More foundational work on extraction

This commit is contained in:
Caleb Gardner
2026-06-11 06:03:45 -05:00
parent 0bac064252
commit f3318e55a3
7 changed files with 410 additions and 81 deletions
+103
View File
@@ -0,0 +1,103 @@
const std = @import("std");
const Io = std.Io;
const Decomp = @import("../decomp.zig");
const DataBlock = @import("../inode.zig").DataBlock;
const Extractor = @This();
data: []u8,
decomp: Decomp.Fn,
block_size: u32,
blocks: []DataBlock,
start: u64,
size: u64,
frag_data: ?[]u8 = null,
frag_offset: u32 = 0,
pub fn init(data: []u8, decomp: Decomp.Fn, block_size: u32, blocks: []DataBlock, start: u64, size: u64) Extractor {
return .{
.data = data,
.decomp = decomp,
.block_size = block_size,
.blocks = blocks,
.start = start,
.size = size,
};
}
pub fn addFrag(self: *Extractor, frag_data: []u8, frag_offset: u32) void {
self.frag_data = frag_data;
self.frag_offset = frag_offset;
}
pub fn extractAsync(self: Extractor, alloc: std.mem.Allocator, io: Io, file: Io.File) Error!void {
if (self.size == 0) return;
try file.writePositionalAll(io, &[0]u8{}, self.size);
var map = try file.createMemoryMap(io, .{
.len = self.size,
.protection = .{ .write = true },
});
defer map.destroy(io);
var err: ?Error = null;
var group: Io.Group = .init;
var read_offset: u64 = self.start;
for (0.., self.blocks) |i, block| {
group.async(io, blockThread, .{ self, alloc, map.memory, read_offset, i, &err });
read_offset += block.size;
}
if (self.frag_data != null)
group.async(io, fragThread, .{ self, map.memory });
try group.await(io);
if (err != null)
return err;
try map.write(io);
}
fn blockThread(self: Extractor, alloc: std.mem.Allocator, map_data: []u8, read_offset: u64, block_idx: u32, err: *?Error) error{Canceled}!void {
const size = if (self.frag_data == null and block_idx == (self.size - 1 / self.block_size))
self.size % self.block_size
else
self.block_size;
const offset = self.block_size * block_idx;
const block = self.blocks[block_idx];
if (block.size == 0) {
@memset(map_data[offset..][0..size], 0);
return;
}
const data = self.data[read_offset..][0..block.size];
if (block.uncompressed) {
@memcpy(map_data[offset..][0..block.size], data);
return;
}
_ = self.decomp(alloc, data, map_data[offset..][0..size]) catch |inner_err| {
err.* = inner_err;
return;
};
}
fn fragThread(self: Extractor, map_data: []u8) error{Canceled}!void {
const size = self.size % self.block_size;
const offset = self.blocks.len * self.block_size;
@memcpy(map_data[offset..][0..size], self.frag_data.?[self.frag_offset..][0..size]);
}
// Types
const Error = Io.File.WritePositionalError || Io.File.MemoryMap.CreateError || Decomp.Error;
+155
View File
@@ -0,0 +1,155 @@
const std = @import("std");
const Io = std.Io;
const Decomp = @import("../decomp.zig");
const DataBlock = @import("../inode.zig").DataBlock;
const Reader = @This();
alloc: std.mem.Allocator,
data: []u8,
decomp: Decomp.Fn,
block_size: u32,
blocks: []DataBlock,
size: u64,
offset: u64,
block_idx: u32 = 0,
sparse_block: bool = false,
frag_data: ?[]u8 = null,
frag_offset: u32 = 0,
block: [1024 * 1024]u8 = undefined,
interface: Io.Reader = .{
.buffer = &[0]u8{},
.end = 0,
.seek = 0,
.vtable = &.{
.stream = stream,
.discard = discard,
.readVec = readVec,
},
},
pub fn init(alloc: std.mem.Allocator, data: []u8, decomp: Decomp.Fn, block_size: u32, blocks: []DataBlock, size: u64, data_start: u64) Reader {
return .{
.alloc = alloc,
.data = data,
.decomp = decomp,
.block_size = block_size,
.blocks = blocks,
.size = size,
.offset = data_start,
};
}
pub fn addFrag(self: *Reader, frag_data: []u8, frag_offset: u32) void {
self.frag_data = frag_data;
self.frag_offset = frag_offset;
}
fn advance(self: *Reader) Io.Reader.Error!void {
if (self.block_idx > self.blocks.len) return error.EndOfStream;
defer self.block_idx += 1;
self.interface.seek = 0;
errdefer self.interface.end = 0;
if (self.block_idx == self.blocks.len) {
if (self.frag_data == null) return error.EndOfStream;
self.sparse_block = false;
const size = self.size % self.block_size;
self.interface.buffer = self.frag_data.?[self.frag_offset..][0..size];
self.interface.end = size;
return;
}
const size = if (self.frag_data == null and self.block_idx == self.blocks.len - 1)
self.size % self.block_size
else
self.block_size;
const block = self.blocks[self.block_idx];
defer self.offset += block.size;
if (block.size == 0) {
self.sparse_block = true;
self.end = size;
return;
} else {
self.sparse_block = false;
}
if (block.uncompressed) {
self.interface.buffer = self.data[self.offset..][0..block.size];
self.interface.end = block.size;
return;
}
_ = self.decomp(self.alloc, self.data[self.offset..][0..block.size], self.block[0..size]) catch return error.ReadFailed;
self.interface.buffer = self.block[0..size];
self.interface.end = size;
}
fn stream(r: *Io.Reader, w: *Io.Writer, limit: Io.Limit) Io.Reader.StreamError!usize {
const self: *Reader = @fieldParentPtr("interface", r);
if (r.seek >= r.end)
try self.advance();
if (limit == .nothing) return;
const to_write = @min(@intFromEnum(limit), r.end - r.seek);
const wrote = if (self.sparse_block)
try w.splatByte(0, to_write)
else
try w.write(r.buffer[r.seek..][0..to_write]);
r.seek += wrote;
return wrote;
}
fn discard(r: *Io.Reader, limit: Io.Limit) Io.Reader.Error!usize {
if (r.seek >= r.end) {
const self: *Reader = @fieldParentPtr("interface", r);
try self.advance();
}
if (limit == .nothing) return;
const to_discard = @min(@intFromEnum(limit), r.end - r.seek);
r.seek += to_discard;
return to_discard;
}
fn readVec(r: *Io.Reader, vec: [][]u8) Io.Reader.Error!usize {
const self: *Reader = @fieldParentPtr("interface", r);
if (r.seek >= r.end)
try self.advance();
var copied: usize = 0;
for (vec) |v| {
const to_cpy = @min(v.len, r.end - r.seek);
if (self.sparse_block)
@memset(v[0..to_cpy], 0)
else
@memcpy(v[0..to_cpy], r.buffer[r.seek..][0..to_cpy]);
copied += to_cpy;
if (r.seek >= r.end) break;
}
return copied;
}
+16 -2
View File
@@ -1,10 +1,12 @@
const std = @import("std");
const Io = std.Io;
const Superblock = @import("archive.zig").Superblock;
const Inode = @import("inode.zig");
const Decomp = @import("decomp.zig");
const ExtractionOptions = @import("options.zig");
const Inode = @import("inode.zig");
const Lookup = @import("lookup.zig");
const Superblock = @import("archive.zig").Superblock;
const XattrTable = @import("xattr.zig");
pub fn extract(
alloc: std.mem.Allocator,
@@ -16,3 +18,15 @@ pub fn extract(
ext_loc: []const u8,
options: ExtractionOptions,
) !void {}
pub fn extractReal(
alloc: std.mem.Allocator,
io: Io,
super: Superblock,
data: []u8,
decomp: Decomp.Fn,
frag_table: Lookup.Table(u64),
inode: Inode,
ext_loc: []const u8,
options: ExtractionOptions,
) !void {}
+18 -2
View File
@@ -1,10 +1,12 @@
const std = @import("std");
const Io = std.Io;
const Superblock = @import("archive.zig").Superblock;
const Inode = @import("inode.zig");
const Decomp = @import("decomp.zig");
const ExtractionOptions = @import("options.zig");
const Inode = @import("inode.zig");
const Lookup = @import("lookup.zig");
const Superblock = @import("archive.zig").Superblock;
const XattrTable = @import("xattr.zig");
pub fn extract(
alloc: std.mem.Allocator,
@@ -16,3 +18,17 @@ pub fn extract(
ext_loc: []const u8,
options: ExtractionOptions,
) !void {}
pub fn extractReal(
alloc: std.mem.Allocator,
io: Io,
super: Superblock,
data: []u8,
decomp: Decomp.Fn,
frag_table: Lookup.Table(u64),
id_table: Lookup.Table(u16),
xattr_table: XattrTable,
inode: Inode,
ext_loc: []const u8,
options: ExtractionOptions,
) !void {}
+31 -77
View File
@@ -2,6 +2,8 @@ const std = @import("std");
const Io = std.Io;
const Decomp = @import("decomp.zig");
const MetadataReader = @import("meta_rdr.zig");
const ProtectedMap = @import("util/protected_map.zig");
pub fn Table(comptime T: anytype) type {
return struct {
@@ -15,91 +17,43 @@ pub fn Table(comptime T: anytype) type {
table_start: u64,
table_num: u32,
pub fn init(alloc: std.mem.Allocator, table_start: u64, table_num: u32) Self {}
pub fn deinit(self: *Self) void {}
table: ProtectedMap(u32, []T, getBlock),
pub fn get(self: *Self, io: Io, idx: u32) !T {}
};
}
pub fn ProtectedMap(comptime K: anytype, comptime T: anytype, comptime create_fn: anytype) type {
std.debug.assert(std.meta.activeTag(@typeInfo(create_fn)) == .@"fn");
const ret_info = @typeInfo(create_fn).@"fn".return_type;
std.debug.assert(ret_info == T or
(std.meta.activeTag(@typeInfo(ret_info)) == .error_union and @typeInfo(ret_info).error_union.payload == T));
return struct {
const Map = @This();
alloc: std.mem.Allocator,
map: std.AutoHashMap(K, ProtectedValue),
mut: Io.RwLock = .init,
pub fn init(alloc: std.mem.Allocator, default_value: T) Map {
return .{
.alloc = alloc,
.map = .init(alloc),
.default_value = default_value,
};
pub fn init(alloc: std.mem.Allocator, data: []u8, decomp: Decomp.Fn, table_start: u64, table_num: u32) Self {
return .{ .data = data, .decomp = decomp, .table_start = table_start, .table_num = table_num, .table = .init(alloc) };
}
pub fn deinit(self: *Map) void {}
pub fn deinit(self: *Self) void {
var iter = self.table.map.valueIterator();
pub fn getOrPut(self: *Map, io: Io, key: K, create_fn_args: std.meta.ArgsTuple(create_fn)) !*T {
{
try self.mut.lockShared(io);
defer self.mut.unlockShared(io);
while (iter.next()) |val| {
if (val.err != null) continue;
const value = self.map.getPtr(key);
if (value != null) {
if (!value.?.filled.isSet())
value.?.filled.wait(io);
if (value.?.err != null) return value.?.err != null;
return value.?.value;
}
}
try self.mut.lock(io);
const res = self.map.getOrPut(key) catch |err| {
self.mut.unlock(io);
return err;
};
if (res.found_existing) {
self.mut.unlock(io);
return self.getOrPut(io, key, create_fn_args);
}
res.value_ptr.* = .{};
defer res.value_ptr.filled.set(io);
res.mut.unlock(io);
self.mut.lockSharedUncancelable(io);
defer self.mut.unlockShared(io);
if (@TypeOf(CreateError) == void) {
res.value_ptr.value = @call(.auto, create_fn, create_fn_args);
} else {
res.value_ptr.value = @call(.auto, create_fn, create_fn_args) catch |err| {
res.value_ptr.err = err;
};
self.table.alloc.free(val.value);
}
self.table.deinit();
}
// Map Types
pub fn get(self: *Self, io: Io, idx: u32) !T {
const block = idx / VALUES_PER_BLOCK;
const block_idx = idx % VALUES_PER_BLOCK;
const CreateError: type = switch (@typeInfo(@typeInfo(create_fn).@"fn".return_type)) {
.error_union => |e| e.error_set,
else => void,
};
const values = try self.table.getOrPut(io, block, .{ self.*, block });
const ProtectedValue = struct {
value: T = undefined,
err: ?CreateError = null,
filled: Io.Event = .unset,
};
return values[block_idx];
}
pub fn getBlock(self: Self, block_idx: u32) ![]T {
const offset: u64 = std.mem.readInt(u64, self.data[self.table_start + (block_idx * 8) ..][0..8], .little);
const block = try self.table.alloc(T, if (block_idx == (self.table_num - 1 / VALUES_PER_BLOCK))
self.table_num % VALUES_PER_BLOCK
else
VALUES_PER_BLOCK);
var meta: MetadataReader = .init(self.table.alloc, self.data, self.decomp, offset);
try meta.interface.readSliceEndian(T, block, .little);
return block;
}
};
}
+84
View File
@@ -0,0 +1,84 @@
const std = @import("std");
const Io = std.Io;
pub fn ProtectedMap(comptime K: anytype, comptime T: anytype, comptime create_fn: anytype) type {
std.debug.assert(std.meta.activeTag(@typeInfo(create_fn)) == .@"fn");
const ret_info = @typeInfo(create_fn).@"fn".return_type;
std.debug.assert(ret_info == T or
(std.meta.activeTag(@typeInfo(ret_info)) == .error_union and @typeInfo(ret_info).error_union.payload == T));
return struct {
const Map = @This();
alloc: std.mem.Allocator,
map: std.AutoHashMap(K, ProtectedValue),
mut: Io.RwLock = .init,
pub fn init(alloc: std.mem.Allocator) Map {
return .{
.alloc = alloc,
.map = .init(alloc),
};
}
pub fn deinit(self: *Map) void {
self.map.deinit();
}
pub fn getOrPut(self: *Map, io: Io, key: K, create_fn_args: std.meta.ArgsTuple(create_fn)) !*T {
{
try self.mut.lockShared(io);
defer self.mut.unlockShared(io);
const value = self.map.getPtr(key);
if (value != null) {
if (!value.?.filled.isSet())
value.?.filled.wait(io);
if (value.?.err != null) return value.?.err != null;
return value.?.value;
}
}
try self.mut.lock(io);
const res = self.map.getOrPut(key) catch |err| {
self.mut.unlock(io);
return err;
};
if (res.found_existing) {
self.mut.unlock(io);
return self.getOrPut(io, key, create_fn_args);
}
res.value_ptr.* = .{};
defer res.value_ptr.filled.set(io);
res.mut.unlock(io);
self.mut.lockSharedUncancelable(io);
defer self.mut.unlockShared(io);
if (@TypeOf(CreateError) == void) {
res.value_ptr.value = @call(.auto, create_fn, create_fn_args);
} else {
res.value_ptr.value = @call(.auto, create_fn, create_fn_args) catch |err| {
res.value_ptr.err = err;
};
}
}
// Map Types
const CreateError: type = switch (@typeInfo(@typeInfo(create_fn).@"fn".return_type)) {
.error_union => |e| e.error_set,
else => void,
};
const ProtectedValue = struct {
value: T = undefined,
err: ?CreateError = null,
filled: Io.Event = .unset,
};
};
}
+3
View File
@@ -0,0 +1,3 @@
const std = @import("std");
const XattrTable = @This();