13 Commits

Author SHA1 Message Date
Caleb Gardner 1a8838b544 Fixed missed merge text (oops)
Changed all *const Decompressor to *Decompressor
Changed all decompressors to only stateless (my queues are borked)
2026-05-24 15:26:37 -05:00
Caleb Gardner a9e50a0ff5 Added dedicated single_threaded mode for extraction
Cleanup
2026-05-24 06:48:04 -05:00
Caleb Gardner 712c4d0a19 Fixed issues when using Threaded.single_threaded 2026-05-24 06:48:04 -05:00
Caleb Gardner 5975bbb4a2 Build is working again (on Zig master branch)
Re-added specifying thread count doing something
Added single-threaded performance to benchmark.sh
Added single-threaded test (currently getting stuck forever)
Various minor fixes revealed now that build is working again.
2026-05-24 06:47:50 -05:00
Caleb Gardner 3ea3d8e9a0 Trying to fix build issues (SEGV)
Fix minor issues with new decomp types
2026-05-23 16:11:59 -05:00
Caleb Gardner 5f1089406e Re-added all C decompressors
Some cleanup
Remove inode arena
Added deinit to Archive to destroy the File.MemoryMap
2026-05-23 06:37:34 -05:00
Caleb Gardner 1dae4d8bb7 Updated README.md 2026-05-22 15:56:06 -05:00
Caleb Gardner 3239bf0e01 IT WORKS AND IS FAST 2026-05-22 15:45:44 -05:00
Caleb Gardner 0df14b8adc Moved to File.MemoryMap instead of direct file I/O 2026-05-22 12:49:07 -05:00
Caleb Gardner 8186c3fe9a Further work tweaking decompression 2026-05-22 07:06:16 -05:00
Caleb Gardner 2b49395ab2 Fixes and optimizations
Added FragManager so each frag block only gets decompressed once
Returned to C for decompression (only zstd stateless ATM)
2026-05-22 06:09:06 -05:00
Caleb Gardner 84a9cf17b9 Fixed some issues with stateless Lookup Table 2026-05-21 05:42:00 -05:00
Caleb Gardner d1d453ac29 Finished an initial version of extraction
It works, but is very slow.
2026-05-21 05:07:02 -05:00
38 changed files with 1614 additions and 798 deletions
+14 -10
View File
@@ -10,9 +10,13 @@ Overall works, but currently is missing some features ([see below](#capabilities
## Build options
> `-Duse_c_libs=true`
> `-Duse_zig_decomp=true`
Instead of using Zig's standard library for decompression, use the system's C libraries. Has the benefit of being much faster and enabling LZO and LZ4 decompression.
Instead of using C libraries for decompression, use Zig's standard library for decompression. If using this option LZO and LZ4 decomrpession types are unsupported and decompression times will be significantly longer.
> `-Ddynamic=true`
Dynamicly link C libraries (if they're used) instead of statically linking them.
> `-Dallow_lzo=true`
@@ -35,22 +39,22 @@ Most features are present except for the following:
## Performance
This is some basic observation's I've made about this library's performance when compared to `unsquashfs`. Unless otherwise stated, most observations were made when extracting my test archive (which is fairly small and uses zstd compression) and with `--release=fast`.
This is some basic observation's I've made about this library's performance when compared to `unsquashfs`. Unless otherwise stated, most observations were made when extracting my test archive which is fairly small and uses zstd compression with `-Doptimize=ReleaseFast`.
Currently, my only performance checks are checking execution time, nothing deeper.
* Under ideal circumstances, my library is ~70% slower (.12s vs .20s).
* Using Zig decompression libraries *significantly* increases decompression time by ~600%. Under ideal circumstances.
* Currently, using my test archive, performance aproximately matches `unsquashfs` when multi-threaded, but significantly slower when single-threaded.
* Using Zig decompression libraries *significantly* increases decompression time.
* Performance improvements/regressions will be common. I'm still learning Zig.
Example Times:
* *unsquashfs, multi-threaded*: .12s
* *unsquashfs, multi-threaded*: .11s
* *unsquashfs, single-threaded*: .13s
* *C-libs, single-threaded*: .45s
* *C-libs, multi-threaded*: .20s
* *Zig-libs, single-threaded*: 5.78s
* *Zig-libs, multi-threaded*: 1.08s
* *C-libs, multi-threaded*: .10s
* *C-libs, single-threaded*: ..28s
* *Zig-libs, single-threaded*: .74s
* *Zig-libs, multi-threaded*: 2.70s
## Build considerations
Executable
+17
View File
@@ -0,0 +1,17 @@
#! /usr/bin/env bash
ARCHIVE="testing/LinuxPATest.sfs"
REF_EXT_LOC="testing/LinuxPAReference"
PROG_EXT_LOC="testing/LinuxPABinTest"
echo "Testing Multi-threaded Performance"
echo ""
hyperfine --warmup 5 --prepare "rm -rf $REF_EXT_LOC && rm -rf $PROG_EXT_LOC" "unsquashfs -d $REF_EXT_LOC $ARCHIVE" "zig-out/bin/unsquashfs -d $PROG_EXT_LOC $ARCHIVE"
echo ""
echo "Testing Single-threaded Performance"
echo ""
hyperfine --warmup 5 --prepare "rm -rf $REF_EXT_LOC && rm -rf $PROG_EXT_LOC" "unsquashfs -p 1 -d $REF_EXT_LOC $ARCHIVE" "zig-out/bin/unsquashfs -p 1 -d $PROG_EXT_LOC $ARCHIVE"
+96 -8
View File
@@ -1,17 +1,23 @@
const std = @import("std");
pub fn build(b: *std.Build) !void {
// const use_zig_decomp = b.option(bool, "use_zig_decomp", "Use zig standard library for decompression.") orelse false;
// const allow_lzo = b.option(bool, "allow_lzo", "Compile with lzo support") orelse false;
const debug = b.option(bool, "debug", "Enable options to make debugging easier.");
const use_zig_decomp = b.option(bool, "use_zig_decomp", "Use zig standard library for decompression.") orelse false;
const allow_lzo = b.option(bool, "allow_lzo", "Compile with lzo support") orelse false;
const dynamic = b.option(bool, "dynamic", "Dynamicly link C decompression libraries") orelse false;
var debug = b.option(bool, "debug", "Enable options to make debugging easier.");
const version_string_option = b.option([]const u8, "version", "Version of the library/binary");
// const zig_squashfs_options = b.addOptions();
// zig_squashfs_options.addOption(bool, "use_zig_decomp", use_zig_decomp);
// zig_squashfs_options.addOption(bool, "allow_lzo", allow_lzo);
const zig_squashfs_options = b.addOptions();
zig_squashfs_options.addOption(bool, "use_zig_decomp", use_zig_decomp);
zig_squashfs_options.addOption(bool, "allow_lzo", allow_lzo);
const target = b.standardTargetOptions(.{});
const optimize = b.standardOptimizeOption(.{});
var optimize = b.standardOptimizeOption(.{});
if (debug == true)
optimize = .Debug;
if (optimize == .Debug)
debug = true;
const lib = b.addLibrary(.{
.name = "squashfs",
@@ -20,10 +26,31 @@ pub fn build(b: *std.Build) !void {
.target = target,
.valgrind = debug,
.root_source_file = b.path("src/root.zig"),
.imports = &.{
.{ .name = "options", .module = zig_squashfs_options.createModule() },
},
}),
.use_llvm = debug,
});
const deps = try dependencies(b, optimize, target, use_zig_decomp, allow_lzo, dynamic);
defer b.allocator.free(deps);
for (deps) |d|
lib.root_module.linkLibrary(d);
if (!use_zig_decomp) {
const c = b.addTranslateC(.{
.optimize = optimize,
.target = target,
.root_source_file = b.path("src/c.h"),
});
if (allow_lzo) c.defineCMacro("ALLOW_LZO", null);
lib.root_module.addImport("c", c.createModule());
if (dynamic)
dynamicLinkLibraries(c, allow_lzo);
}
var version = version_string_option orelse "0.0.0-testing";
if (version[0] == 'v') version = version[1..];
const unsquashfs_options = b.addOptions();
@@ -54,9 +81,31 @@ pub fn build(b: *std.Build) !void {
.root_module = b.createModule(.{
.optimize = optimize,
.target = target,
.root_source_file = b.path("src/test.zig"),
.root_source_file = b.path("src/root.zig"),
.imports = &.{
.{ .name = "options", .module = zig_squashfs_options.createModule() },
},
.valgrind = debug,
}),
.use_llvm = debug,
});
for (deps) |d|
mod_tests.root_module.linkLibrary(d);
if (!use_zig_decomp) {
const c = b.addTranslateC(.{
.optimize = optimize,
.target = target,
.root_source_file = b.path("src/c.h"),
});
mod_tests.root_module.addImport("c", c.createModule());
if (allow_lzo) c.defineCMacro("ALLOW_LZO", null);
if (dynamic)
dynamicLinkLibraries(c, allow_lzo);
}
const run_mod_tests = b.addRunArtifact(mod_tests);
const test_step = b.step("test", "Run tests");
test_step.dependOn(&run_mod_tests.step);
@@ -74,3 +123,42 @@ pub fn build(b: *std.Build) !void {
check.dependOn(&lib_check.step);
check.dependOn(&exe_check.step);
}
pub fn dynamicLinkLibraries(mod: *std.Build.Step.TranslateC, allow_lzo: bool) void {
mod.linkSystemLibrary("zstd", .{});
mod.linkSystemLibrary("zlib-ng", .{});
mod.linkSystemLibrary("lzma", .{});
mod.linkSystemLibrary("lz4", .{});
if (allow_lzo)
mod.linkSystemLibrary("minilzo", .{});
}
fn dependencies(
b: *std.Build,
optimize: std.builtin.OptimizeMode,
target: std.Build.ResolvedTarget,
use_zig_decomp: bool,
allow_lzo: bool,
dynamic: bool,
) ![]*std.Build.Step.Compile {
if (use_zig_decomp or dynamic) return &.{};
var list: std.ArrayList(*std.Build.Step.Compile) = .empty;
const zstd = b.dependency("zstd", .{ .optimize = optimize, .target = target });
try list.append(b.allocator, zstd.artifact("zstd"));
const zng = b.dependency("zlib_ng", .{ .optimize = optimize, .target = target });
try list.append(b.allocator, zng.artifact("zng"));
const xz = b.dependency("xz", .{ .optimize = optimize, .target = target });
try list.append(b.allocator, xz.artifact("lzma"));
const lz4 = b.dependency("lz4", .{ .optimize = optimize, .target = target });
try list.append(b.allocator, lz4.artifact("lz4"));
if (allow_lzo) {
const minilzo = b.dependency("minilzo", .{ .optimize = optimize, .target = target });
try list.append(b.allocator, minilzo.artifact("minilzo"));
}
return list.toOwnedSlice(b.allocator);
}
+4
View File
@@ -20,6 +20,10 @@
.url = "git+https://github.com/CalebQ42/zig-minilzo.git#7cbae997b91a44d74b7cd6c073584dc9562a6c90",
.hash = "minilzo-2.10.0-Ij7BO8wLAADeWI4Pe4jp8XTDsDaquZR14oZ7_9yKKDWP",
},
.xz = .{
.url = "git+https://github.com/akunaakwei/zig-xz.git#e2d389262c8291907e3e4c6fb119819141c16c0f",
.hash = "xz-5.8.2-6v47_JYeAABSL-jonprpL5-E_YaaGc4B5xrbe93WsJ3G",
},
},
.paths = .{
"build.zig",
-10
View File
@@ -1,10 +0,0 @@
#!/bin/sh
zig test \
-lc \
-lz \
-llzma \
-lminilzo \
-llz4 \
-lzstd \
src/test.zig
+147 -37
View File
@@ -1,6 +1,7 @@
const std = @import("std");
const Io = std.Io;
const Decomp = @import("decomp.zig");
const ExtractionOptions = @import("options.zig");
const File = @import("file.zig");
const Inode = @import("inode.zig");
@@ -22,60 +23,44 @@ pub fn init(io: Io, file: std.Io.File, offset: u64) !Archive {
try rdr.seekTo(offset);
var super: Superblock = undefined;
try rdr.interface.readSliceEndian(Superblock, @ptrCast(&super), .little);
try super.validate();
return .{
.file = .init(file, offset),
.file = try .init(io, file, super.size, offset),
.super = super,
.stateless_decomp = switch (super.compression) {
.gzip => @import("decomp/zlib.zig").stateless_decompressor,
.lzma => @import("decomp/lzma.zig").stateless_decompressor,
.lzo => return error.LzoUnsupported,
.xz => @import("decomp/xz.zig").stateless_decompressor,
.lz4 => return error.Lz4Unsupported,
.zstd => @import("decomp/zstd.zig").stateless_decompressor,
},
.stateless_decomp = try Decomp.StatelessDecomp(super.compression),
};
}
pub fn deinit(self: *Archive, io: Io) void {
self.file.deinit(io);
}
/// The root folder of the Archive. Used to open other Files.
pub fn root(self: Archive, alloc: std.mem.Allocator, io: Io) !File {
pub fn root(self: *Archive, alloc: std.mem.Allocator) !File {
const root_inode = try Utils.inodeFromRef(
alloc,
io,
self.file,
&self.stateless_decomp,
self.super.inode_start,
self.super.block_size,
self.super.root_ref,
);
return .init(alloc, self, root_inode, "");
return .init(alloc, self.*, root_inode, "");
}
/// Opens a File within the archive.
pub fn open(self: Archive, alloc: std.mem.Allocator, io: Io, filepath: []const u8) !File {
const root_file = try self.root(alloc, io);
pub fn open(self: *Archive, alloc: std.mem.Allocator, io: Io, filepath: []const u8) !File {
var root_file = try self.root(alloc);
const path = std.mem.trim(u8, filepath, "/");
if (Utils.pathIsSelf(path))
return root_file;
defer root_file.deinit();
return root_file.open(alloc, io, filepath);
}
/// Extract the entire archive contents to the given directory.
pub fn extract(self: Archive, alloc: std.mem.Allocator, io: Io, extract_dir: []const u8, options: ExtractionOptions) !void {
const root_inode = try Utils.inodeFromRef(
alloc,
io,
self.file,
&self.stateless_decomp,
self.super.inode_start,
self.super.block_size,
self.super.root_ref,
);
return root_inode.extract(alloc, io, self.file, self.super, extract_dir, options);
}
/// Returns the inode with the given inode number.
/// Requires that the archive is exportable (has an export lookup table).
pub fn inode(self: Archive, alloc: std.mem.Allocator, io: Io, num: u32) !Inode {
pub fn inode(self: *Archive, alloc: std.mem.Allocator, io: Io, num: u32) !Inode {
if (!self.super.flags.exportable)
return error.NotExportable;
const ref = try LookupTable.lookupValue(
@@ -98,7 +83,7 @@ pub fn inode(self: Archive, alloc: std.mem.Allocator, io: Io, num: u32) !Inode {
);
}
/// Returns a value at the given index from the Archive's id (uid/gid) table.
pub fn idTable(self: Archive, alloc: std.mem.Allocator, io: Io, idx: u32) !u16 {
pub fn idTable(self: *Archive, alloc: std.mem.Allocator, io: Io, idx: u32) !u16 {
return LookupTable.lookupValue(
u16,
alloc,
@@ -110,6 +95,19 @@ pub fn idTable(self: Archive, alloc: std.mem.Allocator, io: Io, idx: u32) !u16 {
);
}
/// Extract the entire archive contents to the given directory.
pub fn extract(self: *Archive, alloc: std.mem.Allocator, io: Io, extract_dir: []const u8, options: ExtractionOptions) !void {
const root_inode = try Utils.inodeFromRef(
alloc,
self.file,
&self.stateless_decomp,
self.super.inode_start,
self.super.block_size,
self.super.root_ref,
);
return root_inode.extract(alloc, io, self.file, self.super, extract_dir, options);
}
// Superblock
const SQUASHFS_MAGIC: u32 = std.mem.readInt(u32, "hsqs", .little);
@@ -128,14 +126,7 @@ pub const Superblock = extern struct {
mod_time: u32,
block_size: u32,
frag_count: u32,
compression: enum(u16) {
gzip = 1,
lzma,
lzo,
xz,
lz4,
zstd,
},
compression: Decomp.Enum,
block_log: u16,
flags: packed struct(u16) {
inode_uncompressed: bool,
@@ -176,3 +167,122 @@ pub const Superblock = extern struct {
return SuperblockError.InvalidBlockLog;
}
};
// Tests
const TestArchive = "testing/LinuxPATest.sfs";
test "Basics" {
std.debug.print("Starting test: Basics...\n", .{});
const alloc = std.testing.allocator;
const io = std.testing.io;
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
defer sfs.deinit(io);
try std.testing.expectEqualDeep(sfs.super, LinuxPATestCorrectSuperblock);
const root_file = try sfs.root(alloc);
defer root_file.deinit();
}
const TestFile = "Start.exe";
const TestFileExtractLocation = "testing/Start.exe";
test "ExtractSingleFile" {
std.debug.print("Starting test: ExtractSingleFile...\n", .{});
const alloc = std.testing.allocator;
const io = std.testing.io;
Io.Dir.cwd().deleteFile(io, TestFileExtractLocation) catch {};
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
defer sfs.deinit(io);
var test_fil = try sfs.open(alloc, io, TestFile);
defer test_fil.deinit();
try test_fil.extract(alloc, io, TestFileExtractLocation, .default);
//TODO: validate extracted file.
}
const TestFullExtractLocation = "testing/TestExtract";
test "ExtractCompleteArchive" {
std.debug.print("Starting test: ExtractCompleteArchive...\n", .{});
const alloc = std.testing.allocator;
const io = std.testing.io;
Io.Dir.cwd().deleteTree(io, TestFullExtractLocation) catch {};
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
defer sfs.deinit(io);
try sfs.extract(alloc, io, TestFullExtractLocation, .default);
}
test "ExtractCompleteArchiveSingleThreaded" {
std.debug.print("Starting test: ExtractCompleteArchive...\n", .{});
const alloc = std.testing.allocator;
const io = std.testing.io;
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
{
std.debug.print("First testing using Threaded.global_single_threaded...\n", .{});
Io.Dir.cwd().deleteTree(io, TestFullExtractLocation) catch {};
var sfs: Archive = try .init(Io.Threaded.global_single_threaded.io(), fil, 0);
defer sfs.deinit(Io.Threaded.global_single_threaded.io());
try sfs.extract(alloc, Io.Threaded.global_single_threaded.io(), TestFullExtractLocation, .default);
}
{
std.debug.print("Next testing using ExtractionOptions.single_threaded...\n", .{});
Io.Dir.cwd().deleteTree(io, TestFullExtractLocation) catch {};
var sfs: Archive = try .init(io, fil, 0);
defer sfs.deinit(io);
try sfs.extract(alloc, io, TestFullExtractLocation, .default_single_threaded);
}
}
const LinuxPATestCorrectSuperblock: Superblock = .{
.magic = std.mem.readInt(u32, "hsqs", .little),
.inode_count = 2974,
.mod_time = 1632696724,
.block_size = 131072,
.frag_count = 264,
.compression = .zstd,
.block_log = 17,
.flags = .{
.inode_uncompressed = false,
.data_uncompressed = false,
.check = false,
.frag_uncompressed = false,
.fragment_never = false,
.fragment_always = false,
.duplicates = true,
.exportable = true,
.xattr_uncompressed = false,
.xattr_never = false,
.compression_options = false,
.ids_uncompressed = false,
._ = 0,
},
.id_count = 1,
.ver_maj = 4,
.ver_min = 0,
.root_ref = .{
.block_offset = 1363,
.block_start = 29237,
._ = 0,
},
.size = 106841744,
.id_start = 106841632,
.xattr_start = 106841720,
.inode_start = 106778274,
.dir_start = 106807998,
.frag_start = 106837747,
.export_start = 106841602,
};
+15 -3
View File
@@ -41,7 +41,10 @@ var force: bool = false;
pub fn main(init: std.process.Init) !void {
const alloc = init.gpa;
const io = init.io;
// const io = init.io;
var evented: Io.Evented = undefined;
try evented.init(alloc, .{});
const io = evented.io();
var stdout = std.Io.File.stdout();
defer stdout.close(io);
@@ -60,7 +63,7 @@ pub fn main(init: std.process.Init) !void {
var arc: squashfs.Archive = try .init(io, fil, offset); //TODO: Handle error gracefully.
const options: squashfs.ExtractionOptions = .{
.threads = if (threads == 0) try std.Thread.getCpuCount() else threads,
.single_threaded = threads == 1,
.verbose = verbose,
.verbose_writer = if (verbose) &out.interface else null,
.ignore_xattr = ignore_xattrs,
@@ -69,7 +72,16 @@ pub fn main(init: std.process.Init) !void {
if (force)
try Io.Dir.cwd().deleteTree(io, extLoc);
try arc.extract(alloc, io, extLoc, options); //TODO: Handle error gracefully.
if (threads != 0) {
var limited_io = Io.Threaded.init(alloc, .{
.async_limit = .limited(threads - 1),
.concurrent_limit = .limited(threads - 1),
.argv0 = .init(init.minimal.args),
.environ = init.minimal.environ,
});
return arc.extract(alloc, limited_io.io(), extLoc, options); //TODO: Handle error gracefully.
}
return arc.extract(alloc, io, extLoc, options); //TODO: Handle error gracefully.
}
fn handleArgs(args: std.process.Args, out: *Writer) !void {
+7
View File
@@ -0,0 +1,7 @@
#include <zstd.h>
#include <zlib-ng.h>
#include <lzma.h>
#ifdef ALLOW_LZO
#include <lzo/minilzo.h>
#endif
#include <lz4.h>
+65 -10
View File
@@ -1,22 +1,76 @@
const std = @import("std");
const Io = std.Io;
const options = @import("options");
const Decompressor = @import("util/decompressor.zig");
pub const Decomp = union(enum) {
gzip: @import("decomp/zlib.zig"),
lzma: @import("decomp/lzma.zig"),
lzo: void,
xz: @import("decomp/xz.zig"),
lz4: void,
zstd: @import("decomp/zstd.zig"),
const zlib = if (options.use_zig_decomp) @import("decomp/zig_zlib.zig") else @import("decomp/c_zlib.zig");
const lzma = if (options.use_zig_decomp) @import("decomp/zig_lzma.zig") else @import("decomp/c_lzma.zig");
const lzo = if (options.use_zig_decomp or !options.allow_lzo) void else @import("decomp/c_lzo.zig");
const xz = if (options.use_zig_decomp) @import("decomp/zig_xz.zig") else @import("decomp/c_xz.zig");
const lz4 = if (options.use_zig_decomp) void else @import("decomp/c_lz4.zig");
const zstd = if (options.use_zig_decomp) @import("decomp/zig_zstd.zig") else @import("decomp/c_zstd.zig");
pub fn deinit(self: *Decomp) void {
pub const Enum = enum(u16) {
gzip = 1,
lzma,
lzo,
xz,
lz4,
zstd,
};
pub fn StatelessDecomp(val: Enum) !Decompressor {
return switch (val) {
.gzip => zlib.stateless_decompressor,
.lzma => lzma.stateless_decompressor,
.lzo => if (options.use_zig_decomp or !options.allow_lzo)
error.LzoUnsupported
else
lzo.stateless_decompressor,
.xz => xz.stateless_decompressor,
.lz4 => if (options.use_zig_decomp)
error.Lz4Unsupported
else
lz4.stateless_decompressor,
.zstd => zstd.stateless_decompressor,
};
}
pub const Decomp = union(enum) {
gzip: zlib,
lzma: lzma,
lzo: lzo,
xz: xz,
lz4: lz4,
zstd: zstd,
pub fn init(val: Enum, alloc: std.mem.Allocator, io: Io, block_size: u32) !Decomp {
return switch (val) {
.gzip => .{ .gzip = if (options.use_zig_decomp) try zlib.init(alloc, io, block_size) else try zlib.init(alloc, io) },
.lzma => .{ .lzma = if (options.use_zig_decomp) try lzma.init(alloc, io, block_size) else .{} },
.lzo => if (options.use_zig_decomp or !options.allow_lzo) error.LzoUnsupported else .{ .lzo = .{} },
.xz => .{ .xz = if (options.use_zig_decomp) try xz.init(alloc, io, block_size) else .{} },
.lz4 => if (options.use_zig_decomp) error.Lz4Unsupported else .{ .lz4 = .{} },
.zstd => .{ .zstd = if (options.use_zig_decomp) try zstd.init(alloc, io, block_size) else try zstd.init(alloc, io) },
};
}
pub fn deinit(self: *Decomp, alloc: std.mem.Allocator) void {
if (options.use_zig_decomp) {
switch (self.*) {
.gzip => self.gzip.deinit(),
.lzma => self.lzma.deinit(),
.xz => self.xz.deinit(),
.zstd => self.zstd.deinit(),
else => unreachable,
else => {},
}
} else {
switch (self.*) {
.gzip => self.gzip.deinit(alloc),
.zstd => self.zstd.deinit(alloc),
else => {},
}
}
}
@@ -24,9 +78,10 @@ pub const Decomp = union(enum) {
return switch (self.*) {
.gzip => &self.gzip.interface,
.lzma => &self.lzma.interface,
.lzo => if (options.use_zig_decomp or !options.allow_lzo) unreachable else &self.lzo.interface,
.xz => &self.xz.interface,
.lz4 => if (options.use_zig_decomp) unreachable else &self.lz4.interface,
.zstd => &self.zstd.interface,
else => unreachable,
};
}
};
+17
View File
@@ -0,0 +1,17 @@
const std = @import("std");
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
interface: Decompressor = .{ .decomp_fn = statelessDecomp },
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
const out_len: c_int = @bitCast(@as(u32, @truncate(out.len)));
const res = c.LZ4_decompress_fast(in.ptr, out.ptr, out_len);
if (res < 0) return Error.ReadFailed;
return @abs(res);
}
+46
View File
@@ -0,0 +1,46 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue(c.lzma_stream);
const Self = @This();
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
interface: Decompressor = .{ .decomp_fn = statelessDecomp },
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var stream: c.lzma_stream = .{
.next_in = in.ptr,
.avail_in = in.len,
.next_out = out.ptr,
.avail_out = out.len,
};
var res = c.lzma_alone_decoder(&stream, stream.avail_out * 2);
if (res != c.LZMA_OK) return Error.ReadFailed;
while (res == c.LZMA_OK)
res = c.lzma_code(&stream, c.LZMA_RUN);
if (res != c.LZMA_FINISH) return Error.ReadFailed;
return stream.total_out;
}
// lzma_allocator
// fn lzmaAlloc(ptr: ?*anyopaque, size: usize, _: usize) callconv(.c) ?*anyopaque {
// var alloc: *std.mem.Allocator = @ptrCast(@alignCast(ptr));
// return alloc.rawAlloc(size, .@"1", 0);
// }
// fn lzmaFree(ptr: ?*anyopaque, mem_ptr: ?*anyopaque) callconv(.c) void {
// if (mem_ptr == null) return;
// var alloc: *std.mem.Allocator = @ptrCast(@alignCast(ptr));
// alloc.free(@as([*]u8, @ptrCast(mem_ptr.?)));
// }
+26
View File
@@ -0,0 +1,26 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue(c.lzma_stream);
const Self = @This();
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
interface: Decompressor = .{ .decomp_fn = statelessDecomp },
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
_ = c.lzo_init();
var out_len = out.len;
const res = c.lzo1x_decompress_safe(in.ptr, in.len, out.ptr, &out_len, null);
if (res != c.LZO_E_OK) return Error.ReadFailed;
return out_len;
}
+47
View File
@@ -0,0 +1,47 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue(c.lzma_stream);
const Self = @This();
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
interface: Decompressor = .{ .decomp_fn = statelessDecomp },
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var stream: c.lzma_stream = .{
.next_in = in.ptr,
.avail_in = in.len,
.next_out = out.ptr,
.avail_out = out.len,
};
var res = c.lzma_alone_decoder(&stream, stream.avail_out * 2);
if (res != c.LZMA_OK) return Error.ReadFailed;
while (res == c.LZMA_OK)
res = c.lzma_code(&stream, c.LZMA_RUN);
if (res != c.LZMA_FINISH) return Error.ReadFailed;
return stream.total_out;
}
// lzma_allocator
// fn lzmaAlloc(ptr: ?*anyopaque, size: usize, _: usize) callconv(.c) ?*anyopaque {
// var alloc: *std.mem.Allocator = @ptrCast(@alignCast(ptr));
// const mem = alloc.alloc(u8, size) catch return null;
// return mem.ptr;
// }
// fn lzmaFree(ptr: ?*anyopaque, mem_ptr: ?*anyopaque) callconv(.c) void {
// if (mem_ptr == null) return;
// var alloc: *std.mem.Allocator = @ptrCast(@alignCast(ptr));
// alloc.free(@as([*]u8, @ptrCast(mem_ptr.?)));
// }
+80
View File
@@ -0,0 +1,80 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue(c.zng_stream);
const Self = @This();
interface: Decompressor = .{ .decomp_fn = decomp },
io: Io,
ctx: []c.zng_stream,
ctx_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io) !Self {
const buf = try alloc.alloc(c.zng_stream, 20); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (0..20) |_|
try queue.putOne(io, .{});
return .{
.io = io,
.ctx = buf,
.ctx_queue = queue,
};
}
pub fn deinit(self: *Self, alloc: std.mem.Allocator) void {
self.ctx_queue.close(self.io);
alloc.free(self.ctx);
}
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
return statelessDecomp(d, alloc, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
var stream = self.ctx_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.ctx_queue.putOne(self.io, stream) catch {};
stream.next_in = in.ptr;
stream.avail_in = @truncate(in.len);
stream.next_out = out.ptr;
stream.avail_out = @truncate(out.len);
try zlibDecomp(&stream);
return stream.total_out;
}
inline fn zlibDecomp(stream: *c.zng_stream) !void {
_ = c.zng_inflateReset(stream);
const res = c.zng_inflate(stream, c.Z_FULL_FLUSH);
if (res != c.Z_OK) return Error.ReadFailed;
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var stream: c.zng_stream = .{
.next_in = in.ptr,
.avail_in = @truncate(in.len),
.next_out = out.ptr,
.avail_out = @truncate(out.len),
};
try zlibDecomp(&stream);
return stream.total_out;
}
+69
View File
@@ -0,0 +1,69 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const c = @import("c");
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue(?*c.ZSTD_DCtx);
const Self = @This();
interface: Decompressor = .{ .decomp_fn = decomp },
io: Io,
ctx: []?*c.ZSTD_DCtx,
ctx_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io) !Self {
const buf = try alloc.alloc(?*c.ZSTD_DCtx, 20); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (0..20) |_|
try queue.putOne(io, c.ZSTD_createDCtx());
return .{
.io = io,
.ctx = buf,
.ctx_queue = queue,
};
}
pub fn deinit(self: *Self, alloc: std.mem.Allocator) void {
self.ctx_queue.close(self.io);
for (self.ctx) |ctx|
_ = c.ZSTD_freeDCtx(ctx);
alloc.free(self.ctx);
}
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
return statelessDecomp(d, alloc, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const ctx = self.ctx_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.ctx_queue.putOne(self.io, ctx) catch {};
_ = c.ZSTD_DCtx_reset(ctx, c.ZSTD_reset_session_only);
const res = c.ZSTD_decompressDCtx(ctx, out.ptr, out.len, in.ptr, in.len);
if (c.ZSTD_isError(res) != 0)
return Error.ReadFailed;
return res;
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*Decompressor, _: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
const res = c.ZSTD_decompress(out.ptr, out.len, in.ptr, in.len);
if (c.ZSTD_isError(res) != 0)
return Error.ReadFailed;
return res;
}
-86
View File
@@ -1,86 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const lzma = std.compress.lzma;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
block_size: u32,
buffers: std.ArrayList(Buffer),
buffer_queue: std.SinglyLinkedList = .{},
pub fn init(alloc: std.mem.Allocator, block_size: u32) !Self {
return .{
.alloc = alloc,
.block_size = block_size,
.buffers = try .initCapacity(alloc, 5),
};
}
pub fn deinit(self: *Self) void {
for (self.buffers.items) |buf|
self.alloc.free(buf.buf);
self.buffers.deinit(self.alloc);
}
fn decomp(d: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
var buf = try alloc.alloc(u8, in.len * 2);
defer alloc.free(buf);
return lzmaDecomp(alloc, &buf, in, out) catch |err| return switch (err) {
error.OutOfMemory => Error.OutOfMemory,
else => Error.ReadFailed,
};
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const buf_node = self.buffer_queue.popFirst();
var buf: *Buffer = undefined;
if (buf_node == null) {
const new_buf = try self.buffers.addOne(self.alloc);
new_buf.* = .{ .node = .{}, .buf = try self.alloc.alloc(u8, self.block_size) };
buf = new_buf;
} else {
buf = @fieldParentPtr("node", buf_node.?);
}
defer self.buffer_queue.prepend(&buf.node);
return lzmaDecomp(self.alloc, &buf.buf, in, out) catch |err| {
// self.err = err;
return switch (err) {
error.OutOfMemory => Error.OutOfMemory,
else => Error.ReadFailed,
};
};
}
inline fn lzmaDecomp(alloc: std.mem.Allocator, buffer: *[]u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = try lzma.Decompress.initOptions(&rdr, alloc, buffer.*, .{ .allow_incomplete = true }, 3 * 1024 * 1024);
defer {
buffer.* = d.takeBuffer();
d.deinit();
}
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var buf = try alloc.alloc(u8, in.len);
defer alloc.free(buf);
return lzmaDecomp(alloc, &buf, in, out) catch return Error.ReadFailed;
}
-80
View File
@@ -1,80 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const xz = std.compress.xz;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
block_size: u32,
buffers: std.ArrayList(Buffer),
buffer_queue: std.SinglyLinkedList = .{},
pub fn init(alloc: std.mem.Allocator, block_size: u32) !Self {
return .{
.alloc = alloc,
.block_size = block_size,
.buffers = try .initCapacity(alloc, 5),
};
}
pub fn deinit(self: *Self) void {
for (self.buffers.items) |buf|
self.alloc.free(buf.buf);
self.buffers.deinit(self.alloc);
}
fn decomp(d: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
var buf = try alloc.alloc(u8, in.len * 2);
defer alloc.free(buf);
return xzDecomp(alloc, &buf, in, out) catch return Error.ReadFailed;
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const buf_node = self.buffer_queue.popFirst();
var buf: *Buffer = undefined;
if (buf_node == null) {
const new_buf = try self.buffers.addOne(self.alloc);
new_buf.* = .{ .node = .{}, .buf = try self.alloc.alloc(u8, self.block_size) };
buf = new_buf;
} else {
buf = @fieldParentPtr("node", buf_node.?);
}
defer self.buffer_queue.prepend(&buf.node);
return xzDecomp(self.alloc, &buf.buf, in, out) catch {
// self.err = err;
return Error.ReadFailed;
};
}
inline fn xzDecomp(alloc: std.mem.Allocator, buffer: *[]u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = try xz.Decompress.init(&rdr, alloc, buffer.*);
defer {
buffer.* = d.takeBuffer();
d.deinit();
}
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var buf = try alloc.alloc(u8, in.len);
defer alloc.free(buf);
return xzDecomp(alloc, &buf, in, out) catch return Error.ReadFailed;
}
+81
View File
@@ -0,0 +1,81 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const lzma = std.compress.lzma;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = Io.Queue([]u8);
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
io: Io,
block_size: u32,
buf: [][]u8,
buf_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io, block_size: u32) !Self {
const buf = try alloc.alloc([]u8, 20); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (0..20) |_|
try queue.putOne(io, try alloc.alloc(u8, block_size));
return .{
.alloc = alloc,
.io = io,
.block_size = block_size,
.buf = buf,
.buf_queue = queue,
};
}
pub fn deinit(self: *Self) void {
self.buf_queue.close(self.io);
for (self.buf) |buf|
self.alloc.free(buf);
self.alloc.free(self.buf);
}
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
return statelessDecomp(d, alloc, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
var buf = self.buf_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.buf_queue.putOne(self.io, buf) catch {};
return lzmaDecomp(self.alloc, &buf, in, out) catch return Error.ReadFailed;
}
inline fn lzmaDecomp(alloc: std.mem.Allocator, buffer: *[]u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = try lzma.Decompress.initOptions(&rdr, alloc, buffer.*, .{}, in.len * 2);
defer {
buffer.* = d.takeBuffer();
d.deinit();
}
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var buf = try alloc.alloc(u8, in.len);
defer alloc.free(buf);
return lzmaDecomp(alloc, &buf, in, out) catch return Error.ReadFailed;
}
+81
View File
@@ -0,0 +1,81 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const xz = std.compress.xz;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = Io.Queue([]u8);
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
io: Io,
block_size: u32,
buf: [][]u8,
buf_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io, block_size: u32) !Self {
const buf = try alloc.alloc([]u8, 20); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (0..20) |_|
try queue.putOne(io, try alloc.alloc(u8, block_size));
return .{
.alloc = alloc,
.io = io,
.block_size = block_size,
.buf = buf,
.buf_queue = queue,
};
}
pub fn deinit(self: *Self) void {
self.buf_queue.close(self.io);
for (self.buf) |buf|
self.alloc.free(buf);
self.alloc.free(self.buf);
}
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
return statelessDecomp(d, alloc, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
var buf = self.buf_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.buf_queue.putOne(self.io, buf) catch {};
return xzDecomp(self.alloc, &buf, in, out) catch return Error.ReadFailed;
}
inline fn xzDecomp(alloc: std.mem.Allocator, buffer: *[]u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = try xz.Decompress.init(&rdr, alloc, buffer.*);
defer {
buffer.* = d.takeBuffer();
d.deinit();
}
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
var buf = try alloc.alloc(u8, in.len);
defer alloc.free(buf);
return xzDecomp(alloc, &buf, in, out) catch return Error.ReadFailed;
}
+77
View File
@@ -0,0 +1,77 @@
const std = @import("std");
const Io = std.Io;
const flate = std.compress.flate;
const Node = std.SinglyLinkedList.Node;
const Reader = Io.Reader;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = Io.Queue([]u8);
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
io: Io,
block_size: u32,
buf: [][]u8,
buf_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io, block_size: u32) !Self {
const buf = try alloc.alloc([]u8, 20); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (0..20) |_|
try queue.putOne(io, try alloc.alloc(u8, block_size));
return .{
.alloc = alloc,
.io = io,
.block_size = block_size,
.buf = buf,
.buf_queue = queue,
};
}
pub fn deinit(self: *Self) void {
self.buf_queue.close(self.io);
for (self.buf) |buf|
self.alloc.free(buf);
self.alloc.free(self.buf);
}
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
return statelessDecomp(d, alloc, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const buf = self.buf_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.buf_queue.putOne(self.io, buf) catch {};
return zlibDecomp(buf, in, out);
}
inline fn zlibDecomp(buffer: []u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = flate.Decompress.init(&rdr, .zlib, buffer);
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
const buf = try alloc.alloc(u8, out.len);
defer alloc.free(buf);
return zlibDecomp(buf, in, out);
}
+26 -26
View File
@@ -1,58 +1,58 @@
const std = @import("std");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Self = @This();
const Queue = std.Io.Queue([]u8);
const Buffer = struct {
node: Node,
buf: []u8,
};
const Self = @This();
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
io: Io,
block_size: u32,
buffers: std.ArrayList(Buffer),
buffer_queue: std.SinglyLinkedList = .{},
buf: [][]u8,
buf_queue: Queue,
pub fn init(alloc: std.mem.Allocator, io: Io, block_size: u32) !Self {
const buf = try alloc.alloc([]u8, 5); // TODO: Choose a better number instead of a random one.
var queue: Queue = .init(buf);
for (buf) |_|
try queue.putOne(io, try alloc.alloc(u8, block_size + zstd.block_size_max));
pub fn init(alloc: std.mem.Allocator, block_size: u32) !Self {
return .{
.alloc = alloc,
.io = io,
.block_size = block_size,
.buffers = try .initCapacity(alloc, 5),
.buf = buf,
.buf_queue = queue,
};
}
pub fn deinit(self: *Self) void {
for (self.buffers.items) |buf|
self.alloc.free(buf.buf);
self.buffers.deinit(self.alloc);
self.buf_queue.close(self.io);
for (self.buf) |buf|
self.alloc.free(buf);
self.alloc.free(self.buf);
}
fn decomp(d: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
fn decomp(d: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
const buf = try alloc.alloc(u8, in.len * 2);
defer alloc.free(buf);
return zstdDecomp(buf, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const buf_node = self.buffer_queue.popFirst();
var buf: *Buffer = undefined;
if (buf_node == null) {
const new_buf = try self.buffers.addOne(self.alloc);
new_buf.* = .{ .node = .{}, .buf = try self.alloc.alloc(u8, self.block_size + zstd.block_size_max) };
buf = new_buf;
} else {
buf = @fieldParentPtr("node", buf_node.?);
}
defer self.buffer_queue.prepend(&buf.node);
return zstdDecomp(buf.buf, in, out);
const buf = self.buf_queue.getOne(self.io) catch return Error.ReadFailed;
defer self.buf_queue.putOne(self.io, buf) catch {};
return zstdDecomp(buf, in, out);
}
inline fn zstdDecomp(buffer: []u8, in: []u8, out: []u8) !usize {
@@ -66,7 +66,7 @@ inline fn zstdDecomp(buffer: []u8, in: []u8, out: []u8) !usize {
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
fn statelessDecomp(_: ?*Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
const buf = try alloc.alloc(u8, out.len + zstd.block_size_max);
defer alloc.free(buf);
return zstdDecomp(buf, in, out);
-73
View File
@@ -1,73 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const flate = std.compress.flate;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Self = @This();
const Buffer = struct {
node: Node,
buf: []u8,
};
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
block_size: u32,
buffers: std.ArrayList(Buffer),
buffer_queue: std.SinglyLinkedList = .{},
pub fn init(alloc: std.mem.Allocator, block_size: u32) !Self {
return .{
.alloc = alloc,
.block_size = block_size,
.buffers = try .initCapacity(alloc, 5),
};
}
pub fn deinit(self: *Self) void {
for (self.buffers.items) |buf|
self.alloc.free(buf.buf);
self.buffers.deinit(self.alloc);
}
fn decomp(d: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
if (d == null) {
const buf = try alloc.alloc(u8, in.len * 2);
defer alloc.free(buf);
return zlibDecomp(buf, in, out);
}
var self: *Self = @fieldParentPtr("interface", @constCast(d.?));
const buf_node = self.buffer_queue.popFirst();
var buf: *Buffer = undefined;
if (buf_node == null) {
const new_buf = try self.buffers.addOne(self.alloc);
new_buf.* = .{ .node = .{}, .buf = try self.alloc.alloc(u8, self.block_size) };
buf = new_buf;
} else {
buf = @fieldParentPtr("node", buf_node.?);
}
defer self.buffer_queue.prepend(&buf.node);
return zlibDecomp(buf.buf, in, out);
}
inline fn zlibDecomp(buffer: []u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = flate.Decompress.init(&rdr, .zlib, buffer);
return d.reader.readSliceShort(out);
}
// Stateless
pub const stateless_decompressor: Decompressor = .{ .decomp_fn = statelessDecomp };
fn statelessDecomp(_: ?*const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
const buf = try alloc.alloc(u8, in.len * 2);
defer alloc.free(buf);
return zlibDecomp(buf, in, out);
}
+3 -1
View File
@@ -3,6 +3,8 @@ const Reader = std.Io.Reader;
const Inode = @import("inode.zig");
pub const Error = error{OutOfMemory} || Reader.Error;
const DirEntry = @This();
block_start: u32,
@@ -14,7 +16,7 @@ pub fn deinit(self: DirEntry, alloc: std.mem.Allocator) void {
alloc.free(self.name);
}
pub fn readDirectory(alloc: std.mem.Allocator, rdr: *Reader, size: u32) ![]DirEntry {
pub fn readDirectory(alloc: std.mem.Allocator, rdr: *Reader, size: u32) Error![]DirEntry {
var hdr: Header = undefined;
var raw: RawEntry = undefined;
var out: std.ArrayList(DirEntry) = try .initCapacity(alloc, 30);
+6 -8
View File
@@ -10,7 +10,6 @@ const Inode = @import("inode.zig");
const DataExtractor = @import("util/data_extractor.zig");
const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
const SharedCache = @import("util/shared_cache.zig");
const File = @This();
@@ -35,24 +34,23 @@ pub fn init(alloc: std.mem.Allocator, archive: Archive, in: Inode, name: []const
.name = new_name,
};
}
pub fn fromDirEntry(alloc: std.mem.Allocator, io: Io, archive: Archive, ent: DirEntry) !File {
var rdr = try archive.file.readerAt(io, archive.super.inode_start + ent.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, &archive.stateless_decomp);
pub fn fromDirEntry(alloc: std.mem.Allocator, archive: *Archive, ent: DirEntry) !File {
var rdr = archive.file.readerAt(archive.super.inode_start + ent.block_start);
var meta: MetadataReader = .init(alloc, &rdr, &archive.stateless_decomp);
try meta.interface.discardAll(ent.block_offset);
var in: Inode = try .read(alloc, &meta.interface, archive.super.block_size);
errdefer in.deinit(alloc);
return .init(alloc, archive, in, ent.name);
return .init(alloc, archive.*, in, ent.name);
}
pub fn deinit(self: File) void {
self.alloc.free(self.name);
self.inode.deinit(self.alloc);
}
pub fn open(self: File, alloc: std.mem.Allocator, io: Io, filepath: []const u8) !File {
pub fn open(self: *File, alloc: std.mem.Allocator, io: Io, filepath: []const u8) !File {
const entries = try self.inode.readDirectory(
alloc,
io,
self.archive.file,
&self.archive.stateless_decomp,
self.archive.super.dir_start,
@@ -77,7 +75,7 @@ pub fn open(self: File, alloc: std.mem.Allocator, io: Io, filepath: []const u8)
}
} else return Error.FileNotFound;
const first_elem_file = try fromDirEntry(alloc, io, self.archive, search_slice[idx]);
var first_elem_file = try fromDirEntry(alloc, &self.archive, search_slice[idx]);
if (first_element.len == path.len)
return first_elem_file;
defer first_elem_file.deinit();
+78
View File
@@ -1,7 +1,85 @@
const std = @import("std");
const Io = std.Io;
const BlockSize = @import("inode_data/file.zig").BlockSize;
const LookupTable = @import("lookup_table.zig");
const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
const OffsetFile = @import("util/offset_file.zig");
const FragManager = @This();
pub const FragEntry = extern struct {
start: u64,
size: BlockSize,
_: u32,
};
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: *Decompressor,
block_size: u32,
entries: []FragEntry,
frag_cache: std.array_hash_map.Auto(u32, []u8),
cache_mut: std.Io.RwLock = .init,
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *Decompressor, frag_start: u64, frag_num: u32, block_size: u32) !FragManager {
const first_offset: u64 = std.mem.readInt(u64, @ptrCast(fil.map.memory[frag_start .. frag_start + 8]), .little);
var rdr = fil.readerAt(first_offset);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
const entries = try alloc.alloc(FragEntry, frag_num);
errdefer alloc.free(entries);
try meta.interface.readSliceEndian(FragEntry, entries, .little);
return .{
.alloc = alloc,
.fil = fil,
.decomp = decomp,
.block_size = block_size,
.entries = entries,
.frag_cache = .empty,
};
}
pub fn deinit(self: *FragManager, io: Io) void {
self.cache_mut.lockUncancelable(io);
self.alloc.free(self.entries);
for (self.frag_cache.values()) |v|
self.alloc.free(v);
self.frag_cache.deinit(self.alloc);
}
pub fn get(self: *FragManager, io: Io, idx: u32) ![]u8 {
{
try self.cache_mut.lockShared(io);
defer self.cache_mut.unlockShared(io);
if (self.frag_cache.contains(idx))
return self.frag_cache.get(idx).?;
}
try self.cache_mut.lock(io);
defer self.cache_mut.unlock(io);
if (self.frag_cache.contains(idx))
return self.frag_cache.get(idx).?;
const entry = self.entries[idx];
const out = try self.alloc.alloc(u8, if (entry.size.uncompressed) entry.size.size else self.block_size);
if (entry.size.uncompressed) {
@memcpy(out, self.fil.map.memory[entry.start .. entry.start + entry.size.size]);
} else {
@branchHint(.likely);
_ = try self.decomp.Decompress(self.alloc, self.fil.map.memory[entry.start .. entry.start + entry.size.size], out);
}
try self.frag_cache.put(self.alloc, idx, out);
return out;
}
+410 -114
View File
@@ -9,6 +9,7 @@ const Decomp = @import("decomp.zig").Decomp;
const DirEntry = @import("directory.zig");
const ExtractionOptions = @import("options.zig");
const FragEntry = @import("frag.zig").FragEntry;
const FragManager = @import("frag.zig");
const dir = @import("inode_data/dir.zig");
const file = @import("inode_data/file.zig");
const misc = @import("inode_data/misc.zig");
@@ -19,7 +20,6 @@ const DataReader = @import("util/data_reader.zig");
const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
const OffsetFile = @import("util/offset_file.zig");
const SharedCache = @import("util/shared_cache.zig");
const XattrTable = @import("xattr_table.zig");
const Inode = @This();
@@ -63,49 +63,49 @@ pub fn deinit(self: Inode, alloc: std.mem.Allocator) void {
// Utility Functions
/// Read the directory entries
pub fn readDirectory(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, dir_offset: u64) ![]DirEntry {
pub fn readDirectory(self: Inode, alloc: std.mem.Allocator, fil: OffsetFile, decomp: *Decompressor, dir_offset: u64) ![]DirEntry {
return switch (self.data) {
.dir => |d| readDirFromData(alloc, io, fil, decomp, dir_offset, d),
.ext_dir => |d| readDirFromData(alloc, io, fil, decomp, dir_offset, d),
.dir => |d| readDirFromData(alloc, fil, decomp, dir_offset, d),
.ext_dir => |d| readDirFromData(alloc, fil, decomp, dir_offset, d),
else => Error.NotDirectory,
};
}
fn readDirFromData(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, dir_offset: u64, d: anytype) ![]DirEntry {
var rdr = try fil.readerAt(io, dir_offset + d.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, decomp);
fn readDirFromData(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *Decompressor, dir_offset: u64, d: anytype) ![]DirEntry {
var rdr = fil.readerAt(dir_offset + d.block_start);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
try meta.interface.discardAll(d.block_offset);
return DirEntry.readDirectory(alloc, &meta.interface, d.size);
}
/// Get a reader for a regular file's data.
pub fn dataReader(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, cache: *SharedCache, decomp: *const Decompressor, block_size: u32) !DataReader {
pub fn dataReader(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, block_size: u32, frag_block: ?[]u8) !DataReader {
return switch (self.data) {
.file => |f| getReaderFromData(alloc, io, fil, cache, decomp, block_size, f),
.ext_file => |f| getReaderFromData(alloc, io, fil, cache, decomp, block_size, f),
.file => |f| getReaderFromData(alloc, io, fil, decomp, block_size, frag_block, f),
.ext_file => |f| getReaderFromData(alloc, io, fil, decomp, block_size, frag_block, f),
else => Error.NotRegularFile,
};
}
fn getReaderFromData(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, cache: *SharedCache, decomp: *const Decompressor, block_size: u32, d: anytype) !DataReader {
const ext: DataReader = .init(alloc, io, fil, cache, decomp, block_size, d.size, d.block_start, d.blocks);
fn getReaderFromData(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, block_size: u32, frag_block: ?[]u8, d: anytype) !DataReader {
const ext: DataReader = .init(alloc, io, fil, decomp, block_size, d.size, d.block_start, d.blocks);
if (d.frag_block_offset == 0xFFFFFFFF) {
// TODO:
return error.TODO;
if (frag_block == null) return error.FragBlockNotProvided;
ext.addFrag(d.frag_block_offset, frag_block.?);
}
return ext;
}
/// Get an extractor for a regular file's data.
pub fn dataExtractor(self: Inode, fil: OffsetFile, cache: *SharedCache, decomp: *const Decompressor, block_size: u32) !DataExtractor {
pub fn dataExtractor(self: Inode, fil: OffsetFile, decomp: *Decompressor, block_size: u32, frag_block: ?[]u8) !DataExtractor {
return switch (self.data) {
.file => |f| getExtractorFromData(fil, cache, decomp, block_size, f),
.ext_file => |f| getExtractorFromData(fil, cache, decomp, block_size, f),
.file => |f| getExtractorFromData(fil, decomp, block_size, frag_block, f),
.ext_file => |f| getExtractorFromData(fil, decomp, block_size, frag_block, f),
else => Error.NotRegularFile,
};
}
fn getExtractorFromData(fil: OffsetFile, cache: *SharedCache, decomp: *const Decompressor, block_size: u32, d: anytype) !DataExtractor {
const ext: DataExtractor = .init(fil, cache, decomp, block_size, d.size, d.block_start, d.blocks);
fn getExtractorFromData(fil: OffsetFile, decomp: *Decompressor, block_size: u32, frag_block: ?[]u8, d: anytype) !DataExtractor {
const ext: DataExtractor = .init(fil, decomp, block_size, d.size, d.block_start, d.blocks);
if (d.frag_block_offset == 0xFFFFFFFF) {
// TODO:
return error.TODO;
if (frag_block == null) return error.FragBlockNotProvided;
ext.addFrag(d.frag_block_offset, frag_block.?);
}
return ext;
}
@@ -118,11 +118,11 @@ pub fn symlinkTarget(self: Inode) ![]const u8 {
};
}
/// Get inode's gid
pub fn gid(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, id_table_start: u64) !u16 {
pub fn gid(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, id_table_start: u64) !u16 {
return LookupTable.lookupValue(u16, alloc, io, decomp, fil, id_table_start, self.hdr.gid_idx);
}
/// Get inode's uid
pub fn uid(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, id_table_start: u64) !u16 {
pub fn uid(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, id_table_start: u64) !u16 {
return LookupTable.lookupValue(u16, alloc, io, decomp, fil, id_table_start, self.hdr.uid_idx);
}
/// Get the inode's xattr values as an index into the Archive's xattr table.
@@ -136,13 +136,13 @@ pub fn xattrIndex(self: Inode) !u32 {
.ext_char_dev => |e| e.xattr_idx,
.ext_fifo => |e| e.xattr_idx,
.ext_socket => |e| e.xattr_idx,
else => Error.NoXattr,
else => return error.NoXattr,
};
if (idx == 0xFFFFFFFF) return Error.NoXattr;
if (idx == 0xFFFFFFFF) return error.NoXattr;
return idx;
}
// Get an inode's xattr values. If the inode does not have xattr values (including if the inode is not an extended type), an empty slice is returned.
pub fn xattrValues(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, xattr_table_start: u64) ![]XattrTable.XattrOwned {
pub fn xattrValues(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, xattr_table_start: u64) ![]XattrTable.XattrOwned {
const idx = self.xattrIndex() catch &[0]XattrTable.XattrOwned{};
return XattrTable.statelessLookup(alloc, io, decomp, fil, xattr_table_start, idx);
}
@@ -207,16 +207,53 @@ pub const Header = extern struct {
// Extract
const ExtractError = error{ MknodFailed, CannotSetXattr } || DataExtractor.Error || DirEntry.Error ||
Decompressor.Error || Io.File.Atomic.InitError || Io.File.Atomic.LinkError || Io.Dir.SymLinkError;
const PathRet = struct {
path: []const u8,
permissions: u16,
uid_idx: u16,
gid_idx: u16,
xattr_idx: ?u32 = null,
inode: Inode,
origin: bool,
fn deinit(self: PathRet, alloc: std.mem.Allocator) void {
if (self.origin) return;
alloc.free(self.path);
self.inode.deinit(alloc);
}
fn setMetadata(self: PathRet, alloc: std.mem.Allocator, io: Io, id_table: *CachedTable(u16), xattr_table: ?*XattrTable, options: ExtractionOptions) !void {
var fil = try Io.Dir.cwd().openFile(io, self.path, .{});
defer fil.close(io);
const inode = self.inode;
if (!options.ignore_permissions) {
try fil.setPermissions(io, @enumFromInt(inode.hdr.permissions));
try fil.setOwner(io, try id_table.get(io, inode.hdr.uid_idx), try id_table.get(io, inode.hdr.gid_idx));
}
if (xattr_table != null) {
const idx = inode.xattrIndex() catch return;
const xattrs = try xattr_table.?.get(alloc, io, idx);
defer {
for (xattrs) |x|
x.deinit(alloc);
alloc.free(xattrs);
}
const sentinel_path = try std.mem.concatWithSentinel(alloc, u8, &[_][]const u8{self.path}, 0);
defer alloc.free(sentinel_path);
for (xattrs) |x| {
const xattr_ret = std.os.linux.fsetxattr(fil.handle, x.key, x.value.ptr, x.value.len, 0);
if (xattr_ret != 0)
return ExtractError.CannotSetXattr;
}
}
}
};
fn DirCompare(_: void, a: PathRet, b: PathRet) std.math.Order {
return std.math.order(std.mem.count(u8, a.path, "/"), std.mem.count(u8, b.path, "/"));
}
const ExtractReturnUnion = union(enum) {
path_ret: anyerror!PathRet, // TODO: convert to concrete error type instead of anyerror.
void_ret: anyerror!void,
path_ret: ExtractError!PathRet,
};
const Tables = struct {
id: LookupTable.CachedTable(u16),
@@ -224,114 +261,373 @@ const Tables = struct {
xattr: XattrTable,
};
pub fn extract(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, super: Archive.Superblock, path: []const u8, options: ExtractionOptions) !void {
var decomp_base: Decomp = switch (super.compression) {
.gzip => .{ .gzip = try .init(alloc, super.block_size) },
.lzma => .{ .lzma = try .init(alloc, super.block_size) },
.xz => .{ .xz = try .init(alloc, super.block_size) },
.zstd => .{ .zstd = try .init(alloc, super.block_size) },
else => unreachable,
};
defer decomp_base.deinit();
const decomp = decomp_base.decompressor();
/// Extracts the given inode to the given path. If the inode not a directory, the given path must not exist.
/// If the inode is a directory the path must not exist or be a directory.
pub fn extract(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
super: Archive.Superblock,
filepath: []const u8,
options: ExtractionOptions,
) !void {
const path = std.mem.trimEnd(u8, filepath, "/");
var frag_table: CachedTable(FragEntry) = .init(alloc, fil, decomp, super.frag_start, super.frag_count);
defer if (!options.ignore_permissions) frag_table.deinit(io);
var decomp_base: Decompressor = try @import("decomp.zig").StatelessDecomp(super.compression); // TODO: Replace with actual Decomp value to share states & caches for efficiency.
const decomp = &decomp_base;
var sel_buf = [1]ExtractReturnUnion{undefined} ** 10;
var frag_mgr: FragManager = try .init(alloc, fil, decomp, super.frag_start, super.frag_count, super.block_size);
defer frag_mgr.deinit(io);
if (options.single_threaded)
return self.extractSinglethreaded(alloc, io, fil, super, path, options, decomp, &frag_mgr);
var sel_buf: [10]ExtractReturnUnion = undefined;
var sel: Io.Select(ExtractReturnUnion) = .init(io, &sel_buf);
defer sel.cancelDiscard();
var loop = io.async(finishLoop, .{ alloc, io, fil, decomp, super, options, &sel });
sel.async(.path_ret, extractRealAsync, .{ self, alloc, io, fil, super, decomp, &sel, &frag_mgr, path, true });
try loop.await(io);
}
fn extractRealAsync(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
super: Archive.Superblock,
decomp: *Decompressor,
sel: *Io.Select(ExtractReturnUnion),
frag_mgr: *FragManager,
path: []const u8,
origin: bool,
) ExtractError!PathRet {
errdefer {
if (!origin) {
self.deinit(alloc);
alloc.free(path);
}
}
switch (self.hdr.inode_type) {
.file, .ext_file => sel.async(.path_ret, extractFile, .{ self, alloc, io, fil, decomp, &frag_table, super.block_size, path }),
else => return error.TODO,
.dir, .ext_dir => {
try Io.Dir.cwd().createDir(io, path, @enumFromInt(0o777));
const entries = self.readDirectory(alloc, fil, decomp, super.dir_start) catch |err| switch (err) {
Error.NotDirectory, Error.NotExtended, Error.NotRegularFile, Error.NotSymlink => unreachable,
else => |e| return e,
};
defer {
for (entries) |e|
e.deinit(alloc);
alloc.free(entries);
}
var xattr_table: ?XattrTable = if (!options.ignore_xattr)
try .init(alloc, io, fil, decomp, super.xattr_start)
else
null;
defer if (!options.ignore_xattr) xattr_table.?.deinit(io);
for (entries) |e| {
const new_path = try std.mem.concat(alloc, u8, &[_][]const u8{ path, "/", e.name });
errdefer alloc.free(new_path);
var id_table: ?CachedTable(u16) = if (!options.ignore_xattr)
.init(alloc, fil, decomp, super.id_start, super.id_count)
var rdr = fil.readerAt(super.inode_start + e.block_start);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
try meta.interface.discardAll(e.block_offset);
const new_inode = try read(alloc, &meta.interface, super.block_size);
errdefer new_inode.deinit(alloc);
sel.async(.path_ret, extractRealAsync, .{ new_inode, alloc, io, fil, super, decomp, sel, frag_mgr, new_path, false });
}
},
.file, .ext_file => {
var atomic = try Io.Dir.cwd().createFileAtomic(io, path, .{ .make_path = true });
defer atomic.deinit(io);
var ext: DataExtractor = switch (self.data) {
.file => |f| blk: {
var ext: DataExtractor = .init(fil, decomp, super.block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
ext.addFrag(f.frag_block_offset, try frag_mgr.get(io, f.frag_idx));
break :blk ext;
},
.ext_file => |f| blk: {
var ext: DataExtractor = .init(fil, decomp, super.block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
ext.addFrag(f.frag_block_offset, try frag_mgr.get(io, f.frag_idx));
break :blk ext;
},
else => unreachable,
};
try ext.extractAsync(alloc, io, atomic.file);
try atomic.link(io);
},
.symlink, .ext_symlink => try Io.Dir.cwd().symLink(io, self.symlinkTarget() catch unreachable, path, .{}),
else => {
var mode: u32 = undefined;
var dev: u32 = 0;
const DT = std.posix.DT;
switch (self.data) {
.char_dev => |d| {
dev = d.dev;
mode = DT.CHR;
},
.ext_char_dev => |d| {
dev = d.dev;
mode = DT.CHR;
},
.block_dev => |d| {
dev = d.dev;
mode = DT.BLK;
},
.ext_block_dev => |d| {
dev = d.dev;
mode = DT.BLK;
},
.fifo, .ext_fifo => mode = DT.FIFO,
.socket, .ext_socket => mode = DT.SOCK,
else => unreachable,
}
const sentinel_path = try std.mem.concatWithSentinel(alloc, u8, &[_][]const u8{path}, 0);
const res = std.os.linux.mknod(sentinel_path, mode, dev);
alloc.free(sentinel_path);
if (res != 0)
return ExtractError.MknodFailed;
},
}
return .{
.path = path,
.inode = self,
.origin = origin,
};
}
fn finishLoop(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, super: Archive.Superblock, options: ExtractionOptions, sel: *Io.Select(ExtractReturnUnion)) !void {
var id_table: CachedTable(u16) = .init(alloc, fil, decomp, super.id_start, super.id_count);
defer id_table.deinit(io);
var xattr_table: ?XattrTable = if (super.flags.xattr_never or options.ignore_xattr or !@hasField(std.os, "linux"))
null
else
null;
defer if (!options.ignore_xattr) id_table.?.deinit(io);
try .init(alloc, fil, decomp, super.xattr_start);
defer if (xattr_table != null) xattr_table.?.deinit(io);
var dir_queue: std.PriorityDequeue(PathRet, void, DirCompare) = .empty;
defer dir_queue.deinit(alloc);
while (true) {
if (sel.group.token.load(.unordered) == null) break;
const ret = try sel.queue.getOne(io);
switch (ret) {
.void_ret => {
try ret.void_ret;
continue;
},
else => {},
}
const ret = try sel.await();
const path_ret = try ret.path_ret;
defer if (path_ret.path.len != path.len) alloc.free(path_ret.path);
if (options.ignore_permissions and options.ignore_xattr) continue;
if (options.ignore_permissions and path_ret.xattr_idx == null) continue;
if (options.ignore_permissions and xattr_table == null) {
path_ret.deinit(alloc);
continue;
}
var ret_file = try Io.Dir.cwd().openFile(io, path_ret.path, .{});
defer ret_file.close(io);
if (path_ret.inode.hdr.inode_type == .dir or path_ret.inode.hdr.inode_type == .ext_dir) {
try dir_queue.push(alloc, path_ret);
continue;
}
defer path_ret.deinit(alloc);
try path_ret.setMetadata(alloc, io, &id_table, if (xattr_table == null) null else &xattr_table.?, options);
}
while (sel.cancel()) |ret| {
const path_ret = try ret.path_ret;
if (options.ignore_permissions and xattr_table == null) {
path_ret.deinit(alloc);
continue;
}
if (path_ret.inode.hdr.inode_type == .dir or path_ret.inode.hdr.inode_type == .ext_dir) {
try dir_queue.push(alloc, path_ret);
continue;
}
defer path_ret.deinit(alloc);
try path_ret.setMetadata(alloc, io, &id_table, if (xattr_table == null) null else &xattr_table.?, options);
}
var iter = dir_queue.iterator();
while (iter.next()) |path_ret| {
defer path_ret.deinit(alloc);
try path_ret.setMetadata(alloc, io, &id_table, if (xattr_table == null) null else &xattr_table.?, options);
}
}
/// Extracts the given inode to the given path. If the inode not a directory, the given path must not exist.
/// If the inode is a directory the path must not exist or be a directory.
fn extractSinglethreaded(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
super: Archive.Superblock,
path: []const u8,
options: ExtractionOptions,
decomp: *Decompressor,
frag: *FragManager,
) !void {
var id_table: CachedTable(u16) = .init(alloc, fil, decomp, super.id_start, super.id_count);
defer id_table.deinit(io);
var xattr_table: ?XattrTable = if (super.flags.xattr_never or options.ignore_xattr or !@hasField(std.os, "linux"))
null
else
try .init(alloc, fil, decomp, super.xattr_start);
defer if (xattr_table != null) xattr_table.?.deinit(io);
return self.extractReal(
alloc,
io,
fil,
super,
decomp,
frag,
&id_table,
if (xattr_table == null) null else &xattr_table.?,
path,
options,
);
}
fn extractReal(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
super: Archive.Superblock,
decomp: *Decompressor,
frag_mgr: *FragManager,
id_table: *CachedTable(u16),
xattr_table: ?*XattrTable,
path: []const u8,
options: ExtractionOptions,
) !void {
switch (self.hdr.inode_type) {
.dir, .ext_dir => {
try Io.Dir.cwd().createDir(io, path, @enumFromInt(0o777));
const entries = self.readDirectory(alloc, fil, decomp, super.dir_start) catch |err| switch (err) {
Error.NotDirectory, Error.NotExtended, Error.NotRegularFile, Error.NotSymlink => unreachable,
else => |e| return e,
};
defer {
for (entries) |e|
e.deinit(alloc);
alloc.free(entries);
}
for (entries) |e| {
const new_path = try std.mem.concat(alloc, u8, &[_][]const u8{ path, "/", e.name });
defer alloc.free(new_path);
var rdr = fil.readerAt(super.inode_start + e.block_start);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
try meta.interface.discardAll(e.block_offset);
const new_inode = try read(alloc, &meta.interface, super.block_size);
defer new_inode.deinit(alloc);
try new_inode.extractReal(alloc, io, fil, super, decomp, frag_mgr, id_table, xattr_table, new_path, options);
}
},
.file, .ext_file => {
var atomic = try Io.Dir.cwd().createFileAtomic(io, path, .{ .make_path = true });
defer atomic.deinit(io);
var rdr: DataReader = switch (self.data) {
.file => |f| blk: {
var ext: DataReader = try .init(alloc, io, fil, decomp, super.block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
ext.addFrag(f.frag_block_offset, try frag_mgr.get(io, f.frag_idx));
break :blk ext;
},
.ext_file => |f| blk: {
var ext: DataReader = try .init(alloc, io, fil, decomp, super.block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
ext.addFrag(f.frag_block_offset, try frag_mgr.get(io, f.frag_idx));
break :blk ext;
},
else => unreachable,
};
defer rdr.deinit();
var buf: [512 * 1024]u8 = undefined;
var wrt = atomic.file.writer(io, &buf);
_ = try rdr.interface.streamRemaining(&wrt.interface);
try wrt.flush();
try atomic.link(io);
},
.symlink, .ext_symlink => try Io.Dir.cwd().symLink(io, self.symlinkTarget() catch unreachable, path, .{}),
else => {
var mode: u32 = undefined;
var dev: u32 = 0;
const DT = std.posix.DT;
switch (self.data) {
.char_dev => |d| {
dev = d.dev;
mode = DT.CHR;
},
.ext_char_dev => |d| {
dev = d.dev;
mode = DT.CHR;
},
.block_dev => |d| {
dev = d.dev;
mode = DT.BLK;
},
.ext_block_dev => |d| {
dev = d.dev;
mode = DT.BLK;
},
.fifo, .ext_fifo => mode = DT.FIFO,
.socket, .ext_socket => mode = DT.SOCK,
else => unreachable,
}
const sentinel_path = try std.mem.concatWithSentinel(alloc, u8, &[_][]const u8{path}, 0);
const res = std.os.linux.mknod(sentinel_path, mode, dev);
alloc.free(sentinel_path);
if (res != 0)
return ExtractError.MknodFailed;
},
}
if (options.ignore_permissions and options.ignore_xattr) return;
var f = try Io.Dir.cwd().openFile(io, path, .{});
defer f.close(io);
if (!options.ignore_permissions) {
try ret_file.setPermissions(io, @enumFromInt(path_ret.permissions));
try ret_file.setOwner(io, try id_table.?.get(io, path_ret.uid_idx), try id_table.?.get(io, path_ret.gid_idx));
try f.setPermissions(io, @enumFromInt(self.hdr.permissions));
try f.setOwner(io, try id_table.get(io, self.hdr.uid_idx), try id_table.get(io, self.hdr.gid_idx));
}
if (!options.ignore_xattr and path_ret.xattr_idx != null) {
const xattrs = try xattr_table.?.get(alloc, io, path_ret.xattr_idx.?);
if (xattr_table != null) {
const idx = self.xattrIndex() catch return;
const xattrs = try xattr_table.?.get(alloc, io, idx);
defer {
for (xattrs) |x|
alloc.free(x.key);
x.deinit(alloc);
alloc.free(xattrs);
}
const sentinel_path = try std.mem.concatWithSentinel(alloc, u8, &[_][]const u8{path}, 0);
defer alloc.free(sentinel_path);
for (xattrs) |x| {
const res = std.os.linux.fsetxattr(ret_file.handle, x.key, x.value.ptr, x.value.len, 0);
if (res != 0)
return error.CannotSetXattr;
const xattr_ret = std.os.linux.fsetxattr(f.handle, x.key, x.value.ptr, x.value.len, 0);
if (xattr_ret != 0)
return ExtractError.CannotSetXattr;
}
}
}
}
pub fn extractFile(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, frag: *CachedTable(FragEntry), block_size: u32, path: []const u8) anyerror!PathRet {
var atomic = try Io.Dir.cwd().createFileAtomic(io, path, .{});
defer atomic.deinit(io);
var ret: PathRet = .{
.gid_idx = self.hdr.gid_idx,
.uid_idx = self.hdr.uid_idx,
.permissions = self.hdr.permissions,
.path = path,
};
const data: DataExtractor = blk: {
switch (self.data) {
.file => |f| {
var data: DataExtractor = .init(fil, decomp, block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
data.addFrag(f.frag_block_offset, try frag.get(io, f.frag_idx));
break :blk data;
},
.ext_file => |f| {
if (f.xattr_idx != 0xFFFFFFFF) ret.xattr_idx = f.xattr_idx;
var data: DataExtractor = .init(fil, decomp, block_size, f.size, f.block_start, f.block_sizes);
if (f.frag_idx != 0xFFFFFFFF)
data.addFrag(f.frag_block_offset, try frag.get(io, f.frag_idx));
break :blk data;
},
else => unreachable,
}
};
try data.extractAsync(alloc, io, atomic.file);
try atomic.link(io);
return ret;
}
+49 -15
View File
@@ -5,17 +5,16 @@ const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
const OffsetFile = @import("util/offset_file.zig");
pub fn lookupValue(comptime T: anytype, alloc: std.mem.Allocator, io: Io, decomp: *Decompressor, file: OffsetFile, table_start: u64, idx: u16) !T {
pub fn lookupValue(comptime T: anytype, alloc: std.mem.Allocator, decomp: *Decompressor, file: OffsetFile, table_start: u64, idx: u32) !T {
const T_PER_BLOCK: u16 = 8192 / @sizeOf(T);
const block = idx / T_PER_BLOCK;
const block_offset = idx % T_PER_BLOCK;
var rdr = try file.readerAt(io, table_start + (8 * block), &[0]u8{});
var offset: u64 = undefined;
try rdr.interface.readSliceEndian(u64, @ptrCast(&offset), .little);
const offset_pos = table_start + (8 * block);
const offset: u64 = std.mem.readInt(u64, @ptrCast(file.map.memory[offset_pos .. offset_pos + 8]), .little);
rdr = try file.readerAt(io, offset, &[0]u8{});
var rdr = file.readerAt(offset);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
try meta.interface.discardAll(@sizeOf(T) * block_offset);
@@ -24,6 +23,8 @@ pub fn lookupValue(comptime T: anytype, alloc: std.mem.Allocator, io: Io, decomp
return out;
}
pub const Error = Io.Cancelable || Io.File.Reader.SeekError || Io.Reader.ReadAllocError;
pub fn CachedTable(comptime T: anytype) type {
return struct {
const T_PER_BLOCK: u16 = 8192 / @sizeOf(T);
@@ -32,16 +33,16 @@ pub fn CachedTable(comptime T: anytype) type {
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: *const Decompressor,
decomp: *Decompressor,
table_start: u64,
total_num: u32,
table: std.AutoHashMap(u32, []T),
mut: Io.Mutex = .init,
mut: Io.RwLock = .init,
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *const Decompressor, offset: u64, total_num: u32) Table {
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *Decompressor, offset: u64, total_num: u32) Table {
return .{
.alloc = alloc,
.fil = fil,
@@ -55,14 +56,48 @@ pub fn CachedTable(comptime T: anytype) type {
}
pub fn deinit(self: *Table, io: Io) void {
self.mut.lockUncancelable(io);
var iter = self.table.valueIterator();
while (iter.next()) |val|
self.alloc.free(val.*);
self.table.deinit();
}
pub fn get(self: *Table, io: Io, idx: u32) !T {
pub fn fill(self: *Table, io: Io) Error!void {
try self.mut.lock(io);
defer self.mut.unlock(io);
var num_blocks = self.total_num / T_PER_BLOCK;
if (self.total_num % T_PER_BLOCK > 0)
num_blocks += 1;
for (0..num_blocks) |block| {
const offset_pos = self.table_start + (8 * block);
const offset: u64 = std.mem.readInt(u64, @ptrCast(self.fil.map.memory[offset_pos .. offset_pos + 8]), .little);
const len: u16 = if (self.total_num % T_PER_BLOCK != 0 and block == (self.total_num - 1) / T_PER_BLOCK)
@truncate(self.total_num % T_PER_BLOCK)
else
T_PER_BLOCK;
var rdr = self.fil.readerAt(offset);
var meta: MetadataReader = .init(self.alloc, &rdr.interface, self.decomp);
const slice = try meta.interface.readSliceEndianAlloc(self.alloc, T, len, .little);
try self.table.put(@truncate(block), slice);
}
}
pub fn get(self: *Table, io: Io, idx: u32) Error!T {
const block = idx / T_PER_BLOCK;
const block_offset = idx % T_PER_BLOCK;
{
try self.mut.lockShared(io);
defer self.mut.unlockShared(io);
if (self.table.contains(block))
return self.table.get(block).?[block_offset];
}
try self.mut.lock(io);
defer self.mut.unlock(io);
@@ -70,20 +105,19 @@ pub fn CachedTable(comptime T: anytype) type {
if (self.table.contains(block))
return self.table.get(block).?[block_offset];
var rdr = try self.fil.readerAt(io, self.table_start + (8 * block), &[0]u8{});
var offset: u64 = undefined;
try rdr.interface.readSliceEndian(u64, @ptrCast(&offset), .little);
const offset_pos = self.table_start + (8 * block);
const offset: u64 = std.mem.readInt(u64, @ptrCast(self.fil.map.memory[offset_pos .. offset_pos + 8]), .little);
const len: u16 = if (self.total_num % T_PER_BLOCK != 0 and block == (self.total_num - 1) / T_PER_BLOCK)
@truncate(self.total_num % T_PER_BLOCK)
else
T_PER_BLOCK;
rdr = try self.fil.readerAt(io, offset, &[0]u8{});
var meta: MetadataReader = .init(self.alloc, &rdr.interface, self.decomp);
var rdr = self.fil.readerAt(offset);
var meta: MetadataReader = .init(self.alloc, &rdr, self.decomp);
const slice = try meta.interface.readSliceEndianAlloc(self.alloc, T, len, .little);
try self.table.put(block, slice);
try self.table.put(@truncate(block), slice);
return slice[block_offset];
}
+7 -9
View File
@@ -5,8 +5,10 @@ const Writer = std.Io.Writer;
const ExtractionOptions = @This();
/// The number of threads used for extraction. 0 implies single threaded.
threads: usize = 1, // TODO: Update to better integrate with zig 0.16 Io. Maybe limit to only single or multi-threaded.
/// Extract single-threaded only.
/// Though not necessary if using Threaded.single_threaded,
/// setting single_threaded is more efficient.
single_threaded: bool = false,
/// Don't set the file's owner & permissions after extraction
ignore_permissions: bool = false,
/// Don't set xattr values. Currently xattrs are never set anyway.
@@ -18,16 +20,12 @@ verbose: bool = false,
/// Where to print verbose log.
verbose_writer: ?*Writer = null,
pub const SingleThreadedDefault: ExtractionOptions = .{};
pub fn Default() !ExtractionOptions {
return .{
.threads = try std.Thread.getCpuCount(),
};
}
pub const default: ExtractionOptions = .{};
pub const default_single_threaded: ExtractionOptions = .{ .single_threaded = true };
pub fn VerboseDefault(wrt: *Writer) !ExtractionOptions {
return .{
.verbose = true,
.verbose_writer = wrt,
.threads = try std.Thread.getCpuCount(),
};
}
+4
View File
@@ -1,2 +1,6 @@
pub const Archive = @import("archive.zig");
pub const ExtractionOptions = @import("options.zig");
test {
@import("std").testing.refAllDecls(@This());
}
-83
View File
@@ -1,83 +0,0 @@
const std = @import("std");
const Io = std.Io;
const io = std.testing.io;
const alloc = std.testing.allocator;
const stuff = @import("builtin");
const Archive = @import("archive.zig");
const Superblock = Archive.Superblock;
const TestArchive = "testing/LinuxPATest.sfs";
test "Basics" {
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
try std.testing.expectEqualDeep(sfs.super, LinuxPATestCorrectSuperblock);
const root_file = try sfs.root(alloc, io);
defer root_file.deinit();
}
const TestFile = "Start.exe";
const TestFileExtractLocation = "testing/Start.exe";
test "ExtractSingleFile" {
Io.Dir.cwd().deleteFile(io, TestFileExtractLocation) catch {};
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
var test_fil = try sfs.open(alloc, io, TestFile);
defer test_fil.deinit();
try test_fil.extract(alloc, io, TestFileExtractLocation, try .Default());
//TODO: validate extracted file.
}
const TestFullExtractLocation = "testing/TestExtract";
test "ExtractCompleteArchive" {
Io.Dir.cwd().deleteTree(io, TestFullExtractLocation) catch {};
var fil = try Io.Dir.cwd().openFile(io, TestArchive, .{});
defer fil.close(io);
var sfs: Archive = try .init(io, fil, 0);
try sfs.extract(alloc, io, TestFullExtractLocation, try .Default());
}
const LinuxPATestCorrectSuperblock: Superblock = .{
.magic = std.mem.readInt(u32, "hsqs", .little),
.inode_count = 2974,
.mod_time = 1632696724,
.block_size = 131072,
.frag_count = 264,
.compression = .zstd,
.block_log = 17,
.flags = .{
.inode_uncompressed = false,
.data_uncompressed = false,
.check = false,
.frag_uncompressed = false,
.fragment_never = false,
.fragment_always = false,
.duplicates = true,
.exportable = true,
.xattr_uncompressed = false,
.xattr_never = false,
.compression_options = false,
.ids_uncompressed = false,
._ = 0,
},
.id_count = 1,
.ver_maj = 4,
.ver_min = 0,
.root_ref = .{
.block_offset = 1363,
.block_start = 29237,
._ = 0,
},
.size = 106841744,
.id_start = 106841632,
.xattr_start = 106841720,
.inode_start = 106778274,
.dir_start = 106807998,
.frag_start = 106837747,
.export_start = 106841602,
};
+50 -66
View File
@@ -10,10 +10,12 @@ const OffsetFile = @import("offset_file.zig");
// const SharedCache = @import("shared_cache.zig");
pub const Error = Decompressor.Error || Io.File.MemoryMap.CreateError || Io.File.WritePositionalError;
const DataExtractor = @This();
fil: OffsetFile,
decomp: *const Decompressor,
decomp: *Decompressor,
block_size: u32,
file_size: u64,
@@ -21,11 +23,11 @@ start: u64,
blocks: []BlockSize,
frag_offset: u32 = 0,
frag_entry: ?FragEntry = null,
frag_block: ?[]u8 = null,
err: ?anyerror = null,
err: ?Error = null,
pub fn init(fil: OffsetFile, decomp: *const Decompressor, block_size: u32, file_size: u64, data_start: u64, blocks: []BlockSize) DataExtractor {
pub fn init(fil: OffsetFile, decomp: *Decompressor, block_size: u32, file_size: u64, data_start: u64, blocks: []BlockSize) DataExtractor {
return .{
.fil = fil,
.decomp = decomp,
@@ -36,35 +38,52 @@ pub fn init(fil: OffsetFile, decomp: *const Decompressor, block_size: u32, file_
.blocks = blocks,
};
}
pub fn addFrag(self: *DataExtractor, frag_offset: u32, entry: FragEntry) void {
pub fn addFrag(self: *DataExtractor, frag_offset: u32, block: []u8) void {
self.frag_offset = frag_offset;
self.frag_entry = entry;
self.frag_block = block;
}
fn numBlocks(self: DataExtractor) usize {
var num = self.blocks.len;
if (self.frag_entry != null) num += 1;
if (self.frag_block != null) num += 1;
return num;
}
/// Starts extracting the data using the given group to spawn async tasks.
pub fn extractAsync(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File) !void {
pub fn extractConcurrent(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File) (Error || Io.ConcurrentError)!void {
var group: Io.Group = .init;
defer group.cancel(io);
var err: ?anyerror = null;
var err: ?Error = null;
var read_offset: u64 = self.start;
for (0..self.blocks.len) |idx| {
try group.concurrent(io, blockThread, .{ self, alloc, io, fil, read_offset, idx, &err });
read_offset += self.blocks[idx].size;
}
if (self.frag_block != null)
try group.concurrent(io, fragThread, .{ self, io, fil, &err });
group.await(io) catch |cancel| return err orelse cancel;
}
/// Starts extracting the data using the given group to spawn async tasks.
pub fn extractAsync(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File) Error!void {
var group: Io.Group = .init;
defer group.cancel(io);
var err: ?Error = null;
var read_offset: u64 = self.start;
for (0..self.blocks.len) |idx| {
group.async(io, blockThread, .{ self, alloc, io, fil, read_offset, idx, &err });
read_offset += self.blocks[idx].size;
}
if (self.frag_entry != null)
group.async(io, fragThread, .{ self, alloc, io, fil, &err });
if (self.frag_block != null)
group.async(io, fragThread, .{ self, io, fil, &err });
group.await(io) catch |cancel| return err orelse cancel;
}
fn blockThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File, read_offset: u64, idx: usize, ret_err: *?anyerror) Io.Cancelable!void {
fn blockThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File, read_offset: u64, idx: usize, ret_err: *?Error) Io.Cancelable!void {
const block = self.blocks[idx];
const cur_block_size = if (idx == self.numBlocks() - 1)
@@ -72,13 +91,14 @@ fn blockThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.Fi
else
self.block_size;
const write_offset = self.block_size * idx;
var wrt = fil.writer(io, &[0]u8{});
wrt.seekTo(self.block_size * idx) catch |err| {
wrt.seekTo(write_offset) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
defer wrt.flush() catch {};
if (block.size == 0) {
wrt.interface.splatByteAll(0, cur_block_size) catch |err| {
@@ -86,37 +106,23 @@ fn blockThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.Fi
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
return;
}
var rdr = self.fil.readerAt(io, read_offset, &[0]u8{}) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
} else {
if (block.uncompressed) {
rdr.interface.streamExact(&wrt.interface, cur_block_size) catch |err| {
wrt.interface.writeAll(self.fil.map.memory[read_offset..][0..cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
return;
} else {
@branchHint(.likely);
var cache: [1024 * 1024]u8 = undefined;
var tmp: [1024 * 1024]u8 = undefined;
rdr.interface.readSliceAll(cache[0..block.size]) catch |err| {
_ = self.decomp.Decompress(alloc, self.fil.map.memory[read_offset..][0..block.size], tmp[0..cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
_ = self.decomp.Decompress(alloc, cache[0..block.size], tmp[0..cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
wrt.interface.writeAll(tmp[0..cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
@@ -124,55 +130,33 @@ fn blockThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.Fi
};
}
}
fn fragThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File, ret_err: *?anyerror) Io.Cancelable!void {
const frag = self.frag_entry.?;
wrt.flush() catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
}
fn fragThread(self: DataExtractor, io: Io, fil: Io.File, ret_err: *?Error) Io.Cancelable!void {
const cur_block_size = self.file_size % self.block_size;
const write_offset = self.blocks.len * self.block_size;
var wrt = fil.writer(io, &[0]u8{});
wrt.seekTo(self.blocks.len * self.block_size) catch |err| {
wrt.seekTo(write_offset) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
defer wrt.flush() catch {};
var rdr = self.fil.readerAt(io, frag.start, &[0]u8{}) catch |err| {
wrt.interface.writeAll(self.frag_block.?[self.frag_offset..][0..cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
if (frag.size.uncompressed) {
rdr.interface.discardAll(self.frag_offset) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
rdr.interface.streamExact(&wrt.interface, cur_block_size) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
return;
} else {
@branchHint(.likely);
var cache: [1024 * 1024]u8 = undefined;
var tmp: [1024 * 1024]u8 = undefined;
rdr.interface.readSliceAll(cache[0..frag.size.size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
_ = self.decomp.Decompress(alloc, cache[0..frag.size.size], tmp[0..self.block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
wrt.interface.writeAll(tmp[0..cur_block_size]) catch |err| {
wrt.flush() catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
}
}
+13 -32
View File
@@ -6,7 +6,6 @@ const Reader = Io.Reader;
const Writer = Io.Writer;
const Limit = Io.Limit;
const FragEntry = @import("../frag.zig").FragEntry;
const BlockSize = @import("../inode_data/file.zig").BlockSize;
const Decompressor = @import("decompressor.zig");
const OffsetFile = @import("offset_file.zig");
@@ -19,8 +18,7 @@ alloc: std.mem.Allocator,
fil: OffsetFile,
io: Io,
decomp: *const Decompressor,
cache: *Io.Queue([]u8),
decomp: *Decompressor,
block_size: u32,
file_size: u64,
@@ -28,21 +26,20 @@ cur_offset: u64,
blocks: []BlockSize,
frag_offset: u32 = 0,
frag_entry: ?FragEntry = null,
frag_block: ?[]u8 = null,
block_idx: usize = 0,
sparse_block: bool = false,
interface: Io.Reader,
pub fn init(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, cache: *Io.Queue([]u8), block_size: u32, file_size: u64, data_start: u64, blocks: []BlockSize) !DataReader {
pub fn init(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *Decompressor, block_size: u32, file_size: u64, data_start: u64, blocks: []BlockSize) !DataReader {
return .{
.alloc = alloc,
.fil = fil,
.io = io,
.decomp = decomp,
.cache = cache,
.block_size = block_size,
.file_size = file_size,
@@ -64,43 +61,31 @@ pub fn init(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const De
pub fn deinit(self: *DataReader) void {
self.alloc.free(self.interface.buffer);
}
pub fn addFrag(self: *DataReader, frag_offset: u32, entry: FragEntry) void {
pub fn addFrag(self: *DataReader, frag_offset: u32, block: []u8) void {
self.frag_offset = frag_offset;
self.frag_entry = entry;
self.frag_block = block;
}
fn numBlocks(self: DataReader) usize {
var num = self.blocks.len;
if (self.frag_entry != null) num += 1;
if (self.frag_block != null) num += 1;
return num;
}
fn advanceBuffer(self: *DataReader) !void {
if (self.block_idx >= self.numBlocks()) {
if (self.block_idx >= self.numBlocks())
return Reader.Error.EndOfStream;
}
errdefer self.interface.end = 0;
defer self.block_idx += 1;
self.interface.end = if (self.block_idx == self.numBlocks() - 1)
self.size % self.block_size
self.file_size % self.block_size
else
self.block_size;
// Fragment
if (self.block_idx == self.blocks.len) {
const entry = self.frag_entry.?;
if (entry.size.uncompressed) {
var rdr = try self.fil.readerAt(self.io, entry.start + self.frag_offset, &[0]u8{});
try rdr.interface.readSliceAll(self.interface.buffer[0..self.interface.end]);
} else {
@branchHint(.likely);
const tmp = try self.cache.getOne(self.io);
defer self.cache.putOne(tmp) catch {};
var rdr = try self.fil.readerAt(self.io, entry.start, &[0]u8{});
try rdr.interface.readSliceAll(tmp.cache[0..entry.size.size]);
_ = try self.decomp.Decompress(self.alloc, tmp.cache[0..entry.size.size], self.interface.buffer[0..self.block_size]);
@memmove(self.interface.buffer[0..self.interface.end], self.interface.buffer[self.frag_offset .. self.frag_offset + self.interface.end]);
}
@memcpy(self.interface.buffer[0..self.interface.end], self.frag_block.?[self.frag_offset .. self.frag_offset + self.interface.end]);
self.interface.seek = 0;
return;
}
@@ -115,17 +100,13 @@ fn advanceBuffer(self: *DataReader) !void {
self.sparse_block = false;
}
if (block.uncompressed) {
try self.fil.readAt(self.io, self.cur_offset, self.interface.buffer[0..self.interface.end]);
@memcpy(self.interface.buffer[0..self.interface.end], self.fil.map.memory[self.cur_offset .. self.cur_offset + self.interface.end]);
self.cur_offset += self.interface.end;
} else {
@branchHint(.likely);
const tmp = try self.cache.getOne(self.io);
defer self.cache.putOne(tmp) catch {};
var rdr = try self.fil.readerAt(self.io, self.cur_offset, &[0]u8{});
try rdr.interface.readSliceAll(tmp.cache[0..block.size]);
_ = try self.decomp.Decompress(self.alloc, self.fil.map.memory[self.cur_offset .. self.cur_offset + block.size], self.interface.buffer[0..self.interface.end]);
self.cur_offset += block.size;
_ = try self.decomp.Decompress(self.alloc, tmp.cache[0..block.size], self.interface.buffer[0..self.interface.end]);
}
self.interface.seek = 0;
}
+2 -2
View File
@@ -8,8 +8,8 @@ pub const Error = std.Io.Reader.StreamError || std.mem.Allocator.Error;
/// The actual decompression function.
/// If the given decompressor is null, then the decompression should be done "stateless" without lasting allocations.
decomp_fn: *const fn (?*const Decompressor, std.mem.Allocator, in: []u8, out: []u8) Error!usize,
decomp_fn: *const fn (?*Decompressor, std.mem.Allocator, in: []u8, out: []u8) Error!usize,
pub fn Decompress(self: *const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
pub fn Decompress(self: *Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
return self.decomp_fn(self, alloc, in, out);
}
+2 -2
View File
@@ -15,7 +15,7 @@ const This = @This();
alloc: std.mem.Allocator,
rdr: *Reader,
decomp: *const Decompressor,
decomp: *Decompressor,
cur_block_start: u32 = 0,
next_start_start: u32 = 0,
@@ -34,7 +34,7 @@ interface: Reader = .{
},
},
pub fn init(alloc: std.mem.Allocator, rdr: *Reader, decomp: *const Decompressor) This {
pub fn init(alloc: std.mem.Allocator, rdr: *Reader, decomp: *Decompressor) This {
return .{
.alloc = alloc,
.rdr = rdr,
+3 -3
View File
@@ -16,9 +16,9 @@ pub fn pathIsSelf(path: []const u8) bool {
return path[0] == '.';
}
/// Creates an Inode from an Inode.Ref.
pub fn inodeFromRef(alloc: std.mem.Allocator, io: Io, file: OffsetFile, decomp: *const Decompressor, inode_start: u64, block_size: u32, ref: Inode.Ref) !Inode {
var rdr = try file.readerAt(io, inode_start + ref.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, decomp);
pub fn inodeFromRef(alloc: std.mem.Allocator, file: OffsetFile, decomp: *Decompressor, inode_start: u64, block_size: u32, ref: Inode.Ref) !Inode {
var rdr = file.readerAt(inode_start + ref.block_start);
var meta: MetadataReader = .init(alloc, &rdr, decomp);
try meta.interface.discardAll(ref.block_offset);
return .read(alloc, &meta.interface, block_size);
+15 -17
View File
@@ -3,27 +3,25 @@
const std = @import("std");
const Io = std.Io;
const File = Io.File;
const Reader = File.Reader;
const Reader = Io.Reader;
const OffsetFile = @This();
fil: File,
offset: u64,
map: Io.File.MemoryMap,
pub fn init(fil: File, init_offset: u64) OffsetFile {
return .{ .fil = fil, .offset = init_offset };
pub fn init(io: Io, fil: File, archive_size: u64, init_offset: u64) !OffsetFile {
return .{
.map = try fil.createMemoryMap(io, .{
.protection = .{ .read = true, .write = false, .execute = false },
.len = archive_size,
.offset = init_offset,
}),
};
}
pub fn deinit(self: *OffsetFile, io: Io) void {
self.map.destroy(io);
}
pub fn readerAt(self: OffsetFile, io: Io, offset: u64, buffer: []u8) !Reader {
var rdr = self.fil.reader(io, buffer);
try rdr.seekTo(self.offset + offset);
return rdr;
}
pub fn readAt(self: OffsetFile, io: Io, offset: u64, buf: []u8) !void {
_ = try self.fil.readPositionalAll(io, buf, self.offset + offset);
}
pub fn readValueAt(self: OffsetFile, comptime T: anytype, io: Io, offset: u64) !void {
//TODO: check for endianess and decode accordingly.
var new: T = undefined;
_ = try self.fil.readPositionalAll(io, @ptrCast(&new), self.offset + offset);
pub fn readerAt(self: OffsetFile, offset: u64) Reader {
return .fixed(self.map.memory[offset..]);
}
-52
View File
@@ -1,52 +0,0 @@
const std = @import("std");
const Io = std.Io;
const Node = std.SinglyLinkedList.Node;
const SharedCache = @This();
pub const CACHE_SIZE = 1024 * 1024;
pub const BufferNode = struct {
node: Node,
cache: [CACHE_SIZE]u8,
};
alloc: std.mem.Allocator,
caches: std.ArrayList(BufferNode),
cache_queue: std.SinglyLinkedList,
queue_mut: Io.Mutex,
pub fn init(alloc: std.mem.Allocator, init_cache_size: u32) !SharedCache {
const caches: std.ArrayList(BufferNode) = try .initCapacity(alloc, init_cache_size);
var queue: std.SinglyLinkedList = .{};
for (caches.items) |item|
queue.prepend(&item.node);
return .{
.alloc = alloc,
.caches = caches,
.cache_queue = queue,
};
}
pub fn deinit(self: *SharedCache) void {
self.caches.deinit(self.alloc);
}
pub fn getCache(self: *SharedCache, io: Io) !*BufferNode {
self.queue_mut.lock(io);
const nxt = self.cache_queue.popFirst();
self.queue_mut.unlock(io);
if (nxt == null) {
const new = try self.caches.addOne(self.alloc);
new.* = .{
.node = .{},
.cache = undefined,
};
return new;
}
return @fieldParentPtr("node", nxt.?);
}
pub fn returnCache(self: *SharedCache, buf: *BufferNode) void {
self.cache_queue.prepend(buf);
}
+26 -20
View File
@@ -12,21 +12,17 @@ const XattrCachedTable = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: *const Decompressor,
decomp: *Decompressor,
kv_start: u64,
table: LookupTable.CachedTable(TableValue),
value_cache: std.AutoHashMap(InodeRef, []const u8),
value_mut: Io.Mutex = .init,
value_mut: Io.RwLock = .init,
pub fn init(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, xattr_start: u64) !XattrCachedTable {
var rdr = try fil.readerAt(io, xattr_start, &[0]u8{});
var start: u64 = undefined;
try rdr.interface.readSliceEndian(u64, @ptrCast(&start), .little);
var num: u32 = undefined;
try rdr.interface.readSliceEndian(u32, @ptrCast(&num), .little);
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *Decompressor, xattr_start: u64) !XattrCachedTable {
const start: u64 = std.mem.readInt(u64, @ptrCast(fil.map.memory[xattr_start .. xattr_start + 8]), .little);
const num: u64 = std.mem.readInt(u64, @ptrCast(fil.map.memory[xattr_start + 8 .. xattr_start + 16]), .little);
return .{
.alloc = alloc,
@@ -41,6 +37,7 @@ pub fn init(alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const De
};
}
pub fn deinit(self: *XattrCachedTable, io: Io) void {
self.value_mut.lockUncancelable(io);
self.table.deinit(io);
self.value_cache.deinit();
}
@@ -48,8 +45,8 @@ pub fn deinit(self: *XattrCachedTable, io: Io) void {
pub fn get(self: *XattrCachedTable, alloc: std.mem.Allocator, io: Io, idx: u32) ![]XattrSemiOwned {
const lookup = try self.table.get(io, idx);
var rdr = try self.fil.readerAt(io, self.kv_start + lookup.ref.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, self.decomp);
var rdr = self.fil.readerAt(self.kv_start + lookup.ref.block_start);
var meta: MetadataReader = .init(alloc, &rdr, self.decomp);
try meta.interface.discardAll(lookup.ref.block_offset);
const out = try alloc.alloc(XattrSemiOwned, lookup.count);
@@ -99,8 +96,20 @@ pub fn get(self: *XattrCachedTable, alloc: std.mem.Allocator, io: Io, idx: u32)
}
const val_ref: InodeRef = .{ .block_start = meta.cur_block_start, .block_offset = @truncate(meta.interface.seek) };
{
try self.value_mut.lockShared(io);
defer self.value_mut.unlockShared(io);
if (self.value_cache.contains(val_ref)) {
out[i] = .{
.key = key,
.value = try self.valueAt(io, val_ref),
};
continue;
}
}
try self.value_mut.lock(io);
defer self.value_mut.unlock(io);
if (self.value_cache.contains(val_ref)) {
out[i] = .{
.key = key,
@@ -131,8 +140,8 @@ fn valueAt(self: *XattrCachedTable, io: Io, ref: InodeRef) ![]const u8 {
if (self.value_cache.contains(ref)) return self.value_cache.get(ref).?;
var rdr = try self.fil.readerAt(io, self.kv_start + ref.block_start, &[0]u8{});
var meta: MetadataReader = .init(self.alloc, &rdr.interface, self.decomp);
var rdr = self.fil.readerAt(self.kv_start + ref.block_start);
var meta: MetadataReader = .init(self.alloc, &rdr, self.decomp);
try meta.interface.discardAll(ref.block_offset);
var val_size: u32 = undefined;
@@ -203,15 +212,12 @@ const XattrPrefix = packed struct(u16) {
// Stateless
pub fn statelessLookup(alloc: std.mem.Allocator, io: Io, decomp: *const Decompressor, fil: OffsetFile, table_start: u64, idx: u16) ![]XattrOwned {
var rdr = try fil.readerAt(io, table_start, &[0]u8{});
var kv_start: u64 = undefined;
try rdr.interface.readSliceEndian(u64, @ptrCast(&kv_start), .little);
pub fn statelessLookup(alloc: std.mem.Allocator, io: Io, decomp: *Decompressor, fil: OffsetFile, table_start: u64, idx: u16) ![]XattrOwned {
const kv_start: u64 = std.mem.readInt(u64, @ptrCast(fil.map.memory[table_start .. table_start + 8]), .little);
const lookup = try LookupTable.lookupValue(TableValue, alloc, io, decomp, fil, table_start + 16, idx);
rdr = try fil.readerAt(io, kv_start + lookup.ref.block_start, &[0]u8{});
var rdr = fil.readerAt(kv_start + lookup.ref.block_start);
var meta: MetadataReader = .init(alloc, &rdr.interface, decomp);
try meta.interface.discardAll(lookup.ref.block_offset);
@@ -252,7 +258,7 @@ pub fn statelessLookup(alloc: std.mem.Allocator, io: Io, decomp: *const Decompre
const value: ValueOutOfLineEntry = undefined;
try meta.interface.readSliceEndian(ValueOutOfLineEntry, @ptrCast(&value), .little);
var ool_rdr = try fil.readerAt(io, kv_start + value.ref.block_start, &[0]u8{});
var ool_rdr = fil.readerAt(kv_start + value.ref.block_start);
var ool_meta: MetadataReader = .init(alloc, &ool_rdr.interface, decomp);
try ool_meta.interface.discardAll(value.ref.block_offset);