stdlib: std::compression::zip and std::compression::deflate (#2930)

* stdlib: implement `std::compression::zip` and `std::compression::deflate`

- C3 implementation of DEFLATE (RFC 1951) and ZIP archive handling.
- Support for reading and writing archives using STORE and DEFLATE
methods.
- Decompression supports both fixed and dynamic Huffman blocks.
- Compression using greedy LZ77 matching.
- Zero dependencies on libc.
- Stream-based entry reading and writing.
- Full unit test coverage.

NOTE: This is an initial implementation. Future improvements could be:

- Optimization of the LZ77 matching (lazy matching).
- Support for dynamic Huffman blocks in compression.
- ZIP64 support for large files/archives.
- Support for encryption and additional compression methods.

* optimizations+refactoring

deflate:
- replace linear search with hash-based match finding.
- implement support for dynamic Huffman blocks using the Package-Merge
algorithm.
- add streaming decompression.
- add buffered StreamBitReader.

zip:
- add ZIP64 support.
- add CP437 and UTF-8 filename encoding detection.
- add DOS date/time conversion and timestamp preservation.
- add ZipEntryReader for streaming entry reads.
- implement ZipArchive.extract and ZipArchive.recover helpers.

other:
- Add `set_modified_time` to std::io;
- Add benchmarks and a few more unit tests.

* zip: add archive comment support

add tests

* forgot to rename the benchmark :(

* detect utf8 names on weird zips

fix method not passed to open_writer

* another edge case where directory doesn't end with /

* testing utilities

- detect encrypted zip
- `ZipArchive.open_writer` default to DEFLATE

* fix zip64 creation, add tests

* fix ZIP header endianness for big-endian compatibility

Update ZipLFH, ZipCDH, ZipEOCD, Zip64EOCD, and Zip64Locator structs to
use little-endian bitstruct types from std::core::bitorder

* fix ZipEntryReader position tracking and seek logic ZIP_METHOD_STORE

added a test to track this

* add package-merge algorithm attribution

Thanks @konimarti

* standalone deflate_benchmark.c3 against `miniz`

* fix integer overflows, leaks and improve safety

* a few safety for 32-bit systems and tests

* deflate compress optimization

* improve match finding, hash updates, and buffer usage

* use ulong for zip offsets

* style changes (#18)

* style changes

* update tests

* style changes in `deflate.c3`

* fix typo

* Allocator first. Some changes to deflate to use `copy_to`

* Fix missing conversion on 32 bits.

* Fix deflate stream. Formatting. Prefer switch over if-elseif

* - Stream functions now use long/ulong rather than isz/usz for seek/available.
- `instream.seek` is replaced by `set_cursor` and `cursor`.
- `instream.available`, `cursor` etc are long/ulong rather than isz/usz to be correct on 32-bit.

* Update to constdef

* Fix test

---------

Co-authored-by: Book-reader <thevoid@outlook.co.nz>
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>

benchmarks/stdlib/compression/deflate.c3

@@ -0,0 +1,51 @@
+module deflate_benchmarks;
+import std::compression::deflate;
+
+const uint SMALL_ITERATIONS = 50000;
+const uint LARGE_ITERATIONS = 100;
+
+// Data to compress
+const char[] SMALL_DATA = { [0..1023] = 'A' };
+const char[] LARGE_DATA = { [0..1048575] = 'B' };
+
+char[] small_compressed;
+char[] large_compressed;
+
+fn void initialize_bench() @init
+{
+	small_compressed = deflate::compress(mem, SMALL_DATA)!!;
+	large_compressed = deflate::compress(mem, LARGE_DATA)!!;
+	set_benchmark_warmup_iterations(2);
+	set_benchmark_max_iterations(10);
+
+	set_benchmark_func_iterations($qnameof(deflate_compress_small), SMALL_ITERATIONS);
+	set_benchmark_func_iterations($qnameof(deflate_decompress_small), SMALL_ITERATIONS);
+	set_benchmark_func_iterations($qnameof(deflate_compress_large), LARGE_ITERATIONS);
+	set_benchmark_func_iterations($qnameof(deflate_decompress_large), LARGE_ITERATIONS);
+}
+
+// =======================================================================================
+module deflate_benchmarks @benchmark;
+
+import std::compression::deflate;
+import std::core::mem;
+
+fn void deflate_compress_small() => @pool()
+{
+	char[]? compressed = deflate::compress(tmem, SMALL_DATA);
+}
+
+fn void deflate_decompress_small() => @pool()
+{
+	char[]? decompressed = deflate::decompress(tmem, small_compressed);
+}
+
+fn void deflate_compress_large() => @pool()
+{
+	char[]? compressed = deflate::compress(tmem, LARGE_DATA);
+}
+
+fn void deflate_decompress_large() => @pool()
+{
+	char[]? decompressed = deflate::decompress(tmem, large_compressed);
+}

benchmarks/stdlib/crypto/crypto_shootout.c3

@@ -32,8 +32,8 @@ fn void initialize_bench() @init
 		$qnameof(sha1_16)[..^4],
 		$qnameof(sha2_256_16)[..^4],
 		$qnameof(sha2_512_16)[..^4],
-		$qnameof(blake2s_256_16)[..^4],
+		//$qnameof(blake2s_256_16)[..^4],
-		$qnameof(blake2b_256_16)[..^4],
+		//$qnameof(blake2b_256_16)[..^4],
 		$qnameof(blake3_16)[..^4],
 		$qnameof(ripemd_160_16)[..^4],
 		$qnameof(whirlpool_16)[..^4],
@@ -68,8 +68,8 @@ fn void md5_16() => md5::hash(common_16);
 fn void sha1_16() => sha1::hash(common_16);
 fn void sha2_256_16() => sha256::hash(common_16);
 fn void sha2_512_16() => sha512::hash(common_16);
-fn void blake2s_256_16() => blake2::s(256, common_16);
+//fn void blake2s_256_16() => blake2::s(256, common_16);
-fn void blake2b_256_16() => blake2::b(256, common_16);
+//fn void blake2b_256_16() => blake2::b(256, common_16);
 fn void blake3_16() => blake3::hash(common_16);
 fn void ripemd_160_16() => ripemd::hash{160}(common_16);
 fn void whirlpool_16() => whirlpool::hash(common_16);
@@ -80,8 +80,8 @@ fn void md5_256() => md5::hash(common_256);
 fn void sha1_256() => sha1::hash(common_256);
 fn void sha2_256_256() => sha256::hash(common_256);
 fn void sha2_512_256() => sha512::hash(common_256);
-fn void blake2s_256_256() => blake2::s(256, common_256);
+//fn void blake2s_256_256() => blake2::s(256, common_256);
-fn void blake2b_256_256() => blake2::b(256, common_256);
+//fn void blake2b_256_256() => blake2::b(256, common_256);
 fn void blake3_256() => blake3::hash(common_256);
 fn void ripemd_160_256() => ripemd::hash{160}(common_256);
 fn void whirlpool_256() => whirlpool::hash(common_256);
@@ -92,8 +92,8 @@ fn void md5_4kib() => md5::hash(common_4kib);
 fn void sha1_4kib() => sha1::hash(common_4kib);
 fn void sha2_256_4kib() => sha256::hash(common_4kib);
 fn void sha2_512_4kib() => sha512::hash(common_4kib);
-fn void blake2s_256_4kib() => blake2::s(256, common_4kib);
+//fn void blake2s_256_4kib() => blake2::s(256, common_4kib);
-fn void blake2b_256_4kib() => blake2::b(256, common_4kib);
+//fn void blake2b_256_4kib() => blake2::b(256, common_4kib);
 fn void blake3_4kib() => blake3::hash(common_4kib);
 fn void ripemd_160_4kib() => ripemd::hash{160}(common_4kib);
 fn void whirlpool_4kib() => whirlpool::hash(common_4kib);
@@ -104,8 +104,8 @@ fn void md5_1mib() => md5::hash(common_1mib);
 fn void sha1_1mib() => sha1::hash(common_1mib);
 fn void sha2_256_1mib() => sha256::hash(common_1mib);
 fn void sha2_512_1mib() => sha512::hash(common_1mib);
-fn void blake2s_256_1mib() => blake2::s(256, common_1mib);
+//fn void blake2s_256_1mib() => blake2::s(256, common_1mib);
-fn void blake2b_256_1mib() => blake2::b(256, common_1mib);
+//fn void blake2b_256_1mib() => blake2::b(256, common_1mib);
 fn void blake3_1mib() => blake3::hash(common_1mib);
 fn void ripemd_160_1mib() => ripemd::hash{160}(common_1mib);
 fn void whirlpool_1mib() => whirlpool::hash(common_1mib);

lib/std/core/dstring.c3

@@ -658,9 +658,10 @@ fn usz? DString.read_from_stream(&self, InStream reader)
 	if (&reader.available)
 	{
 		usz total_read = 0;
-		while (usz available = reader.available()!)
+		while (ulong available = reader.available()!)
 		{
-			self.reserve(available);
+			if (available > isz.max) available = (ulong)isz.max;
+			self.reserve((usz)available);
 			StringData* data = self.data();
 			usz len = reader.read(data.chars[data.len..(data.capacity - 1)])!;
 			total_read += len;

lib/std/core/env.c3

@@ -126,6 +126,7 @@ const bool ARCH_64_BIT = $$REGISTER_SIZE == 64;
 const bool LIBC = $$COMPILER_LIBC_AVAILABLE;
 const bool NO_LIBC = !LIBC && !CUSTOM_LIBC;
 const bool CUSTOM_LIBC = $$CUSTOM_LIBC;
+const bool OLD_IO = $feature(OLD_IO);
 const CompilerOptLevel COMPILER_OPT_LEVEL = CompilerOptLevel.from_ordinal($$COMPILER_OPT_LEVEL);
 const bool BIG_ENDIAN = $$PLATFORM_BIG_ENDIAN;
 const bool I128_NATIVE_SUPPORT = $$PLATFORM_I128_SUPPORTED;

lib/std/core/runtime_benchmark.c3

@@ -125,10 +125,11 @@ fn bool run_benchmarks(BenchmarkUnit[] benchmarks)
 		char[] perc_str = { [0..19] = ' ', [20] = 0 };
 		int perc = 0;
 		uint print_step = current_benchmark_iterations / 100;
+		if (print_step == 0) print_step = 1;
 
 		for (this_iteration = 0; this_iteration < current_benchmark_iterations; ++this_iteration, benchmark_nano_seconds = {})
 		{
-			if (0 == this_iteration % print_step)   // only print right about when the % will update
+			if (this_iteration % print_step == 0)   // only print right about when the % will update
 			{
 				perc_str[0..(uint)math::floor((this_iteration / (float)current_benchmark_iterations) * 20)] = '#';
 				perc = (uint)math::ceil(100 * (this_iteration / (float)current_benchmark_iterations));

lib/std/core/runtime_test.c3

@@ -142,7 +142,7 @@ fn void mute_output() @local
     File* stderr = io::stderr();
 	*stderr = test_context.fake_stdout;
 	*stdout = test_context.fake_stdout;
-	(void)test_context.fake_stdout.seek(0, Seek.SET)!!;
+	(void)test_context.fake_stdout.set_cursor(0)!!;
 }
 
 fn void unmute_output(bool has_error) @local
@@ -155,7 +155,7 @@ fn void unmute_output(bool has_error) @local
 	*stderr = test_context.stored.stderr;
 	*stdout = test_context.stored.stdout;
 
-	usz log_size = test_context.fake_stdout.seek(0, Seek.CURSOR)!!;
+	ulong log_size = test_context.fake_stdout.cursor()!!;
 	if (has_error)
 	{
 		io::printn(test_context.has_ansi_codes ? "[\e[0;31mFAIL\e[0m]" : "[FAIL]");
@@ -165,7 +165,7 @@ fn void unmute_output(bool has_error) @local
 	{
 		test_context.fake_stdout.write_byte('\n')!!;
 		test_context.fake_stdout.write_byte('\0')!!;
-		(void)test_context.fake_stdout.seek(0, Seek.SET)!!;
+		test_context.fake_stdout.set_cursor(0)!!;
 
 		io::printfn("\n========== TEST LOG ============");
 		io::printfn("%s\n", test_context.current_test_name);

lib/std/io/file.c3

@@ -39,11 +39,16 @@ fn bool is_dir(String path)
 	return os::native_is_dir(path);
 }
 
-fn usz? get_size(String path)
+fn ulong? get_size(String path)
 {
 	return os::native_file_size(path);
 }
 
+fn void? set_modified_time(String path, Time_t time)
+{
+	return os::native_set_modified_time(path, time);
+}
+
 fn void? delete(String filename)
 {
 	return os::native_remove(filename) @inline;
@@ -63,10 +68,25 @@ fn void? File.reopen(&self, String filename, String mode)
 *>
 fn usz? File.seek(&self, isz offset, Seek seek_mode = Seek.SET) @dynamic
 {
-	os::native_fseek(self.file, offset, seek_mode)!;
+	os::native_fseek(self.file, offset, (SeekOrigin)seek_mode.ordinal)!;
-	return os::native_ftell(self.file);
+	return (usz)os::native_ftell(self.file);
 }
 
+<*
+ @require self.file != null
+*>
+fn void? File.set_cursor(&self, long offset, SeekOrigin whence = FROM_START) @dynamic
+{
+	return os::native_fseek(self.file, offset, whence);
+}
+
+<*
+ @require self.file != null
+*>
+fn long? File.cursor(&self) @dynamic
+{
+	return os::native_ftell(self.file);
+}
 
 /*
 Implement later
@@ -118,6 +138,14 @@ fn void? File.close(&self) @inline @dynamic
 	self.file = null;
 }
 
+fn ulong? File.size(&self) @dynamic
+{
+	long curr = self.cursor()!;
+	defer (void)self.set_cursor(curr);
+	self.set_cursor(0, FROM_END)!;
+	return self.cursor()!;
+}
+
 <*
  @require self.file != null
 *>
@@ -171,9 +199,8 @@ fn char[]? load_buffer(String filename, char[] buffer)
 {
 	File file = open(filename, "rb")!;
 	defer (void)file.close();
-	usz len = file.seek(0, END)!;
+	long len = file.size()!;
 	if (len > buffer.len) return io::OVERFLOW~;
-	file.seek(0, SET)!;
 	usz read = 0;
 	while (read < len)
 	{
@@ -187,16 +214,16 @@ fn char[]? load(Allocator allocator, String filename)
 {
 	File file = open(filename, "rb")!;
 	defer (void)file.close();
-	usz len = file.seek(0, END)!;
+	ulong len = file.size()!;
-	file.seek(0, SET)!;
+	if (len > usz.max) return io::OUT_OF_SPACE~;
-	char* data = allocator::malloc_try(allocator, len)!;
+	char* data = allocator::malloc_try(allocator, (usz)len)!;
 	defer catch allocator::free(allocator, data);
 	usz read = 0;
-	while (read < len)
+	while (read < (usz)len)
 	{
-		read += file.read(data[read:len - read])!;
+		read += file.read(data[read:(usz)len - read])!;
 	}
-	return data[:len];
+	return data[:(usz)len];
 }
 
 fn char[]? load_path(Allocator allocator, Path path) => load(allocator, path.str_view());

lib/std/io/file_mmap.c3

@@ -45,10 +45,9 @@ fn FileMmap? mmap_file(File file, usz offset = 0, usz len = 0, VirtualMemoryAcce
 {
 	if (len == 0)
 	{
-		usz cur = file.seek(0, CURSOR)!;
+		ulong new_len = file.size()! - offset;
-		defer file.seek(cur, SET)!!;
+		if (new_len > (ulong)isz.max) return mem::OUT_OF_MEMORY~;
-		usz file_size = file.seek(0, END)!;
+		len = (usz)new_len;
-		len = file_size - offset;
 	}
 
 	// get the page size

lib/std/io/io.c3

@@ -11,6 +11,14 @@ enum Seek
 	END
 }
 
+enum SeekOrigin
+{
+	FROM_START,
+	FROM_CURSOR,
+	FROM_END
+}
+
+
 faultdef
 	ALREADY_EXISTS,
 	BUSY,

lib/std/io/os/file_libc.c3

@@ -49,16 +49,16 @@ fn void*? native_freopen(void* file, String filename, String mode) @inline  => @
 		return file ?: file_open_errno()~;
 }
 
-fn void? native_fseek(void* file, isz offset, Seek seek_mode) @inline
+fn void? native_fseek(void* file, long offset, SeekOrigin seek_mode) @inline
 {
 	if (libc::fseek(file, (SeekIndex)offset, seek_mode.ordinal)) return file_seek_errno()~;
 }
 
 
-fn usz? native_ftell(CFile file) @inline
+fn long? native_ftell(CFile file) @inline
 {
 	long index = libc::ftell(file);
-	return index >= 0 ? (usz)index : file_seek_errno()~;
+	return index >= 0 ? index : file_seek_errno()~;
 }
 
 fn usz? native_fwrite(CFile file, char[] buffer) @inline
@@ -123,3 +123,22 @@ macro fault file_seek_errno() @local
 	}
 }
 
+
+struct Utimbuf
+{
+	Time_t actime;
+	Time_t modtime;
+}
+
+extern fn int utime(char* filename, void* times) @if(!env::WIN32);
+extern fn int _wutime(WChar* filename, void* times) @if(env::WIN32);
+
+fn void? native_set_modified_time(String filename, libc::Time_t time) => @stack_mem(256; Allocator smem)
+{
+	Utimbuf times = { time, time };
+	$if env::WIN32:
+		if (_wutime(filename.to_wstring(smem)!, &times)) return io::GENERAL_ERROR~;
+	$else
+		if (utime(filename.zstr_copy(smem), &times)) return io::GENERAL_ERROR~;
+	$endif
+}

lib/std/io/os/file_nolibc.c3

@@ -4,12 +4,13 @@ import libc;
 alias FopenFn = fn void*?(String, String);
 alias FreopenFn = fn void*?(void*, String, String);
 alias FcloseFn = fn void?(void*);
-alias FseekFn = fn void?(void*, isz, Seek);
+alias FseekFn = fn void?(void*, long, SeekOrigin);
-alias FtellFn = fn usz?(void*);
+alias FtellFn = fn long?(void*);
 alias FwriteFn = fn usz?(void*, char[] buffer);
 alias FreadFn = fn usz?(void*, char[] buffer);
 alias RemoveFn = fn void?(String);
 alias FputcFn = fn void?(int, void*);
+alias SetModifiedTimeFn = fn void?(String, libc::Time_t);
 
 FopenFn native_fopen_fn @weak @if(!$defined(native_fopen_fn));
 FcloseFn native_fclose_fn @weak @if(!$defined(native_fclose_fn));
@@ -20,6 +21,7 @@ FwriteFn native_fwrite_fn @weak @if(!$defined(native_fwrite_fn));
 FreadFn native_fread_fn @weak @if(!$defined(native_fread_fn));
 RemoveFn native_remove_fn @weak @if(!$defined(native_remove_fn));
 FputcFn native_fputc_fn @weak @if(!$defined(native_fputc_fn));
+SetModifiedTimeFn native_set_modified_time_fn @weak @if(!$defined(native_set_modified_time_fn));
 
 <*
  @require mode.len > 0
@@ -52,13 +54,13 @@ fn void*? native_freopen(void* file, String filename, String mode) @inline
 	return io::UNSUPPORTED_OPERATION~;
 }
 
-fn void? native_fseek(void* file, isz offset, Seek seek_mode) @inline
+fn void? native_fseek(void* file, long offset, SeekOrigin whence) @inline
 {
-	if (native_fseek_fn) return native_fseek_fn(file, offset, seek_mode);
+	if (native_fseek_fn) return native_fseek_fn(file, offset, whence);
 	return io::UNSUPPORTED_OPERATION~;
 }
 
-fn usz? native_ftell(CFile file) @inline
+fn ulong? native_ftell(CFile file) @inline
 {
 	if (native_ftell_fn) return native_ftell_fn(file);
 	return io::UNSUPPORTED_OPERATION~;
@@ -81,3 +83,9 @@ fn void? native_fputc(CInt c, CFile stream) @inline
 	if (native_fputc_fn) return native_fputc_fn(c, stream);
 	return io::UNSUPPORTED_OPERATION~;
 }
+
+fn void? native_set_modified_time(String filename, libc::Time_t time) @inline
+{
+	if (native_set_modified_time_fn) return native_set_modified_time_fn(filename, time);
+	return io::UNSUPPORTED_OPERATION~;
+}

lib/std/io/os/fileinfo.c3

@@ -47,14 +47,15 @@ fn usz? native_file_size(String path) @if(env::WIN32) => @pool()
 	return (usz)size.quadPart;
 }
 
-fn usz? native_file_size(String path) @if(!env::WIN32 && !env::DARWIN && !env::LINUX && !env::ANDROID && !env::BSD_FAMILY)
+fn ulong? native_file_size(String path) @if(!env::WIN32 && !env::DARWIN && !env::LINUX && !env::ANDROID && !env::BSD_FAMILY)
 {
 	File f = file::open(path, "r")!;
 	defer (void)f.close();
-	return f.seek(0, Seek.END)!;
+	f.set_cursor(0, FROM_END)!;
+	return f.cursor();
 }
 
-fn usz? native_file_size(String path) @if(env::DARWIN || env::LINUX || env::ANDROID || env::BSD_FAMILY)
+fn ulong? native_file_size(String path) @if(env::DARWIN || env::LINUX || env::ANDROID || env::BSD_FAMILY)
 {
 	Stat stat;
 	native_stat(&stat, path)!;

lib/std/io/path.c3

@@ -36,7 +36,7 @@ fn Path? cwd(Allocator allocator)
 
 fn bool is_dir(Path path) => os::native_is_dir(path.str_view());
 fn bool is_file(Path path) => os::native_is_file(path.str_view());
-fn usz? file_size(Path path) => os::native_file_size(path.str_view());
+fn ulong? file_size(Path path) => os::native_file_size(path.str_view());
 fn bool exists(Path path) => os::native_file_or_dir_exists(path.str_view());
 fn Path? tcwd() => cwd(tmem) @inline;
 

lib/std/io/stream.c3

@@ -1,12 +1,20 @@
 module std::io;
 import std::math;
 
+
+
+alias SetCursorFn = fn void?(void*, long offset, SeekOrigin whence = START);
+
+
 interface InStream
 {
 	fn void? close() @optional;
+	fn long? cursor() @optional;
+	fn void? set_cursor(long offset, SeekOrigin whence = FROM_START) @optional;
 	fn usz? seek(isz offset, Seek seek) @optional;
 	fn usz len() @optional;
-	fn usz? available() @optional;
+	fn ulong? size() @optional;
+	fn ulong? available() @optional;
 	fn usz? read(char[] buffer);
 	fn char? read_byte();
 	fn usz? write_to(OutStream out) @optional;
@@ -24,15 +32,23 @@ interface OutStream
 	fn usz? read_to(InStream in) @optional;
 }
 
-fn usz? available(InStream s)
+fn ulong? available(InStream s)
 {
 	if (&s.available) return s.available();
+	if (&s.set_cursor && &s.cursor)
+	{
+		long curr = s.cursor()!;
+		s.set_cursor(0, FROM_END)!;
+		ulong len = s.cursor()!;
+		s.set_cursor(curr)!;
+		return len - curr;
+	}
 	if (&s.seek)
 	{
 		usz curr = s.seek(0, Seek.CURSOR)!;
 		usz len = s.seek(0, Seek.END)!;
 		s.seek(curr, Seek.SET)!;
-		return len - curr;
+		return (ulong)len - (ulong)curr;
 	}
 	return io::UNSUPPORTED_OPERATION~;
 }
@@ -177,6 +193,11 @@ macro usz? write_using_write_byte(s, char[] bytes)
 
 macro void? pushback_using_seek(s)
 {
+	if (&s.set_cursor)
+	{
+		s.set_cursor(-1, FROM_CURSOR)!;
+		return;
+	}
 	s.seek(-1, CURSOR)!;
 }
 
@@ -407,11 +428,11 @@ macro ulong? read_le_ulong(stream)
 {
 	ulong val = (ulong)stream.read_byte()!;
 	val += (ulong)stream.read_byte()! << 8;
-    val += (ulong)stream.read_byte()! << 16;
+	val += (ulong)stream.read_byte()! << 16;
 	val += (ulong)stream.read_byte()! << 24;
 	val += (ulong)stream.read_byte()! << 32;
 	val += (ulong)stream.read_byte()! << 40;
-    val += (ulong)stream.read_byte()! << 48;
+	val += (ulong)stream.read_byte()! << 48;
 	return val + (ulong)stream.read_byte()! << 56;
 }
 
@@ -621,24 +642,30 @@ macro void? skip(stream, usz bytes)
 {
 	if (!bytes) return;
 	$switch:
-		$case !$defined(stream.seek):
-	        for (usz i = 0; i < bytes; i++)
-	        {
-	            stream.read()!;
-	        }
-	        return;
 		$case $typeof(stream) == InStream:
-	        if (!&stream.seek)
+			if (!&stream.seek && !&stream.set_cursor)
-	        {
+			{
-	            for (usz i = 0; i < bytes; i++)
+				for (usz i = 0; i < bytes; i++)
-	            {
+				{
-	                stream.read()!;
+					stream.read()!;
-	            }
+				}
-	            return;
+				return;
-	        }
+			}
+			if (!&stream.set_cursor)
+			{
+				stream.seek(bytes, CURSOR)!;
+				return;
+			}
+			stream.set_cursor(bytes, FROM_CURSOR)!;
+		$case $defined(stream.set_cursor):
+			stream.set_cursor(bytes, FROM_CURSOR)!;
+		$case $defined(stream.seek):
 			stream.seek(bytes, CURSOR)!;
 		$default:
-			stream.seek(bytes, CURSOR)!;
+			for (usz i = 0; i < bytes; i++)
+			{
+				stream.read()!;
+			}
 	$endswitch
 }
 

lib/std/io/stream/bytebuffer.c3

@@ -104,28 +104,37 @@ fn void? ByteBuffer.pushback_byte(&self) @dynamic
 	self.has_last = false;
 }
 
-fn usz? ByteBuffer.seek(&self, isz offset, Seek seek) @dynamic
+fn long? ByteBuffer.cursor(&self) @dynamic
 {
-	switch (seek)
-	{
-		case SET:
-			if (offset < 0 || offset > self.write_idx) return INVALID_POSITION~;
-			self.read_idx = offset;
-			return offset;
-		case CURSOR:
-			if ((offset < 0 && self.read_idx < -offset) ||
-				(offset > 0 && self.read_idx + offset > self.write_idx)) return INVALID_POSITION~;
-			self.read_idx += offset;
-		case END:
-			if (offset < 0 || offset > self.write_idx) return INVALID_POSITION~;
-			self.read_idx = self.write_idx - offset;
-	}
 	return self.read_idx;
 }
 
-fn usz? ByteBuffer.available(&self) @inline @dynamic
+fn void? ByteBuffer.set_cursor(&self, long offset, SeekOrigin whence = FROM_START) @dynamic
 {
-	return self.write_idx - self.read_idx;
+	switch (whence)
+	{
+		case FROM_START:
+			if (offset < 0 || offset > self.write_idx) return INVALID_POSITION~;
+			self.read_idx = (usz)offset;
+		case FROM_CURSOR:
+			if ((offset < 0 && self.read_idx < -offset) ||
+				(offset > 0 && self.read_idx + offset > self.write_idx)) return INVALID_POSITION~;
+			self.read_idx += (usz)offset;
+		case FROM_END:
+			if (offset < 0 || offset > self.write_idx) return INVALID_POSITION~;
+			self.read_idx = self.write_idx - (usz)offset;
+	}
+}
+
+fn usz? ByteBuffer.seek(&self, isz offset, Seek seek) @dynamic
+{
+	self.set_cursor(offset, (SeekOrigin)seek.ordinal)!;
+	return (usz)self.cursor();
+}
+
+fn ulong? ByteBuffer.available(&self) @inline @dynamic
+{
+	return (ulong)self.write_idx - self.read_idx;
 }
 
 fn void ByteBuffer.grow(&self, usz n)

lib/std/io/stream/bytereader.c3

@@ -41,16 +41,26 @@ fn void? ByteReader.pushback_byte(&self) @dynamic
 
 fn usz? ByteReader.seek(&self, isz offset, Seek seek) @dynamic
 {
-	isz new_index;
+	self.set_cursor((long)offset, (SeekOrigin)seek.ordinal)!;
-	switch (seek)
+	return (usz)self.cursor();
+}
+
+fn long? ByteReader.cursor(&self) @dynamic
+{
+	return self.index;
+}
+
+fn void? ByteReader.set_cursor(&self, long offset, SeekOrigin whence = FROM_START) @dynamic
+{
+	long new_index;
+	switch (whence)
 	{
-		case SET: new_index = offset;
+		case FROM_START: new_index = offset;
-		case CURSOR: new_index = self.index + offset;
+		case FROM_CURSOR: new_index = self.index + offset;
-		case END: new_index = self.bytes.len + offset;
+		case FROM_END: new_index = self.bytes.len + offset;
 	}
-	if (new_index < 0) return INVALID_POSITION~;
+	if (new_index < 0 || new_index > self.bytes.len) return INVALID_POSITION~;
-	self.index = new_index;
+	self.index = (usz)new_index;
-	return new_index;
 }
 
 fn usz? ByteReader.write_to(&self, OutStream writer) @dynamic
@@ -62,7 +72,7 @@ fn usz? ByteReader.write_to(&self, OutStream writer) @dynamic
 	return written;
 }
 
-fn usz? ByteReader.available(&self) @inline @dynamic
+fn ulong? ByteReader.available(&self) @inline @dynamic
 {
 	return max(0, self.bytes.len - self.index);
 }

lib/std/io/stream/bytewriter.c3

@@ -86,9 +86,10 @@ fn usz? ByteWriter.read_from(&self, InStream reader) @dynamic
 	usz start_index = self.index;
 	if (&reader.available)
 	{
-		while (usz available = reader.available()!)
+		while (ulong available = reader.available()!)
 		{
-			self.ensure_capacity(self.index + available)!;
+			if (available > usz.max) return OUT_OF_SPACE~;
+			self.ensure_capacity(self.index + (usz)available)!;
 			usz read = reader.read(self.bytes[self.index..])!;
 			self.index += read;
 		}

lib/std/io/stream/limitreader.c3

@@ -38,7 +38,7 @@ fn char? LimitReader.read_byte(&self) @dynamic
 	return self.wrapped_stream.read_byte();
 }
 
-fn usz? LimitReader.available(&self) @inline @dynamic
+fn ulong? LimitReader.available(&self) @inline @dynamic
 {
 	return self.limit;
 }

lib/std/os/linux/linux.c3

@@ -186,7 +186,7 @@ fn ulong? elf_module_image_base(String path) @local
 	bool is_little_endian = file.read_byte()! == 1;
 	// Actually, not supported.
 	if (!is_little_endian) return backtrace::IMAGE_NOT_FOUND~;
-	file.seek(0)!;
+	file.set_cursor(0)!;
 	if (is_64)
 	{
 		Elf64_Ehdr file_header;
@@ -195,7 +195,7 @@ fn ulong? elf_module_image_base(String path) @local
 		for (isz i = 0; i < file_header.e_phnum; i++)
 		{
 			Elf64_Phdr header;
-			file.seek((usz)file_header.e_phoff + (usz)file_header.e_phentsize * i)!;
+			file.set_cursor(file_header.e_phoff + (long)file_header.e_phentsize * i)!;
 			io::read_any(&file, &header)!;
 			if (header.p_type == PT_PHDR) return header.p_vaddr - header.p_offset;
 		}
@@ -207,7 +207,7 @@ fn ulong? elf_module_image_base(String path) @local
 	for (isz i = 0; i < file_header.e_phnum; i++)
 	{
 		Elf32_Phdr header;
-		file.seek(file_header.e_phoff + (usz)file_header.e_phentsize * i)!;
+		file.set_cursor(file_header.e_phoff + (long)file_header.e_phentsize * i)!;
 		io::read_any(&file, &header)!;
 		if (header.p_type == PT_PHDR) return (ulong)header.p_vaddr - header.p_offset;
 	}

releasenotes.md

@@ -33,6 +33,9 @@
 - Add `array::even`, `array::odd`, and `array::unlace` macros. #2892
 - Add discrete and continuous distributions in `std::math::distributions`.
 - Add bitorder functions `store_le`, `load_le`, `store_be`, `store_le`.
+- Stream functions now use long/ulong rather than isz/usz for seek/available. 
+- `instream.seek` is replaced by `set_cursor` and `cursor`.
+- `instream.available`, `cursor` etc are long/ulong rather than isz/usz to be correct on 32-bit.
 
 ### Fixes
 - Add error message if directory with output file name already exists

test/compression/deflate_benchmark.c3

@@ -0,0 +1,207 @@
+// 1. `gcc -O3 -c dependencies/miniz/miniz.c -o build/miniz.o`
+// 2. `build/c3c -O3 compile-run test/compression/deflate_benchmark.c3 build/miniz.o`
+
+module deflate_benchmark;
+import std, std::time::clock;
+
+const int AMOUNT_OF_WORK = 10; // Increase this to scale test data sizes
+
+fn int main(String[] args)
+{
+	io::printf("\n%s%s DEFLATE BENCHMARK %s", Ansi.BOLD, Ansi.BG_CYAN, Ansi.RESET);
+	io::printfn(" Comparing C3 std::compression::deflate with miniz (in-process)\n");
+	io::printfn(" Work Scale: %dx\n", AMOUNT_OF_WORK);
+
+	io::printfn("%-26s | %7s | %7s | %7s | %7s | %-10s", "Test Case", "C3 Rat.", "Miz Rat.", "C3 MB/s", "Miz MB/s", "Winner");
+	io::printfn("---------------------------+---------+---------+---------+---------+-----------");
+
+	// Test 1: Redundant data
+	usz redundant_size = 10_000_000 * (usz)AMOUNT_OF_WORK;
+	char[] redundant = allocator::alloc_array(tmem, char, redundant_size);
+	mem::set(redundant.ptr, 'A', redundant_size);
+	run_bench(string::tformat("Redundant (%dMB 'A')", (int)(redundant_size / 1_000_000)), redundant);
+
+	// Test 2: Large Source Project (All .c files in src/compiler)
+	DString project_src;
+	Path src_dir = path::new(tmem, "src/compiler")!!;
+	PathList? compiler_files = path::ls(tmem, src_dir);
+	if (try files = compiler_files) {
+		for (int i = 0; i < AMOUNT_OF_WORK; i++) {
+			foreach (p : files) {
+				if (p.basename().ends_with(".c")) {
+					Path full_p = src_dir.tappend(p.str_view())!!;
+					if (try data = file::load_path(tmem, full_p)) {
+						project_src.append(data);
+					}
+				}
+			}
+		}
+	}
+	run_bench("Compiler Source (Bulk)", project_src.str_view());
+
+	// Test 3: Standard Library (All .c3 files in lib/std)
+	DString std_src;
+	for (int i = 0; i < AMOUNT_OF_WORK; i++) {
+		collect_files(path::new(tmem, "lib/std")!!, ".c3", &std_src);
+	}
+	run_bench("Stdlib Source (Bulk)", std_src.str_view());
+
+	// Test 4: Log Files (Simulated)
+	DString log_data;
+	for (int i = 0; i < 50_000 * AMOUNT_OF_WORK; i++) {
+		log_data.appendf("2024-02-13 21:30:%02d.%03d [INFO] Connection established from 192.168.1.%d\n", i % 60, i % 1000, i % 255);
+		log_data.appendf("2024-02-13 21:30:%02d.%03d [DEBUG] Buffer size: %d bytes\n", i % 60, i % 1000, (i * 123) % 4096);
+	}
+	run_bench("Log Files (Simulated)", log_data.str_view());
+
+	// Test 5: Web Content (Simulated HTML/CSS)
+	DString web_data;
+	web_data.append("<!DOCTYPE html><html><head><style>.item { color: red; margin: 10px; }</style></head><body>");
+	for (int i = 0; i < 1000 * AMOUNT_OF_WORK; i++) {
+		web_data.appendf("<div class='item' id='obj%d'>", i);
+		web_data.append("<h1>Title of the item</h1><p>This is some repetitive descriptive text that might appear on a web page.</p>");
+		web_data.append("<ul><li>Feature 1</li><li>Feature 2</li><li>Feature 3</li></ul></div>");
+	}
+	web_data.append("</body></html>");
+	run_bench("Web Content (Simulated)", web_data.str_view());
+
+	// Test 6: CSV Data (Simulated)
+	DString csv_data;
+	csv_data.append("id,name,value1,value2,status,category\n");
+	for (int i = 0; i < 20_000 * AMOUNT_OF_WORK; i++) {
+		csv_data.appendf("%d,Product_%d,%d.5,%d,\"%s\",\"%s\"\n", 
+			i, i % 100, i * 10, i % 500, 
+			i % 3 == 0 ? "Active" : "Inactive", 
+			i % 5 == 0 ? "Electronics" : "Home");
+	}
+	run_bench("CSV Data (Simulated)", csv_data.str_view());
+
+	// Test 7: Binary Data (Structured)
+	usz binary_size = 2_000_000 * (usz)AMOUNT_OF_WORK;
+	char[] binary = allocator::alloc_array(tmem, char, binary_size);
+	for (usz i = 0; i < binary.len; i += 4) {
+		uint val = (uint)i ^ 0xDEADBEEF;
+		mem::copy(binary.ptr + i, &val, 4);
+	}
+	run_bench("Binary Data (Structured)", binary);
+
+	// Test 8: Random Noise (1MB)
+	usz noise_size = 1_000_000 * (usz)AMOUNT_OF_WORK;
+	DString noise;
+	for (usz i = 0; i < noise_size; i++) {
+		noise.append((char)rand('z' - 'a' + 1) + 'a');
+	}
+	run_bench("Random Noise (Scaled)", noise.str_view());
+
+	// Test 9: Tiny File (Check overhead)
+	run_bench("Tiny File (asd.c3)", "module asd; fn void main() {}\n");
+
+	// Test 10: Natural Language (Repetitive)
+	String text = "The quick brown fox jumps over the lazy dog. ";
+	DString long_text;
+	for (int i = 0; i < 50_000 * AMOUNT_OF_WORK; i++) long_text.append(text);
+	run_bench("Natural Text (Scaled)", long_text.str_view());
+
+	if (args.len > 1) {
+		Path custom_p = path::new(tmem, args[1])!!;
+		if (try custom_data = file::load_path(tmem, custom_p)) {
+			run_bench(string::tformat("Custom: %s", custom_p.basename()), custom_data);
+		}
+	}
+	// Final Summary
+	double avg_c3 = totals.c3_speed_sum / totals.count;
+	double avg_miniz = totals.miniz_speed_sum / totals.count;
+	double total_factor = avg_c3 / avg_miniz;
+
+	io::printfn("\n%sOVERALL SUMMARY%s", Ansi.BOLD, Ansi.RESET);
+	io::printfn("  Average Throughput C3:    %8.1f MB/s", avg_c3);
+	io::printfn("  Average Throughput Miniz: %8.1f MB/s", avg_miniz);
+	io::printfn("  %sC3 is %.1fx faster on average!%s\n", Ansi.BOLD, total_factor, Ansi.RESET);
+
+	return 0;
+}
+
+struct BenchResult {
+	long time_ns;
+	usz size;
+	double ratio;
+	double throughput_mbs;
+}
+
+struct BenchTotal {
+	double c3_speed_sum;
+	double miniz_speed_sum;
+	int count;
+}
+BenchTotal totals;
+
+fn void run_bench(String title, char[] data)
+{
+	// C3 Bench
+	Clock start = clock::now();
+	char[] c3_compressed = deflate::compress(data, tmem)!!;
+	Clock end = clock::now();
+	BenchResult c3 = calculate_metrics(data.len, (long)(end - start), c3_compressed.len);
+
+	// Miniz Bench
+	usz miniz_size;
+	start = clock::now();
+	void* miniz_ptr = tdefl_compress_mem_to_heap(data.ptr, data.len, &miniz_size, MINIZ_FLAGS);
+	end = clock::now();
+	BenchResult miniz = calculate_metrics(data.len, (long)(end - start), miniz_size);
+	if (miniz_ptr) mz_free(miniz_ptr);
+
+	// Performance Delta
+	double speed_factor = c3.throughput_mbs / miniz.throughput_mbs;
+	
+	io::printf("%-26s | %6.2f%% | %6.2f%% | %7.1f | %7.1f | %s%s (%.1fx)%s\n",
+		title[:(min(title.len, 26))],
+		c3.ratio, miniz.ratio,
+		c3.throughput_mbs, miniz.throughput_mbs,
+		speed_factor > 1.0 ? Ansi.CYAN : Ansi.WHITE,
+		speed_factor > 1.0 ? "C3" : "Miniz",
+		speed_factor > 1.0 ? speed_factor : 1.0 / speed_factor,
+		Ansi.RESET);
+
+	totals.c3_speed_sum += c3.throughput_mbs;
+	totals.miniz_speed_sum += miniz.throughput_mbs;
+	totals.count++;
+}
+
+fn void collect_files(Path dir, String suffix, DString* out)
+{
+	PathList? items = path::ls(tmem, dir);
+	if (catch items) return;
+	foreach (p : items) {
+		Path full = dir.tappend(p.str_view())!!;
+		if (path::is_dir(full)) {
+			if (p.basename() != "." && p.basename() != "..") {
+				collect_files(full, suffix, out);
+			}
+		} else if (p.basename().ends_with(suffix)) {
+			if (try data = file::load_path(tmem, full)) {
+				out.append(data);
+			}
+		}
+	}
+}
+
+fn BenchResult calculate_metrics(usz original_len, long time_ns, usz compressed_len)
+{
+	BenchResult res;
+	res.time_ns = time_ns;
+	res.size = compressed_len;
+	res.ratio = (double)compressed_len / (double)original_len * 100.0;
+	res.throughput_mbs = (double)original_len / (1024.0 * 1024.0) / ((double)time_ns / 1_000_000_000.0);
+	return res;
+}
+
+// External Miniz bindings
+extern fn void* tdefl_compress_mem_to_heap(void* pSrc_buf, usz src_buf_len, usz* pOut_len, int flags);
+extern fn void mz_free(void* p);
+
+const int TDEFL_GREEDY_PARSING_FLAG = 0x04000;
+const int TDEFL_NONDETERMINISTIC_PARSING_FLAG = 0x08000; // Fastest init for miniz for a fair comparisson
+const int C3_EQUIVALENT_PROBES = 16; // C3 uses MAX_CHAIN = 16 as default (this should be exposed)
+
+const int MINIZ_FLAGS = C3_EQUIVALENT_PROBES | TDEFL_GREEDY_PARSING_FLAG | TDEFL_NONDETERMINISTIC_PARSING_FLAG;

test/compression/zip_compare_7z.c3

@@ -0,0 +1,194 @@
+<*
+ Compare `C3 zip` vs `7z` extraction
+ External dependencies: 7z, diff
+*>
+module verify_zip;
+import std;
+import libc;
+
+fn int main(String[] args)
+{
+	if (args.len < 2)
+	{
+		io::printfn("Usage: %s [-r|--recursive] [-o|--output <dir>] <zip_dir>", args[0]);
+		return 1;
+	}
+
+	bool recursive = false;
+	String zip_dir;
+	String output_dir;
+
+	for (int i = 1; i < args.len; i++)
+	{
+		String arg = args[i];
+		switch (arg)
+		{
+			case "-r":
+			case "--recursive":
+				recursive = true;
+			case "-o":
+			case "--output":
+				if (++i >= args.len)
+				{
+					io::printfn("Error: %s requires a directory path", arg);
+					return 1;
+				}
+				output_dir = args[i];
+			default:
+				if (arg.starts_with("-"))
+				{
+					io::printfn("Error: unknown option %s", arg);
+					return 1;
+				}
+				if (zip_dir)
+				{
+					io::printfn("Error: multiple zip directories specified ('%s' and '%s')", zip_dir, arg);
+					return 1;
+				}
+				zip_dir = arg;
+		}
+	}
+
+	if (!zip_dir)
+	{
+		io::printfn("Error: no zip directory specified.");
+		return 1;
+	}
+
+	return process_dir(zip_dir, recursive, output_dir);
+}
+
+fn int process_dir(String dir, bool recursive, String output_dir)
+{
+	PathList? files = path::ls(tmem, path::temp(dir)!!);
+	if (catch excuse = files)
+	{
+		io::printfn("Could not open directory: %s (Excuse: %s)", dir, excuse);
+		return 1;
+	}
+
+	foreach (p : files)
+	{
+		String name = p.basename();
+		if (name == "." || name == "..") continue;
+
+		String zip_path = path::temp(dir)!!.tappend(name)!!.str_view();
+
+		if (file::is_dir(zip_path))
+		{
+			if (recursive)
+			{
+				if (process_dir(zip_path, recursive, output_dir) != 0) return 1;
+			}
+			continue;
+		}
+
+		if (!name.ends_with(".zip")) continue;
+
+		ulong size = 0;
+		File? f = file::open(zip_path, "rb");
+		if (try fh = f)
+		{
+			(void)fh.seek(0, Seek.END);
+			size = fh.seek(0, Seek.CURSOR) ?? 0;
+			fh.close()!!;
+		}
+		io::printf("Verifying %-40s [%7d KB] ", name[:(min(name.len, 40))], size / 1024);
+
+		switch (verify_one(zip_path, output_dir))
+		{
+			case 0:
+				io::printfn("%sFAILED%s ❌", Ansi.RED, Ansi.RESET);
+				return 1;
+			case 1:
+				io::printfn("%sPASSED%s ✅", Ansi.GREEN, Ansi.RESET);
+			default:
+				io::printn();
+		}
+	}
+
+	return 0;
+}
+
+fn int verify_one(String zip_path, String output_dir)
+{
+	Path extract_root;
+	if (output_dir)
+	{
+		extract_root = path::temp(output_dir)!!;
+	}
+	else
+	{
+		extract_root = path::temp_directory(tmem)!!;
+	}
+
+	String name = (String)path::temp(zip_path)!!.basename();
+
+	Path temp_c3 = extract_root.tappend(name.tconcat("_c3"))!!;
+	Path temp_7z = extract_root.tappend(name.tconcat("_7z"))!!;
+
+	(void)path::mkdir(temp_c3, true);
+	(void)path::mkdir(temp_7z, true);
+
+	ZipArchive? archive = zip::open(zip_path, "r");
+	if (catch excuse = archive)
+	{
+		io::printfn("%sFAIL%s (open: %s)", Ansi.RED, Ansi.RESET, excuse);
+		return 0;
+	}
+	defer (void)archive.close();
+
+	Time start = time::now();
+	if (catch excuse = archive.extract(temp_c3.str_view()))
+	{
+		if (excuse == zip::ENCRYPTED_FILE)
+		{
+			io::printf("%sSKIPPED%s (Encrypted)", Ansi.YELLOW, Ansi.RESET);
+			return 2;
+		}
+		io::printfn("%sFAIL%s (extract: %s)", Ansi.RED, Ansi.RESET, excuse);
+		return 0;
+	}
+	Duration c3_time = time::now() - start;
+
+	start = time::now();
+	if (!extract_7z(zip_path, temp_7z.str_view()))
+	{
+		io::printfn("%sFAIL%s (7z extract)", Ansi.RED, Ansi.RESET);
+		return 0;
+	}
+	Duration p7_time = time::now() - start;
+
+	io::printf(" [C3: %5d ms, 7z: %5d ms]", (long)c3_time / 1000, (long)p7_time / 1000);
+
+	io::print(" Comparing... ");
+	if (!compare_dirs(temp_c3.str_view(), temp_7z.str_view()))
+	{
+		io::printfn("%sFAIL%s (Differences found)", Ansi.RED, Ansi.RESET);
+		return 0;
+	}
+
+	// keep files on error for manual verification
+	(void)path::rmtree(temp_c3);
+	(void)path::rmtree(temp_7z);
+
+	return 1;
+}
+
+fn bool extract_7z(String zip_path, String output_dir)
+{
+	String out_opt = "-o".tconcat(output_dir);
+	String[] cmd = { "7z", "x", zip_path, out_opt, "-y", "-bb0" };
+	SubProcess? proc = process::create(cmd, { .search_user_path = true });
+	if (catch excuse = proc) return false;
+	return (int)proc.join()!! == 0;
+}
+
+fn bool compare_dirs(String dir1, String dir2)
+{
+	String[] cmd = { "diff", "-r", dir1, dir2 };
+	SubProcess? proc = process::create(cmd, { .search_user_path = true, .inherit_stdio = true });
+	if (catch excuse = proc) return false;
+	int res = (int)proc.join()!!;
+	return res == 0;
+}

test/unit/stdlib/compression/deflate.c3

@@ -0,0 +1,210 @@
+module deflate_test @test;
+
+import std::compression::deflate, std::io, std::math;
+
+fn void test_deflate_basic()
+{
+	String original = "Hello, world! This is a test of the DEFLATE compression algorithm.";
+	char[] compressed = deflate::compress(mem, original)!!;
+	defer free(compressed.ptr);
+
+	char[] decompressed = deflate::decompress(mem, compressed)!!;
+	defer free(decompressed.ptr);
+
+	assert((String)decompressed == original, "Decompressed data does not match original");
+}
+
+fn void test_deflate_repetitive()
+{
+	// 5000 bytes of repetitive data should compress very well
+	usz len = 5000;
+	char[] original = mem::malloc(len)[:len];
+	defer free(original.ptr);
+
+	for (usz i = 0; i < len; i++)
+	{
+		original[i] = (char)((i % 10) + '0');
+	}
+
+	char[] compressed = deflate::compress(mem, original)!!;
+	defer free(compressed.ptr);
+
+	// Check that we actually achieved some compression
+	assert(compressed.len < len / 10, "Repetitive data should compress well");
+
+	char[] decompressed = deflate::decompress(mem, compressed)!!;
+	defer free(decompressed.ptr);
+
+	assert(decompressed.len == original.len, "Length mismatch");
+	assert((String)decompressed == (String)original, "Data mismatch");
+}
+
+fn void test_deflate_empty()
+{
+	char[] original = {};
+	char[] compressed = deflate::compress(mem, original)!!;
+	defer free(compressed.ptr);
+
+	char[] decompressed = deflate::decompress(mem, compressed)!!;
+	defer free(decompressed.ptr);
+
+	assert(decompressed.len == 0, "Expected empty decompression");
+}
+
+fn void test_deflate_large_repetitive() @if($feature(SLOW_TESTS))
+{
+	// Test larger buffer to trigger reallocs in inflater
+	usz len = 100000;
+	char[] original = mem::malloc(len)[:len];
+	defer free(original.ptr);
+
+	mem::set(original.ptr, (char)'A', len);
+
+	char[] compressed = deflate::compress(mem, original)!!;
+	defer free(compressed.ptr);
+
+	char[] decompressed = deflate::decompress(mem, compressed)!!;
+	defer free(decompressed.ptr);
+
+	assert(decompressed.len == len, "Length mismatch");
+	assert(decompressed[0] == 'A' && decompressed[len-1] == 'A', "Data mismatch");
+}
+
+fn void test_deflate_random_ish()
+{
+	// Data that doesn't compress well
+	usz len = 1024;
+	char[] original = mem::malloc(len)[:len];
+	defer free(original.ptr);
+
+	for (usz i = 0; i < len; i++)
+	{
+		original[i] = (char)(i & 0xFF);
+	}
+
+	char[] compressed = deflate::compress(mem, original)!!;
+	defer free(compressed.ptr);
+
+	char[] decompressed = deflate::decompress(mem, compressed)!!;
+	defer free(decompressed.ptr);
+
+	assert((String)decompressed == (String)original, "Data mismatch");
+}
+
+fn void test_deflate_corrupted()
+{
+	char[] compressed = deflate::compress(mem, "Some data")!!;
+	defer free(compressed.ptr);
+
+	// Corrupt the block type (bits 1-2 of first byte) to 3 (reserved/invalid)
+	compressed[0] |= 0x06;
+
+	char[]? decompressed = deflate::decompress(mem, compressed);
+	assert(!@ok(decompressed), "Expected decompression to fail for corrupted data");
+}
+
+fn void test_deflate_stream()
+{
+	@pool()
+	{
+		String base = "This is a streaming test for DEFLATE. ";
+		usz base_len = base.len;
+		usz count = 50;
+		char[] original_arr = mem::malloc(base_len * count)[:base_len * count];
+		defer free(original_arr.ptr);
+		for (usz i = 0; i < count; i++) {
+			mem::copy(original_arr.ptr + i * base_len, base.ptr, base_len);
+		}
+		String original = (String)original_arr;
+
+		char[] compressed = deflate::compress(mem, original_arr)!!;
+		defer free(compressed.ptr);
+
+		// Use a temporary file on disk to test the streaming interface
+		File f = file::open("unittest_stream_deflate.bin", "wb+")!!;
+		defer { (void)f.close(); (void)file::delete("unittest_stream_deflate.bin"); }
+
+		f.write(compressed)!!;
+		f.seek(0, Seek.SET)!!;
+
+		// Decompress using stream
+		File out_f = file::open("unittest_stream_out.bin", "wb+")!!;
+		defer { (void)out_f.close(); (void)file::delete("unittest_stream_out.bin"); }
+
+		deflate::decompress_stream(&f, &out_f)!!;
+
+		usz out_size = out_f.seek(0, Seek.CURSOR)!!;
+		assert(out_size == original.len, "Length mismatch in streaming decompression");
+
+		out_f.seek(0, Seek.SET)!!;
+		char[] result = mem::malloc(out_size)[:out_size];
+		defer free(result.ptr);
+		out_f.read(result)!!;
+
+		assert((String)result == original, "Data mismatch in streaming decompression");
+	};
+}
+
+fn void test_deflate_embedded_stream()
+{
+	String base = "This is a streaming test for DEFLATE. ";
+
+	char[] compressed = deflate::compress(mem, base[..])!!;
+	defer free(compressed.ptr);
+
+	usz append_len = compressed.len + 1;
+	char[] append = mem::malloc(append_len)[:append_len];
+	defer free(append.ptr);
+
+	append[:compressed.len] = compressed[..];
+	append[compressed.len..] = 'c';
+
+	ByteReader reader;
+	reader.init(append);
+
+	ByteWriter writer;
+	writer.tinit();
+
+	deflate::decompress_stream(&reader, &writer)!!;
+
+	assert(writer.str_view() == base);
+
+	assert(reader.read_byte()!! == 'c');
+}
+
+fn void test_deflate_incremental()
+{
+	@pool()
+	{
+		String original = "This is a test of incremental decompression. We will read it byte by byte.";
+		char[] compressed = deflate::compress(mem, original)!!;
+		defer free(compressed.ptr);
+
+		// Use a ByteReader for the compressed data
+		io::ByteReader in_stream;
+		in_stream.init(compressed);
+
+		Inflater* inflater = mem::new(Inflater);
+		char[] bit_buf = mem::malloc(8192)[:8192];
+		inflater.init(&in_stream, bit_buf);
+		defer free(bit_buf.ptr);
+		defer free(inflater);
+
+		char[] decompressed = mem::malloc(original.len)[:original.len];
+		defer free(decompressed.ptr);
+
+		for (usz i = 0; i < original.len; i++)
+		{
+			char[1] one_byte;
+			usz n = inflater.read(one_byte[..])!!;
+			assert(n == 1, "Expected 1 byte");
+			decompressed[i] = one_byte[0];
+		}
+
+		// One more read should return 0 (or EOF)
+		char[1] extra;
+		assert(inflater.read(extra[..])!! == 0, "Expected EOF");
+
+		assert((String)original == (String)decompressed, "Incremental decompression failed");
+	};
+}

test/unit/stdlib/compression/zip.c3

@@ -0,0 +1,549 @@
+module zip_test @test;
+
+import std::io;
+import std::compression::zip;
+
+fn void test_zip_store()
+{
+	@pool()
+	{
+		// Create archive with uncompressed file
+		ZipArchive zip = zip::open(mem, "unittest_store.zip", "w")!!;
+		zip.write_file("test.txt", "Hello, World!", STORE)!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_store.zip");
+
+		// Read and verify
+		ZipArchive read_zip = zip::open(mem, "unittest_store.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 1, "Expected 1 entry");
+
+		ZipEntry entry = read_zip.stat("test.txt")!!;
+		assert(entry.method == STORE, "Expected STORE method");
+		assert(entry.uncompressed_size == 13, "Expected 13 bytes");
+
+		char[] data = read_zip.read_file_all(mem, "test.txt")!!;
+		defer free(data);
+		assert((String)data == "Hello, World!", "Data mismatch");
+
+	};
+}
+
+fn void test_zip_deflate()
+{
+	@pool()
+	{
+		// Create archive with compressed file
+		ZipArchive zip = zip::open(mem, "unittest_deflate.zip", "w")!!;
+
+		ZipEntryWriter writer = zip.open_writer("compressed.txt", DEFLATE)!!;
+		String data = "This is a test. ";
+		for (int i = 0; i < 100; i++)
+		{
+			writer.write((char[])data)!!;
+		}
+		writer.close()!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_deflate.zip");
+
+		// Read and verify
+		ZipArchive read_zip = zip::open(mem, "unittest_deflate.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 1, "Expected 1 entry");
+
+		ZipEntry entry = read_zip.stat("compressed.txt")!!;
+		assert(entry.method == DEFLATE, "Expected DEFLATE method");
+		assert(entry.uncompressed_size == 1600, "Expected 1600 bytes");
+
+		char[] decompressed = read_zip.read_file_all(mem, "compressed.txt")!!;
+		defer free(decompressed);
+		assert(decompressed.len == 1600, "Decompressed size mismatch");
+
+	};
+}
+
+fn void test_zip_directory()
+{
+	@pool()
+	{
+		// Create archive with directory
+		ZipArchive zip = zip::open(mem, "unittest_dir.zip", "w")!!;
+		zip.add_directory("docs")!!;
+		zip.write_file("docs/readme.txt", "README")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_dir.zip");
+
+		// Read and verify
+		ZipArchive read_zip = zip::open(mem, "unittest_dir.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 2, "Expected 2 entries");
+
+		ZipEntry dir_entry = read_zip.stat("docs/")!!;
+		assert(dir_entry.is_directory, "Expected directory");
+		assert(dir_entry.uncompressed_size == 0, "Directory should have 0 size");
+
+		ZipEntry file_entry = read_zip.stat("docs/readme.txt")!!;
+		assert(!file_entry.is_directory, "Expected file");
+
+		char[] data = read_zip.read_file_all(mem, "docs/readme.txt")!!;
+		defer free(data);
+		assert((String)data == "README", "Data mismatch");
+
+	};
+}
+
+fn void test_zip_crc32_verification()
+{
+	@pool()
+	{
+		// Create archive
+		ZipArchive zip = zip::open(mem, "unittest_crc.zip", "w")!!;
+		zip.write_file("data.txt", "Test data for CRC32")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_crc.zip");
+
+		ZipArchive read_zip = zip::open(mem, "unittest_crc.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		char[] data = read_zip.read_file_all(mem, "data.txt")!!;
+		defer free(data);
+		assert((String)data == "Test data for CRC32", "Data mismatch");
+
+	};
+}
+
+fn void test_zip_multiple_files()
+{
+	@pool()
+	{
+		// Create archive with multiple files
+		ZipArchive zip = zip::open(mem, "unittest_multi.zip", "w")!!;
+		zip.write_file("file1.txt", "First file")!!;
+		zip.write_file("file2.txt", "Second file")!!;
+		zip.write_file("file3.txt", "Third file")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_multi.zip");
+
+		// Read and verify
+		ZipArchive read_zip = zip::open(mem, "unittest_multi.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 3, "Expected 3 entries");
+
+		for (usz i = 0; i < read_zip.count(); i++)
+		{
+			ZipEntry entry = read_zip.stat_at(i)!!;
+			assert(!entry.is_directory, "Expected files only");
+		}
+
+		char[] data1 = read_zip.read_file_all(mem, "file1.txt")!!;
+		defer free(data1);
+		assert((String)data1 == "First file", "File1 mismatch");
+
+		char[] data2 = read_zip.read_file_all(mem, "file2.txt")!!;
+		defer free(data2);
+		assert((String)data2 == "Second file", "File2 mismatch");
+
+		char[] data3 = read_zip.read_file_all(mem, "file3.txt")!!;
+		defer free(data3);
+		assert((String)data3 == "Third file", "File3 mismatch");
+
+	};
+}
+
+fn void test_zip_streaming()
+{
+	@pool()
+	{
+		// Test streaming write
+		ZipArchive zip = zip::open(mem, "unittest_stream.zip", "w")!!;
+
+		ZipEntryWriter writer = zip.open_writer("stream.txt", DEFLATE)!!;
+		writer.write("Part 1. ")!!;
+		writer.write("Part 2. ")!!;
+		writer.write("Part 3.")!!;
+		writer.close()!!;
+
+		(void)zip.close();
+		defer (void)file::delete("unittest_stream.zip");
+
+		// Read and verify
+		ZipArchive read_zip = zip::open(mem, "unittest_stream.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		char[] data = read_zip.read_file_all(mem, "stream.txt")!!;
+		defer free(data);
+		assert((String)data == "Part 1. Part 2. Part 3.", "Streaming write failed");
+
+	};
+}
+fn void test_zip_invalid_access()
+{
+	@pool()
+	{
+		// Test non-existent archive
+		ZipArchive? opt = zip::open(mem, "non_existent.zip", "r");
+		assert(!@ok(opt), "Expected error when opening non-existent file");
+
+		// Test non-existent entry
+		ZipArchive zip = zip::open(mem, "unittest_edge.zip", "w")!!;
+		zip.write_file("exists.txt", "data")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_edge.zip");
+
+		ZipArchive read_zip = zip::open(mem, "unittest_edge.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntry? entry_opt = read_zip.stat("does_not_exist.txt");
+		assert(!@ok(entry_opt), "Expected ENTRY_NOT_FOUND");
+
+		char[]? data_opt = read_zip.read_file_all(mem, "does_not_exist.txt");
+		assert(!@ok(data_opt), "Expected error when reading non-existent file");
+		assert(!@ok(data_opt), "Expected error when reading non-existent file");
+
+	};
+}
+
+fn void test_zip_empty_archive()
+{
+	@pool()
+	{
+		// Create empty archive
+		ZipArchive zip = zip::open(mem, "unittest_empty.zip", "w")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_empty.zip");
+
+		// Read empty archive
+		ZipArchive read_zip = zip::open(mem, "unittest_empty.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 0, "Expected 0 entries");
+	};
+}
+
+fn void test_zip_recovery()
+{
+	@pool()
+	{
+		String path = "unittest_embedded_broken.zip";
+		// Create a "broken" ZIP (LFH + Data, but no Central Directory)
+		// Filename: "a", Data: "bc"
+		char[] broken_zip = {0x50,0x4B,0x03,0x04,0x14,0x00,0x00,0x08,0x00,0x00,0x00,0x00,0x00,0x00,0x38,0x2B,0xA9,0xC2,0x02,0x00,0x00,0x00,0x02,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x61,0x62,0x63};
+		file::save(path, broken_zip)!!;
+		defer (void)file::delete(path);
+
+		ZipArchive? normal = zip::open(mem, path, "r");
+		assert(!@ok(normal), "Normal open should fail on broken ZIP");
+
+		ZipArchive recovered = zip::recover(mem, path)!!;
+		defer (void)recovered.close();
+
+		assert(recovered.count() == 1, "Should have recovered 1 file");
+		char[] data = recovered.read_file_all(mem, "a")!!;
+		defer free(data);
+		assert((String)data == "bc", "Recovered data mismatch");
+	};
+}
+
+fn void test_zip_cp437()
+{
+	@pool()
+	{
+		String path = "unittest_embedded_cp437.zip";
+		// Create a ZIP with CP437 encoding (Bit 11 NOT set)
+		// Filename: 0x80 (Ç in CP437), Data: "x"
+		char[] cp437_zip = {0x50,0x4B,0x03,0x04,0x14,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x83,0x16,0xDC,0x8C,0x01,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x80,0x78};
+
+		file::save(path, cp437_zip)!!;
+		defer (void)file::delete(path);
+
+		ZipArchive recovered = zip::recover(mem, path)!!;
+		defer (void)recovered.close();
+
+		ZipEntry entry = recovered.stat_at(0)!!;
+		assert(entry.name == "Ç", "CP437 decoding failed");
+
+		char[] data = recovered.read_file_all(mem, "Ç")!!;
+		defer free(data);
+		assert((String)data == "x", "Data mismatch in CP437 test");
+	};
+}
+
+fn void test_zip_with_comment()
+{
+	@pool()
+	{
+		// Create a ZIP file with a comment
+		ZipArchive zip = zip::open(mem, "unittest_comment.zip", "w")!!;
+		zip.write_file("test.txt", "Hello, World!")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_comment.zip");
+
+		char[] zip_data = file::load(mem, "unittest_comment.zip")!!;
+		defer free(zip_data);
+
+		isz eocd_pos = -1;
+		for (isz i = (isz)zip_data.len - 22; i >= 0; i--)
+		{
+			uint sig = mem::load((uint*)&zip_data[i], 1);
+			if (sig == 0x06054b50)
+			{
+				eocd_pos = i;
+				break;
+			}
+		}
+		assert(eocd_pos >= 0, "EOCD not found");
+
+		String comment = "This is a test comment!";
+		mem::store((ushort*)&zip_data[eocd_pos + 20], (ushort)comment.len, 1);
+
+		char[] new_zip = mem::new_array(char, zip_data.len + comment.len);
+		defer free(new_zip);
+		mem::copy(new_zip.ptr, zip_data.ptr, zip_data.len);
+		mem::copy(new_zip.ptr + zip_data.len, comment.ptr, comment.len);
+
+		file::save("unittest_comment.zip", new_zip[:zip_data.len + comment.len])!!;
+
+		// Try to open it
+		ZipArchive read_zip = zip::open(mem, "unittest_comment.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.count() == 1, "Expected 1 entry");
+		assert(read_zip.comment == comment, "Comment mismatch");
+
+		char[] data = read_zip.read_file_all(mem, "test.txt")!!;
+		defer free(data);
+		assert((String)data == "Hello, World!", "Data mismatch with comment");
+	};
+}
+
+fn void test_zip_write_comment()
+{
+	@pool()
+	{
+		ZipArchive zip = zip::open(mem, "unittest_write_comment.zip", "w")!!;
+		zip.comment = String.copy("Created by C3 ZIP library", zip.allocator);
+		zip.write_file("test.txt", "Hello!")!!;
+		(void)zip.close();
+		defer (void)file::delete("unittest_write_comment.zip");
+
+		ZipArchive read_zip = zip::open(mem, "unittest_write_comment.zip", "r")!!;
+		defer (void)read_zip.close();
+
+		assert(read_zip.comment == "Created by C3 ZIP library", "Comment not preserved");
+		assert(read_zip.count() == 1, "Expected 1 entry");
+	};
+}
+
+fn void test_zip64_headers()
+{
+	@pool()
+	{
+		String filename = "unittest_zip64.zip";
+		ZipArchive zip = zip::open(mem, filename, "w")!!;
+
+		ZipEntryWriter writer = zip.open_writer("large.txt", STORE)!!;
+		writer.write("data")!!;
+
+		// Manually set the size to > 4GB to trigger ZIP64 headers in the Central Directory.
+		// This tests the fix for ZIP64 extra field serialization (ensuring no byte truncation).
+		writer.entry.uncompressed_size = 0x100000001;
+		writer.entry.compressed_size = 0x100000001;
+
+		writer.close()!!;
+		(void)zip.close();
+		defer (void)file::delete(filename);
+
+		ZipArchive read_zip = zip::open(mem, filename, "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntry entry = read_zip.stat("large.txt")!!;
+		assert(entry.uncompressed_size == 0x100000001, "Failed to read ZIP64 uncompressed size");
+		assert(entry.compressed_size == 0x100000001, "Failed to read ZIP64 compressed size");
+	};
+}
+
+fn void test_zip_utf8()
+{
+	@pool()
+	{
+		String filename = "unittest_utf8.zip";
+		String utf8_name = "测试_🚀.txt";
+		ZipArchive zip = zip::open(mem, filename, "w")!!;
+		zip.write_file(utf8_name, "content")!!;
+		(void)zip.close();
+		defer (void)file::delete(filename);
+
+		ZipArchive read_zip = zip::open(mem, filename, "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntry entry = read_zip.stat(utf8_name)!!;
+		assert(entry.name == utf8_name, "UTF-8 filename mismatch");
+	};
+}
+
+fn void test_zip_zero_length()
+{
+	@pool()
+	{
+		String filename = "unittest_zero.zip";
+		ZipArchive zip = zip::open(mem, filename, "w")!!;
+		zip.write_file("empty.txt", "")!!;
+		(void)zip.close();
+		defer (void)file::delete(filename);
+
+		ZipArchive read_zip = zip::open(mem, filename, "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntry entry = read_zip.stat("empty.txt")!!;
+		assert(entry.uncompressed_size == 0, "Size should be 0");
+
+		char[] data = read_zip.read_file_all(mem, "empty.txt")!!;
+		defer free(data);
+		assert(data.len == 0, "Read data should be empty");
+	};
+}
+
+fn void test_zip64_offset()
+{
+	@pool()
+	{
+		String filename = "unittest_zip64_offset.zip";
+		ZipArchive zip = zip::open(mem, filename, "w")!!;
+
+		ZipEntryWriter writer = zip.open_writer("offset_test.txt", STORE)!!;
+		writer.write("data")!!;
+
+		// Manually set offset to > 4GB to trigger ZIP64 headers in the Central Directory
+		writer.entry.offset = 0x100000005;
+
+		writer.close()!!;
+		(void)zip.close();
+		defer (void)file::delete(filename);
+
+		ZipArchive read_zip = zip::open(mem, filename, "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntry entry = read_zip.stat("offset_test.txt")!!;
+		assert(entry.offset == 0x100000005, "Failed to read ZIP64 offset");
+	};
+}
+
+fn void test_zip_reader_pos_and_seek()
+{
+	@pool()
+	{
+		String path = "unittest_reader.zip";
+		ZipArchive zip = zip::open(mem, path, "w")!!;
+		zip.write_file("test.txt", "0123456789", STORE)!!;
+		(void)zip.close();
+		defer (void)file::delete(path);
+
+		ZipArchive read_zip = zip::open(mem, path, "r")!!;
+		defer (void)read_zip.close();
+
+		ZipEntryReader reader = read_zip.open_reader("test.txt")!!;
+		defer (void)reader.close();
+
+		assert(reader.len() == 10, "Expected length 10");
+		assert(reader.available()!! == 10, "Expected 10 bytes available");
+		assert(reader.pos == 0, "Expected pos 0");
+
+		char[3] buf;
+		assert(reader.read(buf[..])!! == 3);
+		assert((String)buf[..] == "012", "Expected '012'");
+		assert(reader.pos == 3, "Expected pos 3");
+		assert(reader.available()!! == 7, "Expected 7 bytes available");
+
+		assert(reader.seek(2, Seek.CURSOR)!! == 5, "Expected seek to 5");
+		assert(reader.pos == 5, "Expected pos 5 after seek");
+		assert(reader.available()!! == 5, "Expected 5 bytes available after seek");
+
+		assert(reader.read(buf[..])!! == 3);
+		assert((String)buf[..] == "567", "Expected '567'");
+
+		assert(reader.seek(1, Seek.SET)!! == 1, "Expected seek to 1");
+		assert(reader.read(buf[..])!! == 3);
+		assert((String)buf[..] == "123", "Expected '123'");
+
+		assert(reader.seek(-2, Seek.END)!! == 8, "Expected seek to 8");
+		assert(reader.read(buf[..])!! == 2);
+		assert((String)buf[:2] == "89", "Expected '89'");
+		assert(reader.available()!! == 0, "Expected 0 bytes available at end");
+
+		// Edge case: Negative seek SET
+		assert(!@ok(reader.seek(-1, Seek.SET)), "Negative seek SET should fail");
+
+		// Edge case: Seek past end
+		assert(reader.seek(100, Seek.SET)!! == 10, "Seek past end should cap at size");
+		assert(reader.pos == 10, "Pos should be 10");
+	};
+}
+
+fn void test_zip_comment_boundary()
+{
+	@pool()
+	{
+		String filename = "unittest_comment_limit.zip";
+
+		// 1. Test exactly 65535 bytes (Should pass)
+		{
+			ZipArchive zip = zip::open(mem, filename, "w")!!;
+			char[] huge_comment = allocator::malloc(tmem, 65535)[:65535];
+			mem::set(huge_comment.ptr, (char)'C', 65535);
+			zip.comment = String.copy((String)huge_comment, zip.allocator);
+			zip.write_file("t.txt", "d")!!;
+			(void)zip.close();
+
+			ZipArchive read_zip = zip::open(mem, filename, "r")!!;
+			assert(read_zip.comment.len == 65535, "Comment length mismatch at 65535");
+			(void)read_zip.close();
+			(void)file::delete(filename);
+		}
+
+		// 2. Test 65536 bytes (Should fail with INVALID_ARGUMENT)
+		{
+			ZipArchive zip = zip::open(mem, filename, "w")!!;
+			char[] too_huge = allocator::malloc(tmem, 65536)[:65536];
+			mem::set(too_huge.ptr, (char)'X', 65536);
+			zip.comment = String.copy((String)too_huge, zip.allocator);
+			zip.write_file("t.txt", "d")!!;
+
+			fault res = @catch(zip.close());
+			assert(res == zip::INVALID_ARGUMENT, "Expected INVALID_ARGUMENT for 64k+1 comment");
+			(void)file::delete(filename);
+		}
+	};
+}
+
+fn void test_zip_reader_available_capping()
+{
+	@pool()
+	{
+		// We manually construct a reader to test the capping logic for huge entry sizes
+		// that might exist on 32-bit systems (where usz < 64-bit).
+		ZipEntryReader reader;
+		mem::set(&reader, 0, ZipEntryReader.sizeof);
+		reader.size = 0xFFFFFFFFFFFFFFFF;
+		reader.pos = 0;
+
+		usz avail = reader.available()!!;
+		assert(avail == usz.max, "Expected available to be capped at usz.max");
+
+		reader.pos = 100;
+		avail = reader.available()!!;
+		if (usz.max < 0xFFFFFFFFFFFFFFFF)
+		{
+			// triggers on 32-bit
+			assert(avail == usz.max, "Expected available to still be capped at usz.max");
+		}
+		else
+		{
+			// on 64-bit
+			assert(avail == usz.max - (usz)100, "Expected available size to be correct on 64-bit");
+		}
+	};
+}