From d3878bc8ae855e7134ee26afaa05c5a8bd99e750 Mon Sep 17 00:00:00 2001
From: Eric Biggers <ebiggers3@gmail.com>
Date: Fri, 28 Dec 2018 10:25:44 -0600
Subject: [PATCH] programs: new test program - test_incomplete_codes

---
 .gitignore                       |   3 +-
 Makefile                         |   1 +
 programs/test_incomplete_codes.c | 299 +++++++++++++++++++++++++++++++
 programs/test_util.c             |  14 ++
 programs/test_util.h             |   1 +
 5 files changed, 316 insertions(+), 2 deletions(-)
 create mode 100644 programs/test_incomplete_codes.c

diff --git a/.gitignore b/.gitignore
index 27bfc9d..cf39421 100644
--- a/.gitignore
+++ b/.gitignore
@@ -15,7 +15,6 @@
 /gzip
 /gunzip
 /run_tests.log
-/test_checksums
-/test_slow_decompression
+/test_*
 tags
 cscope*
diff --git a/Makefile b/Makefile
index bc9f005..fe81798 100644
--- a/Makefile
+++ b/Makefile
@@ -178,6 +178,7 @@ TEST_PROG_COMMON_SRC := programs/test_util.c
 TEST_PROG_SRC        := programs/benchmark.c \
 			programs/checksum.c \
 			programs/test_checksums.c \
+			programs/test_incomplete_codes.c \
 			programs/test_slow_decompression.c
 
 NONTEST_PROGRAMS := $(NONTEST_PROG_SRC:programs/%.c=%$(PROG_SUFFIX))
diff --git a/programs/test_incomplete_codes.c b/programs/test_incomplete_codes.c
new file mode 100644
index 0000000..cbeb338
--- /dev/null
+++ b/programs/test_incomplete_codes.c
@@ -0,0 +1,299 @@
+/*
+ * test_incomplete_codes.c
+ *
+ * Test that the decompressor accepts incomplete Huffman codes in certain
+ * specific cases.
+ */
+
+#include "test_util.h"
+
+static void
+verify_decompression_libdeflate(const u8 *in, size_t in_nbytes,
+				u8 *out, size_t out_nbytes_avail,
+				const u8 *expected_out,
+				size_t expected_out_nbytes)
+{
+	struct libdeflate_decompressor *d;
+	enum libdeflate_result res;
+	size_t actual_out_nbytes;
+
+	d = libdeflate_alloc_decompressor();
+	ASSERT(d != NULL);
+
+	res = libdeflate_deflate_decompress(d, in, in_nbytes,
+					    out, out_nbytes_avail,
+					    &actual_out_nbytes);
+	ASSERT(res == LIBDEFLATE_SUCCESS);
+	ASSERT(actual_out_nbytes == expected_out_nbytes);
+	ASSERT(memcmp(out, expected_out, actual_out_nbytes) == 0);
+
+	libdeflate_free_decompressor(d);
+}
+
+static void
+verify_decompression_zlib(const u8 *in, size_t in_nbytes,
+			  u8 *out, size_t out_nbytes_avail,
+			  const u8 *expected_out, size_t expected_out_nbytes)
+{
+	z_stream z;
+	int res;
+	size_t actual_out_nbytes;
+
+	memset(&z, 0, sizeof(z));
+	res = inflateInit2(&z, -15);
+	ASSERT(res == Z_OK);
+
+	z.next_in = (void *)in;
+	z.avail_in = in_nbytes;
+	z.next_out = (void *)out;
+	z.avail_out = out_nbytes_avail;
+	res = inflate(&z, Z_FINISH);
+	ASSERT(res == Z_STREAM_END);
+	actual_out_nbytes = out_nbytes_avail - z.avail_out;
+	ASSERT(actual_out_nbytes == expected_out_nbytes);
+	ASSERT(memcmp(out, expected_out, actual_out_nbytes) == 0);
+
+	inflateEnd(&z);
+}
+
+static void
+verify_decompression(const u8 *in, size_t in_nbytes,
+		     u8 *out, size_t out_nbytes_avail,
+		     const u8 *expected_out, size_t expected_out_nbytes)
+{
+	verify_decompression_libdeflate(in, in_nbytes, out, out_nbytes_avail,
+					expected_out, expected_out_nbytes);
+	verify_decompression_zlib(in, in_nbytes, out, out_nbytes_avail,
+				  expected_out, expected_out_nbytes);
+
+}
+
+/* Test that an empty offset code is accepted. */
+static void
+test_empty_offset_code(void)
+{
+	static const u8 expected_out[] = { 'A', 'B', 'A', 'A' };
+	u8 in[128];
+	u8 out[128];
+	struct output_bitstream os = { .next = in, .end = in + sizeof(in) };
+	int i;
+
+	/*
+	 * Generate a DEFLATE stream containing a "dynamic Huffman" block
+	 * containing literals, but no offsets; and having an empty offset code
+	 * (all codeword lengths set to 0).
+	 *
+	 * Litlen code:
+	 *	litlensym_A			freq=3 len=1 codeword= 0
+	 *	litlensym_B			freq=1 len=2 codeword=01
+	 *	litlensym_256 (end-of-block)	freq=1 len=2 codeword=11
+	 * Offset code:
+	 *	(empty)
+	 *
+	 * Litlen and offset codeword lengths:
+	 *	[0..'A'-1]	= 0	presym_18
+	 *	['A']		= 1	presym_1
+	 *	['B']		= 2	presym_2
+	 *	['B'+1..255]	= 0	presym_18 presym_18
+	 *	[256]		= 2	presym_2
+	 *	[257]		= 0	presym_0
+	 *
+	 * Precode:
+	 *	presym_0	freq=1 len=3 codeword=011
+	 *	presym_1	freq=1 len=3 codeword=111
+	 *	presym_2	freq=2 len=2 codeword= 01
+	 *	presym_18	freq=3 len=1 codeword=  0
+	 */
+
+	ASSERT(put_bits(&os, 1, 1));	/* BFINAL: 1 */
+	ASSERT(put_bits(&os, 2, 2));	/* BTYPE: DYNAMIC_HUFFMAN */
+	ASSERT(put_bits(&os, 0, 5));	/* num_litlen_syms: 0 + 257 */
+	ASSERT(put_bits(&os, 0, 5));	/* num_offset_syms: 0 + 1 */
+	ASSERT(put_bits(&os, 14, 4));	/* num_explicit_precode_lens: 14 + 4 */
+
+	/*
+	 * Precode codeword lengths: order is
+	 * [16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15]
+	 */
+	for (i = 0; i < 2; i++)		/* presym_{16,17}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 1, 3));	/* presym_18: len=1 */
+	ASSERT(put_bits(&os, 3, 3));	/* presym_0: len=3 */
+	for (i = 0; i < 11; i++)	/* presym_{8,...,13}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 2, 3));	/* presym_2: len=2 */
+	ASSERT(put_bits(&os, 0, 3));	/* presym_14: len=0 */
+	ASSERT(put_bits(&os, 3, 3));	/* presym_1: len=3 */
+
+	/* Litlen and offset codeword lengths */
+	ASSERT(put_bits(&os, 0x0, 1) &&
+	       put_bits(&os, 54, 7));	/* presym_18, 65 zeroes */
+	ASSERT(put_bits(&os, 0x7, 3));	/* presym_1 */
+	ASSERT(put_bits(&os, 0x1, 2));	/* presym_2 */
+	ASSERT(put_bits(&os, 0x0, 1) &&
+	       put_bits(&os, 89, 7));	/* presym_18, 100 zeroes */
+	ASSERT(put_bits(&os, 0x0, 1) &&
+	       put_bits(&os, 78, 7));	/* presym_18, 89 zeroes */
+	ASSERT(put_bits(&os, 0x1, 2));	/* presym_2 */
+	ASSERT(put_bits(&os, 0x3, 3));	/* presym_0 */
+
+	/* Litlen symbols */
+	ASSERT(put_bits(&os, 0x0, 1));	/* litlensym_A */
+	ASSERT(put_bits(&os, 0x1, 2));	/* litlensym_B */
+	ASSERT(put_bits(&os, 0x0, 1));	/* litlensym_A */
+	ASSERT(put_bits(&os, 0x0, 1));	/* litlensym_A */
+	ASSERT(put_bits(&os, 0x3, 2));	/* litlensym_256 (end-of-block) */
+
+	ASSERT(flush_bits(&os));
+
+	verify_decompression(in, os.next - in, out, sizeof(out),
+			     expected_out, sizeof(expected_out));
+}
+
+/* Test that a litrunlen code containing only one symbol is accepted. */
+static void
+test_singleton_litrunlen_code(void)
+{
+	u8 in[128];
+	u8 out[128];
+	struct output_bitstream os = { .next = in, .end = in + sizeof(in) };
+	int i;
+
+	/*
+	 * Litlen code:
+	 *	litlensym_256 (end-of-block)	freq=1 len=1 codeword=0
+	 * Offset code:
+	 *	(empty)
+	 *
+	 * Litlen and offset codeword lengths:
+	 *	[0..256]	= 0	presym_18 presym_18
+	 *	[256]		= 1	presym_1
+	 *	[257]		= 0	presym_0
+	 *
+	 * Precode:
+	 *	presym_0	freq=1 len=2 codeword=01
+	 *	presym_1	freq=1 len=2 codeword=11
+	 *	presym_18	freq=2 len=1 codeword= 0
+	 */
+
+	ASSERT(put_bits(&os, 1, 1));	/* BFINAL: 1 */
+	ASSERT(put_bits(&os, 2, 2));	/* BTYPE: DYNAMIC_HUFFMAN */
+	ASSERT(put_bits(&os, 0, 5));	/* num_litlen_syms: 0 + 257 */
+	ASSERT(put_bits(&os, 0, 5));	/* num_offset_syms: 0 + 1 */
+	ASSERT(put_bits(&os, 14, 4));	/* num_explicit_precode_lens: 14 + 4 */
+
+	/*
+	 * Precode codeword lengths: order is
+	 * [16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15]
+	 */
+	for (i = 0; i < 2; i++)		/* presym_{16,17}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 1, 3));	/* presym_18: len=1 */
+	ASSERT(put_bits(&os, 2, 3));	/* presym_0: len=2 */
+	for (i = 0; i < 13; i++)	/* presym_{8,...,14}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 2, 3));	/* presym_1: len=2 */
+
+	/* Litlen and offset codeword lengths */
+	for (i = 0; i < 2; i++) {
+		ASSERT(put_bits(&os, 0, 1) &&	/* presym_18, 128 zeroes */
+		       put_bits(&os, 117, 7));
+	}
+	ASSERT(put_bits(&os, 0x3, 2));	/* presym_1 */
+	ASSERT(put_bits(&os, 0x1, 2));	/* presym_0 */
+
+	/* Litlen symbols */
+	ASSERT(put_bits(&os, 0x0, 1));	/* litlensym_256 (end-of-block) */
+
+	ASSERT(flush_bits(&os));
+
+	verify_decompression(in, os.next - in, out, sizeof(out), in, 0);
+}
+
+/* Test that an offset code containing only one symbol is accepted. */
+static void
+test_singleton_offset_code(void)
+{
+	static const u8 expected_out[] = { 255, 255, 255, 255 };
+	u8 in[128];
+	u8 out[128];
+	struct output_bitstream os = { .next = in, .end = in + sizeof(in) };
+	int i;
+
+	ASSERT(put_bits(&os, 1, 1));	/* BFINAL: 1 */
+	ASSERT(put_bits(&os, 2, 2));	/* BTYPE: DYNAMIC_HUFFMAN */
+
+	/*
+	 * Litlen code:
+	 *	litlensym_255			freq=1 len=1 codeword= 0
+	 *	litlensym_256 (end-of-block)	freq=1 len=2 codeword=01
+	 *	litlensym_257 (len 3)		freq=1 len=2 codeword=11
+	 * Offset code:
+	 *	offsetsym_0 (offset 0)		freq=1 len=1 codeword=0
+	 *
+	 * Litlen and offset codeword lengths:
+	 *	[0..254] = 0	presym_{18,18}
+	 *	[255]	 = 1	presym_1
+	 *	[256]	 = 1	presym_2
+	 *	[257]	 = 1	presym_2
+	 *	[258]	 = 1	presym_1
+	 *
+	 * Precode:
+	 *	presym_1	freq=2 len=2 codeword=01
+	 *	presym_2	freq=2 len=2 codeword=11
+	 *	presym_18	freq=2 len=1 codeword= 0
+	 */
+
+	ASSERT(put_bits(&os, 1, 5));	/* num_litlen_syms: 1 + 257 */
+	ASSERT(put_bits(&os, 0, 5));	/* num_offset_syms: 0 + 1 */
+	ASSERT(put_bits(&os, 14, 4));	/* num_explicit_precode_lens: 14 + 4 */
+	/*
+	 * Precode codeword lengths: order is
+	 * [16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15]
+	 */
+	for (i = 0; i < 2; i++)		/* presym_{16,17}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 1, 3));	/* presym_18: len=1 */
+	for (i = 0; i < 12; i++)	/* presym_{0,...,13}: len=0 */
+		ASSERT(put_bits(&os, 0, 3));
+	ASSERT(put_bits(&os, 2, 3));	/* presym_2: len=2 */
+	ASSERT(put_bits(&os, 0, 3));	/* presym_14: len=0 */
+	ASSERT(put_bits(&os, 2, 3));	/* presym_1: len=2 */
+
+	/* Litlen and offset codeword lengths */
+	ASSERT(put_bits(&os, 0x0, 1) &&	/* presym_18, 128 zeroes */
+	       put_bits(&os, 117, 7));
+	ASSERT(put_bits(&os, 0x0, 1) &&	/* presym_18, 127 zeroes */
+	       put_bits(&os, 116, 7));
+	ASSERT(put_bits(&os, 0x1, 2));	/* presym_1 */
+	ASSERT(put_bits(&os, 0x3, 2));	/* presym_2 */
+	ASSERT(put_bits(&os, 0x3, 2));	/* presym_2 */
+	ASSERT(put_bits(&os, 0x1, 2));	/* presym_1 */
+
+	/* Literal */
+	ASSERT(put_bits(&os, 0x0, 1));	/* litlensym_255 */
+
+	/* Match */
+	ASSERT(put_bits(&os, 0x3, 2));	/* litlensym_257 */
+	ASSERT(put_bits(&os, 0x0, 1));	/* offsetsym_0 */
+
+	/* End of block */
+	ASSERT(put_bits(&os, 0x1, 2));	/* litlensym_256 */
+
+	ASSERT(flush_bits(&os));
+
+	verify_decompression(in, os.next - in, out, sizeof(out),
+			     expected_out, sizeof(expected_out));
+}
+
+int
+tmain(int argc, tchar *argv[])
+{
+	program_invocation_name = get_filename(argv[0]);
+
+	test_empty_offset_code();
+	test_singleton_litrunlen_code();
+	test_singleton_offset_code();
+
+	return 0;
+}
diff --git a/programs/test_util.c b/programs/test_util.c
index 189035d..8bd88e6 100644
--- a/programs/test_util.c
+++ b/programs/test_util.c
@@ -217,3 +217,17 @@ put_bits(struct output_bitstream *os, machine_word_t bits, int num_bits)
 	}
 	return true;
 }
+
+bool
+flush_bits(struct output_bitstream *os)
+{
+	while (os->bitcount > 0) {
+		if (os->next == os->end)
+			return false;
+		*os->next++ = os->bitbuf;
+		os->bitcount -= 8;
+		os->bitbuf >>= 8;
+	}
+	os->bitcount = 0;
+	return true;
+}
diff --git a/programs/test_util.h b/programs/test_util.h
index bbdc546..c88fc9c 100644
--- a/programs/test_util.h
+++ b/programs/test_util.h
@@ -61,5 +61,6 @@ struct output_bitstream {
 
 extern bool put_bits(struct output_bitstream *os, machine_word_t bits,
 		     int num_bits);
+extern bool flush_bits(struct output_bitstream *os);
 
 #endif /* PROGRAMS_TEST_UTIL_H */