Several fixes for File.extract

Tweaks and fixes.
This commit is contained in:
Caleb Gardner
2025-05-25 05:34:32 -05:00
parent 82011a092c
commit dd452060cb
7 changed files with 101 additions and 71 deletions
+50 -24
View File
@@ -1,6 +1,7 @@
const std = @import("std"); const std = @import("std");
const io = std.io; const io = std.io;
const fs = std.fs; const fs = std.fs;
const builtin = @import("builtin");
const inode = @import("inode/inode.zig"); const inode = @import("inode/inode.zig");
const directory = @import("directory.zig"); const directory = @import("directory.zig");
@@ -36,16 +37,17 @@ pub const File = struct {
); );
defer meta_rdr.deinit(); defer meta_rdr.deinit();
try meta_rdr.skip(ent.offset); try meta_rdr.skip(ent.offset);
const out: File = .{ const name = try rdr.alloc.alloc(u8, ent.name.len);
.name = try rdr.alloc.alloc(u8, ent.name.len), errdefer rdr.alloc.free(name);
@memcpy(name, ent.name);
return .{
.name = name,
.inode = try .init( .inode = try .init(
rdr.alloc, rdr.alloc,
meta_rdr.any(), meta_rdr.any(),
rdr.super.block_size, rdr.super.block_size,
), ),
}; };
@memcpy(out.name, ent.name);
return out;
} }
pub fn deinit(self: *File, alloc: std.mem.Allocator) void { pub fn deinit(self: *File, alloc: std.mem.Allocator) void {
@@ -159,7 +161,7 @@ pub const File = struct {
return self.data_rdr.?.read(bytes); return self.data_rdr.?.read(bytes);
} }
pub const FileReader = io.GenericReader(*File, (FileError || anyerror), read); const FileReader = io.GenericReader(*File, (FileError || anyerror), read);
pub fn reader(self: *File) FileReader { pub fn reader(self: *File) FileReader {
return .{ return .{
@@ -199,30 +201,54 @@ pub const File = struct {
} }
var iter = try self.iterator(rdr); var iter = try self.iterator(rdr);
defer iter.deinit(); defer iter.deinit();
while (iter.next()) |f| { while (iter.next()) |*f| {
//TODO: Threading const extr_path = try std.mem.concat(rdr.alloc, u8, &[3][]const u8{ real_path, "/", f.name });
try f.extractReal(rdr, path + "/" + f.name, false); defer rdr.alloc.free(extr_path);
try @constCast(f).extractReal(rdr, extr_path, false);
} }
}, },
.file, .ext_file => { .file, .ext_file => {
if (exists) { if ((!first and exists) or
if (!first) { (first and exists and stat.?.kind != .directory)) return ExtractError.FileExists;
return ExtractError.FileExists; const extr_path = if (first and exists and stat.?.kind == .directory) blk: {
} else { break :blk try std.mem.concat(rdr.alloc, u8, &[3][]const u8{ real_path, "/", self.name });
if (stat.?.kind == .directory) { } else blk: {
real_path += "/" + self.name; const tmp = try rdr.alloc.alloc(u8, real_path.len);
} else { @memcpy(tmp, real_path);
return ExtractError.FileExists; break :blk tmp;
} };
} defer rdr.alloc.free(extr_path);
} var ext = try self.extractor(rdr);
const ext = try self.extractor(rdr);
defer ext.deinit(); defer ext.deinit();
const fil = try fs.cwd().createFile(real_path, .{}); const fil = try fs.cwd().createFile(extr_path, .{});
try ext.writeToFile(fil); defer fil.close();
try ext.writeToFile(try .init(), &fil);
}, },
.sym, .ext_sym => {}, .sym, .ext_sym => {
.block, .ext_block, .char, .ext_char => {}, if (exists) return ExtractError.FileExists;
try fs.cwd().symLink(try self.symPath(), real_path, .{});
},
.block, .ext_block, .char, .ext_char, .fifo, .ext_fifo => {
if (exists) return ExtractError.FileExists;
comptime if (builtin.os.tag != .linux) return;
const IFCHR: u32 = 0o020000;
const IFBLK: u32 = 0o060000;
const IFIFO: u32 = 0o010000;
const mode = switch (self.inode.header.inode_type) {
.block, .ext_block => IFBLK,
.char, .ext_char => IFCHR,
.fifo, .ext_fifo => IFIFO,
else => unreachable,
};
const dev = switch (self.inode.data) {
.block, .char => |b| b.device,
.ext_block, .ext_char => |b| b.device,
.fifo, .ext_fifo => 0,
else => unreachable,
};
_ = std.os.linux.mknod(@ptrCast(real_path), mode, dev);
},
.sock, .ext_sock => {},
} }
//TODO: permissions //TODO: permissions
} }
+2 -2
View File
@@ -15,7 +15,7 @@ pub const FileInode = struct {
blocks: []const BlockSize, blocks: []const BlockSize,
pub fn init(alloc: std.mem.Allocator, rdr: io.AnyReader, block_size: u32) !FileInode { pub fn init(alloc: std.mem.Allocator, rdr: io.AnyReader, block_size: u32) !FileInode {
const fixed_buf = [16]u8{}; var fixed_buf = [1]u8{0} ** 16;
_ = try rdr.readAll(&fixed_buf); _ = try rdr.readAll(&fixed_buf);
const frag_idx = std.mem.bytesToValue(u32, fixed_buf[4..8]); const frag_idx = std.mem.bytesToValue(u32, fixed_buf[4..8]);
const size = std.mem.bytesToValue(u32, fixed_buf[12..16]); const size = std.mem.bytesToValue(u32, fixed_buf[12..16]);
@@ -49,7 +49,7 @@ pub const ExtFileInode = struct {
blocks: []const BlockSize, blocks: []const BlockSize,
pub fn init(alloc: std.mem.Allocator, rdr: io.AnyReader, block_size: u32) !ExtFileInode { pub fn init(alloc: std.mem.Allocator, rdr: io.AnyReader, block_size: u32) !ExtFileInode {
var fixed_buf = [40]u8{}; var fixed_buf = [1]u8{0} ** 40;
_ = try rdr.readAll(&fixed_buf); _ = try rdr.readAll(&fixed_buf);
const size = std.mem.bytesToValue(u64, fixed_buf[8..16]); const size = std.mem.bytesToValue(u64, fixed_buf[8..16]);
const frag_idx = std.mem.bytesToValue(u32, fixed_buf[28..32]); const frag_idx = std.mem.bytesToValue(u32, fixed_buf[28..32]);
+9 -2
View File
@@ -80,8 +80,6 @@ pub const Reader = struct {
test "root iter" { test "root iter" {
const test_sfs_path = "testing/LinuxPATest.sfs"; const test_sfs_path = "testing/LinuxPATest.sfs";
// const test_file_path = "PortableApps/PortableApps.com/Data/PortableAppsMenu.ini";
var rdr: Reader = try .init(std.testing.allocator, test_sfs_path, 0); var rdr: Reader = try .init(std.testing.allocator, test_sfs_path, 0);
defer rdr.deinit(); defer rdr.deinit();
var rootIter = try rdr.root.iterator(&rdr); var rootIter = try rdr.root.iterator(&rdr);
@@ -90,3 +88,12 @@ test "root iter" {
std.debug.print("{s}\n", .{f.name}); std.debug.print("{s}\n", .{f.name});
} }
} }
test "extract" {
const test_sfs_path = "testing/LinuxPATest.sfs";
const extract_path = "testing/testExtract";
try std.fs.cwd().deleteTree(extract_path);
var rdr: Reader = try .init(std.testing.allocator, test_sfs_path, 0);
defer rdr.deinit();
try rdr.root.extract(&rdr, extract_path);
}
+11 -12
View File
@@ -28,7 +28,7 @@ pub const DataExtractor = struct {
max_mem: u64, max_mem: u64,
pub fn init() !Config { pub fn init() !Config {
return .{ return .{
.thread_count = try std.Thread.getCpuCount(), .thread_count = @truncate(try std.Thread.getCpuCount()),
.max_mem = comptime 1024 * 1024 * 1024, .max_mem = comptime 1024 * 1024 * 1024,
}; };
} }
@@ -42,17 +42,17 @@ pub const DataExtractor = struct {
var frag_offset: u32 = 0; var frag_offset: u32 = 0;
switch (fil.inode.data) { switch (fil.inode.data) {
.file => |f| { .file => |f| {
data_start = f.data_start;
sizes = try reader.alloc.alloc(BlockSize, f.blocks.len); sizes = try reader.alloc.alloc(BlockSize, f.blocks.len);
@memcpy(sizes, f.blocks); @memcpy(sizes, f.blocks);
data_start = f.data_start;
size = f.size; size = f.size;
frag_idx = f.frag_idx; frag_idx = f.frag_idx;
frag_offset = f.frag_offset; frag_offset = f.frag_offset;
}, },
.ext_file => |f| { .ext_file => |f| {
data_start = f.data_start;
sizes = try reader.alloc.alloc(BlockSize, f.blocks.len); sizes = try reader.alloc.alloc(BlockSize, f.blocks.len);
@memcpy(sizes, f.blocks); @memcpy(sizes, f.blocks);
data_start = f.data_start;
size = f.size; size = f.size;
frag_idx = f.frag_idx; frag_idx = f.frag_idx;
frag_offset = f.frag_offset; frag_offset = f.frag_offset;
@@ -62,25 +62,25 @@ pub const DataExtractor = struct {
var out: DataExtractor = .{ var out: DataExtractor = .{
.alloc = reader.alloc, .alloc = reader.alloc,
.decomp = reader.super.decomp, .decomp = reader.super.decomp,
.holder = reader.holder, .holder = &reader.holder,
.block_size = reader.super.block_size, .block_size = reader.super.block_size,
.sizes = sizes, .sizes = sizes,
.block_offset = try reader.alloc.alloc(u64, sizes.len), .block_offset = try reader.alloc.alloc(u64, sizes.len),
.data_start = data_start,
}; };
errdefer out.deinit(); errdefer out.deinit();
var offset: u64 = data_start; var offset: u64 = data_start;
for (0..out.block_offset) |i| { for (0.., out.block_offset) |i, _| {
out.block_offset[i] = offset; out.block_offset[i] = offset;
offset += out.sizes[i].size; offset += out.sizes[i].size;
} }
if (frag_idx != 0xFFFFFFFF) { if (frag_idx != 0xFFFFFFFF) {
const frag_entry = try reader.frag_table.getValue(frag_idx); const frag_entry = try reader.frag_table.getValue(reader, frag_idx);
var frag_rdr: DataReader = try .fromFragEntry(reader, frag_entry); var frag_rdr: DataReader = try .fromFragEntry(reader, frag_entry);
std.debug.print("{} {}\n", .{ frag_offset, frag_entry });
defer frag_rdr.deinit(); defer frag_rdr.deinit();
try frag_rdr.skip(frag_offset); try frag_rdr.skip(frag_offset);
out.frag_data = try reader.alloc.alloc(u8, size % out.block_size); out.frag_data = try reader.alloc.alloc(u8, size % out.block_size);
_ = try frag_rdr.any().readAll(out.frag_data); _ = try frag_rdr.any().readAll(out.frag_data.?);
} }
return out; return out;
} }
@@ -88,8 +88,7 @@ pub const DataExtractor = struct {
pub fn deinit(self: *DataExtractor) void { pub fn deinit(self: *DataExtractor) void {
self.alloc.free(self.sizes); self.alloc.free(self.sizes);
self.alloc.free(self.block_offset); self.alloc.free(self.block_offset);
if (self.cur_bloc.len > 0) self.alloc.free(self.cur_bloc); if (self.frag_data != null) self.alloc.free(self.frag_data.?);
if (self.frag_data != null) self.alloc.free(self.frag_data);
} }
fn processBlock(self: DataExtractor, block_ind: u32) ![]u8 { fn processBlock(self: DataExtractor, block_ind: u32) ![]u8 {
@@ -98,7 +97,7 @@ pub const DataExtractor = struct {
//TODO //TODO
} }
fn processBlockToFile(self: DataExtractor, block_ind: u32, fil: *fs.File) !void { fn processBlockToFile(self: DataExtractor, block_ind: u32, fil: *const fs.File) !void {
_ = self; _ = self;
_ = block_ind; _ = block_ind;
_ = fil; _ = fil;
@@ -110,7 +109,7 @@ pub const DataExtractor = struct {
/// Returns the amount of bytes written. /// Returns the amount of bytes written.
/// ///
/// Optimized for lower memory usage by using File.pwrite. /// Optimized for lower memory usage by using File.pwrite.
pub fn writeToFile(self: DataExtractor, conf: Config, fil: *fs.File) !void { pub fn writeToFile(self: DataExtractor, conf: Config, fil: *const fs.File) !void {
_ = self; _ = self;
_ = fil; _ = fil;
_ = conf; _ = conf;
+17 -17
View File
@@ -23,7 +23,7 @@ pub const DataReader = struct {
frag_data: ?[]u8 = null, frag_data: ?[]u8 = null,
next_block_num: u32 = 0, next_block_num: u32 = 0,
cur_bloc: []u8 = undefined, cur_bloc: []u8 = &[0]u8{},
cur_offset: u32 = 0, cur_offset: u32 = 0,
pub fn init(fil: *File, reader: *Reader) !DataReader { pub fn init(fil: *File, reader: *Reader) !DataReader {
@@ -84,17 +84,17 @@ pub const DataReader = struct {
pub fn deinit(self: *DataReader) void { pub fn deinit(self: *DataReader) void {
self.alloc.free(self.sizes); self.alloc.free(self.sizes);
if (self.cur_bloc.len > 0) self.alloc.free(self.cur_bloc); if (self.cur_bloc.len > 0) self.alloc.free(self.cur_bloc);
if (self.frag_data != null) self.alloc.free(self.frag_data); if (self.frag_data != null) self.alloc.free(self.frag_data.?);
} }
pub fn skip(self: *DataReader, offset: u32) !void { pub fn skip(self: *DataReader, offset: u32) !void {
var cur_skip: u32 = 0; var cur_skip: u32 = 0;
var to_skip: u32 = 0; var to_skip: u32 = 0;
while (cur_skip < offset) { while (cur_skip < offset) {
if (self.offset >= self.block.len) try self.readNextBlock(); if (self.cur_offset >= self.cur_bloc.len) try self.readNextBlock();
to_skip = @min(offset - cur_skip, self.block.len - self.offset); to_skip = @min(offset - cur_skip, self.cur_bloc.len - self.cur_offset);
cur_skip += to_skip; cur_skip += to_skip;
self.offset += to_skip; self.cur_offset += to_skip;
} }
} }
@@ -106,8 +106,8 @@ pub const DataReader = struct {
const siz = self.sizes[self.next_block_num]; const siz = self.sizes[self.next_block_num];
self.next_block_num += 1; self.next_block_num += 1;
if (self.next_block_num == self.sizes.len - 1 and self.frag_data != null) { if (self.next_block_num == self.sizes.len - 1 and self.frag_data != null) {
try self.sizeBlock(self.frag_data.len); try self.sizeBlock(self.frag_data.?.len);
@memcpy(self.cur_bloc, self.frag_data); @memcpy(self.cur_bloc, self.frag_data.?);
return; return;
} }
if (siz.size == 0) { if (siz.size == 0) {
@@ -120,14 +120,14 @@ pub const DataReader = struct {
_ = try self.rdr.any().readAll(self.cur_bloc); _ = try self.rdr.any().readAll(self.cur_bloc);
} else { } else {
self.alloc.free(self.cur_bloc); self.alloc.free(self.cur_bloc);
var limit = std.io.limitedReader(self.reader, siz.size); var limit = std.io.limitedReader(self.rdr, siz.size);
var dat = try self.decomp.decompress(self.alloc, limit.reader().any()); var dat = try self.decomp.decompress(self.alloc, limit.reader().any());
self.block = try dat.toOwnedSlice(); self.cur_bloc = try dat.toOwnedSlice();
} }
} }
fn sizeBlock(self: *DataReader, size: u32) !void { fn sizeBlock(self: *DataReader, size: usize) !void {
if (!self.alloc.resize(u8, size)) { if (!self.alloc.resize(self.cur_bloc, size)) {
self.alloc.free(self.cur_bloc); self.alloc.free(self.cur_bloc);
self.cur_bloc = try self.alloc.alloc(u8, size); self.cur_bloc = try self.alloc.alloc(u8, size);
} }
@@ -137,15 +137,15 @@ pub const DataReader = struct {
var cur_read: usize = 0; var cur_read: usize = 0;
var to_read: usize = 0; var to_read: usize = 0;
while (cur_read < bytes.len) { while (cur_read < bytes.len) {
if (self.offset >= self.block.len) { if (self.cur_offset >= self.cur_bloc.len) {
if (self.readNextBlock()) |err| { self.readNextBlock() catch |err| {
if (err == DataReaderError.EOF) return cur_read; if (err == DataReaderError.EOF) return cur_read;
return err; return err;
} };
} }
to_read = @min(bytes.len - cur_read, self.block.len - self.offset); to_read = @min(bytes.len - cur_read, self.cur_bloc.len - self.cur_offset);
@memcpy(bytes[cur_read..], self.block[self.offset .. @as(usize, self.offset) + to_read]); @memcpy(bytes[cur_read..], self.cur_bloc[self.cur_offset .. @as(usize, self.cur_offset) + to_read]);
self.offset += @truncate(to_read); self.cur_offset += @truncate(to_read);
cur_read += to_read; cur_read += to_read;
} }
return cur_read; return cur_read;
+2
View File
@@ -33,6 +33,8 @@ pub const FileOffsetReader = struct {
file: *File, file: *File,
offset: u64, offset: u64,
pub const Error = fs.File.PReadError;
pub fn read(self: *FileOffsetReader, bytes: []u8) !usize { pub fn read(self: *FileOffsetReader, bytes: []u8) !usize {
const red = try self.file.preadAll(bytes, self.offset); const red = try self.file.preadAll(bytes, self.offset);
self.offset += red; self.offset += red;
+10 -14
View File
@@ -13,7 +13,6 @@ pub fn Table(
comptime T: type, comptime T: type,
) type { ) type {
return struct { return struct {
alloc: std.mem.Allocator,
decomp: DecompressType, decomp: DecompressType,
table: []T = &[0]T{}, table: []T = &[0]T{},
offset: u64, offset: u64,
@@ -21,14 +20,13 @@ pub fn Table(
pub fn init(read: *Reader, offset: u64, item_count: u32) Self { pub fn init(read: *Reader, offset: u64, item_count: u32) Self {
return .{ return .{
.alloc = read.alloc,
.decomp = read.super.decomp, .decomp = read.super.decomp,
.offset = offset, .offset = offset,
.item_count = item_count, .item_count = item_count,
}; };
} }
pub fn deinit(self: *Self, alloc: std.mem.Allocator) void { pub fn deinit(self: *Self, alloc: std.mem.Allocator) void {
alloc.free(self.table); if (self.table.len == 0) alloc.free(self.table);
} }
pub fn getValue(self: *Self, read: *Reader, i: u64) !T { pub fn getValue(self: *Self, read: *Reader, i: u64) !T {
@@ -36,23 +34,21 @@ pub fn Table(
if (self.table.len > i) return self.table[i]; if (self.table.len > i) return self.table[i];
var meta_rdr: MetadataReader = undefined; var meta_rdr: MetadataReader = undefined;
var offset_rdr: FileOffsetReader = undefined; var offset_rdr: FileOffsetReader = undefined;
var meta_buf: [8]u8 = [1]u8{0} ** 8;
var meta_offset: u64 = 0; var meta_offset: u64 = 0;
var to_read: u32 = 0; var to_read: u32 = 0;
while (self.table.len <= i) { while (self.table.len <= i) {
_ = try read.holder.file.preadAll(std.mem.sliceAsBytes(&meta_offset), self.offset); _ = try read.holder.file.preadAll(&meta_buf, self.offset);
self.offset += 8; self.offset += 8;
meta_offset = std.mem.bytesToValue(u8, &meta_buf);
offset_rdr = read.holder.readerAt(meta_offset); offset_rdr = read.holder.readerAt(meta_offset);
meta_rdr = .init(self.alloc, self.decomp, offset_rdr.any()); meta_rdr = .init(read.alloc, self.decomp, offset_rdr.any());
defer meta_rdr.deinit(); defer meta_rdr.deinit();
to_read = @min(self.item_count - self.table.len, comptime blk: { to_read = @min(self.item_count - self.table.len, comptime 8192 / @sizeOf(T));
break :blk 8192 / @sizeOf(T); const alloc_size = self.table.len + to_read;
}); if (self.table.len != 0) read.alloc.free(self.table);
if (!self.alloc.resize(self.table, self.table.len + to_read)) { self.table = try read.alloc.alloc(T, alloc_size);
const alloc_size = self.table.len + to_read; _ = try meta_rdr.any().readAll(@ptrCast(self.table[self.table.len - to_read ..]));
self.alloc.free(self.table);
self.table = try self.alloc.alloc(T, alloc_size);
}
_ = try meta_rdr.any().readAll(std.mem.asBytes(self.table[self.table.len - to_read ..]));
} }
return self.table[i]; return self.table[i];
} }