+			header_length++
+		}
+		header_length++
+	}
+	if data[4] & 0b0000_0010 > 0 { // FHCRC, flag header crc
+		if header_length + 12 > data.len {
+			return error('data too short')
+		}
+		checksum_header := crc32.sum(data[..header_length])
+		checksum_header_expected := (u32(data[header_length]) << 24) | (u32(data[header_length + 1]) << 16) | (u32(data[
+			header_length + 2]) << 8) | data[header_length + 3]
+		if checksum_header != checksum_header_expected {
+			return error('header checksum verification failed')
+		}
+		header_length += 4
+	}
+	if header_length + 8 > data.len {
+		return error('data too short')
+	}
+
+	decompressed := compress.decompress(data[header_length..data.len - 8], 0)?
+	length_expected := (u32(data[data.len - 4]) << 24) | (u32(data[data.len - 3]) << 16) | (u32(data[data.len - 2]) << 8) | data[data.len - 1]
+	if decompressed.len != length_expected {
+		return error('length verification failed, got $decompressed.len, expected $length_expected')
+	}
+	checksum := crc32.sum(decompressed)
+	checksum_expected := (u32(data[data.len - 8]) << 24) | (u32(data[data.len - 7]) << 16) | (u32(data[data.len - 6]) << 8) | data[data.len - 5]
+	if checksum != checksum_expected {
+		return error('checksum verification failed')
+	}
+	return decompressed
+}

vlib/compress/gzip/gzip_test.v

@@ -0,0 +1,134 @@
+module gzip
+
+import hash.crc32
+
+fn test_gzip() ? {
+	uncompressed := 'Hello world!'
+	compressed := compress(uncompressed.bytes())?
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn assert_decompress_error(data []u8, reason string) ? {
+	decompress(data) or {
+		assert err.msg() == reason
+		return
+	}
+	return error('did not error')
+}
+
+fn test_gzip_invalid_too_short() ? {
+	assert_decompress_error([]u8{}, 'data is too short, not gzip compressed?')?
+}
+
+fn test_gzip_invalid_magic_numbers() ? {
+	assert_decompress_error([]u8{len: 100}, 'wrong magic numbers, not gzip compressed?')?
+}
+
+fn test_gzip_invalid_compression() ? {
+	mut data := []u8{len: 100}
+	data[0] = 0x1f
+	data[1] = 0x8b
+	assert_decompress_error(data, 'gzip data is not compressed with DEFLATE')?
+}
+
+fn test_gzip_with_ftext() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0000_0001 // FTEXT
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_fname() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0000_1000
+	compressed.insert(10, `h`)
+	compressed.insert(11, `i`)
+	compressed.insert(12, 0x00)
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_fcomment() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0001_0000
+	compressed.insert(10, `h`)
+	compressed.insert(11, `i`)
+	compressed.insert(12, 0x00)
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_fname_fcomment() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0001_1000
+	compressed.insert(10, `h`)
+	compressed.insert(11, `i`)
+	compressed.insert(12, 0x00)
+	compressed.insert(10, `h`)
+	compressed.insert(11, `i`)
+	compressed.insert(12, 0x00)
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_fextra() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0000_0100
+	compressed.insert(10, 2)
+	compressed.insert(11, `h`)
+	compressed.insert(12, `i`)
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_hcrc() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0000_0010
+	checksum := crc32.sum(compressed[..10])
+	compressed.insert(10, u8(checksum >> 24))
+	compressed.insert(11, u8(checksum >> 16))
+	compressed.insert(12, u8(checksum >> 8))
+	compressed.insert(13, u8(checksum))
+	decompressed := decompress(compressed)?
+	assert decompressed == uncompressed.bytes()
+}
+
+fn test_gzip_with_invalid_hcrc() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b0000_0010
+	checksum := crc32.sum(compressed[..10])
+	compressed.insert(10, u8(checksum >> 24))
+	compressed.insert(11, u8(checksum >> 16))
+	compressed.insert(12, u8(checksum >> 8))
+	compressed.insert(13, u8(checksum + 1))
+	assert_decompress_error(compressed, 'header checksum verification failed')?
+}
+
+fn test_gzip_with_invalid_checksum() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[compressed.len - 5] += 1
+	assert_decompress_error(compressed, 'checksum verification failed')?
+}
+
+fn test_gzip_with_invalid_length() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[compressed.len - 1] += 1
+	assert_decompress_error(compressed, 'length verification failed, got 12, expected 13')?
+}
+
+fn test_gzip_with_invalid_flags() ? {
+	uncompressed := 'Hello world!'
+	mut compressed := compress(uncompressed.bytes())?
+	compressed[4] |= 0b1000_0000
+	assert_decompress_error(compressed, 'reserved flags are set, unsupported field detected')?
+}

vlib/compress/zlib/zlib.v

@@ -4,7 +4,6 @@ import compress
 
 // compresses an array of bytes using zlib and returns the compressed bytes in a new array
 // Example: compressed := zlib.compress(b)?
-[manualfree]
 pub fn compress(data []u8) ?[]u8 {
 	// flags = TDEFL_WRITE_ZLIB_HEADER (0x01000)
 	return compress.compress(data, 0x01000)
@@ -12,7 +11,6 @@ pub fn compress(data []u8) ?[]u8 {
 
 // decompresses an array of bytes using zlib and returns the decompressed bytes in a new array
 // Example: decompressed := zlib.decompress(b)?
-[manualfree]
 pub fn decompress(data []u8) ?[]u8 {
 	// flags = TINFL_FLAG_PARSE_ZLIB_HEADER (0x1)
 	return compress.decompress(data, 0x1)