19 Commits

Author SHA1 Message Date
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
Caleb Gardner 69ce562b6c Re-doing extraction with learning about Io 2026-05-17 12:32:58 -05:00
Caleb Gardner 10e9b66ac6 Further work on extraction 2026-05-16 23:41:28 -05:00
Caleb Gardner 3c57a2d1e4 Cleanup & fixes 2026-05-16 06:14:08 -05:00
Caleb Gardner 700993b0e3 Finished up some errors
Kinda finished extraction
2026-05-13 06:21:01 -05:00
Caleb Gardner 78d1ee2937 Started concrete implementation of extraction 2026-05-13 01:12:02 -05:00
Caleb Gardner 2b0625e178 Finished (?) Data extractor 2026-05-12 12:22:48 -05:00
Caleb J. Gardner ad05e5dff1 Added xattr function to inode 2026-05-10 14:34:05 -05:00
Caleb J. Gardner 688ca53206 Finished (?) Xattr Cached table 2026-05-10 14:23:56 -05:00
Caleb Gardner 93a55aa5c7 Started work on xattr decoding
Added more utility to Inode's
2026-05-10 12:31:40 -05:00
Caleb Gardner d76b164e45 More work on extraction, especially for regular files 2026-05-08 06:06:33 -05:00
Caleb Gardner 5521b2ce6a Started working on file extraction 2026-05-03 05:31:04 -05:00
Caleb Gardner cbd2697c19 File openning
Start on data extraction
2026-05-03 04:40:49 -05:00
Caleb Gardner a3f7b86e67 Updated a lot of packed structs to extern struct
Specified int types for remaining packed structs
Instead of manually decoding File & ExtFile structs, decode an extern struct first
Fixed some zstd issues
Some more File stuff
2026-05-02 06:10:24 -05:00
Caleb Gardner ab606bdfa5 Fished decompression (maybe)
Added the necessary skeleton functions to re-add test
2026-05-01 07:05:52 -05:00
Caleb Gardner 274d088490 Further work on getting everything working again
Mainly working on decompression interface
2026-04-30 07:00:46 -05:00
Caleb Gardner b67d02074d Started re-write (once again)
Main reason for this re-write is to be compatible with zig 0.16.0
2026-04-29 03:48:34 -05:00
Caleb J. Gardner 4b2b7021c7 Moved & organized decompression
Fully implemented Decompressor vtable
2026-04-02 06:27:34 -05:00
41 changed files with 2127 additions and 2376 deletions
+1
View File
@@ -2,3 +2,4 @@ testing/
.zig-cache/
zig-out/
zig-pkg/
+1 -1
View File
@@ -11,7 +11,7 @@
"build": {
"command": "zig",
"args": ["build", "-Duse_c_libs=true", "-Ddebug=true"],
"args": ["build", "-Ddebug=true"],
},
"program": "zig-out/bin/unsquashfs",
+31 -68
View File
@@ -1,54 +1,28 @@
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 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.") orelse false;
const 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);
// 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(.{});
const mod = b.addModule("zig_squashfs", .{
.root_source_file = b.path("src/root.zig"),
.target = target,
.optimize = if (debug == true) .Debug else optimize,
.link_libc = !use_zig_decomp,
.valgrind = debug,
.error_tracing = debug,
.strip = if (debug == true) false else null,
const lib = b.addLibrary(.{
.name = "squashfs",
.root_module = b.createModule(.{
.optimize = if (debug == true) .Debug else optimize,
.target = target,
.valgrind = debug,
.root_source_file = b.path("src/root.zig"),
}),
.use_llvm = debug,
});
mod.addOptions("config", zig_squashfs_options);
if (!use_zig_decomp) {
var zlib_ng = b.dependency("zlib_ng", .{
.target = target,
.optimize = optimize,
});
mod.linkLibrary(zlib_ng.artifact("zng"));
mod.linkSystemLibrary("lzma", .{ .preferred_link_mode = .static });
var minilzo = b.dependency("minilzo", .{
.target = target,
.optimize = optimize,
});
mod.linkLibrary(minilzo.artifact("minilzo"));
var lz4 = b.dependency("lz4", .{
.target = target,
.optimize = optimize,
});
mod.linkLibrary(lz4.artifact("lz4"));
var zstd = b.dependency("zstd", .{
.target = target,
.optimize = optimize,
});
mod.linkLibrary(zstd.artifact("zstd"));
}
var version = version_string_option orelse "0.0.0-testing";
if (version[0] == 'v') version = version[1..];
@@ -58,41 +32,30 @@ pub fn build(b: *std.Build) !void {
"version",
try std.SemanticVersion.parse(version),
);
var exe_mod = b.createModule(.{
.root_source_file = b.path("src/bin/unsquashfs.zig"),
.target = target,
.optimize = if (debug == true) .Debug else optimize,
.link_libc = !use_zig_decomp,
.imports = &.{
.{ .name = "zig_squashfs", .module = mod },
},
.valgrind = debug,
.error_tracing = debug,
.strip = if (debug == true) false else null,
});
exe_mod.addOptions("config", unsquashfs_options);
const exe = b.addExecutable(.{
.name = "unsquashfs",
.root_module = exe_mod,
.use_llvm = debug,
});
const lib = b.addLibrary(.{
.name = "squashfs",
.root_module = mod,
.root_module = b.createModule(.{
.optimize = if (debug == true) .Debug else optimize,
.target = target,
.valgrind = debug,
.root_source_file = b.path("src/bin/unsquashfs.zig"),
.imports = &.{
.{ .name = "zig_squashfs", .module = lib.root_module },
},
}),
.use_llvm = debug,
});
exe.root_module.addOptions("config", unsquashfs_options);
b.installArtifact(lib);
b.installArtifact(exe);
const mod_tests = b.addTest(.{
.root_module = mod,
.test_runner = .{
.mode = .simple,
.path = b.path("src/test.zig"),
},
.root_module = b.createModule(.{
.optimize = optimize,
.target = target,
.root_source_file = b.path("src/test.zig"),
}),
});
const run_mod_tests = b.addRunArtifact(mod_tests);
const test_step = b.step("test", "Run tests");
@@ -101,11 +64,11 @@ pub fn build(b: *std.Build) !void {
// zls build check steps
const lib_check = b.addLibrary(.{
.name = "squashfs",
.root_module = mod,
.root_module = exe.root_module,
});
const exe_check = b.addExecutable(.{
.name = "unsquashfs",
.root_module = exe_mod,
.root_module = lib.root_module,
});
const check = b.step("check", "Check if unsquashfs compiles");
check.dependOn(&lib_check.step);
+155 -98
View File
@@ -1,121 +1,178 @@
//! A squashfs archive read from a file.
//! Can be used to directly access File's contents or extract to the filesystem.
const std = @import("std");
const File = std.fs.File;
const builtin = @import("builtin");
const Io = std.Io;
const config = @import("config");
const cDecomp = @import("decomp/misc_c.zig");
const Decomp = @import("decomp/zig_decomp.zig");
const ExtractionOptions = @import("options.zig");
const File = @import("file.zig");
const Inode = @import("inode.zig");
const InodeRef = Inode.Ref;
const BlockSize = @import("inode_data/file.zig").BlockSize;
const SfsFile = @import("file.zig");
const Superblock = @import("super.zig").Superblock;
const Tables = @import("tables.zig");
const LookupTable = @import("lookup_table.zig");
const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
const Utils = @import("util/misc.zig");
const OffsetFile = @import("util/offset_file.zig");
const XattrTable = @import("xattr.zig");
const Archive = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
file: OffsetFile,
super: Superblock,
tables: ?Tables = null,
stateless_decomp: Decompressor,
decomp: if (config.use_zig_decomp) Decomp.DecompFn else union(enum) {
gzip: @import("decomp/gzip.zig"),
lzma: @import("decomp/lzma.zig"),
lzo: void,
xz: @import("decomp/lzma.zig"),
lz4: void,
zstd: @import("decomp/zstd.zig"),
pub fn decomp(self: @This(), in: []u8, out: []u8) !usize {
return switch (self) {
.gzip => |*g| g.decompress(in, out),
.zstd => |*z| z.decompress(in, out),
.lzma, .xz => |*d| d.decompress(in, out),
.lz4 => cDecomp.lz4(in, out),
.lzo => cDecomp.lzo(in, out),
};
}
},
/// Default settings using std.Thread.getCpuCount() threads and the minimum of 4gb or half of system memory for memory usage.
pub fn init(alloc: std.mem.Allocator, fil: File, offset: u64) !Archive {
pub fn init(io: Io, file: std.Io.File, offset: u64) !Archive {
var rdr = file.reader(io, &[0]u8{});
try rdr.seekTo(offset);
var super: Superblock = undefined;
const red = try fil.pread(@ptrCast(&super), offset);
std.debug.assert(red == @sizeOf(Superblock));
try super.validate();
const off_fil: OffsetFile = .init(fil, offset);
try rdr.interface.readSliceEndian(Superblock, @ptrCast(&super), .little);
return .{
.alloc = alloc,
.fil = off_fil,
.decomp = if (config.use_zig_decomp)
switch (super.compression) {
.lz4 => return error.Lz4Unsupported,
.lzo => return error.LzoUnsupported,
.gzip => Decomp.gzip,
.lzma => Decomp.lzma,
.xz => Decomp.xz,
.zstd => Decomp.zstd,
}
else switch (super.compression) {
.gzip => .{ .gzip = .init(alloc) },
.zstd => .{ .zstd = .init(alloc) },
.xz => .{ .xz = .init(alloc, true) },
.lzma => .{ .lzma = .init(alloc, false) },
.lzo => .{ .lzo = .{} },
.lz4 => .{ .lz4 = .{} },
},
.file = .init(file, 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,
},
};
}
pub fn deinit(self: *Archive) void {
if (self.tables != null)
self.tables.?.deinit();
/// The root folder of the Archive. Used to open other Files.
pub fn root(self: Archive, alloc: std.mem.Allocator, io: Io) !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, "");
}
/// 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);
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);
}
pub fn inode(self: *Archive, alloc: std.mem.Allocator, num: u32) !Inode {
if (self.tables == null)
self.tables = try .init(alloc, self);
const ref = try self.export_table.get(num - 1);
var rdr = try self.fil.readerAt(ref.block_start + self.super.inode_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, &self.decomp);
try meta.interface.discardAll(ref.block_offset);
return try .read(alloc, &meta.interface, self.super.block_size);
/// 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 {
if (!self.super.flags.exportable)
return error.NotExportable;
const ref = try LookupTable.lookupValue(
Inode.Ref,
alloc,
io,
&self.stateless_decomp,
self.file,
self.super.export_start,
num + 1,
);
return Utils.inodeFromRef(
alloc,
io,
self.file,
&self.stateless_decomp,
self.super.inode_start,
self.super.block_size,
ref,
);
}
/// 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 {
return LookupTable.lookupValue(
u16,
alloc,
io,
&self.stateless_decomp,
self.file,
self.super.id_start,
idx,
);
}
pub fn root(self: *Archive, alloc: std.mem.Allocator) !SfsFile {
if (self.tables == null)
self.tables = try .init(alloc, self);
var rdr = try self.fil.readerAt(self.super.root_ref.block_start + self.super.inode_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, self.decomp);
try meta.interface.discardAll(self.super.root_ref.block_offset);
const in: Inode = try .read(alloc, &meta.interface, self.super.block_size);
return .init(self, in, "");
}
// Superblock
pub fn open(self: *Archive, alloc: std.mem.Allocator, path: []const u8) !SfsFile {
if (self.tables == null)
self.tables = try .init(alloc, self);
var root_fil = try self.root(alloc);
defer if (!SfsFile.pathIsSelf(path)) root_fil.deinit();
return root_fil.open(path);
}
const SQUASHFS_MAGIC: u32 = std.mem.readInt(u32, "hsqs", .little);
pub fn extract(self: Archive, alloc: std.mem.Allocator, path: []const u8, options: ExtractionOptions) !void {
var rdr = try self.fil.readerAt(self.super.root_ref.block_start + self.super.inode_start, &[0]u8{});
var meta: MetadataReader = .init(self.alloc, &rdr.interface, self.decomp);
try meta.interface.discardAll(self.super.root_ref.block_offset);
const in: Inode = try .read(self.alloc, &meta.interface, self.super.block_size);
try in.extractTo(alloc, self, path, options);
}
const SuperblockError = error{
InvalidMagic,
InvalidBlockLog,
InvalidVersion,
InvalidCheck,
};
/// A squashfs Superblock
pub const Superblock = extern struct {
magic: u32,
inode_count: u32,
mod_time: u32,
block_size: u32,
frag_count: u32,
compression: enum(u16) {
gzip = 1,
lzma,
lzo,
xz,
lz4,
zstd,
},
block_log: u16,
flags: packed struct(u16) {
inode_uncompressed: bool,
data_uncompressed: bool,
check: bool,
frag_uncompressed: bool,
fragment_never: bool,
fragment_always: bool,
duplicates: bool,
exportable: bool,
xattr_uncompressed: bool,
xattr_never: bool,
compression_options: bool,
ids_uncompressed: bool,
_: u4,
},
id_count: u16,
ver_maj: u16,
ver_min: u16,
root_ref: Inode.Ref,
size: u64,
id_start: u64,
xattr_start: u64,
inode_start: u64,
dir_start: u64,
frag_start: u64,
export_start: u64,
/// Validate the Superblock. If an error is returned, it's likely the archive is corrupted or not a squashfs archive.
pub fn validate(self: Superblock) !void {
if (self.magic != SQUASHFS_MAGIC)
return SuperblockError.InvalidMagic;
if (self.flags.check)
return SuperblockError.InvalidCheck;
if (self.ver_maj != 4 or self.ver_min != 0)
return SuperblockError.InvalidVersion;
if (std.math.log2(self.block_size) != self.block_log)
return SuperblockError.InvalidBlockLog;
}
};
+28 -21
View File
@@ -1,5 +1,6 @@
const std = @import("std");
const Writer = std.Io.Writer;
const Io = std.Io;
const Writer = Io.Writer;
const builtin = @import("builtin");
const config = @import("config");
@@ -17,7 +18,7 @@ const help_mgs =
\\ -dx Don't set xattr values
\\ -dp Don't set permissions (includes setting uid & gid owner)
\\
\\ -p <threads> Specify how many threads to use. If no present or zero, the system's logical cores count is used.
\\ -p <threads> Specify how many threads to use. If not present or zero, the system's logical cores count is used.
\\ -v Verbose
\\
\\ --force Force extraction. If the destination already exists, it will be deleted.
@@ -38,21 +39,26 @@ var ignore_xattrs: bool = false;
var ignore_permissions: bool = false;
var force: bool = false;
pub fn main() !void {
const alloc = std.heap.smp_allocator;
var stdout = std.fs.File.stdout();
var out = stdout.writer(&[0]u8{});
pub fn main(init: std.process.Init) !void {
const alloc = init.gpa;
const io = init.io;
var stdout = std.Io.File.stdout();
defer stdout.close(io);
var out = stdout.writer(io, &[0]u8{});
defer out.interface.flush() catch {};
try handleArgs(alloc, &out.interface);
try handleArgs(init.minimal.args, &out.interface);
if (archive.len == 0) {
try out.interface.print("You must provide a squashfs archive\n", .{});
try out.interface.print(help_mgs, .{});
return;
}
var fil: std.fs.File = try std.fs.cwd().openFile(archive, .{}); //TODO: Handle error gracefully.
defer fil.close();
var arc: squashfs.Archive = try .init(alloc, fil, offset); //TODO: Update when memory size matters. //TODO: Handle error gracefully.
defer arc.deinit();
var fil = try Io.Dir.cwd().openFile(io, archive, .{}); //TODO: Handle error gracefully.
defer fil.close(io);
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,
.verbose = verbose,
@@ -60,18 +66,19 @@ pub fn main() !void {
.ignore_xattr = ignore_xattrs,
.ignore_permissions = ignore_permissions,
};
if (force)
try std.fs.cwd().deleteTree(extLoc);
try arc.extract(alloc, extLoc, options); //TODO: Handle error gracefully.
try Io.Dir.cwd().deleteTree(io, extLoc);
try arc.extract(alloc, io, extLoc, options); //TODO: Handle error gracefully.
}
fn handleArgs(alloc: std.mem.Allocator, out: *Writer) !void {
var args = try std.process.argsWithAllocator(alloc);
defer args.deinit();
_ = args.next(); // args[0] is the application launch command.
while (args.next()) |arg| {
fn handleArgs(args: std.process.Args, out: *Writer) !void {
var arg_iter = args.iterate();
defer arg_iter.deinit();
_ = arg_iter.next(); // args[0] is the application launch command.
while (arg_iter.next()) |arg| {
if (std.mem.eql(u8, arg, "-o")) {
const nxt = args.next();
const nxt = arg_iter.next();
if (nxt == null or nxt.?.len == 0) {
try out.print("-o must be followed by a number\n", .{});
return errors.InvalidArguments;
@@ -82,7 +89,7 @@ fn handleArgs(alloc: std.mem.Allocator, out: *Writer) !void {
};
continue;
} else if (std.mem.eql(u8, arg, "-d")) {
const nxt = args.next();
const nxt = arg_iter.next();
if (nxt == null or nxt.?.len == 0) {
try out.print("-d must be followed by a location\n", .{});
return errors.InvalidArguments;
@@ -90,7 +97,7 @@ fn handleArgs(alloc: std.mem.Allocator, out: *Writer) !void {
extLoc = nxt.?;
continue;
} else if (std.mem.eql(u8, arg, "-p")) {
const nxt = args.next();
const nxt = arg_iter.next();
if (nxt == null or nxt.?.len == 0) {
try out.print("-p must be followed by a number\n", .{});
return errors.InvalidArguments;
+28 -21
View File
@@ -1,25 +1,32 @@
const std = @import("std");
const Decompressor = @This();
const Decompressor = @import("util/decompressor.zig");
pub const Error = error{
OutOfMemory,
BadInput,
OutputTooSmall,
ReadFailed,
WriteFailed,
EndOfStream,
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"),
pub fn deinit(self: *Decomp) void {
switch (self.*) {
.gzip => self.gzip.deinit(),
.lzma => self.lzma.deinit(),
.xz => self.xz.deinit(),
.zstd => self.zstd.deinit(),
else => unreachable,
}
}
pub fn decompressor(self: *Decomp) *Decompressor {
return switch (self.*) {
.gzip => &self.gzip.interface,
.lzma => &self.lzma.interface,
.xz => &self.xz.interface,
.zstd => &self.zstd.interface,
else => unreachable,
};
}
};
vtable: *const struct {
decompress: *const fn (*Decompressor, []u8, []u8) Error!usize = DefaultDecompress,
stateless: *const fn (std.mem.Allocator, []u8, []u8) Error!usize,
},
pub fn decompress(self: *Decompressor, in: []u8, out: []u8) Error!usize {
return self.vtable.decompress(self, in, out);
}
fn DefaultDecompress(self: *Decompressor, in: []u8, out: []u8) Error!usize {
return self.vtable.stateless(std.heap.smp_allocator, in, out);
}
-14
View File
@@ -1,14 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const builtin = @import("builtin");
const config = @import("config");
pub const c = @cImport({
@cInclude("zlib-ng.h");
@cInclude("lzma.h");
@cInclude("lz4.h");
@cInclude("zstd.h");
@cInclude("zstd_errors.h");
@cInclude("lzo/minilzo.h");
});
-114
View File
@@ -1,114 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const builtin = @import("builtin");
const Decompressor = @import("../decomp.zig");
const c = @import("c.zig").c;
const zng_stream = c.zng_stream;
const ZlibErrors = error{
OutOfMemory,
OutputBufferTooSmall,
BadData,
StreamError,
Unknown,
};
const Self = @This();
alloc: std.mem.Allocator,
streams: std.AutoHashMap(std.Thread.Id, zng_stream),
interface: Decompressor = .{ .vtable = &.{
.decompress = decompress,
.stateless = stateless,
} },
err: ?ZlibErrors = null,
pub fn init(alloc: std.mem.Allocator) !Self {
return .{
.alloc = alloc,
.streams = .init(alloc),
};
}
pub fn deinit(self: Self) void {
var iter = self.streams.keyIterator();
while (iter.next()) |key| {
_ = c.inflateEnd(self.streams.getPtr(key).?);
}
self.streams.deinit(self.alloc);
}
fn decompress(decomp: *Decompressor, in: []u8, out: []u8) Decompressor.Error!usize {
var self: *Self = @fieldParentPtr("interface", decomp);
var stream = try self.getOrCreate();
stream.next_in = in.ptr;
stream.avail_in = @truncate(in.len);
stream.next_out = out.ptr;
stream.avail_out = @truncate(out.len);
var res = c.zng_inflateReset(stream);
switch (res) {
c.Z_OK => {},
c.Z_STREAM_ERROR => {
self.err = ZlibErrors.StreamError;
return Decompressor.Error.BadInput;
},
else => {
self.err = ZlibErrors.Unknown;
return Decompressor.Error.BadInput;
},
}
res = c.zng_inflate(stream, c.Z_FINISH);
switch (res) {
c.Z_OK => return stream.total_out,
c.Z_MEM_ERROR => {
self.err = ZlibErrors.OutOfMemory;
return Decompressor.Error.OutOfMemory;
},
c.Z_BUF_ERROR => {
self.err = ZlibErrors.OutputBufferTooSmall;
return Decompressor.Error.OutputTooSmall;
},
c.Z_DATA_ERROR => {
self.err = ZlibErrors.BadData;
return Decompressor.Error.BadInput;
},
else => {
self.err = ZlibErrors.Unknown;
return Decompressor.Error.BadInput;
},
}
}
inline fn getOrCreate(self: *Self) Decompressor.Error!*zng_stream {
const res = try self.streams.getOrPut(std.Thread.getCurrentId());
if (res.found_existing) return res.value_ptr;
res.value_ptr.* = .{
.zalloc = zalloc,
.zfree = zfree,
.@"opaque" = @ptrCast(self),
};
return res.value_ptr;
}
fn zalloc(self_ptr: ?*anyopaque, items: c_uint, size: c_uint) callconv(.c) ?*anyopaque {
var self: *Self = @ptrCast(@alignCast(self_ptr));
return self.alloc.rawAlloc(items * size, .@"1", 0);
}
fn zfree(self_ptr: ?*anyopaque, alloc_ptr: ?*anyopaque) callconv(.c) void {
var self: *Self = @ptrCast(@alignCast(self_ptr));
self.alloc.rawFree(@ptrCast(alloc_ptr), .@"1", 0);
}
fn stateless(_: std.mem.Allocator, in: []u8, out: []u8) Decompressor.Error!usize {
var out_len = out.len;
const res = c.zng_uncompress(out.ptr, &out_len, in.ptr, in.len);
return switch (res) {
c.Z_OK => out_len,
c.Z_MEM_ERROR => Decompressor.Error.OutOfMemory,
c.Z_BUF_ERROR => Decompressor.Error.OutputTooSmall,
c.Z_DATA_ERROR => Decompressor.Error.BadInput,
else => Decompressor.Error.BadInput,
};
}
+64 -124
View File
@@ -1,146 +1,86 @@
const std = @import("std");
const builtin = @import("builtin");
const Reader = std.Io.Reader;
const lzma = std.compress.lzma;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../decomp.zig");
const c = @import("c.zig").c;
const stream = c.lzma_stream;
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,
streams: std.AutoHashMap(std.Thread.Id, stream),
xz: bool,
interface: Decompressor = .{ .vtable = &.{
.decompress = decompress,
.stateless = stateless,
} },
block_size: u32,
buffers: std.ArrayList(Buffer),
buffer_queue: std.SinglyLinkedList = .{},
err: ?LzmaError = null,
pub fn init(alloc: std.mem.Allocator, xz: bool) !Self {
pub fn init(alloc: std.mem.Allocator, block_size: u32) !Self {
return .{
.alloc = alloc,
.streams = .init(alloc),
.xz = xz,
.block_size = block_size,
.buffers = try .initCapacity(alloc, 5),
};
}
pub fn deinit(self: Self) void {
var iter = self.streams.keyIterator();
while (iter.next()) |key|
c.lzma_end(self.streams.getPtr(key));
self.streams.deinit();
pub fn deinit(self: *Self) void {
for (self.buffers.items) |buf|
self.alloc.free(buf.buf);
self.buffers.deinit(self.alloc);
}
pub fn decompress(decomp: *Decompressor, in: []u8, out: []u8) Decompressor.Error!usize {
var self: *Self = @fieldParentPtr("interface", decomp);
var strm = try self.getOrCreate();
strm.next_in = in.ptr;
strm.avail_in = in.len;
strm.next_out = out.ptr;
strm.avail_out = out.len;
var res = if (self.xz)
c.lzma_stream_decoder(strm, in.len * 2, 0)
else
c.lzma_alone_decoder(strm, in.len * 2);
switch (res) {
c.LZMA_OK => {},
c.LZMA_MEM_ERROR => {
self.err = LzmaError.LzmaMemoryError;
return Decompressor.Error.OutOfMemory;
},
c.LZMA_PROG_ERROR => {
self.err = LzmaError.LzmaProgramError;
return Decompressor.Error.BadInput;
},
else => {
self.err = LzmaError.Unknown;
return Decompressor.Error.BadInput;
},
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,
};
}
while (res == c.LZMA_OK)
res = c.lzma_code(strm, c.LZMA_RUN);
switch (res) {
c.LZMA_STREAM_END => return strm.total_out,
c.LZMA_MEM_ERROR => {
self.err = LzmaError.LzmaMemoryError;
return Decompressor.Error.OutOfMemory;
},
c.LZMA_MEMLIMIT_ERROR => {
self.err = LzmaError.LzmaMemoryLimit;
return Decompressor.Error.OutOfMemory;
},
c.LZMA_FORMAT_ERROR => {
self.err = LzmaError.LzmaBadFormat;
return Decompressor.Error.BadInput;
},
c.LZMA_DATA_ERROR => {
self.err = LzmaError.LzmaDataCorrupt;
return Decompressor.Error.BadInput;
},
c.LZMA_BUF_ERROR => {
self.err = LzmaError.LzmaCannotProgress;
return Decompressor.Error.BadInput;
},
c.LZMA_PROG_ERROR => {
self.err = LzmaError.LzmaProgramError;
return Decompressor.Error.BadInput;
},
else => {
self.err = LzmaError.Unknown;
return Decompressor.Error.BadInput;
},
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.?);
}
}
pub fn stateless(_: std.mem.Allocator, in: []u8, out: []u8) Decompressor.Error!usize {
var strm = c.lzma_stream{
.next_in = in.ptr,
.avail_in = in.len,
.next_out = out.ptr,
.avail_out = out.len,
};
var res = c.lzma_auto_decoder(&strm, out.len * 2, 0);
switch (res) {
c.LZMA_OK => {},
c.LZMA_MEM_ERROR => return Decompressor.Error.OutOfMemory,
c.LZMA_PROG_ERROR => return Decompressor.Error.BadInput,
else => return Decompressor.Error.BadInput,
}
while (res == c.LZMA_OK)
res = c.lzma_code(&strm, c.LZMA_RUN);
return switch (res) {
c.LZMA_STREAM_END => strm.total_out,
c.LZMA_MEM_ERROR => Decompressor.Error.OutOfMemory,
c.LZMA_MEMLIMIT_ERROR => Decompressor.Error.OutOfMemory,
c.LZMA_FORMAT_ERROR => Decompressor.Error.BadInput,
c.LZMA_DATA_ERROR => Decompressor.Error.BadInput,
c.LZMA_BUF_ERROR => Decompressor.Error.BadInput,
c.LZMA_PROG_ERROR => Decompressor.Error.BadInput,
else => Decompressor.Error.BadInput,
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 getOrCreate(self: *Self) Decompressor.Error!*stream {
const res = try self.streams.getOrPut(std.Thread.getCurrentId());
if (res.found_existing) return res.value_ptr;
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();
}
// Ideally, the zero value should be LZMA_STREAM_INIT, but translate-c can't handle it properly.
// According to lzma, setting it to entirely zero values *should* work.
// res.value_ptr.* = c.LZMA_STREAM_INIT;
res.value_ptr.* = std.mem.zeroInit(stream, .{});
return res.value_ptr;
return d.reader.readSliceShort(out);
}
pub const LzmaError = error{
OutOfMemory,
LzmaMemoryError,
LzmaMemoryLimit,
LzmaBadFormat,
LzmaDataCorrupt,
LzmaCannotProgress,
LzmaProgramError,
Unknown,
};
// 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;
}
-43
View File
@@ -1,43 +0,0 @@
const std = @import("std");
const Decompressor = @import("../decomp.zig");
const c = @import("c.zig").c;
pub const Lzo = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = lzo } },
};
fn lzo(_: std.mem.Allocator, in: []u8, out: []u8) Decompressor.Error!usize {
var res = c.lzo_init();
if (res != 0) return Decompressor.Error.BadInput;
var out_len: usize = out.len;
res = c.lzo1x_decompress(in.ptr, in.len, out.ptr, &out_len, null);
return switch (res) {
c.LZO_E_OK => out_len,
c.LZO_E_OUT_OF_MEMORY => Decompressor.Error.OutOfMemory,
c.LZO_E_ERROR,
c.LZO_E_INPUT_OVERRUN,
c.LZO_E_LOOKBEHIND_OVERRUN,
c.LZO_E_EOF_NOT_FOUND,
c.LZO_E_NOT_YET_IMPLEMENTED,
c.LZO_E_INVALID_ARGUMENT,
c.LZO_E_INVALID_ALIGNMENT,
=> Decompressor.Error.BadInput,
c.LZO_E_INPUT_NOT_CONSUMED,
c.LZO_E_OUTPUT_NOT_CONSUMED,
c.LZO_E_OUTPUT_OVERRUN,
=> Decompressor.Error.OutputTooSmall,
else => Decompressor.Error.BadInput,
};
}
pub const Lz4 = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = lz4 } },
};
fn lz4(_: std.mem.Allocator, in: []u8, out: []u8) Decompressor.Error!usize {
const res = c.LZ4_decompress_safe(in.ptr, out.ptr, @intCast(in.len), @intCast(out.len));
if (res > 0) return @abs(res);
return Decompressor.Error.BadInput; // TODO: Find out what error values it can return.
}
+80
View File
@@ -0,0 +1,80 @@
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;
}
-51
View File
@@ -1,51 +0,0 @@
const std = @import("std");
const Reader = std.Io.Reader;
const builtin = @import("builtin");
const Decompressor = @import("../decomp.zig");
pub const Gzip = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = gzip } },
};
pub fn gzip(alloc: std.mem.Allocator, in: []u8, out: []u8) anyerror!usize {
var rdr: Reader = .fixed(in);
const buf = try alloc.alloc(u8, out.len);
defer alloc.free(buf);
var decomp = std.compress.flate.Decompress.init(&rdr, .zlib, buf);
return decomp.reader.readSliceShort(out);
}
pub const Lzma = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = lzma } },
};
pub fn lzma(alloc: std.mem.Allocator, in: []u8, out: []u8) anyerror!usize {
var rdr: Reader = .fixed(in);
var decomp = try std.compress.lzma.decompress(alloc, rdr.adaptToOldInterface());
return decomp.read(out);
}
pub const Xz = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = xz } },
};
pub fn xz(alloc: std.mem.Allocator, in: []u8, out: []u8) anyerror!usize {
var rdr: Reader = .fixed(in);
var decomp = try std.compress.xz.decompress(alloc, rdr.adaptToOldInterface());
return decomp.read(out);
}
pub const Zstd = struct {
interface: Decompressor = .{ .vtable = &.{ .stateless = zstd } },
};
pub fn zstd(alloc: std.mem.Allocator, in: []u8, out: []u8) anyerror!usize {
var rdr: Reader = .fixed(in);
const buf = try alloc.alloc(u8, 1024 * 1024);
defer alloc.free(buf);
var decomp = std.compress.zstd.Decompress.init(&rdr, buf, .{});
return decomp.reader.readSliceShort(out) catch |err| {
return decomp.err orelse err;
};
}
+73
View File
@@ -0,0 +1,73 @@
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);
}
+54 -196
View File
@@ -1,216 +1,74 @@
const std = @import("std");
const builtin = @import("builtin");
const Io = std.Io;
const Reader = std.Io.Reader;
const zstd = std.compress.zstd;
const Node = std.SinglyLinkedList.Node;
const Decompressor = @import("../decomp.zig");
const c = @import("c.zig").c;
const DCtx = c.ZSTD_DCtx;
const Decompressor = @import("../util/decompressor.zig");
const Error = Decompressor.Error;
const Queue = std.Io.Queue([]u8);
const Self = @This();
interface: Decompressor = .{ .decomp_fn = decomp },
alloc: std.mem.Allocator,
ctx: std.AutoHashMap(std.Thread.Id, *DCtx),
io: Io,
interface: Decompressor = .{ .vtable = &.{
.decompress = decompress,
.stateless = stateless,
} },
err: ?ZstdError = null,
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 + zstd.block_size_max));
pub fn init(alloc: std.mem.Allocator) !Self {
return .{
.alloc = alloc,
.ctx = .init(alloc),
.io = io,
.block_size = block_size,
.buf = buf,
.buf_queue = queue,
};
}
pub fn deinit(self: Self) void {
var iter = self.ctx.keyIterator();
while (iter.next()) |key| {
_ = c.ZSTD_freeDCtx(self.ctx.getPtr(key));
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: ?*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 zstdDecomp(buf, in, out);
}
self.ctx.deinit(self.alloc);
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 zstdDecomp(buf, in, out);
}
pub fn decompress(decomp: *Decompressor, in: []u8, out: []u8) Decompressor.Error!usize {
var self: *Self = @fieldParentPtr("interface", decomp);
inline fn zstdDecomp(buffer: []u8, in: []u8, out: []u8) !usize {
var rdr: Reader = .fixed(in);
var d = zstd.Decompress.init(&rdr, buffer, .{ .window_len = @truncate(out.len) });
const ctx = try self.getOrCreate();
const res = c.ZSTD_decompressDCtx(ctx, out.ptr, out.len, in.ptr, in.len);
try self.checkError(res);
return res;
}
inline fn getOrCreate(self: *Self) Decompressor.Error!*DCtx {
const res = try self.ctx.getOrPut(std.Thread.getCurrentId());
if (res.found_existing) {
try self.checkError(c.ZSTD_DCtx_reset(res.value_ptr.*, c.ZSTD_reset_session_only));
return res.value_ptr.*;
}
res.value_ptr.* = c.ZSTD_createDCtx() orelse return Decompressor.Error.OutOfMemory;
return res.value_ptr.*;
return d.reader.readSliceShort(out);
}
fn stateless(_: std.mem.Allocator, in: []u8, out: []u8) Decompressor.Error!usize {
const res = c.ZSTD_decompress(out.ptr, out.len, in.ptr, in.len);
if (c.ZSTD_isError(res) == 0) return res;
return switch (c.ZSTD_getErrorCode(res)) {
c.ZSTD_error_memory_allocation => Decompressor.Error.OutOfMemory,
c.ZSTD_error_workSpace_tooSmall,
c.ZSTD_error_dstSize_tooSmall,
c.ZSTD_error_dstBuffer_null,
c.ZSTD_error_noForwardProgress_destFull,
=> Decompressor.Error.OutputTooSmall,
else => Decompressor.Error.BadInput,
};
}
// Stateless
inline fn checkError(self: *Self, res: usize) Decompressor.Error!void {
if (res == 0) return;
if (c.ZSTD_isError(res) == 0) return;
switch (c.ZSTD_getErrorCode(res)) {
c.ZSTD_error_prefix_unknown => {
self.err = ZstdError.PrefixUnknown;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_version_unsupported => {
self.err = ZstdError.VersionUnsupported;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_frameParameter_unsupported => {
self.err = ZstdError.FrameParameterUnsupported;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_frameParameter_windowTooLarge => {
self.err = ZstdError.FrameParameterWindowTooLarge;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_corruption_detected => {
self.err = ZstdError.CorruptionDetected;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_checksum_wrong => {
self.err = ZstdError.ChecksumWrong;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_literals_headerWrong => {
self.err = ZstdError.LiteralsHeaderWrong;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_dictionary_corrupted => {
self.err = ZstdError.DictionaryCorrupted;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_dictionary_wrong => {
self.err = ZstdError.DictionaryWrong;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_dictionaryCreation_failed => {
self.err = ZstdError.DictionaryCreationFailed;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_parameter_unsupported => {
self.err = ZstdError.ParameterUnsupported;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_parameter_combination_unsupported => {
self.err = ZstdError.ParameterCombinationUnsupported;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_parameter_outOfBound => {
self.err = ZstdError.ParameterOutOfBound;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_tableLog_tooLarge => {
self.err = ZstdError.TableLogTooLarge;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_maxSymbolValue_tooLarge => {
self.err = ZstdError.MaxSymbolValueTooLarge;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_maxSymbolValue_tooSmall => {
self.err = ZstdError.MaxSymbolValueTooSmall;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_stabilityCondition_notRespected => {
self.err = ZstdError.StabilityConditionNotRespected;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_stage_wrong => {
self.err = ZstdError.StageWrong;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_init_missing => {
self.err = ZstdError.InitMissing;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_memory_allocation => {
self.err = ZstdError.MemoryAllocation;
return Decompressor.Error.OutOfMemory;
},
c.ZSTD_error_workSpace_tooSmall => {
self.err = ZstdError.WorkSpaceTooSmall;
return Decompressor.Error.OutputTooSmall;
},
c.ZSTD_error_dstSize_tooSmall => {
self.err = ZstdError.DstSizeTooSmall;
return Decompressor.Error.OutputTooSmall;
},
c.ZSTD_error_srcSize_wrong => {
self.err = ZstdError.SrcSizeWrong;
return Decompressor.Error.BadInput;
},
c.ZSTD_error_dstBuffer_null => {
self.err = ZstdError.DstBufferNull;
return Decompressor.Error.OutputTooSmall;
},
c.ZSTD_error_noForwardProgress_destFull => {
self.err = ZstdError.NoForwardProgressDestFull;
return Decompressor.Error.OutputTooSmall;
},
c.ZSTD_error_noForwardProgress_inputEmpty => {
self.err = ZstdError.NoForwardProgressInputEmpty;
return Decompressor.Error.BadInput;
},
else => {
self.err = ZstdError.Generic;
return Decompressor.Error.BadInput;
},
}
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, out.len + zstd.block_size_max);
defer alloc.free(buf);
return zstdDecomp(buf, in, out);
}
pub const ZstdError = error{
OutOfMemory,
Generic,
PrefixUnknown,
VersionUnsupported,
FrameParameterUnsupported,
FrameParameterWindowTooLarge,
CorruptionDetected,
ChecksumWrong,
LiteralsHeaderWrong,
DictionaryCorrupted,
DictionaryWrong,
DictionaryCreationFailed,
ParameterUnsupported,
ParameterCombinationUnsupported,
ParameterOutOfBound,
TableLogTooLarge,
MaxSymbolValueTooLarge,
MaxSymbolValueTooSmall,
CannotProduceUncompressedBlock,
StabilityConditionNotRespected,
StageWrong,
InitMissing,
MemoryAllocation,
WorkSpaceTooSmall,
DstSizeTooSmall,
SrcSizeWrong,
DstBufferNull,
NoForwardProgressDestFull,
NoForwardProgressInputEmpty,
FrameIndexTooLarge,
SeekableIo,
DstBufferWrong,
SrcBufferWrong,
SequenceProducerFailed,
ExternalSequencesInvalid,
MaxCode,
};
-60
View File
@@ -1,60 +0,0 @@
//! Directory entry from the directory table.
const std = @import("std");
const Reader = std.Io.Reader;
const InodeType = @import("inode.zig").InodeType;
const Entry = @This();
const Header = extern struct { // use extern due to bad alignment with packed.
count: u32,
block_start: u32,
num: u32,
};
const RawEntry = packed struct {
offset: u16,
inode_offset: i16,
inode_type: InodeType,
name_size: u16,
};
block_start: u32,
block_offset: u16,
num: u32,
inode_type: InodeType,
name: []const u8,
pub fn readDir(alloc: std.mem.Allocator, rdr: *Reader, size: u32) ![]Entry {
var cur_red: u32 = 3; // start at 3 due to "." & ".." being counted in the dir size.
var hdr: Header = undefined;
var raw: RawEntry = undefined;
var out: std.ArrayList(Entry) = try .initCapacity(alloc, 100); // Start out with a decent capacity instead of needing to allocate per header.
errdefer out.deinit(alloc);
while (cur_red < size) {
try rdr.readSliceEndian(Header, @ptrCast(&hdr), .little);
cur_red += @sizeOf(Header);
try out.ensureUnusedCapacity(alloc, hdr.count + 1);
for (0..hdr.count + 1) |_| {
try rdr.readSliceEndian(RawEntry, @ptrCast(&raw), .little);
const name = try alloc.alloc(u8, raw.name_size + 1);
errdefer alloc.free(name);
try rdr.readSliceEndian(u8, name, .little);
const val = out.addOneAssumeCapacity();
val.* = .{
.block_start = hdr.block_start,
.block_offset = raw.offset,
.num = @abs(hdr.num + raw.offset),
.inode_type = raw.inode_type,
.name = name,
};
cur_red += @sizeOf(RawEntry) + raw.name_size + 1;
}
}
return out.toOwnedSlice(alloc);
}
pub fn deinit(self: Entry, alloc: std.mem.Allocator) void {
alloc.free(self.name);
}
+69
View File
@@ -0,0 +1,69 @@
const std = @import("std");
const Reader = std.Io.Reader;
const Inode = @import("inode.zig");
pub const Error = error{OutOfMemory} || Reader.Error;
const DirEntry = @This();
block_start: u32,
block_offset: u16,
type: Inode.Type,
name: []const u8,
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) Error![]DirEntry {
var hdr: Header = undefined;
var raw: RawEntry = undefined;
var out: std.ArrayList(DirEntry) = try .initCapacity(alloc, 30);
errdefer {
for (out.items) |ent|
alloc.free(ent.name);
out.deinit(alloc);
}
var tot_red: u32 = 3;
while (tot_red < size) {
try rdr.readSliceEndian(Header, @ptrCast(&hdr), .little);
try out.ensureUnusedCapacity(alloc, hdr.count + 1);
tot_red += @sizeOf(Header);
for (0..hdr.count + 1) |_| {
try rdr.readSliceEndian(RawEntry, @ptrCast(&raw), .little);
const new_name = try alloc.alloc(u8, raw.name_size + 1);
try rdr.readSliceEndian(u8, new_name, .little);
const new = out.addOneAssumeCapacity();
new.* = .{
.block_start = hdr.block_start,
.block_offset = raw.block_offset,
.type = raw.type,
.name = new_name,
};
tot_red += @sizeOf(RawEntry) + raw.name_size + 1;
}
}
return out.toOwnedSlice(alloc);
}
// Types
const Header = extern struct {
count: u32,
block_start: u32,
num: u32,
};
const RawEntry = extern struct {
block_offset: u16,
num_offset: i16,
type: Inode.Type,
name_size: u16,
};
+64 -171
View File
@@ -1,202 +1,95 @@
//! A file/directory within the squashfs archive.
//! An easier to use wrapper around an inode.
const std = @import("std");
const File = std.fs.File;
const WaitGroup = std.Thread.WaitGroup;
const Mutex = std.Thread.Mutex;
const Io = std.Io;
const Archive = @import("archive.zig");
const DirEntry = @import("dir_entry.zig");
const DirEntry = @import("directory.zig");
const ExtractionOptions = @import("options.zig");
const Inode = @import("inode.zig");
const BlockSize = @import("inode_data/file.zig").BlockSize;
const DataReader = @import("util/data.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 FileError = error{
NotDirectory,
NotRegularFile,
NotSymlink,
NotDevice,
NotFound,
ExtractionPathExists,
};
const File = @This();
const SfsFile = @This();
alloc: std.mem.Allocator,
archive: *Archive,
archive: Archive,
inode: Inode,
name: []const u8,
/// Initialize a new File.
/// name is copied to the File so can be safely freed afterwards.
pub fn init(archive: *Archive, inode: Inode, name: []const u8) !SfsFile {
const new_name = try archive.allocator().alloc(u8, name.len);
/// Creates a new File from an inode. Takes ownership of the Inode and creates a copy of the given name.
/// Requires the given allocator was used to create the Inode.
pub fn init(alloc: std.mem.Allocator, archive: Archive, in: Inode, name: []const u8) !File {
const new_name = try alloc.alloc(u8, name.len);
@memcpy(new_name, name);
return .{
.alloc = alloc,
.archive = archive,
.inode = inode,
.inode = in,
.name = new_name,
};
}
pub fn fromEntry(archive: *Archive, entry: DirEntry) !SfsFile {
var rdr = try archive.fil.readerAt(entry.block_start + archive.super.inode_start, &[0]u8{});
var meta: MetadataReader = .init(archive.allocator(), &rdr.interface, archive.decomp);
try meta.interface.discardAll(entry.block_offset);
const inode: Inode = try .read(archive.allocator(), &meta.interface, archive.super.block_size);
errdefer inode.deinit(archive.allocator());
return .init(archive, inode, entry.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);
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);
}
pub fn deinit(self: File) void {
self.alloc.free(self.name);
self.inode.deinit(self.alloc);
}
pub fn deinit(self: SfsFile) void {
var alloc = self.archive.allocator();
alloc.free(self.name);
self.inode.deinit(alloc);
}
fn getEntries(self: SfsFile) ![]DirEntry {
return self.inode.dirEntries(self.archive.allocator(), self.archive.*);
}
pub fn ownerUid(self: SfsFile) !u16 {
return self.archive.id(self.inode.hdr.uid_idx);
}
pub fn ownerGid(self: SfsFile) !u16 {
return self.archive.id(self.inode.hdr.gid_idx);
}
pub fn permissions(self: SfsFile) u16 {
return self.inode.hdr.permissions;
}
pub fn isRegular(self: SfsFile) bool {
return switch (self.inode.hdr.inode_type) {
.file, .ext_file => true,
else => false,
};
}
/// The returned DataReader will no longer work if the File's deinit function is called
/// or, more specifically, it's inode's deinit function is called.
pub fn dataReader(self: SfsFile) !DataReader {
return self.inode.dataReader(self.archive);
}
pub fn isDir(self: SfsFile) bool {
return switch (self.inode.hdr.inode_type) {
.dir, .ext_dir => true,
else => false,
};
}
pub fn iterate(self: SfsFile) !Iterator {
if (!self.isDir()) return FileError.NotDirectory;
return .{
.entries = try self.getEntries(),
.archive = self.archive,
};
}
/// Open a sub-file/folder within a directory at the given path.
/// If path is "", ".", "/", or "./", this File is returned.
pub fn open(self: SfsFile, alloc: std.mem.Allocator, path: []const u8) !SfsFile {
if (!self.isDir()) return FileError.NotDirectory;
if (pathIsSelf(path)) return self;
// Recursively stip ending & leading path separators.
if (path[0] == '/') return self.open(path[1..]);
if (path[path.len - 1] == '/') return self.open(path[0 .. path.len - 1]);
const idx = std.mem.indexOf(u8, path, "/") orelse path.len;
const first_element = path[0..idx];
if (std.mem.eql(u8, first_element, ".")) return self.open(path[idx + 1 ..]);
const entries = try self.getEntries();
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,
);
defer {
for (entries) |e| {
e.deinit(alloc);
}
for (entries) |ent|
alloc.free(ent.name);
alloc.free(entries);
}
var cur_slice = entries;
var split = cur_slice.len / 2;
while (cur_slice.len > 0) {
split = cur_slice.len / 2;
const comp = std.mem.order(u8, first_element, cur_slice[split].name);
switch (comp) {
.eq => {
var fil: SfsFile = try .fromEntry(self.archive, cur_slice[split]);
if (idx == path.len) {
return fil;
}
defer fil.deinit();
return fil.open(alloc, path[idx + 1 ..]);
},
.lt => cur_slice = cur_slice[0..split],
.gt => cur_slice = cur_slice[split + 1 ..],
const path = std.mem.trim(u8, filepath, "/");
const first_element: []const u8 = std.mem.sliceTo(path, '/');
var search_slice = entries;
var idx: usize = undefined;
while (search_slice.len > 0) {
idx = search_slice.len / 2;
const middle = search_slice[idx];
switch (std.mem.order(u8, first_element, middle.name)) {
.eq => break,
.lt => search_slice = search_slice[0..idx],
.gt => search_slice = search_slice[idx + 1 ..],
}
}
return FileError.NotFound;
} else return Error.FileNotFound;
const first_elem_file = try fromDirEntry(alloc, io, self.archive, search_slice[idx]);
if (first_element.len == path.len)
return first_elem_file;
defer first_elem_file.deinit();
return first_elem_file.open(alloc, io, path[first_element.len + 1 ..]);
}
pub fn isSymlink(self: SfsFile) bool {
return switch (self.inode.hdr.inode_type) {
.symlink, .ext_symlink => true,
else => false,
};
}
pub fn symlinkPath(self: SfsFile) ![]const u8 {
if (!self.isSymlink()) return FileError.NotSymlink;
return switch (self.inode.data) {
.symlink => |s| s.target,
.ext_symlink => |s| s.target,
else => unreachable,
};
pub fn extract(self: File, alloc: std.mem.Allocator, io: Io, filepath: []const u8, options: ExtractionOptions) !void {
return self.inode.extract(alloc, io, self.archive.file, self.archive.super, filepath, options);
}
/// Check if the File is a block or character device.
pub fn isDevice(self: SfsFile) bool {
return switch (self.inode.hdr.inode_type) {
.block_dev, .char_dev, .ext_block_dev, .ext_char_dev => true,
else => false,
};
}
/// If the File is a block or character device, get's it's device number.
pub fn devNum(self: SfsFile) !u32 {
if (!self.isDevice()) return FileError.NotDevice;
return switch (self.inode.data) {
.block_dev, .char_dev => |d| d.dev,
.ext_block_dev, .ext_char_dev => |d| d.dev,
else => unreachable,
};
}
// Types
/// Extract the given File to the path. If File is a regular file, the path must be a directory or not exist.
/// If the gievn path is a folder, the File's contents will be extracted within.
pub fn extract(self: *SfsFile, alloc: std.mem.Allocator, path: []const u8, options: ExtractionOptions) !void {
return self.inode.extractToThreaded(alloc, self.archive, path, options);
}
/// Utility function.
pub fn pathIsSelf(path: []const u8) bool {
if (path.len == 0) return true;
if (path.len == 1 and (path[0] == '/' or path[0] == '.')) return true;
if (path.len == 2 and (path[0] == '.' and path[1] == '/')) return true;
return false;
}
pub const Iterator = struct {
entries: []DirEntry,
archive: *Archive,
idx: u32 = 0,
pub fn next(self: *Iterator) !?SfsFile {
if (self.idx >= self.entries.len) return null;
defer self.idx += 1;
return try SfsFile.fromEntry(self.archive, self.entries[self.idx]);
}
pub fn deinit(self: Iterator) void {
var alloc = self.archive.allocator();
for (self.entries) |e| {
e.deinit(alloc);
}
alloc.free(self.entries);
}
};
pub const Error = error{
FileNotFound,
} || Inode.Error;
+7
View File
@@ -0,0 +1,7 @@
const BlockSize = @import("inode_data/file.zig").BlockSize;
pub const FragEntry = extern struct {
start: u64,
size: BlockSize,
_: u32,
};
+455 -132
View File
@@ -2,77 +2,30 @@
const std = @import("std");
const Reader = std.Io.Reader;
const WaitGroup = std.Thread.WaitGroup;
const Pool = std.Thread.Pool;
const Mutex = std.Thread.Mutex;
const Io = std.Io;
const Archive = @import("archive.zig");
const DirEntry = @import("dir_entry.zig");
const Decomp = @import("decomp.zig").Decomp;
const DirEntry = @import("directory.zig");
const ExtractionOptions = @import("options.zig");
const FragEntry = @import("frag.zig").FragEntry;
const dir = @import("inode_data/dir.zig");
const file = @import("inode_data/file.zig");
const misc = @import("inode_data/misc.zig");
const Tables = @import("tables.zig");
const DataReader = @import("util/data.zig");
const ThreadedDataReader = @import("util/data_threaded.zig");
const InodeExtract = @import("util/extract.zig");
const InodeFinish = @import("util/inode_finish.zig");
const FinishUnion = InodeFinish.FinishUnion;
const LookupTable = @import("lookup_table.zig");
const CachedTable = LookupTable.CachedTable;
const DataExtractor = @import("util/data_extractor.zig");
const DataReader = @import("util/data_reader.zig");
const Decompressor = @import("util/decompressor.zig");
const MetadataReader = @import("util/metadata.zig");
pub const Ref = packed struct {
block_offset: u16,
block_start: u32,
_: u16,
};
pub const InodeType = enum(u16) {
dir = 1,
file,
symlink,
block_dev,
char_dev,
fifo,
socket,
ext_dir,
ext_file,
ext_symlink,
ext_block_dev,
ext_char_dev,
ext_fifo,
ext_socket,
};
pub const InodeData = union(InodeType) {
dir: dir.Dir,
file: file.File,
symlink: misc.Symlink,
block_dev: misc.Dev,
char_dev: misc.Dev,
fifo: misc.IPC,
socket: misc.IPC,
ext_dir: dir.ExtDir,
ext_file: file.ExtFile,
ext_symlink: misc.ExtSymlink,
ext_block_dev: misc.ExtDev,
ext_char_dev: misc.ExtDev,
ext_fifo: misc.ExtIPC,
ext_socket: misc.ExtIPC,
};
pub const Header = packed struct {
inode_type: InodeType,
permissions: u16,
uid_idx: u16,
gid_idx: u16,
mod_time: u32,
num: u32,
};
const OffsetFile = @import("util/offset_file.zig");
const SharedCache = @import("util/shared_cache.zig");
const XattrTable = @import("xattr_table.zig");
const Inode = @This();
hdr: Header,
data: InodeData,
data: Data,
pub fn read(alloc: std.mem.Allocator, rdr: *Reader, block_size: u32) !Inode {
var hdr: Header = undefined;
@@ -97,93 +50,463 @@ pub fn read(alloc: std.mem.Allocator, rdr: *Reader, block_size: u32) !Inode {
},
};
}
pub fn readFromEntry(alloc: std.mem.Allocator, archive: Archive, entry: DirEntry) !Inode {
var rdr = try archive.fil.readerAt(archive.super.inode_start + entry.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, archive.decomp);
try meta.interface.discardAll(entry.block_offset);
return read(alloc, &meta.interface, archive.super.block_size);
}
pub fn deinit(self: Inode, alloc: std.mem.Allocator) void {
switch (self.data) {
.file => |f| alloc.free(f.block_sizes),
.ext_file => |f| alloc.free(f.block_sizes),
.symlink => |s| alloc.free(s.target),
.ext_symlink => |s| alloc.free(s.target),
.file => |d| d.deinit(alloc),
.symlink => |d| d.deinit(alloc),
.ext_file => |d| d.deinit(alloc),
.ext_symlink => |d| d.deinit(alloc),
else => {},
}
}
/// Get the data reader for a file inode.
pub fn dataReader(self: Inode, alloc: std.mem.Allocator, archive: Archive, tables: *Tables) !DataReader {
return switch (self.hdr.inode_type) {
.file => readerFromData(alloc, archive, tables, self.data.file),
.ext_file => readerFromData(alloc, archive, tables, self.data.ext_file),
else => error.NotRegularFile,
};
}
fn readerFromData(alloc: std.mem.Allocator, archive: Archive, tables: *Tables, data: anytype) !DataReader {
var out: DataReader = .init(alloc, archive, data.block_sizes, data.block_start, data.size);
if (data.frag_idx != 0xFFFFFFFF)
out.addFragment(try tables.frag_table.get(data.frag_idx), data.frag_block_offset);
return out;
}
// Utility Functions
/// Get the directory entries for a directory inode.
pub fn dirEntries(self: Inode, alloc: std.mem.Allocator, archive: Archive) ![]DirEntry {
return switch (self.hdr.inode_type) {
.dir => entriesFromData(alloc, archive, self.data.dir),
.ext_dir => entriesFromData(alloc, archive, self.data.ext_dir),
else => error.NotDirectory,
};
}
fn entriesFromData(alloc: std.mem.Allocator, archive: Archive, data: anytype) ![]DirEntry {
var rdr = try archive.fil.readerAt(archive.super.dir_start + data.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, archive.decomp);
try meta.interface.discardAll(data.block_offset);
return DirEntry.readDir(alloc, &meta.interface, data.size);
}
/// Returns the xattr index for the given inode. If the inode isn't an extended variant or doesn't have any, the u32 max is returned (0xFFFFFFFF).
pub fn xattrIdx(self: Inode) u32 {
/// Read the directory entries
pub fn readDirectory(self: Inode, alloc: std.mem.Allocator, io: Io, fil: OffsetFile, decomp: *const Decompressor, dir_offset: u64) ![]DirEntry {
return switch (self.data) {
.ext_dir => |d| d.xattr_id,
.ext_file => |f| f.xattr_idx,
.ext_symlink => |s| s.xattr_idx,
.ext_block_dev, .ext_char_dev => |d| d.xattr_idx,
.ext_fifo, .ext_socket => |i| i.xattr_idx,
else => 0xFFFFFFFF,
.dir => |d| readDirFromData(alloc, io, fil, decomp, dir_offset, d),
.ext_dir => |d| readDirFromData(alloc, io, 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);
try meta.interface.discardAll(d.block_offset);
/// Applies the Inode's metadata to the given File.
/// Mod time is always set, but permissions and xattrs are set based on the given ExtractionOptions.
pub fn setMetadata(self: Inode, alloc: std.mem.Allocator, tables: *Tables, fil: std.fs.File, options: ExtractionOptions) !void {
const time = @as(i128, self.hdr.mod_time) * 1000000000;
try fil.updateTimes(time, time);
if (!options.ignore_permissions) {
try fil.chmod(self.hdr.permissions);
try fil.chown(try tables.id_table.get(self.hdr.uid_idx), try tables.id_table.get(self.hdr.gid_idx));
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 {
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),
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);
if (d.frag_block_offset == 0xFFFFFFFF) {
// TODO:
return error.TODO;
}
if (!options.ignore_xattr) {
const idx = self.xattrIdx();
if (idx == 0xFFFFFFFF) return;
const xattrs = try tables.xattr_table.get(alloc, idx);
defer alloc.free(xattrs);
for (xattrs) |kv| {
const res = std.os.linux.fsetxattr(fil.handle, kv.key, kv.value.ptr, kv.value.len, 0);
alloc.free(kv.key);
alloc.free(kv.value);
if (res != 0) {
if (options.verbose)
options.verbose_writer.?.print("fsetxattr has result of: {}\n", .{res}) catch {};
return error.SetXattr;
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 {
return switch (self.data) {
.file => |f| getExtractorFromData(fil, cache, decomp, block_size, f),
.ext_file => |f| getExtractorFromData(fil, cache, decomp, block_size, 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);
if (d.frag_block_offset == 0xFFFFFFFF) {
// TODO:
return error.TODO;
}
return ext;
}
/// Get a symlink's target path
pub fn symlinkTarget(self: Inode) ![]const u8 {
return switch (self.data) {
.symlink => |s| s.target,
.ext_symlink => |s| s.target,
else => Error.NotSymlink,
};
}
/// 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 {
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 {
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.
/// Returns error.NoXattr if the inode doesn't have extended attributes.
pub fn xattrIndex(self: Inode) !u32 {
const idx = switch (self.data) {
.ext_dir => |e| e.xattr_idx,
.ext_file => |e| e.xattr_idx,
.ext_symlink => |e| e.xattr_idx,
.ext_block_dev => |e| e.xattr_idx,
.ext_char_dev => |e| e.xattr_idx,
.ext_fifo => |e| e.xattr_idx,
.ext_socket => |e| e.xattr_idx,
else => 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 {
const idx = self.xattrIndex() catch &[0]XattrTable.XattrOwned{};
return XattrTable.statelessLookup(alloc, io, decomp, fil, xattr_table_start, idx);
}
// Types
pub const Error = error{
NotDirectory,
NotRegularFile,
NotSymlink,
NotExtended,
};
pub const Ref = packed struct(u64) {
block_offset: u16,
block_start: u32,
_: u16 = 0,
};
pub const Type = enum(u16) {
dir = 1,
file,
symlink,
block_dev,
char_dev,
fifo,
socket,
ext_dir,
ext_file,
ext_symlink,
ext_block_dev,
ext_char_dev,
ext_fifo,
ext_socket,
};
pub const Data = union(Type) {
dir: dir.Dir,
file: file.File,
symlink: misc.Symlink,
block_dev: misc.Dev,
char_dev: misc.Dev,
fifo: misc.IPC,
socket: misc.IPC,
ext_dir: dir.ExtDir,
ext_file: file.ExtFile,
ext_symlink: misc.ExtSymlink,
ext_block_dev: misc.ExtDev,
ext_char_dev: misc.ExtDev,
ext_fifo: misc.ExtIPC,
ext_socket: misc.ExtIPC,
};
pub const Header = extern struct {
inode_type: Type,
permissions: u16,
uid_idx: u16,
gid_idx: u16,
mod_time: u32,
num: u32,
};
// Extract
const ExtractError = error{ MknodFailed, CannotSetXattr, ConcurrencyUnavailable } || DataExtractor.Error || Io.Dir.CreateFileAtomicError || LookupTable.Error ||
Io.File.Reader.SeekError || Io.File.Atomic.LinkError || Io.Dir.CreateDirError || Io.File.OpenError ||
Io.File.SetPermissionsError || Io.File.SetOwnerError || Io.Dir.SymLinkError || Io.Dir.CreateDirPathError;
const PathRet = struct {
path: []const u8,
inode: Inode,
xattr_idx: ?u32 = null,
};
const ExtractReturnUnion = union(enum) {
path_ret: ExtractError!PathRet,
};
const Tables = struct {
id: LookupTable.CachedTable(u16),
frag: LookupTable.CachedTable(FragEntry),
xattr: XattrTable,
};
/// 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 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, io, super.block_size) },
else => unreachable,
};
defer decomp_base.deinit();
const decomp = decomp_base.decompressor();
var frag_table: CachedTable(FragEntry) = .init(alloc, fil, decomp, super.frag_start, super.frag_count);
defer if (!options.ignore_permissions) frag_table.deinit(io);
try frag_table.fill(io);
var arena: std.heap.ArenaAllocator = .init(alloc);
defer arena.deinit();
var sel_buf = [1]ExtractReturnUnion{undefined} ** 10;
var sel: Io.Select(ExtractReturnUnion) = .init(io, &sel_buf);
defer sel.cancelDiscard();
switch (self.hdr.inode_type) {
.dir, .ext_dir => try sel.concurrent(
.path_ret,
extractDir,
.{ self, alloc, io, fil, decomp, &sel, &arena, super.dir_start, super.inode_start, &frag_table, super.block_size, path },
),
.file, .ext_file => try sel.concurrent(
.path_ret,
extractFile,
.{ self, alloc, io, fil, decomp, &frag_table, super.block_size, path },
),
.symlink, .ext_symlink => try sel.concurrent(.path_ret, extractSymlink, .{ self, io, path }),
else => if (@hasField(std.os, "linux"))
try sel.concurrent(.path_ret, extractDevOrIPC, .{ self, alloc, path }),
}
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);
if (xattr_table != null) try xattr_table.?.table.fill(io);
var id_table: ?CachedTable(u16) = if (!options.ignore_xattr)
.init(alloc, fil, decomp, super.id_start, super.id_count)
else
null;
defer if (!options.ignore_xattr) id_table.?.deinit(io);
if (id_table != null) try id_table.?.fill(io);
while (true) {
const group_token = sel.group.token.load(.acquire);
if (group_token == null) break;
// std.debug.print("{any}\n", .{sel.group.state});
// std.debug.print("Waiting for return...", .{});
const ret = try sel.await();
// std.debug.print("Got One...\n", .{});
const path_ret = try ret.path_ret;
if (options.ignore_permissions and options.ignore_xattr) continue;
if (options.ignore_permissions and path_ret.xattr_idx == null) continue;
var ret_file = try Io.Dir.cwd().openFile(io, path_ret.path, .{});
defer ret_file.close(io);
if (!options.ignore_permissions) {
try ret_file.setPermissions(io, @enumFromInt(path_ret.inode.hdr.permissions));
try ret_file.setOwner(
io,
try id_table.?.get(io, path_ret.inode.hdr.uid_idx),
try id_table.?.get(io, path_ret.inode.hdr.gid_idx),
);
}
if (@hasField(std.os, "linux") and !options.ignore_xattr and path_ret.xattr_idx != null) {
const xattrs = try xattr_table.?.get(alloc, io, path_ret.xattr_idx.?);
defer {
for (xattrs) |x|
alloc.free(x.key);
alloc.free(xattrs);
}
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;
}
}
}
}
fn extractDir(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
decomp: *const Decompressor,
parent_select: *Io.Select(ExtractReturnUnion),
arena: *std.heap.ArenaAllocator,
dir_start: u64,
inode_start: u64,
frag: *CachedTable(FragEntry),
block_size: u32,
path: []const u8,
) ExtractError!PathRet {
try Io.Dir.cwd().createDirPath(io, path);
/// Extract the inode to the given path.
pub fn extractTo(self: Inode, alloc: std.mem.Allocator, archive: Archive, path: []const u8, options: ExtractionOptions) !void {
return InodeExtract.extractTo(alloc, self, archive, path, options);
var sel_buf = [1]ExtractReturnUnion{undefined} ** 10;
var sel: Io.Select(ExtractReturnUnion) = .init(io, &sel_buf);
defer sel.cancelDiscard();
var num: usize = 0;
{
const dir_entries = self.readDirectory(alloc, io, fil, decomp, dir_start) catch |err| switch (err) {
Error.NotDirectory => unreachable,
else => return @errorCast(err),
};
num = dir_entries.len;
defer {
for (dir_entries) |d|
d.deinit(alloc);
alloc.free(dir_entries);
}
for (dir_entries) |d| {
var rdr = try fil.readerAt(io, d.block_start + inode_start, &[0]u8{});
var meta_rdr: MetadataReader = .init(alloc, &rdr.interface, decomp);
try meta_rdr.interface.discardAll(d.block_offset);
const inode = try read(arena.allocator(), &meta_rdr.interface, block_size);
errdefer inode.deinit(arena.allocator());
const new_path = try std.mem.concat(arena.allocator(), u8, &[_][]const u8{ path, "/", d.name });
errdefer arena.allocator().free(new_path);
switch (d.type) {
.dir => try sel.concurrent(
.path_ret,
extractDir,
.{ inode, alloc, io, fil, decomp, parent_select, arena, dir_start, inode_start, frag, block_size, new_path },
),
.file => try sel.concurrent(
.path_ret,
extractFile,
.{ inode, alloc, io, fil, decomp, frag, block_size, new_path },
),
.symlink => try sel.concurrent(.path_ret, extractSymlink, .{ inode, io, new_path }),
else => try sel.concurrent(.path_ret, extractDevOrIPC, .{ inode, alloc, new_path }),
}
}
}
while (num > 0) {
const ret = sel.await() catch break;
num -= 1;
parent_select.queue.putOne(io, ret) catch |err| switch (err) {
error.Canceled => return error.Canceled,
else => break,
};
}
return .{
.path = path,
.inode = self,
};
}
fn extractFile(
self: Inode,
alloc: std.mem.Allocator,
io: Io,
fil: OffsetFile,
decomp: *const Decompressor,
frag: *CachedTable(FragEntry),
block_size: u32,
path: []const u8,
) ExtractError!PathRet {
var atomic = try Io.Dir.cwd().createFileAtomic(io, path, .{});
defer atomic.deinit(io);
var ret: PathRet = .{
.inode = self,
.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;
}
fn extractSymlink(self: Inode, io: Io, path: []const u8) ExtractError!PathRet {
const target = switch (self.data) {
.symlink => |s| s.target,
.ext_symlink => |s| s.target,
else => unreachable,
};
try Io.Dir.cwd().symLink(io, target, path, .{});
return .{
.path = path,
.inode = self,
};
}
fn extractDevOrIPC(self: Inode, alloc: std.mem.Allocator, path: []const u8) ExtractError!PathRet {
var dev_num: u32 = 0;
var mode: u32 = 0;
const DT = std.posix.DT;
var ret: PathRet = .{
.inode = self,
.path = path,
};
switch (self.data) {
.block_dev => |d| {
dev_num = d.dev;
mode = DT.BLK;
},
.ext_block_dev => |d| {
dev_num = d.dev;
mode = DT.BLK;
if (d.xattr_idx != 0xFFFFFFFF) ret.xattr_idx = d.xattr_idx;
},
.char_dev => |d| {
dev_num = d.dev;
mode = DT.CHR;
},
.ext_char_dev => |d| {
dev_num = d.dev;
mode = DT.CHR;
if (d.xattr_idx != 0xFFFFFFFF) ret.xattr_idx = d.xattr_idx;
},
.fifo => mode = DT.FIFO,
.ext_fifo => |f| {
mode = DT.FIFO;
if (f.xattr_idx != 0xFFFFFFFF) ret.xattr_idx = f.xattr_idx;
},
.socket => mode = DT.SOCK,
.ext_socket => |s| {
mode = DT.SOCK;
if (s.xattr_idx != 0xFFFFFFFF) ret.xattr_idx = s.xattr_idx;
},
else => unreachable,
}
const sentinel_path = try std.mem.concatWithSentinel(alloc, u8, &[_][]const u8{path}, 0);
defer alloc.free(sentinel_path);
const res = std.os.linux.mknod(sentinel_path, mode, dev_num);
if (res != 0) return ExtractError.MknodFailed;
return ret;
}
+4 -4
View File
@@ -1,6 +1,6 @@
const Reader = @import("std").Io.Reader;
pub const Dir = packed struct {
pub const Dir = extern struct {
block_start: u32,
hard_links: u32,
size: u16,
@@ -14,19 +14,19 @@ pub const Dir = packed struct {
}
};
pub const ExtDir = packed struct {
pub const ExtDir = extern struct {
hard_links: u32,
size: u32,
block_start: u32,
parent_num: u32,
idx_count: u16,
block_offset: u16,
xattr_id: u32,
xattr_idx: u32,
// index: []DirIndex
pub fn read(rdr: *Reader) !ExtDir {
var d: ExtDir = undefined;
try rdr.readSliceEndian(Dir, @ptrCast(&d), .little);
try rdr.readSliceEndian(ExtDir, @ptrCast(&d), .little);
return d;
}
};
+56 -35
View File
@@ -1,34 +1,43 @@
const std = @import("std");
const Reader = std.Io.Reader;
pub const BlockSize = packed struct {
pub const BlockSize = packed struct(u32) {
size: u24,
uncompressed: bool,
_: u7,
};
const FileRawRead = extern struct {
block_start: u32,
frag_idx: u32,
frag_block_offset: u32,
size: u32,
};
pub const File = struct {
block_start: u32, // bytes 0-3
frag_idx: u32, // bytes 4-7
frag_block_offset: u32, // bytes 8-11
size: u32, // bytes 12-15
block_start: u32,
frag_idx: u32,
frag_block_offset: u32,
size: u32,
block_sizes: []BlockSize,
pub fn read(alloc: std.mem.Allocator, rdr: *Reader, block_size: u32) !File {
var start: [16]u8 = undefined;
try rdr.readSliceAll(&start);
const frag_idx: u32 = std.mem.readInt(u32, start[4..8], .little);
const size: u32 = std.mem.readInt(u32, start[12..16], .little);
var num_blocks: u32 = size / block_size;
if (size % block_size != 0 and frag_idx == 0xFFFFFFFF) num_blocks += 1;
var raw: FileRawRead = undefined;
try rdr.readSliceEndian(FileRawRead, @ptrCast(&raw), .little);
var num_blocks: u32 = raw.size / block_size;
if (raw.size % block_size != 0 and raw.frag_idx == 0xFFFFFFFF)
num_blocks += 1;
const sizes = try alloc.alloc(BlockSize, num_blocks);
errdefer alloc.free(sizes);
try rdr.readSliceEndian(BlockSize, sizes, .little);
return .{
.block_start = std.mem.readInt(u32, start[0..4], .little),
.frag_idx = frag_idx,
.frag_block_offset = std.mem.readInt(u32, start[8..12], .little),
.size = size,
.block_start = raw.block_start,
.frag_idx = raw.frag_idx,
.frag_block_offset = raw.frag_block_offset,
.size = raw.size,
.block_sizes = sizes,
};
}
@@ -38,34 +47,46 @@ pub const File = struct {
}
};
const ExtFileRawRead = extern struct {
block_start: u64,
size: u64,
sparse: u64,
hard_links: u32,
frag_idx: u32,
frag_block_offset: u32,
xattr_idx: u32,
};
pub const ExtFile = struct {
block_start: u64, // bytes 0-7
size: u64, // bytes 8-15
sparse: u64, // bytes 16-23
hard_links: u32, // bytes 24-27
frag_idx: u32, // bytes 28-31
frag_block_offset: u32, // bytes 32-35
xattr_idx: u32, // bytes 36-39
block_start: u64,
size: u64,
sparse: u64,
hard_links: u32,
frag_idx: u32,
frag_block_offset: u32,
xattr_idx: u32,
block_sizes: []BlockSize,
pub fn read(alloc: std.mem.Allocator, rdr: *Reader, block_size: u32) !ExtFile {
var start: [40]u8 = undefined;
try rdr.readSliceAll(&start);
const frag_idx: u32 = std.mem.readInt(u32, start[28..32], .little);
const size: u64 = std.mem.readInt(u64, start[8..16], .little);
var num_blocks: u32 = @truncate(size / block_size);
if (size % block_size != 0 and frag_idx == 0xFFFFFFFF) num_blocks += 1;
var raw: ExtFileRawRead = undefined;
try rdr.readSliceEndian(ExtFileRawRead, @ptrCast(&raw), .little);
var num_blocks: u32 = @truncate(raw.size / block_size);
if (raw.size % block_size != 0 and raw.frag_idx == 0xFFFFFFFF)
num_blocks += 1;
const sizes = try alloc.alloc(BlockSize, num_blocks);
errdefer alloc.free(sizes);
try rdr.readSliceEndian(BlockSize, sizes, .little);
return .{
.block_start = std.mem.readInt(u64, start[0..8], .little),
.size = size,
.sparse = std.mem.readInt(u64, start[16..24], .little),
.hard_links = std.mem.readInt(u32, start[24..28], .little),
.frag_idx = frag_idx,
.frag_block_offset = std.mem.readInt(u32, start[32..36], .little),
.xattr_idx = std.mem.readInt(u32, start[36..40], .little),
.block_start = raw.block_start,
.size = raw.size,
.sparse = raw.sparse,
.hard_links = raw.hard_links,
.frag_idx = raw.frag_idx,
.frag_block_offset = raw.frag_block_offset,
.xattr_idx = raw.xattr_idx,
.block_sizes = sizes,
};
}
+4 -4
View File
@@ -50,7 +50,7 @@ pub const ExtSymlink = struct {
};
/// A block or character device.
pub const Dev = packed struct {
pub const Dev = extern struct {
hard_links: u32,
dev: u32,
@@ -62,7 +62,7 @@ pub const Dev = packed struct {
};
/// An extended block or character device.
pub const ExtDev = packed struct {
pub const ExtDev = extern struct {
hard_links: u32,
dev: u32,
xattr_idx: u32,
@@ -75,7 +75,7 @@ pub const ExtDev = packed struct {
};
/// A socket or FIFO file.
pub const IPC = packed struct {
pub const IPC = extern struct {
hard_links: u32,
pub fn read(rdr: *Reader) !IPC {
@@ -86,7 +86,7 @@ pub const IPC = packed struct {
};
/// An extended socket or FIFO file.
pub const ExtIPC = packed struct {
pub const ExtIPC = extern struct {
hard_links: u32,
xattr_idx: u32,
+122
View File
@@ -0,0 +1,122 @@
const std = @import("std");
const Io = std.Io;
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: *const 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);
rdr = try file.readerAt(io, offset, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, decomp);
try meta.interface.discardAll(@sizeOf(T) * block_offset);
var out: T = undefined;
try meta.interface.readSliceEndian(T, @ptrCast(&out), .little);
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);
const Table = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: *const Decompressor,
table_start: u64,
total_num: u32,
table: std.AutoHashMap(u32, []T),
mut: Io.Mutex = .init,
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: *const Decompressor, offset: u64, total_num: u32) Table {
return .{
.alloc = alloc,
.fil = fil,
.decomp = decomp,
.table_start = offset,
.total_num = total_num,
.table = .init(alloc),
};
}
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 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| {
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 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);
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;
if (self.table.contains(block))
return self.table.get(block).?[block_offset];
try self.mut.lock(io);
defer self.mut.unlock(io);
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 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);
const slice = try meta.interface.readSliceEndianAlloc(self.alloc, T, len, .little);
try self.table.put(block, slice);
return slice[block_offset];
}
};
}
+1 -1
View File
@@ -6,7 +6,7 @@ const Writer = std.Io.Writer;
const ExtractionOptions = @This();
/// The number of threads used for extraction. 0 implies single threaded.
threads: usize = 1,
threads: usize = 1, // TODO: Update to better integrate with zig 0.16 Io. Maybe limit to only single or multi-threaded.
/// 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.
-69
View File
@@ -1,69 +0,0 @@
const std = @import("std");
const math = std.math;
const InodeRef = @import("inode.zig").Ref;
const SQUASHFS_MAGIC: u32 = std.mem.readInt(u32, "hsqs", .little);
const SuperblockError = error{
InvalidMagic,
InvalidBlockLog,
InvalidVersion,
InvalidCheck,
};
/// A squashfs Superblock
pub const Superblock = packed struct {
magic: u32,
inode_count: u32,
mod_time: u32,
block_size: u32,
frag_count: u32,
compression: enum(u16) {
gzip = 1,
lzma,
lzo,
xz,
lz4,
zstd,
},
block_log: u16,
flags: packed struct {
inode_uncompressed: bool,
data_uncompressed: bool,
check: bool,
frag_uncompressed: bool,
fragment_never: bool,
fragment_always: bool,
duplicates: bool,
exportable: bool,
xattr_uncompressed: bool,
xattr_never: bool,
compression_options: bool,
ids_uncompressed: bool,
_: u4,
},
id_count: u16,
ver_maj: u16,
ver_min: u16,
root_ref: InodeRef,
size: u64,
id_start: u64,
xattr_start: u64,
inode_start: u64,
dir_start: u64,
frag_start: u64,
export_start: u64,
/// Validate the Superblock. If an error is returned, it's likely the archive is corrupted or not a squashfs archive.
pub fn validate(self: Superblock) !void {
if (self.magic != SQUASHFS_MAGIC)
return SuperblockError.InvalidMagic;
if (self.flags.check)
return SuperblockError.InvalidCheck;
if (self.ver_maj != 4 or self.ver_min != 0)
return SuperblockError.InvalidVersion;
if (math.log2(self.block_size) != self.block_log)
return SuperblockError.InvalidBlockLog;
}
};
-107
View File
@@ -1,107 +0,0 @@
const std = @import("std");
const Mutex = std.Thread.Mutex;
const Archive = @import("archive.zig");
const BlockSize = @import("inode_data/file.zig").BlockSize;
const InodeRef = @import("inode.zig").Ref;
const Superblock = @import("super.zig").Superblock;
const MetadataReader = @import("util/metadata.zig");
const OffsetFile = @import("util/offset_file.zig");
const XattrTable = @import("xattr.zig");
/// Information about a fragment section. Multiple fragments are contained in the block described by a single FragEntry.
/// The offset into the block and fragment size is stored in the file's inode.
pub const FragEntry = packed struct {
start: u64,
size: BlockSize,
_: u32,
};
const Tables = @This();
frag_table: Table(FragEntry),
id_table: Table(u16),
export_table: Table(InodeRef),
xattr_table: XattrTable,
pub fn init(alloc: std.mem.Allocator, archive: Archive) !Tables {
return .{
.frag_table = try .init(alloc, archive.fil, archive.decomp, archive.super.frag_start, archive.super.frag_count),
.id_table = try .init(alloc, archive.fil, archive.decomp, archive.super.id_start, archive.super.id_count),
.export_table = try .init(alloc, archive.fil, archive.decomp, archive.super.export_start, archive.super.inode_count),
.xattr_table = try .init(alloc, archive.fil, archive.decomp, archive.super.xattr_start),
};
}
pub fn deinit(self: *Tables) void {
self.frag_table.deinit();
self.id_table.deinit();
self.export_table.deinit();
self.xattr_table.deinit();
}
/// A two-layer metadata table.
pub fn Table(T: anytype) type {
return struct {
const Self = @This();
const VALS_PER_BLOCK = 8192 / @sizeOf(T);
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: DecompFn,
tab_start: u64,
tab: std.AutoHashMap(u32, []T),
values: u32,
mut: Mutex = .{},
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: DecompFn, tab_start: u64, values: u32) !Self {
return .{
.alloc = alloc,
.fil = fil,
.decomp = decomp,
.tab_start = tab_start,
.tab = .init(alloc),
.values = values,
};
}
pub fn deinit(self: *Self) void {
var iter = self.tab.valueIterator();
while (iter.next()) |s| {
self.alloc.free(s.*);
}
self.tab.deinit();
}
pub fn get(self: *Self, idx: u32) !T {
if (idx >= self.values) return error.InvalidIndex;
const block_num = idx / VALS_PER_BLOCK;
const idx_offset = idx - (block_num * VALS_PER_BLOCK);
if (self.tab.contains(block_num)) {
const block = self.tab.get(block_num).?;
return block[idx_offset];
}
self.mut.lock();
defer self.mut.unlock();
// Double check in case of race condition..
if (self.tab.contains(block_num)) {
const block = self.tab.get(block_num).?;
return block[idx_offset];
}
const is_last = (self.values - 1) / VALS_PER_BLOCK == block_num;
const slice_size = if (is_last) self.values - (block_num * VALS_PER_BLOCK) else VALS_PER_BLOCK;
const slice = try self.alloc.alloc(T, slice_size);
var rdr = try self.fil.readerAt(self.tab_start + (8 * block_num), &[0]u8{});
var offset: u64 = 0;
try rdr.interface.readSliceEndian(u64, @ptrCast(&offset), .little);
rdr = try self.fil.readerAt(offset, &[0]u8{});
var meta: MetadataReader = .init(self.alloc, &rdr.interface, self.decomp);
try meta.interface.readSliceEndian(T, @ptrCast(slice), .little);
try self.tab.put(block_num, slice);
return slice[idx_offset];
}
};
}
+27 -22
View File
@@ -1,46 +1,51 @@
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 = @import("super.zig").Superblock;
const Superblock = Archive.Superblock;
const TestArchive = "testing/LinuxPATest.sfs";
test "Basics" {
var fil = try std.fs.cwd().openFile(TestArchive, .{});
defer fil.close();
var sfs: Archive = try .init(std.testing.allocator, fil);
defer sfs.deinit();
if (sfs.super != LinuxPATestCorrectSuperblock) {
std.debug.print("Superblock wrong\nShould be: {}\n\nis: {}\n", .{ LinuxPATestCorrectSuperblock, sfs.super });
return error.BadSuperblock;
}
std.debug.print("Starting test: Basics...\n", .{});
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" {
std.fs.cwd().deleteFile(TestFileExtractLocation) catch {};
var fil = try std.fs.cwd().openFile(TestArchive, .{});
defer fil.close();
var sfs: Archive = try .init(std.testing.allocator, fil);
defer sfs.deinit();
var test_fil = try sfs.open(TestFile);
std.debug.print("Starting test: ExtractSingleFile...\n", .{});
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(TestFileExtractLocation, .Default);
try test_fil.extract(alloc, io, TestFileExtractLocation, try .Default());
//TODO: validate extracted file.
}
const TestFullExtractLocation = "testing/TestExtract";
test "ExtractCompleteArchive" {
std.fs.cwd().deleteTree(TestFullExtractLocation) catch {};
var fil = try std.fs.cwd().openFile(TestArchive, .{});
defer fil.close();
var sfs: Archive = try .init(std.testing.allocator, fil);
defer sfs.deinit();
try sfs.extract(TestFullExtractLocation, .Default);
std.debug.print("Starting test: ExtractCompleteArchive...\n", .{});
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 = .{
-174
View File
@@ -1,174 +0,0 @@
//! A reader for a regular file.
const std = @import("std");
const Reader = std.Io.Reader;
const Writer = std.Io.Writer;
const Limit = std.Io.Limit;
const Archive = @import("../archive.zig");
const DecompFn = @import("../decomp.zig").DecompFn;
const BlockSize = @import("../inode_data/file.zig").BlockSize;
const FragEntry = @import("../tables.zig").FragEntry;
const OffsetFile = @import("offset_file.zig");
const DataReader = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: DecompFn,
block_size: u32,
blocks: []BlockSize,
frag: ?FragEntry = null, // TODO: do something better?
frag_offset: u32 = 0,
size: u64,
interface: Reader,
cur_offset: u64,
block_idx: u32 = 0,
pub fn init(alloc: std.mem.Allocator, archive: Archive, blocks: []BlockSize, start: u64, size: u64) DataReader {
return .{
.alloc = alloc,
.fil = archive.fil,
.decomp = archive.decomp,
.block_size = archive.super.block_size,
.blocks = blocks,
.size = size,
.cur_offset = start,
.interface = .{
.end = 0,
.seek = 0,
.buffer = &[0]u8{},
.vtable = &.{
.stream = stream,
.discard = discard,
.readVec = readVec,
},
},
};
}
pub fn deinit(self: *DataReader) void {
self.alloc.free(self.interface.buffer);
self.interface.end = 0;
self.interface.seek = 0;
}
pub fn addFragment(self: *DataReader, entry: FragEntry, frag_offset: u32) void {
self.frag = entry;
self.frag_offset = frag_offset;
}
fn numBlocks(self: DataReader) usize {
var res = self.blocks.len;
if (self.frag != null) res += 1;
return res;
}
fn advance(self: *DataReader) !void {
if (self.block_idx > self.blocks.len or (self.block_idx == self.blocks.len and self.frag == null)) {
if (self.interface.buffer.len > 0) {
self.alloc.free(self.interface.buffer);
self.interface.buffer = &[0]u8{};
self.interface.end = 0;
self.interface.seek = 0;
}
return Reader.Error.EndOfStream;
}
defer self.block_idx += 1;
const cur_block_size = if (self.block_idx == self.numBlocks() - 1) self.size % self.block_size else self.block_size;
try self.resizeBuffer(cur_block_size);
self.interface.seek = 0;
self.interface.end = cur_block_size;
if (self.block_idx == self.blocks.len) { // fragment
var rdr = try self.fil.readerAt(self.frag.?.start, &[0]u8{});
if (self.frag.?.size.uncompressed) {
try rdr.interface.discardAll(self.frag_offset);
try rdr.interface.readSliceAll(self.interface.buffer);
return;
}
const tmp_buf = try self.alloc.alloc(u8, self.frag.?.size.size);
defer self.alloc.free(tmp_buf);
try rdr.interface.readSliceAll(tmp_buf);
const needed_block = try self.alloc.alloc(u8, self.block_size);
defer self.alloc.free(needed_block);
_ = try self.decomp(self.alloc, tmp_buf, needed_block);
@memcpy(self.interface.buffer, needed_block[self.frag_offset .. self.frag_offset + cur_block_size]);
return;
}
const block = self.blocks[self.block_idx];
if (block.size == 0) {
@memset(self.interface.buffer, 0);
return;
}
var rdr = try self.fil.readerAt(self.cur_offset, &[0]u8{});
self.cur_offset += block.size;
if (block.uncompressed) {
try rdr.interface.readSliceAll(self.interface.buffer);
return;
}
const tmp_buf = try self.alloc.alloc(u8, block.size);
defer self.alloc.free(tmp_buf);
try rdr.interface.readSliceAll(tmp_buf);
_ = try self.decomp(self.alloc, tmp_buf, self.interface.buffer);
}
/// Does not guarentee that data currently in the buffer is retained.
fn resizeBuffer(self: *DataReader, size: usize) !void {
if (self.interface.buffer.len == size) return;
if (!self.alloc.resize(self.interface.buffer, size)) {
self.alloc.free(self.interface.buffer);
self.interface.buffer = self.alloc.alloc(u8, size) catch |err| {
self.interface.buffer = &[0]u8{};
return err;
};
} else {
self.interface.buffer.len = size;
}
}
fn stream(rdr: *Reader, wrt: *Writer, limit: Limit) Reader.StreamError!usize {
var self: *DataReader = @alignCast(@fieldParentPtr("interface", rdr));
if (rdr.seek >= rdr.end) self.advance() catch |err| {
if (err == error.EndOfStream) return error.EndOfStream;
std.log.err("Error advancing data reader: {}\n", .{err});
return Reader.Error.ReadFailed;
};
if (limit == .nothing) return 0;
const to_read = @min(rdr.end - rdr.seek, @intFromEnum(limit));
const res = try wrt.write(rdr.buffer[rdr.seek .. rdr.seek + to_read]);
rdr.seek += res;
return res;
}
fn discard(rdr: *Reader, limit: Limit) Reader.Error!usize {
var self: *DataReader = @alignCast(@fieldParentPtr("interface", rdr));
if (rdr.seek >= rdr.end) self.advance() catch |err| {
if (err == error.EndOfStream) return error.EndOfStream;
std.log.err("Error advancing data reader: {}\n", .{err});
return Reader.Error.ReadFailed;
};
if (limit == .nothing) return 0;
const to_adv = @min(rdr.end - rdr.seek, @intFromEnum(limit));
rdr.seek += to_adv;
return to_adv;
}
fn readVec(rdr: *Reader, vec: [][]u8) Reader.Error!usize {
var self: *DataReader = @alignCast(@fieldParentPtr("interface", rdr));
if (rdr.seek >= rdr.end) self.advance() catch |err| {
if (err == error.EndOfStream) return error.EndOfStream;
std.log.err("Error advancing data reader: {}\n", .{err});
return Reader.Error.ReadFailed;
};
var cur_red: usize = 0;
for (vec) |s| {
const to_copy: usize = @min(rdr.end - rdr.seek, s.len);
@memcpy(s[0..to_copy], rdr.buffer[rdr.seek .. rdr.seek + to_copy]);
rdr.seek += to_copy;
cur_red += to_copy;
if (rdr.end == rdr.seek) break;
}
return cur_red;
}
+180
View File
@@ -0,0 +1,180 @@
//! The DataExtractor is meant to extract a regular file's data to a given file asyncronously.
const std = @import("std");
const Io = std.Io;
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");
// const SharedCache = @import("shared_cache.zig");
pub const Error = error{OutOfMemory} || Io.File.Reader.SeekError || Io.Writer.Error;
const DataExtractor = @This();
fil: OffsetFile,
decomp: *const Decompressor,
block_size: u32,
file_size: u64,
start: u64,
blocks: []BlockSize,
frag_offset: u32 = 0,
frag_entry: ?FragEntry = null,
err: ?Error = null,
pub fn init(fil: OffsetFile, decomp: *const Decompressor, block_size: u32, file_size: u64, data_start: u64, blocks: []BlockSize) DataExtractor {
return .{
.fil = fil,
.decomp = decomp,
.block_size = block_size,
.file_size = file_size,
.start = data_start,
.blocks = blocks,
};
}
pub fn addFrag(self: *DataExtractor, frag_offset: u32, entry: FragEntry) void {
self.frag_offset = frag_offset;
self.frag_entry = entry;
}
fn numBlocks(self: DataExtractor) usize {
var num = self.blocks.len;
if (self.frag_entry != 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) 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 });
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: *?Error) Io.Cancelable!void {
const block = self.blocks[idx];
const cur_block_size = if (idx == self.numBlocks() - 1)
self.file_size % self.block_size
else
self.block_size;
var wrt = fil.writer(io, &[0]u8{});
wrt.seekTo(self.block_size * idx) 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| {
ret_err.* = err;
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;
};
if (block.uncompressed) {
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..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();
return Io.Cancelable.Canceled;
};
}
}
fn fragThread(self: DataExtractor, alloc: std.mem.Allocator, io: Io, fil: Io.File, ret_err: *?Error) Io.Cancelable!void {
const frag = self.frag_entry.?;
const cur_block_size = self.file_size % self.block_size;
var wrt = fil.writer(io, &[0]u8{});
wrt.seekTo(self.blocks.len * self.block_size) 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| {
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[self.frag_offset .. self.frag_offset + cur_block_size]) catch |err| {
ret_err.* = err;
if (err == error.Canceled) io.recancel();
return Io.Cancelable.Canceled;
};
}
}
+208
View File
@@ -0,0 +1,208 @@
//! DataReader reads a regular file's data linearly from start to finish using Io.Reader interface.
const std = @import("std");
const Io = std.Io;
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");
// const SharedCache = @import("shared_cache.zig");
const DataReader = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
io: Io,
decomp: *const Decompressor,
cache: *Io.Queue([]u8),
block_size: u32,
file_size: u64,
cur_offset: u64,
blocks: []BlockSize,
frag_offset: u32 = 0,
frag_entry: ?FragEntry = 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 {
return .{
.alloc = alloc,
.fil = fil,
.io = io,
.decomp = decomp,
.cache = cache,
.block_size = block_size,
.file_size = file_size,
.cur_offset = data_start,
.blocks = blocks,
.interface = .{
.buffer = try alloc.alloc(u8, block_size),
.seek = 0,
.end = 0,
.vtable = &.{
.stream = stream,
.discard = discard,
.readVec = readVec,
},
},
};
}
pub fn deinit(self: *DataReader) void {
self.alloc.free(self.interface.buffer);
}
pub fn addFrag(self: *DataReader, frag_offset: u32, entry: FragEntry) void {
self.frag_offset = frag_offset;
self.frag_entry = entry;
}
fn numBlocks(self: DataReader) usize {
var num = self.blocks.len;
if (self.frag_entry != null) num += 1;
return num;
}
fn advanceBuffer(self: *DataReader) !void {
if (self.block_idx >= self.numBlocks()) {
return Reader.Error.EndOfStream;
}
defer self.block_idx += 1;
self.interface.end = if (self.block_idx == self.numBlocks() - 1)
self.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]);
}
self.interface.seek = 0;
return;
}
// Normal Block
const block = self.blocks[self.block_idx];
if (block.size == 0) {
self.interface.seek = 0;
self.sparse_block = true;
return;
} else {
self.sparse_block = false;
}
if (block.uncompressed) {
try self.fil.readAt(self.io, self.cur_offset, self.interface.buffer[0..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]);
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;
}
fn stream(rdr: *Reader, wrt: *Writer, limit: Limit) Reader.StreamError!usize {
var data: *DataReader = @fieldParentPtr("interface", rdr);
if (rdr.seek == rdr.end)
data.advanceBuffer() catch |err| return switch (err) {
error.ReadFailed => error.ReadFailed,
error.EndOfStream => error.EndOfStream,
else => error.ReadFailed,
};
switch (limit) {
.nothing => return 0,
.unlimited => {
const wrote = if (data.sparse_block)
try wrt.splatByte(0, rdr.end - rdr.seek)
else
try wrt.write(rdr.buffer[rdr.seek..rdr.end]);
rdr.seek += wrote;
return wrote;
},
else => {
const to_read = @min(rdr.end - rdr.seek, @intFromEnum(limit));
const wrote = if (data.sparse_block)
try wrt.splatByte(0, to_read)
else
try wrt.write(rdr.buffer[rdr.seek .. rdr.seek + to_read]);
rdr.seek += wrote;
return wrote;
},
}
}
fn discard(rdr: *Reader, limit: Limit) Reader.Error!usize {
var data: *DataReader = @fieldParentPtr("interface", rdr);
if (rdr.seek == rdr.end)
data.advanceBuffer() catch |err| return switch (err) {
error.ReadFailed => error.ReadFailed,
error.EndOfStream => error.EndOfStream,
else => error.ReadFailed,
};
switch (limit) {
.nothing => return 0,
.unlimited => {
const adv = rdr.end - rdr.seek;
rdr.seek = rdr.end;
return adv;
},
else => {
const adv = @min(rdr.end - rdr.seek, @intFromEnum(limit));
rdr.seek += adv;
return adv;
},
}
}
fn readVec(rdr: *Reader, vec: [][]u8) Reader.Error!usize {
var data: *DataReader = @fieldParentPtr("interface", rdr);
if (rdr.seek == rdr.end)
data.advanceBuffer() catch |err| return switch (err) {
error.ReadFailed => error.ReadFailed,
error.EndOfStream => error.EndOfStream,
else => error.ReadFailed,
};
var wrote: usize = 0;
for (vec) |buf| {
if (rdr.seek == rdr.end) break;
const to_copy = @min(rdr.end - rdr.seek, buf.len);
if (data.sparse_block)
@memset(buf[0..to_copy], 0)
else
@memcpy(buf[0..to_copy], rdr.buffer[rdr.seek .. rdr.seek + to_copy]);
rdr.seek += to_copy;
wrote += to_copy;
}
return wrote;
}
-207
View File
@@ -1,207 +0,0 @@
//! Similiar to DataReader, but set-up for threaded writing to files.
const std = @import("std");
const Reader = std.Io.Reader;
const Writer = std.Io.Writer;
const Limit = std.Io.Limit;
const WaitGroup = std.Thread.WaitGroup;
const Pool = std.Thread.Pool;
const Archive = @import("../archive.zig");
const DecompFn = @import("../decomp.zig").DecompFn;
const BlockSize = @import("../inode_data/file.zig").BlockSize;
const FragEntry = @import("../tables.zig").FragEntry;
const InodeFinish = @import("inode_finish.zig");
const OffsetFile = @import("offset_file.zig");
const ThreadedDataReader = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: DecompFn,
block_size: u32,
blocks: []BlockSize,
frag: ?FragEntry = null, // TODO: do something better?
frag_offset: u32 = 0,
size: u64,
num_blocks: usize,
start_offset: u64,
pub fn init(alloc: std.mem.Allocator, archive: Archive, blocks: []BlockSize, start: u64, size: u64) ThreadedDataReader {
return .{
.alloc = alloc,
.fil = archive.fil,
.decomp = archive.decomp,
.block_size = archive.super.block_size,
.blocks = blocks,
.size = size,
.num_blocks = blocks.len,
.start_offset = start,
};
}
pub fn addFragment(self: *ThreadedDataReader, entry: FragEntry, frag_offset: u32) void {
self.frag = entry;
self.frag_offset = frag_offset;
self.num_blocks = self.blocks.len + 1;
}
/// Extract the data to the file threadedly, using pool to spawn threads.
/// If errors occur, they are set to finish.out_err.
pub fn extractThreaded(self: ThreadedDataReader, file: std.fs.File, pool: *Pool, finish: *InodeFinish) void {
var cur_write_offset: u64 = 0;
var cur_read_offset: u64 = self.start_offset;
for (0..self.blocks.len) |i| {
const cur_block_size = if (i == self.num_blocks - 1) self.size % self.block_size else self.block_size;
pool.spawn(workThreadBlocks, .{ self, file, cur_write_offset, cur_read_offset, self.blocks[i], cur_block_size, finish }) catch |res_err| {
finish.logError("Can't spawn pool task: {}", .{res_err});
finish.out_err.* = res_err;
finish.finish();
};
cur_write_offset += cur_block_size;
cur_read_offset += self.blocks[i].size;
}
if (self.frag != null)
pool.spawn(workThreadFragment, .{ self, file, cur_write_offset, finish }) catch |res_err| {
finish.logError("Can't spawn pool task: {}", .{res_err});
finish.out_err.* = res_err;
finish.finish();
};
}
fn workThreadBlocks(
self: ThreadedDataReader,
fil: std.fs.File,
write_offset: u64,
read_offset: u64,
block: BlockSize,
cur_block_size: u64,
finish: *InodeFinish,
) void {
defer finish.finish();
var wrt = fil.writer(&[0]u8{});
wrt.seekTo(write_offset) catch |err| {
finish.logError("Error seeking file writer: {}", .{err});
finish.out_err.* = err;
return;
};
defer wrt.interface.flush() catch |err| {
finish.logError("Error flushing file writer: {}", .{err});
finish.out_err.* = err;
};
if (block.size == 0) {
wrt.interface.splatByteAll(0, cur_block_size) catch |err| {
finish.logError("Error writing zeroes: {}", .{err});
finish.out_err.* = err;
return;
};
return;
}
var rdr = self.fil.readerAt(read_offset, &[0]u8{}) catch |err| {
finish.logError("Error creating file reader: {}", .{err});
finish.out_err.* = err;
return;
};
if (block.uncompressed) {
rdr.interface.streamExact(&wrt.interface, block.size) catch |err| {
finish.logError("Error streaming data: {}", .{err});
finish.out_err.* = err;
return;
};
return;
}
// TODO: shared buffers
const read_buf = self.alloc.alloc(u8, block.size) catch |err| {
finish.logError("Error creating reader buffer: {}", .{err});
finish.out_err.* = err;
return;
};
defer self.alloc.free(read_buf);
rdr.interface.readSliceAll(read_buf) catch |err| {
finish.logError("Error reading data into reader buffer: {}", .{err});
finish.out_err.* = err;
return;
};
// TODO: shared buffers
const res_buf = self.alloc.alloc(u8, cur_block_size) catch |err| {
finish.logError("Error creating result buffer: {}", .{err});
finish.out_err.* = err;
return;
};
defer self.alloc.free(res_buf);
_ = self.decomp(self.alloc, read_buf, res_buf) catch |err| {
finish.logError("Error decompressing data block: {}", .{err});
finish.out_err.* = err;
return;
};
wrt.interface.writeAll(res_buf) catch |err| {
finish.logError("Error writing to file: {}", .{err});
finish.out_err.* = err;
return;
};
}
fn workThreadFragment(self: ThreadedDataReader, fil: std.fs.File, write_offset: u64, finish: *InodeFinish) void {
defer finish.finish();
var wrt = fil.writer(&[0]u8{});
wrt.seekTo(write_offset) catch |err| {
finish.logError("Error seeking file writer for file fragment: {}", .{err});
finish.out_err.* = err;
return;
};
defer wrt.interface.flush() catch |err| {
finish.out_err.* = err;
};
var rdr = self.fil.readerAt(self.frag.?.start, &[0]u8{}) catch |err| {
finish.logError("Error creating file reader for file fragment: {}", .{err});
finish.out_err.* = err;
return;
};
if (self.frag.?.size.uncompressed) {
rdr.interface.discardAll(self.frag_offset) catch |err| {
finish.logError("Error discarding useless fragment data: {}", .{err});
finish.out_err.* = err;
return;
};
rdr.interface.streamExact(&wrt.interface, self.size % self.block_size) catch |err| {
finish.logError("Error streaming fragment data: {}", .{err});
finish.out_err.* = err;
return;
};
return;
}
const tmp_buf = self.alloc.alloc(u8, self.frag.?.size.size) catch |err| {
finish.logError("Error creating a temporary buffer for a file fragment: {}", .{err});
finish.out_err.* = err;
return;
};
defer self.alloc.free(tmp_buf);
rdr.interface.readSliceAll(tmp_buf) catch |err| {
finish.logError("Error reading data into fragment buffer: {}", .{err});
finish.out_err.* = err;
return;
};
const needed_block = self.alloc.alloc(u8, self.block_size) catch |err| {
finish.logError("Error allocating fragment decompression results: {}", .{err});
finish.out_err.* = err;
return;
};
defer self.alloc.free(needed_block);
_ = self.decomp(self.alloc, tmp_buf, needed_block) catch |err| {
finish.logError("Error decompressing fragment: {}", .{err});
finish.out_err.* = err;
return;
};
wrt.interface.writeAll(needed_block[self.frag_offset .. self.frag_offset + (self.size % self.block_size)]) catch |err| {
finish.logError("Error writing fragment: {}", .{err});
finish.out_err.* = err;
return;
};
}
+15
View File
@@ -0,0 +1,15 @@
//! A decompression interface
const std = @import("std");
const Decompressor = @This();
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,
pub fn Decompress(self: *const Decompressor, alloc: std.mem.Allocator, in: []u8, out: []u8) Error!usize {
return self.decomp_fn(self, alloc, in, out);
}
-353
View File
@@ -1,353 +0,0 @@
const std = @import("std");
const Allocator = std.mem.Allocator;
const Pool = std.Thread.Pool;
const WaitGroup = std.Thread.WaitGroup;
const Archive = @import("../archive.zig");
const DirEntry = @import("../dir_entry.zig");
const Inode = @import("../inode.zig");
const ExtractionOptions = @import("../options.zig");
const Tables = @import("../tables.zig");
const InodeFinish = @import("inode_finish.zig");
const FinishUnion = InodeFinish.FinishUnion;
const ThreadedDataReader = @import("data_threaded.zig");
// 1 MB
const STACK_ALLOC_SIZE = 1024 * 1024;
pub fn extractTo(
allocator: Allocator,
inode: Inode,
archive: Archive,
path: []const u8,
options: ExtractionOptions,
) !void {
if (path[path.len - 1] == '/')
return extractTo(allocator, inode, archive, path[0 .. path.len - 2], options);
var stack_alloc = std.heap.stackFallback(STACK_ALLOC_SIZE, allocator);
var arena: std.heap.ArenaAllocator = .init(stack_alloc.get());
defer arena.deinit();
if (options.threads <= 1) {
const alloc = arena.allocator();
var tables: Tables = try .init(alloc, archive);
return extractSingleThread(arena.allocator(), inode, archive, &tables, path, options);
}
var thread_alloc = std.heap.ThreadSafeAllocator{ .child_allocator = arena.allocator() };
const alloc = thread_alloc.allocator();
var tables: Tables = try .init(alloc, archive);
var pool_alloc = std.heap.stackFallback(10 * 1024, alloc);
var pool: Pool = undefined;
try pool.init(.{ .allocator = pool_alloc.get(), .n_jobs = options.threads - 1 });
var wg: WaitGroup = .{};
var err: ?anyerror = null;
wg.start();
try pool.spawn(extractMultiThread, .{
alloc,
inode,
archive,
&tables,
path,
options,
&pool,
FinishUnion{ .wg = &wg },
&err,
});
pool.waitAndWork(&wg);
if (err != null) return err.?;
}
fn extractSingleThread(
alloc: Allocator,
inode: Inode,
archive: Archive,
tables: *Tables,
path: []const u8,
options: ExtractionOptions,
) !void {
switch (inode.hdr.inode_type) {
.dir, .ext_dir => {
_ = std.fs.cwd().makeDir(path) catch |err| switch (err) {
std.fs.Dir.MakeError.PathAlreadyExists => {},
else => return err,
};
// Currently we are ignoring any deinit or free calls since we know we are under an ArenaAllocator.
// Possibly in the future, do some simple math to see if it would be safe to ONLY deinit via Arena,
// otherwise be more conscientious about freeing memory.
// For now, this is good enough.
const entries = try inode.dirEntries(alloc, archive);
for (entries) |ent| {
const sub_inode: Inode = try .readFromEntry(alloc, archive, ent);
const new_path = try std.mem.concat(alloc, u8, &[_][]const u8{ path, "/", ent.name });
try extractSingleThread(alloc, sub_inode, archive, tables, new_path, options);
}
const fil = try std.fs.cwd().openFile(path, .{});
defer fil.close();
try inode.setMetadata(alloc, tables, fil, options);
},
.file, .ext_file => {
var fil = try std.fs.cwd().createFile(path, .{ .exclusive = true });
defer fil.close();
var wrt = fil.writer(&[0]u8{});
var dat_rdr = try inode.dataReader(alloc, archive, tables);
defer dat_rdr.deinit();
_ = try dat_rdr.interface.streamRemaining(&wrt.interface);
try wrt.interface.flush();
try inode.setMetadata(alloc, tables, fil, options);
},
.symlink, .ext_symlink => return extractSymlink(inode, path),
else => return extractDeviceAndIPC(inode, alloc, tables, path, options),
}
}
fn extractMultiThread(
alloc: Allocator,
inode: Inode,
archive: Archive,
tables: *Tables,
path: []const u8,
options: ExtractionOptions,
pool: *Pool,
fin: FinishUnion,
err: *?anyerror,
) void {
if (err.* != null) {
fin.finish();
return;
}
switch (inode.hdr.inode_type) {
.dir, .ext_dir => {
_ = std.fs.cwd().makeDir(path) catch |res_err| switch (res_err) {
std.fs.Dir.MakeError.PathAlreadyExists => {},
else => {
err.* = res_err;
fin.finish();
return;
},
};
// Currently we are ignoring any deinit or free calls since we know we are under an ArenaAllocator.
// Possibly in the future, do some simple math to see if it would be safe to ONLY deinit via Arena,
// otherwise be more conscientious about freeing memory.
// For now, this is good enough.
const entries = inode.dirEntries(alloc, archive) catch |res_err| {
err.* = res_err;
fin.finish();
return;
};
if (entries.len == 0) {
fin.finish();
return;
}
var dir_fin = InodeFinish.create(
alloc,
inode,
path,
tables,
options,
fin,
err,
null,
entries.len,
) catch |res_err| {
err.* = res_err;
fin.finish();
return;
};
for (entries) |ent| {
if (ent.inode_type == .dir) {
extractEntry(
alloc,
ent,
archive,
tables,
path,
options,
pool,
.{ .fin = dir_fin },
err,
);
continue;
}
pool.spawn(
extractEntry,
.{ alloc, ent, archive, tables, path, options, pool, FinishUnion{ .fin = dir_fin }, err },
) catch |res_err| {
err.* = res_err;
dir_fin.finish();
return;
};
}
},
.file, .ext_file => {
const fil = std.fs.cwd().createFile(path, .{ .exclusive = true }) catch |res_err| {
if (options.verbose)
options.verbose_writer.?.print("Can't create file at {s}: {}\n", .{ path, res_err }) catch {};
err.* = res_err;
fin.finish();
return;
};
var data_rdr = threadedDataReader(inode, alloc, archive, tables) catch |res_err| {
if (options.verbose)
options.verbose_writer.?.print("Can't create data reader for inode #{} (extracting to {s}): {}\n", .{ inode.hdr.num, path, res_err }) catch {};
err.* = res_err;
fin.finish();
return;
};
if (data_rdr == null) {
inode.setMetadata(alloc, tables, fil, options) catch |res_err| {
if (options.verbose)
options.verbose_writer.?.print("Can't set metadata to {s}: {}\n", .{ path, res_err }) catch {};
err.* = res_err;
};
fin.finish();
return;
}
const file_fin = InodeFinish.create(
alloc,
inode,
path,
tables,
options,
fin,
err,
fil,
data_rdr.?.num_blocks,
) catch |res_err| {
if (options.verbose)
options.verbose_writer.?.print("Can't create callback for inode #{} (extracting to {s}): {}\n", .{ inode.hdr.num, path, res_err }) catch {};
err.* = res_err;
fin.finish();
return;
};
data_rdr.?.extractThreaded(fil, pool, file_fin);
},
.symlink, .ext_symlink => {
extractSymlink(inode, path) catch |res_err| {
err.* = res_err;
};
fin.finish();
},
else => {
extractDeviceAndIPC(inode, alloc, tables, path, options) catch |res_err| {
err.* = res_err;
};
fin.finish();
},
}
}
fn extractEntry(
alloc: Allocator,
ent: DirEntry,
archive: Archive,
tables: *Tables,
path: []const u8,
options: ExtractionOptions,
pool: *Pool,
fin: FinishUnion,
err: *?anyerror,
) void {
const new_path = std.mem.concat(alloc, u8, &[_][]const u8{ path, "/", ent.name }) catch |res_err| {
err.* = res_err;
fin.finish();
return;
};
const inode = Inode.readFromEntry(alloc, archive, ent) catch |res_err| {
err.* = res_err;
fin.finish();
return;
};
extractMultiThread(alloc, inode, archive, tables, new_path, options, pool, fin, err);
}
/// Get a threaded data reader for a file inode.
fn threadedDataReader(self: Inode, alloc: std.mem.Allocator, archive: Archive, tables: *Tables) !?ThreadedDataReader {
return switch (self.hdr.inode_type) {
.file => threadedReaderFromData(alloc, archive, tables, self.data.file),
.ext_file => threadedReaderFromData(alloc, archive, tables, self.data.ext_file),
else => error.NotRegularFile,
};
}
fn threadedReaderFromData(alloc: std.mem.Allocator, archive: Archive, tables: *Tables, data: anytype) !?ThreadedDataReader {
if (data.block_sizes.len == 0 and data.frag_idx == 0xFFFFFFFF) return null;
var out: ThreadedDataReader = .init(alloc, archive, data.block_sizes, data.block_start, data.size);
if (data.frag_idx != 0xFFFFFFFF)
out.addFragment(try tables.frag_table.get(data.frag_idx), data.frag_block_offset);
return out;
}
/// Creates the symlink described by the inode.
/// Sets metadata.
fn extractSymlink(self: Inode, path: []const u8) !void {
const target = switch (self.data) {
.symlink => |s| s.target,
.ext_symlink => |s| s.target,
else => unreachable,
};
try std.fs.cwd().symLink(target, path, .{});
}
/// Creates the device described by the inode.
/// Sets metadata.
fn extractDeviceAndIPC(self: Inode, alloc: std.mem.Allocator, tables: *Tables, path: []const u8, options: ExtractionOptions) !void {
var mode: u32 = undefined;
var dev: u32 = 0;
switch (self.data) {
.char_dev => |d| {
mode = std.posix.S.IFCHR;
dev = d.dev;
},
.ext_char_dev => |d| {
mode = std.posix.S.IFCHR;
dev = d.dev;
},
.block_dev => |d| {
mode = std.posix.S.IFBLK;
dev = d.dev;
},
.ext_block_dev => |d| {
mode = std.posix.S.IFBLK;
dev = d.dev;
},
.fifo, .ext_fifo => mode = std.posix.S.IFIFO,
.socket, .ext_socket => mode = std.posix.S.IFSOCK,
else => unreachable,
}
const res: std.os.linux.E = @enumFromInt(std.os.linux.mknod(@ptrCast(path), mode, dev));
switch (res) {
.SUCCESS => {},
.ACCES => return std.fs.Dir.MakeError.AccessDenied,
.DQUOT => return std.fs.Dir.MakeError.DiskQuota,
.EXIST => return std.fs.Dir.MakeError.PathAlreadyExists,
.FAULT, .NOENT => return std.fs.Dir.MakeError.BadPathName,
.LOOP => return std.fs.Dir.MakeError.SymLinkLoop,
.NAMETOOLONG => return std.fs.Dir.MakeError.NameTooLong,
.NOMEM => return std.fs.Dir.MakeError.SystemResources,
.NOSPC => return std.fs.Dir.MakeError.NoSpaceLeft,
.NOTDIR => return std.fs.Dir.MakeError.NotDir,
.PERM => return std.fs.Dir.MakeError.PermissionDenied,
.ROFS => return std.fs.Dir.MakeError.ReadOnlyFileSystem,
else => return blk: {
std.debug.print("unhandled mknod result: {}\n", .{res});
break :blk std.fs.Dir.MakeError.Unexpected;
},
}
var fil = try std.fs.cwd().openFile(path, .{});
defer fil.close();
try self.setMetadata(alloc, tables, fil, options);
}
-101
View File
@@ -1,101 +0,0 @@
const std = @import("std");
const WaitGroup = std.Thread.WaitGroup;
const Mutex = std.Thread.Mutex;
const Archive = @import("../archive.zig");
const Inode = @import("../inode.zig");
const ExtractionOptions = @import("../options.zig");
const Tables = @import("../tables.zig");
const InodeFinish = @This();
const FinishEnum = enum {
wg,
fin,
};
pub const FinishUnion = union(FinishEnum) {
wg: *WaitGroup,
fin: *InodeFinish,
pub fn finish(self: FinishUnion) void {
switch (self) {
.wg => |wg| wg.finish(),
.fin => |fin| fin.finish(),
}
}
};
alloc: std.mem.Allocator,
inode: Inode,
path: []const u8,
tables: *Tables,
options: ExtractionOptions,
parent_finish: FinishUnion,
fil: ?std.fs.File,
out_err: *?anyerror,
wg: WaitGroup = .{},
mut: Mutex = .{},
pub fn create(
alloc: std.mem.Allocator,
inode: Inode,
path: []const u8,
tables: *Tables,
options: ExtractionOptions,
parent_finish: FinishUnion,
out_err: *?anyerror,
fil: ?std.fs.File,
work_size: usize,
) !*InodeFinish {
if (work_size == 0)
return error.InvalidWorkSize;
const out = try alloc.create(InodeFinish);
errdefer alloc.destroy(out);
out.* = .{
.alloc = alloc,
.inode = inode,
.path = path,
.tables = tables,
.options = options,
.parent_finish = parent_finish,
.out_err = out_err,
.fil = fil,
};
out.wg.startMany(work_size);
return out;
}
pub fn logError(self: *InodeFinish, comptime fmt: []const u8, args: anytype) void {
if (self.options.verbose)
self.options.verbose_writer.?.print(fmt, args) catch {};
}
pub fn finish(self: *InodeFinish) void {
self.mut.lock();
{
defer self.mut.unlock();
self.wg.finish();
if (!self.wg.isDone()) return;
}
defer {
self.parent_finish.finish();
self.alloc.destroy(self);
}
if (self.fil == null)
self.fil = std.fs.cwd().openFile(self.path, .{}) catch |err| {
if (self.options.verbose)
self.options.verbose_writer.?.print("Error opening {s} to set metadata: {}\n", .{ self.path, err }) catch {};
self.out_err.* = err;
return;
};
defer self.fil.?.close();
self.inode.setMetadata(self.alloc, self.tables, self.fil.?, self.options) catch |err| {
if (self.options.verbose)
self.options.verbose_writer.?.print("Error setting metadata to {s}: {}\n", .{ self.path, err }) catch {};
self.out_err.* = err;
return;
};
}
-41
View File
@@ -1,41 +0,0 @@
const std = @import("std");
const OffsetFile = @import("offset_file.zig");
const MetadataCache = @This();
alloc: std.mem.Allocator,
buf: []u8,
fixed_alloc: std.heap.FixedBufferAllocator,
cache: std.AutoArrayHashMap(u64, [8192]u8),
mut: std.Thread.Mutex = .{},
cache_mut: std.AutoArrayHashMap(u64, std.Thread.Mutex),
fil: OffsetFile,
pub fn init(alloc: std.mem.Allocator, cache_size: u64) !MetadataCache {}
pub fn deinit(self: *MetadataCache) void {
self.mut.lock();
defer self.mut.unlock();
self.cache.deinit();
self.cache_mut.deinit();
self.alloc.free(self.buf);
}
pub fn getChunk(self: *MetadataCache, offset: u64) ![8192]u8 {
var res = self.cache.get(offset);
if (res != null) return res.?;
var offset_mut = blk: {
self.mut.lock();
defer self.mut.unlock();
const mut = try self.cache_mut.getOrPut(offset);
if (!mut.found_existing)
mut.value_ptr.* = .{};
break :blk mut.value_ptr;
};
offset_mut.lock();
defer offset_mut.unlock();
}
+29 -23
View File
@@ -4,9 +4,9 @@ const Writer = std.Io.Writer;
const Limit = std.Io.Limit;
const StreamError = std.Io.Reader.StreamError;
const DecompFn = @import("../decomp.zig").DecompFn;
const Decompressor = @import("decompressor.zig");
const BlockHeader = packed struct {
const BlockHeader = packed struct(u16) {
size: u15,
uncompressed: bool,
};
@@ -15,28 +15,30 @@ const This = @This();
alloc: std.mem.Allocator,
rdr: *Reader,
decomp: DecompFn,
decomp: *const Decompressor,
cur_block_start: u32 = 0,
next_start_start: u32 = 0,
buf: [8192]u8 = undefined,
interface: Reader,
err: ?anyerror = null,
interface: Reader = .{
.buffer = &[0]u8{},
.end = 0,
.seek = 0,
.vtable = &.{
.stream = stream,
.discard = discard,
.readVec = readVec,
// TODO: Potentially add rebase so that we can guarentee that self.block_start & interface.seek is correct.
},
},
pub fn init(alloc: std.mem.Allocator, rdr: *Reader, decomp: DecompFn) This {
pub fn init(alloc: std.mem.Allocator, rdr: *Reader, decomp: *const Decompressor) This {
return .{
.alloc = alloc,
.rdr = rdr,
.decomp = decomp,
.interface = .{
.buffer = &[0]u8{},
.end = 0,
.seek = 0,
.vtable = &.{
.stream = stream,
.discard = discard,
.readVec = readVec,
},
},
};
}
@@ -44,16 +46,20 @@ fn advance(self: *This) !void {
self.interface.seek = 0;
var hdr: BlockHeader = undefined;
try self.rdr.readSliceEndian(BlockHeader, @ptrCast(&hdr), .little);
self.cur_block_start = self.next_start_start;
self.next_start_start += hdr.size;
if (hdr.uncompressed) {
try self.rdr.readSliceEndian(u8, self.buf[0..hdr.size], .little);
self.interface.end = hdr.size;
self.interface.buffer = self.buf[0..hdr.size];
return;
} else {
@branchHint(.likely);
var tmp_buf: [8192]u8 = undefined;
try self.rdr.readSliceAll(tmp_buf[0..hdr.size]);
self.interface.end = try self.decomp.Decompress(self.alloc, tmp_buf[0..hdr.size], &self.buf);
self.interface.buffer = self.buf[0..self.interface.end];
}
var tmp_buf: [8192]u8 = undefined;
try self.rdr.readSliceAll(tmp_buf[0..hdr.size]);
self.interface.end = try self.decomp(self.alloc, tmp_buf[0..hdr.size], &self.buf);
self.interface.buffer = self.buf[0..self.interface.end];
}
fn stream(rdr: *Reader, wrt: *Writer, limit: Limit) StreamError!usize {
@@ -68,22 +74,22 @@ fn stream(rdr: *Reader, wrt: *Writer, limit: Limit) StreamError!usize {
self.interface.seek += wrote;
return wrote;
}
fn discard(rdr: *Reader, limit: Limit) error{ EndOfStream, ReadFailed }!usize {
fn discard(rdr: *Reader, limit: Limit) Reader.Error!usize {
const self: *This = @fieldParentPtr("interface", rdr);
if (rdr.end == rdr.seek) self.advance() catch |err| {
self.err = err;
return StreamError.ReadFailed;
return error.ReadFailed;
};
if (@intFromEnum(limit) == 0) return 0;
const to_skip = @min(rdr.end - rdr.seek, @intFromEnum(limit));
rdr.seek += to_skip;
return to_skip;
}
fn readVec(rdr: *Reader, vec: [][]u8) error{ EndOfStream, ReadFailed }!usize {
fn readVec(rdr: *Reader, vec: [][]u8) Reader.Error!usize {
const self: *This = @fieldParentPtr("interface", rdr);
if (rdr.end == rdr.seek) self.advance() catch |err| {
self.err = err;
return StreamError.ReadFailed;
return error.ReadFailed;
};
var cur_red: usize = 0;
for (vec) |s| {
+25
View File
@@ -0,0 +1,25 @@
//! Miscellaneous utility functions.
const std = @import("std");
const Io = std.Io;
const Inode = @import("../inode.zig");
const Decompressor = @import("decompressor.zig");
const MetadataReader = @import("metadata.zig");
const OffsetFile = @import("offset_file.zig");
/// check is the path is referencing itself ("" or ".").
/// separators must be trimmed before calling this function for it to work properly.
pub fn pathIsSelf(path: []const u8) bool {
if (path.len == 0) return true;
if (path.len > 1) return false;
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);
try meta.interface.discardAll(ref.block_offset);
return .read(alloc, &meta.interface, block_size);
}
+8 -4
View File
@@ -1,8 +1,9 @@
//! A File where it's meaningful (to us) content starts at a given offset.
const std = @import("std");
const File = std.fs.File;
const Reader = std.fs.File.Reader;
const Io = std.Io;
const File = Io.File;
const Reader = File.Reader;
const OffsetFile = @This();
@@ -13,8 +14,11 @@ pub fn init(fil: File, init_offset: u64) OffsetFile {
return .{ .fil = fil, .offset = init_offset };
}
pub fn readerAt(self: OffsetFile, offset: u64, buffer: []u8) !Reader {
var rdr = self.fil.reader(buffer);
pub fn readerAt(self: OffsetFile, io: Io, offset: u64, buffer: []u8) Reader.SeekError!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) File.ReadPositionalError!void {
_ = try self.fil.readPositionalAll(io, buf, self.offset + offset);
}
+52
View File
@@ -0,0 +1,52 @@
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);
}
-117
View File
@@ -1,117 +0,0 @@
const std = @import("std");
const DecompFn = @import("decomp.zig").DecompFn;
const Table = @import("tables.zig").Table;
const MetadataReader = @import("util/metadata.zig");
const OffsetFile = @import("util/offset_file.zig");
const Ref = packed struct {
block_offset: u16,
block_start: u32,
_: u16,
};
const Entry = packed struct {
ref: Ref,
count: u32,
size: u32,
};
const KeyPrefix = enum(u8) {
user,
trusted,
security,
};
const KeyRaw = packed struct {
type: packed struct {
prefix: KeyPrefix,
out_of_line: bool,
_: u7,
},
name_size: u16,
};
pub const KeyValue = struct {
key: [:0]u8,
value: []u8,
};
const XattrTable = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: DecompFn,
count: u32,
start: u64,
table: Table(Entry),
pub fn init(alloc: std.mem.Allocator, fil: OffsetFile, decomp: DecompFn, table_start: u64) !XattrTable {
var info = packed struct {
start: u64 = undefined,
count: u32 = undefined,
_: u32 = undefined,
}{};
var rdr = try fil.readerAt(table_start, &[0]u8{});
try rdr.interface.readSliceEndian(@TypeOf(info), @ptrCast(&info), .little);
return .{
.alloc = alloc,
.fil = fil,
.decomp = decomp,
.count = info.count,
.start = info.start,
.table = try .init(alloc, fil, decomp, table_start + 16, info.count),
};
}
pub fn deinit(self: *XattrTable) void {
self.table.deinit();
}
pub fn get(self: *XattrTable, alloc: std.mem.Allocator, idx: u32) ![]KeyValue {
const entry: Entry = try self.table.get(idx);
const out = try alloc.alloc(KeyValue, entry.count);
for (out) |*kv| {
var rdr = try self.fil.readerAt(self.start + entry.ref.block_start, &[0]u8{});
var meta: MetadataReader = .init(alloc, &rdr.interface, self.decomp);
try meta.interface.discardAll(entry.ref.block_offset);
var key_raw: KeyRaw = undefined;
try meta.interface.readSliceEndian(KeyRaw, @ptrCast(&key_raw), .little);
switch (key_raw.type.prefix) {
.user => {
kv.key = @ptrCast(try alloc.alloc(u8, key_raw.name_size + 5 + 1));
@memcpy(kv.key[0..5], "user.");
try meta.interface.readSliceAll(kv.key[5 .. kv.key.len - 1]);
kv.key[kv.key.len - 1] = 0;
},
.security => {
kv.key = @ptrCast(try alloc.alloc(u8, key_raw.name_size + 9 + 1));
@memcpy(kv.key[0..9], "security.");
try meta.interface.readSliceAll(kv.key[9 .. kv.key.len - 1]);
kv.key[kv.key.len - 1] = 0;
},
.trusted => {
kv.key = @ptrCast(try alloc.alloc(u8, key_raw.name_size + 8 + 1));
@memcpy(kv.key[0..8], "trusted.");
try meta.interface.readSliceAll(kv.key[8 .. kv.key.len - 1]);
kv.key[kv.key.len - 1] = 0;
},
}
if (key_raw.type.out_of_line) {
try meta.interface.discardAll(4);
var ref: Ref = undefined;
try meta.interface.readSliceEndian(Ref, @ptrCast(&ref), .little);
rdr = try self.fil.readerAt(self.start + ref.block_start, &[0]u8{});
meta = .init(alloc, &rdr.interface, self.decomp);
try meta.interface.discardAll(ref.block_offset);
}
var value_size: u32 = undefined;
try meta.interface.readSliceEndian(u32, @ptrCast(&value_size), .little);
kv.value = try alloc.alloc(u8, value_size);
try meta.interface.readSliceAll(kv.value);
}
return out;
}
+286
View File
@@ -0,0 +1,286 @@
const std = @import("std");
const Io = std.Io;
const InodeRef = @import("inode.zig").Ref;
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 XattrCachedTable = @This();
alloc: std.mem.Allocator,
fil: OffsetFile,
decomp: *const Decompressor,
kv_start: u64,
table: LookupTable.CachedTable(TableValue),
value_cache: std.AutoHashMap(InodeRef, []const u8),
value_mut: Io.Mutex = .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);
return .{
.alloc = alloc,
.fil = fil,
.decomp = decomp,
.kv_start = start,
.table = .init(alloc, fil, decomp, xattr_start + 16, num),
.value_cache = .init(alloc),
};
}
pub fn deinit(self: *XattrCachedTable, io: Io) void {
self.table.deinit(io);
self.value_cache.deinit();
}
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);
try meta.interface.discardAll(lookup.ref.block_offset);
const out = try alloc.alloc(XattrSemiOwned, lookup.count);
errdefer alloc.free(out);
for (0..lookup.count) |i| {
var key_entry: KeyEntry = undefined;
try meta.interface.readSliceEndian(KeyEntry, @ptrCast(&key_entry), .little);
const key: [:0]u8 = switch (key_entry.type.namespace) {
.user => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 5);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[5 .. tmp.len - 1], .little);
@memcpy(tmp[0..5], "user.");
tmp[tmp.len - 1] = 0;
break :blk @ptrCast(tmp);
},
.trusted => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 8);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[8 .. tmp.len - 1], .little);
@memcpy(tmp[0..8], "trusted.");
tmp[tmp.len - 1] = 0;
break :blk @ptrCast(tmp);
},
.security => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 9);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[9 .. tmp.len - 1], .little);
@memcpy(tmp[0..9], "security.");
tmp[tmp.len - 1] = 0;
break :blk @ptrCast(tmp);
},
};
errdefer alloc.free(key);
if (key_entry.type.out_of_line) {
var value: ValueOutOfLineEntry = undefined;
try meta.interface.readSliceEndian(ValueOutOfLineEntry, @ptrCast(&value), .little);
out[i] = .{
.key = key,
.value = try self.valueAt(io, value.ref),
};
continue;
}
const val_ref: InodeRef = .{ .block_start = meta.cur_block_start, .block_offset = @truncate(meta.interface.seek) };
try self.value_mut.lock(io);
defer self.value_mut.unlock(io);
if (self.value_cache.contains(val_ref)) {
out[i] = .{
.key = key,
.value = try self.valueAt(io, val_ref),
};
continue;
}
var val_size: u32 = undefined;
try meta.interface.readSliceEndian(u32, @ptrCast(&val_size), .little);
const val = try self.alloc.alloc(u8, val_size);
errdefer alloc.free(val);
try meta.interface.readSliceEndian(u8, val, .little);
try self.value_cache.put(val_ref, val);
out[i] = .{
.key = key,
.value = val,
};
}
return out;
}
fn valueAt(self: *XattrCachedTable, io: Io, ref: InodeRef) ![]const u8 {
try self.value_mut.lock(io);
defer self.value_mut.unlock(io);
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);
try meta.interface.discardAll(ref.block_offset);
var val_size: u32 = undefined;
try meta.interface.readSliceEndian(u32, @ptrCast(&val_size), .little);
const val = try self.alloc.alloc(u8, val_size);
errdefer self.alloc.free(val);
try meta.interface.readSliceEndian(u8, val, .little);
try self.value_cache.put(ref, val);
return val;
}
// Types
/// An Xattr return value where the reciever only owns the key value.
pub const XattrSemiOwned = struct {
key: [:0]const u8,
value: []const u8,
pub fn deinit(self: XattrSemiOwned, alloc: std.mem.Allocator) void {
alloc.free(self.key);
}
};
/// An Xattr return value where the reciever owns both the key & value.
pub const XattrOwned = struct {
key: [:0]const u8,
value: []const u8,
pub fn deinit(self: XattrSemiOwned, alloc: std.mem.Allocator) void {
alloc.free(self.key);
alloc.free(self.value);
}
};
const TableValue = extern struct {
ref: InodeRef,
count: u32,
size: u32,
};
const KeyEntry = extern struct {
type: XattrPrefix,
name_size: u16,
};
const ValueOutOfLineEntry = extern struct {
_: u32,
ref: InodeRef,
};
const XattrPrefix = packed struct(u16) {
namespace: enum(u8) {
user,
trusted,
security,
fn prefixSize(self: @This()) u16 {
return switch (self) {
.user => 5,
.trusted => 8,
.security => 9,
};
}
},
out_of_line: bool,
_: u7,
};
// 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);
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 meta: MetadataReader = .init(alloc, &rdr.interface, decomp);
try meta.interface.discardAll(lookup.ref.block_offset);
const out = try alloc.alloc(XattrOwned, lookup.count);
errdefer alloc.free(out);
for (0..lookup.count) |i| {
const key_entry: KeyEntry = undefined;
try meta.interface.readSliceEndian(KeyEntry, @ptrCast(&key_entry), .little);
const key = switch (key_entry.type.namespace) {
.user => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 5);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[5 .. tmp.len - 1], .little);
@memset(tmp[0..5], "user.");
break :blk tmp;
},
.trusted => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 8);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[8 .. tmp.len - 1], .little);
@memset(tmp[0..8], "trusted.");
break :blk tmp;
},
.security => blk: {
const tmp = try alloc.alloc(u8, key_entry.name_size + 1 + 9);
errdefer alloc.free(tmp);
try meta.interface.readSliceEndian(u8, tmp[9 .. tmp.len - 1], .little);
@memset(tmp[0..9], "security.");
break :blk tmp;
},
};
key[key.len - 1] = 0;
errdefer alloc.free(key);
if (key_entry.type.out_of_line) {
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_meta: MetadataReader = .init(alloc, &ool_rdr.interface, decomp);
try ool_meta.interface.discardAll(value.ref.block_offset);
var val_size: u32 = undefined;
try ool_meta.interface.readSliceEndian(val_size, @ptrCast(&val_size), .little);
const val = try alloc.alloc(u8, val_size);
errdefer alloc.free(val);
try ool_meta.interface.readSliceEndian(u8, val, .little);
out[i] = .{
.key = key,
.value = val,
};
continue;
}
var val_size: u32 = undefined;
try meta.interface.readSliceEndian(val_size, @ptrCast(&val_size), .little);
const val = try alloc.alloc(u8, val_size);
errdefer alloc.free(val);
try meta.interface.readSliceEndian(u8, val, .little);
out[i] = .{
.key = key,
.value = val,
};
}
return out;
}