Work on extraction

Created DataExtractor & DataReader
Created Lookup tables

src/data/extractor.zig

@@ -0,0 +1,102 @@
+const std = @import("std");
+const Io = std.Io;
+
+const DecompCache = @import("../decomp_cache.zig");
+const DataBlock = @import("../inode.zig").DataBlock;
+
+const Extractor = @This();
+
+cache: *DecompCache,
+block_size: u32,
+
+start: u64,
+size: u64,
+blocks: []DataBlock,
+
+frag_data: ?[]u8 = null,
+frag_offset: u32 = 0,
+
+pub fn init(cache: *DecompCache, block_size: u32, size: u64, start: u64, blocks: []DataBlock) Extractor {
+    return .{
+        .cache = cache,
+        .block_size = block_size,
+
+        .start = start,
+        .size = size,
+        .blocks = blocks,
+    };
+}
+
+pub fn addFragment(self: *Extractor, data: []u8, offset: u32) void {
+    self.frag_data = data;
+    self.frag_offset = offset;
+}
+
+pub fn asyncExtract(self: Extractor, io: Io, fil: Io.File) Error!void {
+    try fil.writePositionalAll(io, &.{&.{0}}, self.size - 1);
+
+    var map = try fil.createMemoryMap(io, .{ .len = self.size, .protection = .{ .write = true } });
+    defer map.destroy(io);
+
+    var group: Io.Group = .init;
+    defer group.cancel(io);
+
+    var ret_err: ?Error = null;
+
+    var offset = self.start;
+    for (0..self.blocks.len) |i| {
+        group.async(io, blockThread, .{ self, io, map, offset, i, &ret_err });
+
+        offset += self.blocks[i].size;
+    }
+    if (self.frag_data != null)
+        group.async(io, fragThread, .{ self, map });
+
+    group.await(io) catch |err| return ret_err orelse err;
+
+    try map.write(io);
+}
+
+fn blockThread(self: Extractor, io: Io, map: Io.File.MemoryMap, read_offset: u64, idx: usize, ret_err: *?Error) error{Canceled}!void {
+    const write_pos = idx * self.block_size;
+    const size = if (self.frag_data == null and idx == self.block_size.len - 1)
+        self.size % self.block_size
+    else
+        self.block_size;
+    const block = self.blocks[idx];
+
+    if (block.size == 0) {
+        @memset(map.memory[write_pos..][0..size], 0);
+        return;
+    }
+    if (block.uncompressed) {
+        @memcpy(map[write_pos..][0..size], self.cache.map.memory[read_offset..][0..size]);
+        return;
+    }
+    const data = self.cache.get(io, read_offset, block.size, size) catch |err| switch (err) {
+        error.Canceled => {
+            io.recancel();
+            return error.Canceled;
+        },
+        else => |e| {
+            ret_err.* = e;
+            return error.Canceled;
+        },
+    };
+    defer self.cache.finished(io, read_offset);
+    if (data.len != size) {
+        std.debug.print("Size of decompression at {} is {} and should be {}\n", .{ read_offset, data.len, size });
+        return Error.BadDecompressionSize;
+    }
+    @memcpy(map[write_pos..][0..size], data);
+}
+fn fragThread(self: Extractor, map: Io.File.MemoryMap) error{Canceled}!void {
+    const write_pos = self.blocks.len * self.block_size;
+    const size = self.size % self.block_size;
+
+    @memcpy(map.memory[write_pos..][0..size], self.frag_data.?[self.frag_offset..][0..size]);
+}
+
+// Types
+
+pub const Error = error{BadDecompressionSize} || Io.File.WritePositionalError || Io.File.MemoryMap.CreateError;

src/data/reader.zig

@@ -0,0 +1,151 @@
+const std = @import("std");
+const Io = std.Io;
+
+const DecompCache = @import("../decomp_cache.zig");
+const DataBlock = @import("../inode.zig").DataBlock;
+
+const Reader = @This();
+
+io: Io,
+
+cache: *DecompCache,
+block_size: u32,
+
+size: u64,
+blocks: []DataBlock,
+
+frag_data: ?[]u8 = null,
+frag_offset: u32 = 0,
+
+cur_offset: u64 = 0,
+next_offset: u64,
+
+idx: u32 = 0,
+cur_block_sparse: bool = false,
+
+interface: Io.Reader = .{
+    .buffer = &[0]u8{},
+    .end = 0,
+    .seek = 0,
+    .vtable = &.{
+        .stream = stream,
+        .discard = discard,
+        .readVec = readVec,
+    },
+},
+
+pub fn init(io: Io, cache: *DecompCache, block_size: u32, size: u64, start: u64, blocks: []DataBlock) Reader {
+    return .{
+        .io = io,
+
+        .cache = cache,
+        .block_size = block_size,
+
+        .size = size,
+        .blocks = blocks,
+
+        .next_offset = start,
+    };
+}
+pub fn deinit(self: Reader) void {
+    self.cache.finished(self.io);
+}
+
+pub fn addFragment(self: *Reader, data: []u8, offset: u32) void {
+    self.frag_data = data;
+    self.frag_offset = offset;
+}
+
+fn advance(self: *Reader) Io.Reader.Error!void {
+    errdefer self.interface.end = 0;
+    self.interface.seek = 0;
+
+    if (self.idx > self.blocks.len) return error.EndOfStream;
+    defer self.idx += 1;
+    self.cache.finished(self.io, self.cur_offset);
+
+    if (self.idx == self.blocks.len) {
+        if (self.frag_data == null) return error.EndOfStream;
+        self.cur_offset = 0;
+
+        const size = self.size % self.block_size;
+        self.interface.buffer = self.frag_data.?[self.frag_offset..][0..size];
+        self.interface.end = size;
+        return;
+    }
+
+    const block = self.blocks[self.idx];
+
+    const size = if (self.idx == self.blocks.len - 1 and self.frag_data == null)
+        self.size % self.block_size
+    else
+        self.block_size;
+
+    if (block.size == 0) {
+        self.interface.buffer = &[0]u8{};
+        self.cur_block_sparse = true;
+        self.interface.end = size;
+        return;
+    } else {
+        self.cur_block_sparse = false;
+    }
+
+    self.cur_offset = self.next_offset;
+    self.next_offset = self.cur_offset + block.size;
+
+    if (block.uncompressed) {
+        self.interface.buffer = self.cache.map.memory[self.cur_offset..][0..size];
+        self.interface.end = size;
+        return;
+    }
+    const data = self.cache.get(self.io, self.cur_offset, block.size, size);
+    if (data.len != size) {
+        std.debug.print("Size of decompression at {} is {} and should be {}\n", .{ self.cur_offset, data.len, size });
+        return Io.Reader.Error.ReadFailed;
+    }
+    self.interface.buffer = data;
+    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();
+    }
+    const to_write = @min(@intFromEnum(limit), r.end - r.seek);
+    const wrote = try if (self.cur_block_sparse)
+        w.splatByte(0, to_write)
+    else
+        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();
+    }
+    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 total: usize = 0;
+    for (vec) |v| {
+        const to_copy = @min(v.len, r.end - r.seek);
+        if (self.cur_block_sparse) {
+            @memset(v[0..to_copy], 0);
+        } else {
+            @memcpy(v[0..to_copy], r.buffer[r.seek..][0..to_copy]);
+        }
+        total += to_copy;
+        r.seek += to_copy;
+
+        if (r.seek >= r.end) break;
+    }
+    return total;
+}

src/directory.zig

@@ -8,7 +8,7 @@ const Directory = @This();
 
 entries: []Entry,
 
-pub fn init(alloc: std.mem.Allocator, rdr: *Reader, size: u32) !Directory {
+pub fn init(alloc: std.mem.Allocator, rdr: *Reader, size: u32) Error!Directory {
     if (size <= 3) return .{ .entries = &[0]Entry{} };
 
     var entries: std.ArrayList(Entry) = try .initCapacity(alloc, 50);
@@ -56,6 +56,8 @@ pub fn deinit(self: Directory, alloc: std.mem.Allocator) void {
 
 // Types
 
+pub const Error = Reader.Error || std.mem.Allocator.Error;
+
 pub const Entry = struct {
     inode_num: u32,
     block_start: u32,

src/extract.zig

@@ -1,51 +1,242 @@
 const std = @import("std");
 const Io = std.Io;
 
+const Atomic = std.atomic.Value;
+
 const DecompCache = @import("decomp_cache.zig");
 const ExtractionOptions = @import("options.zig");
 const Inode = @import("inode.zig");
 const Superblock = @import("archive.zig").Superblock;
+const Directory = @import("directory.zig");
+const DataExtractor = @import("data/extractor.zig");
+const DataReader = @import("data/reader.zig");
 
 pub fn extract(alloc: std.mem.Allocator, io: Io, inode: Inode, cache: *DecompCache, super: Superblock, ext_loc: []const u8, options: ExtractionOptions) !void {
-    _ = alloc;
-    _ = io;
-    _ = inode;
-    _ = cache;
-    _ = super;
-    _ = ext_loc;
-    _ = options;
-    return error.TODO;
+    const path = std.mem.trim(u8, ext_loc, "/");
+
+    var buf: [50]ReturnUnion = undefined;
+    var sel: Io.Select(ReturnUnion) = .init(io, &buf);
+    defer sel.cancelDiscard();
+
+    var ret_loop = io.async(returnLoop, .{ alloc, &sel, options });
+
+    try extractReal(alloc, io, cache, super, &sel, path, inode, null, false);
+
+    ret_loop.await(io) catch |err| {
+        // TODO: Drain sel
+        return err;
+    };
 }
 
-pub fn extractDir(alloc: std.mem.Allocator, io: Io, path: []const u8, d: anytype) Error!PathReturn {}
-pub fn extractFile(alloc: std.mem.Allocator, io: Io, path: []const u8, d: anytype) Error!PathReturn {
+fn extractReal(
+    alloc: std.mem.Allocator,
+    io: Io,
+    cache: *DecompCache,
+    super: Superblock,
+    sel: *Io.Select(ReturnUnion),
+    path: []const u8,
+    inode: Inode,
+    parent: ?*Atomic(usize),
+    origin: bool,
+) Error!void {
+    try io.checkCancel();
+
+    switch (inode.data) {
+        .dir, .ext_dir => sel.async(
+            .dir_ret,
+            extractDir,
+            .{ alloc, io, cache, super, sel, path, inode, parent, origin },
+        ),
+        else => return error.Canceled,
+    }
+}
+
+fn extractDir(
+    alloc: std.mem.Allocator,
+    io: Io,
+    cache: *DecompCache,
+    super: Superblock,
+    sel: *Io.Select(ReturnUnion),
+    path: []const u8,
+    inode: Inode,
+    parent: ?*Atomic(usize),
+    origin: bool,
+) Error!DirReturn {
+    defer {
+        if (parent != null)
+            _ = parent.?.fetchSub(1, .acquire);
+        if (!origin) inode.deinit(alloc);
+    }
+    errdefer if (!origin) alloc.free(path);
+
+    const dir = inode.directory(alloc, io, cache, super.dir_start) catch |err| switch (err) {
+        error.NotDirectory => unreachable,
+        else => |e| return e,
+    };
+    defer dir.deinit(alloc);
+
+    const sub_files = try alloc.create(Atomic(usize));
+    sub_files.* = .init(dir.entries.len);
+
+    const ret: DirReturn = .{
+        .path = path,
+        .sub_files = sub_files,
+        .origin = origin,
+
+        .uid_idx = inode.hdr.uid_idx,
+        .gid_idx = inode.hdr.gid_idx,
+        .mod_time = inode.hdr.mod_time,
+        .permissions = inode.hdr.permission,
+
+        .xattr_idx = switch (inode.data) {
+            .ext_dir => |d| if (d.xattr_idx != 0xFFFFFFFF) d.xattr_idx else null,
+            else => null,
+        },
+    };
+
+    for (dir.entries) |entry| {
+        const new_inode: Inode = try .initDirEntry(alloc, io, cache, super.inode_start, super.block_size, entry);
+        errdefer new_inode.deinit(alloc);
+
+        const new_path = try std.mem.concat(alloc, u8, &.{ path, "/", entry.name });
+
+        try extractReal(
+            alloc,
+            io,
+            cache,
+            super,
+            sel,
+            new_path,
+            new_inode,
+            sub_files,
+            false,
+        );
+    }
+    return ret;
+}
+fn extractFile(
+    alloc: std.mem.Allocator,
+    io: Io,
+    cache: *DecompCache,
+    block_size: u32,
+    path: []const u8,
+    inode: Inode,
+    parent: ?*Atomic(usize),
+    origin: bool,
+) Error!FileReturn {
+    defer {
+        if (parent != null)
+            _ = parent.?.fetchSub(1, .acquire);
+        if (!origin) inode.deinit(alloc);
+    }
+    errdefer if (!origin) alloc.free(path);
+
     const atomic = try Io.Dir.cwd().createFileAtomic(io, path, .{});
     defer atomic.deinit(io);
 
-    // TODO
+    var ret: FileReturn = .{
+        .path = path,
+        .origin = origin,
+
+        .uid_idx = inode.hdr.uid_idx,
+        .gid_idx = inode.hdr.gid_idx,
+        .permissions = inode.hdr.permission,
+        .mod_time = inode.hdr.mod_time,
+    };
+
+    var data: DataExtractor = switch (inode.data) {
+        .file => |f| blk: {
+            var data: DataExtractor = .init(cache, block_size, f.size, f.data_start, f.blocks);
+            if (f.frag_idx != 0xFFFFFFFF) {
+                // TODO
+            }
+            break :blk data;
+        },
+        .ext_file => |f| blk: {
+            var data: DataExtractor = .init(cache, block_size, f.size, f.data_start, f.blocks);
+            if (f.frag_idx != 0xFFFFFFFF) {
+                //TODO
+            }
+            break :blk data;
+        },
+        else => unreachable,
+    };
 
     try atomic.link(io);
     // return .{
     //     .path = path,
     // };
-    return error.TODO;
+    return error.Canceled;
+}
+
+// Loop
+
+fn returnLoop(alloc: std.mem.Allocator, sel: *Io.Select(ReturnUnion), options: ExtractionOptions) !void {
+    while (true) {
+        const finished = try sel.await();
+
+        switch (finished) {
+            .dir_ret => |d| {
+                const ret = try d;
+                if (ret.sub_files.load(.unordered) != 0) {
+                    sel.queue.putOne(sel.io, .{ .dir_ret = ret }) catch |err| {
+                        if (!ret.origin) alloc.free(ret.path);
+                        return err;
+                    };
+                    continue;
+                }
+                if (!ret.origin) alloc.free(ret.path);
+                alloc.destroy(ret.sub_files);
+
+                if (!options.ignore_permissions and !options.ignore_xattr) {
+                    // TODO: set permissions & xattr.
+                }
+            },
+            .file_ret => |f| {
+                const ret = try f;
+                if (!ret.origin) alloc.free(ret.path);
+
+                if (!options.ignore_permissions and !options.ignore_xattr) {
+                    // TODO: set permissions & xattr.
+                }
+            },
+            .void_ret => |v| try v,
+        }
+
+        if (sel.group.token.load(.unordered) == null) break;
+    }
 }
 
 // Utility types
 
-const ReturnUnion = union {
-    path_ret: Error!PathReturn,
+const ReturnUnion = union(enum) {
+    file_ret: Error!FileReturn,
+    dir_ret: Error!DirReturn,
+    void_ret: Error!void,
 };
 
-const Error = error{};
+const Error = error{Canceled} || Directory.Error;
 
-const PathReturn = struct {
+const FileReturn = struct {
     path: []const u8,
+    origin: bool,
 
     uid_idx: u32,
     gid_idx: u32,
     mod_time: u32,
-    permission: u16,
+    permissions: u16,
 
-    xattr_idx: ?u32,
+    xattr_idx: ?u32 = null,
+};
+const DirReturn = struct {
+    path: []const u8,
+    sub_files: *Atomic(usize),
+    origin: bool,
+
+    uid_idx: u32,
+    gid_idx: u32,
+    mod_time: u32,
+    permissions: u16,
+
+    xattr_idx: ?u32 = null,
 };

src/lookup.zig

@@ -0,0 +1,104 @@
+const std = @import("std");
+const Io = std.Io;
+
+const DecompCache = @import("decomp_cache.zig");
+const MetadataReader = @import("meta_rdr.zig");
+
+pub fn stateless(comptime T: anytype, io: Io, cache: *DecompCache, table_start: u64, idx: u32) !T {
+    const PER_BLOCK = 8192 / @sizeOf(T);
+
+    const block = idx / PER_BLOCK;
+    const block_idx = idx % PER_BLOCK;
+
+    const offset_offset = table_start + (block * 8);
+    const offset: u64 = std.mem.readInt(u64, cache.map.memory[offset_offset..][0..2], .little);
+
+    var meta: MetadataReader = .init(io, cache, offset);
+    defer meta.deinit(io);
+    try meta.discardAll(block_idx * @sizeOf(T));
+
+    var new: T = undefined;
+    try meta.interface.readSliceEndian(T, @ptrCast(&new), .little);
+    return new;
+}
+
+pub fn Table(comptime T: anytype) type {
+    return struct {
+        const PER_BLOCK = 8192 / @sizeOf(T);
+
+        const Table = @This();
+
+        alloc: std.mem.Allocator,
+
+        cache: *DecompCache,
+        table_start: u64,
+
+        num: u32,
+        values: std.AutoHashMap(u32, []T),
+        mut: Io.RwLock,
+
+        pub fn init(alloc: std.mem.Allocator, cache: *DecompCache, table_start: u64, num_values: u32) Table {
+            return .{
+                .alloc = alloc,
+
+                .cache = cache,
+                .table_start = table_start,
+
+                .num = num_values,
+                .values = .init(alloc),
+            };
+        }
+        pub fn deinit(self: *Table) void {
+            var iter = self.values.valueIterator();
+            while (iter.next()) |v|
+                self.alloc.free(v);
+            self.values.deinit();
+        }
+
+        pub fn get(self: *Table, io: Io, idx: u32) Error!T {
+            const block = idx / PER_BLOCK;
+            const block_idx = idx % PER_BLOCK;
+            {
+                try self.mut.lockShared(io);
+                defer self.mut.unlockShared(io);
+
+                const val = self.values.get(block);
+                if (val != null) return val.*[block_idx];
+            }
+            try self.mut.lock(io);
+            defer self.mut.unlock(io);
+
+            const val = try self.values.getOrPut(block);
+            if (val.found_existing)
+                return val.value_ptr.*[block_idx];
+            errdefer self.values.removeByPtr(val.key_ptr);
+
+            const offset_offset = self.table_start + (block * 8);
+            const offset: u64 = std.mem.readInt(u64, self.cache.map.memory[offset_offset..][0..2], .little);
+
+            var meta: MetadataReader = .init(io, self.cache, offset);
+            defer meta.deinit(io);
+
+            const size = if (block == ((self.num - 1) / PER_BLOCK))
+                self.num % PER_BLOCK
+            else
+                PER_BLOCK;
+
+            const new_block = try self.alloc.alloc(T, size);
+            errdefer self.alloc.free(new_block);
+            try meta.interface.readSliceEndian(T, new_block, .little);
+
+            val.value_ptr.* = new_block;
+
+            return new_block[block_idx];
+        }
+    };
+}
+
+// Types
+
+pub const Error = error{} || std.mem.Allocator.Error;
+
+pub const FragmentEntry = extern struct {};
+
+pub const XattrEntry = extern struct {};