turbonss/lib/PackedUser.zig

const std = @import("std");
const assert = std.debug.assert;
const mem = std.mem;
const math = std.math;
const Allocator = mem.Allocator;
const ArrayList = std.ArrayList;
const StringHashMap = std.StringHashMap;
const fieldInfo = std.meta.fieldInfo;

const pad = @import("padding.zig");
const validate = @import("validate.zig");
const compress = @import("compress.zig");
const shellImport = @import("shell.zig");
const InvalidRecord = validate.InvalidRecord;
const User = @import("User.zig");

const PackedUser = @This();
pub const alignment_bits = 3;

const Inner = packed struct {
    uid: u32,
    gid: u32,
    shell_len_or_idx: u8,
    shell_here: bool,
    name_is_a_suffix: bool,
    home_len: u6,
    name_len: u5,
    gecos_len: u11,

    fn homeLen(self: *const Inner) usize {
        return @as(u32, self.home_len) + 1;
    }

    fn nameStart(self: *const Inner) usize {
        const name_len = self.nameLen();
        if (self.name_is_a_suffix) {
            return self.homeLen() - name_len;
        } else return self.homeLen();
    }

    fn nameLen(self: *const Inner) usize {
        return @as(u32, self.name_len) + 1;
    }

    fn gecosStart(self: *const Inner) usize {
        if (self.name_is_a_suffix) {
            return self.homeLen();
        } else return self.homeLen() + self.nameLen();
    }

    fn gecosLen(self: *const Inner) usize {
        return self.gecos_len;
    }

    fn maybeShellStart(self: *const Inner) usize {
        assert(self.shell_here);
        return self.gecosStart() + self.gecosLen();
    }

    fn shellLen(self: *const Inner) usize {
        return @as(u32, self.shell_len_or_idx) + 1;
    }

    // stringLength returns the length of the blob storing string values.
    fn stringLength(self: *const Inner) usize {
        var result: usize = self.homeLen() + self.gecosLen();
        if (!self.name_is_a_suffix)
            result += self.nameLen();
        if (self.shell_here)
            result += self.shellLen();
        return result;
    }
};

// PackedUser does not allocate; it re-interprets the "bytes" blob
// field. Both of those fields are pointers to "our representation" of
// that field.
inner: *const Inner,
bytes: []const u8,
additional_gids_offset: u64,

pub const Entry = struct {
    user: PackedUser,
    next: ?[]const u8,
};

// TODO(motiejus) provide a way to return an entry without decoding the
// additional_gids_offset:
// - will not return the 'next' slice.
// - cannot throw an Overflow error.
pub fn fromBytes(bytes: []const u8) Entry {
    const inner = mem.bytesAsValue(Inner, bytes[0..@sizeOf(Inner)]);
    const start_blob = @sizeOf(Inner);
    const end_strings = start_blob + inner.stringLength();
    const gids_offset = compress.uvarint(bytes[end_strings..]) catch |err| switch (err) {
        error.Overflow => unreachable,
    };
    const end_blob = end_strings + gids_offset.bytes_read;

    const nextStart = pad.roundUp(usize, alignment_bits, end_blob);
    var next: ?[]const u8 = null;
    if (nextStart < bytes.len)
        next = bytes[nextStart..];

    return Entry{
        .user = PackedUser{
            .inner = inner,
            .bytes = bytes[start_blob..end_blob],
            .additional_gids_offset = gids_offset.value,
        },
        .next = next,
    };
}

pub const Iterator = struct {
    section: ?[]const u8,
    shell_reader: shellImport.ShellReader,

    pub fn next(it: *Iterator) ?PackedUser {
        if (it.section) |section| {
            const entry = PackedUser.fromBytes(section);
            it.section = entry.next;
            return entry.user;
        }
        return null;
    }
};

pub fn iterator(section: []const u8, shell_reader: shellImport.ShellReader) Iterator {
    return Iterator{ .section = section, .shell_reader = shell_reader };
}

// packTo packs the User record and copies it to the given arraylist.
pub fn packTo(
    arr: *ArrayList(u8),
    user: User,
    additional_gids_offset: u64,
    idxFn: StringHashMap(u8),
) error{ InvalidRecord, OutOfMemory }!void {
    std.debug.assert(arr.items.len & 7 == 0);
    // function arguments are consts. We need to mutate the underlying
    // slice, so passing it via pointer instead.
    //const home_len = try validate.downCast(u6, user.home.len - 1);
    const home_len = try validate.downCast(fieldInfo(Inner, .home_len).field_type, user.home.len - 1);
    const name_len = try validate.downCast(fieldInfo(Inner, .name_len).field_type, user.name.len - 1);
    const shell_len = try validate.downCast(fieldInfo(Inner, .shell_len_or_idx).field_type, user.shell.len - 1);
    const gecos_len = try validate.downCast(fieldInfo(Inner, .gecos_len).field_type, user.gecos.len);

    try validate.utf8(user.home);
    try validate.utf8(user.name);
    try validate.utf8(user.shell);
    try validate.utf8(user.gecos);

    const inner = Inner{
        .uid = user.uid,
        .gid = user.gid,
        .shell_here = idxFn.get(user.shell) == null,
        .shell_len_or_idx = idxFn.get(user.shell) orelse shell_len,
        .home_len = home_len,
        .name_is_a_suffix = mem.endsWith(u8, user.home, user.name),
        .name_len = name_len,
        .gecos_len = gecos_len,
    };
    const innerBytes = mem.asBytes(&inner);

    try arr.*.appendSlice(innerBytes[0..@sizeOf(Inner)]);
    try arr.*.appendSlice(user.home);

    if (!inner.name_is_a_suffix)
        try arr.*.appendSlice(user.name);
    try arr.*.appendSlice(user.gecos);
    if (inner.shell_here)
        try arr.*.appendSlice(user.shell);
    try compress.appendUvarint(arr, additional_gids_offset);
}

pub fn uid(self: PackedUser) u32 {
    return self.inner.uid;
}

pub fn gid(self: PackedUser) u32 {
    return self.inner.gid;
}

pub fn additionalGidsOffset(self: PackedUser) u64 {
    return self.additional_gids_offset;
}

pub fn home(self: PackedUser) []const u8 {
    return self.bytes[0..self.inner.homeLen()];
}

pub fn name(self: PackedUser) []const u8 {
    const name_pos = self.inner.nameStart();
    const name_len = self.inner.nameLen();
    return self.bytes[name_pos .. name_pos + name_len];
}

pub fn gecos(self: PackedUser) []const u8 {
    const gecos_pos = self.inner.gecosStart();
    const gecos_len = self.inner.gecosLen();
    return self.bytes[gecos_pos .. gecos_pos + gecos_len];
}

pub fn shell(self: PackedUser, shell_reader: shellImport.ShellReader) []const u8 {
    if (self.inner.shell_here) {
        const shell_pos = self.inner.maybeShellStart();
        const shell_len = self.inner.shellLen();
        return self.bytes[shell_pos .. shell_pos + shell_len];
    }
    return shell_reader.get(self.inner.shell_len_or_idx);
}

pub const max_str_len = math.maxInt(fieldInfo(Inner, .shell_len_or_idx).field_type) + 1 +
    math.maxInt(fieldInfo(Inner, .home_len).field_type) + 1 +
    math.maxInt(fieldInfo(Inner, .name_len).field_type) + 1 +
    math.maxInt(fieldInfo(Inner, .gecos_len).field_type);

const testing = std.testing;

test "PackedUser internal and external alignment" {
    try testing.expectEqual(
        @sizeOf(PackedUser.Inner) * 8,
        @bitSizeOf(PackedUser.Inner),
    );
}

fn testShellIndex(allocator: Allocator) StringHashMap(u8) {
    var result = StringHashMap(u8).init(allocator);
    result.put("/bin/bash", 0) catch unreachable;
    result.put("/bin/zsh", 1) catch unreachable;
    return result;
}

const test_shell_reader = shellImport.ShellReader{
    .blob = "/bin/bash/bin/zsh",
    .index = &[_]u16{ 0, 9, 17 },
};

const max_user = User{
    .uid = 0,
    .gid = math.maxInt(u32),
    .name = "Name" ** 8,
    .gecos = "realname" ** 255 ++ "realnam",
    .home = "Home" ** 16,
    .shell = "She.LllL" ** 32,
};

test "max_user and max_str_len are consistent" {
    const total_len = max_user.name.len +
        max_user.gecos.len +
        max_user.home.len +
        max_user.shell.len;
    try testing.expectEqual(total_len, max_str_len);
}

test "pack max_user" {
    var arr = ArrayList(u8).init(testing.allocator);
    defer arr.deinit();

    var idx_noop = StringHashMap(u8).init(testing.allocator);
    defer idx_noop.deinit();

    try packTo(&arr, max_user, 0, idx_noop);
}

test "construct PackedUser section" {
    var buf = ArrayList(u8).init(testing.allocator);
    defer buf.deinit();

    const users = [_]User{ User{
        .uid = 1000,
        .gid = 1000,
        .name = "vidmantas",
        .gecos = "Vidmantas Kaminskas",
        .home = "/home/vidmantas",
        .shell = "/bin/bash",
    }, User{
        .uid = 1001,
        .gid = 1001,
        .name = "svc-foo",
        .gecos = "Service Account",
        .home = "/home/service1",
        .shell = "/usr/bin/nologin",
    }, max_user, User{
        .uid = 1002,
        .gid = 1002,
        .name = "svc-bar",
        .gecos = "",
        .home = "/",
        .shell = "/bin/zsh",
    } };
    var shellIndex = testShellIndex(testing.allocator);
    const additional_gids = math.maxInt(u64);
    defer shellIndex.deinit();
    for (users) |user| {
        try PackedUser.packTo(&buf, user, additional_gids, shellIndex);
        try pad.arrayList(&buf, PackedUser.alignment_bits);
    }

    var i: u29 = 0;
    var it1 = PackedUser.iterator(buf.items, test_shell_reader);
    while (it1.next()) |user| : (i += 1) {
        try testing.expectEqual(users[i].uid, user.uid());
        try testing.expectEqual(users[i].gid, user.gid());
        try testing.expectEqual(user.additionalGidsOffset(), additional_gids);
        try testing.expectEqualStrings(users[i].name, user.name());
        try testing.expectEqualStrings(users[i].gecos, user.gecos());
        try testing.expectEqualStrings(users[i].home, user.home());
        try testing.expectEqualStrings(users[i].shell, user.shell(test_shell_reader));
    }
    try testing.expectEqual(users.len, i);
}