turbonss/src/user.zig

const std = @import("std");

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

const assert = std.debug.assert;
const mem = std.mem;
const math = std.math;
const Allocator = mem.Allocator;
const ArrayList = std.ArrayList;

// Idx2ShellProto is a function prototype that, given a shell's index (in
// global shell section), will return a shell string. Matches ShellReader.get.
const Idx2ShellProto = fn (u6) []const u8;

// User is a convenient public struct for record construction and
// serialization.
pub const User = struct {
    uid: u32,
    gid: u32,
    name: []const u8,
    gecos: []const u8,
    home: []const u8,
    shell: []const u8,

    // deep-clones a User record with a given Allocator.
    pub fn clone(
        self: *const User,
        allocator: Allocator,
    ) Allocator.Error!User {
        const stringdata = try allocator.alloc(u8, self.strlen());
        const gecos_start = self.name.len;
        const home_start = gecos_start + self.gecos.len;
        const shell_start = home_start + self.home.len;
        mem.copy(u8, stringdata[0..self.name.len], self.name);
        mem.copy(u8, stringdata[gecos_start..], self.gecos);
        mem.copy(u8, stringdata[home_start..], self.home);
        mem.copy(u8, stringdata[shell_start..], self.shell);

        return User{
            .uid = self.uid,
            .gid = self.gid,
            .name = stringdata[0..self.name.len],
            .gecos = stringdata[gecos_start .. gecos_start + self.gecos.len],
            .home = stringdata[home_start .. home_start + self.home.len],
            .shell = stringdata[shell_start .. shell_start + self.shell.len],
        };
    }

    fn strlen(self: *const User) usize {
        return self.name.len +
            self.gecos.len +
            self.home.len +
            self.shell.len;
    }

    pub fn deinit(self: *User, allocator: Allocator) void {
        const slice = self.home.ptr[0..self.strlen()];
        allocator.free(slice);
        self.* = undefined;
    }
};

pub fn Shell2Index(T: type) type {
    return struct {
        const Self = @This();
        data: T,

        pub fn init(data: T) Self {
            return Self{ .data = data };
        }

        pub fn get(self: *const Self, str: []const u8) ?u6 {
            return self.data.get(str);
        }
    };
}

pub const PackedUserHash = packedUser(std.StringHashMap(u6));

fn packedUser(comptime ShellIndexType: type) type {
    return struct {
        const Self = @This();

        const alignmentBits = 3;

        const Inner = packed struct {
            uid: u32,
            gid: u32,
            padding: u2 = 0,
            shell_len_or_idx: u6,
            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: Self,
            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) error{Overflow}!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 = try compress.uvarint(bytes[end_strings..]);
            const end_blob = end_strings + gids_offset.bytes_read;

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

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

        pub const Iterator = struct {
            section: ?[]const u8,
            shellIndex: Idx2ShellProto,

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

        pub fn iterator(section: []const u8, idxFn: Idx2ShellProto) Iterator {
            return Iterator{ .section = section, .shellIndex = idxFn };
        }

        // packTo packs the User record and copies it to the given byte slice.
        // The slice must have at least maxRecordSize() bytes available. The
        // slice is passed as a pointer, so it can be mutated.
        pub fn packTo(
            arr: *ArrayList(u8),
            user: User,
            additional_gids_offset: u64,
            idxFn: ShellIndexType,
        ) error{ InvalidRecord, OutOfMemory }!void {
            // 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 name_len = try validate.downCast(u5, user.name.len - 1);
            const shell_len = try validate.downCast(u6, user.shell.len - 1);
            const gecos_len = try validate.downCast(u8, 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);

            // innerBytes.len is longer than @sizeOf(Inner). We want to copy
            // only the @sizeOf(Inner)-number of bytes.
            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);
            try pad.arrayList(arr, alignmentBits);
        }

        // maxSize is the maximum number of records a PackedUser can take
        // (struct + strings).
        pub fn maxSize() usize {
            comptime {
                const unpadded = @sizeOf(Inner) +
                    math.maxInt(u6) + 1 + // home
                    math.maxInt(u5) + 1 + // name
                    math.maxInt(u6) + 1 + // shell
                    math.maxInt(u8); // gecos
                return pad.roundUp(u64, alignmentBits, unpadded);
            }
        }

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

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

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

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

        pub fn name(self: Self) []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: Self) []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: Self, idxFn: Idx2ShellProto) []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 idxFn(self.inner.shell_len_or_idx);
        }
    };
}

const testing = std.testing;

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

const TestShellIndex = struct {
    pub fn get(_: *const TestShellIndex, shell: []const u8) ?u6 {
        if (mem.eql(u8, shell, "/bin/bash")) {
            return 0;
        } else if (mem.eql(u8, shell, "/bin/zsh")) {
            return 1;
        }
        return null;
    }
};

const PackedUserTest = packedUser(TestShellIndex);

fn testShell(index: u6) []const u8 {
    return switch (index) {
        0 => "/bin/bash",
        1 => "/bin/zsh",
        else => unreachable,
    };
}

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",
    }, User{
        .uid = 0,
        .gid = math.maxInt(u32),
        .name = "Name" ** 8,
        .gecos = "Gecos" ** 51,
        .home = "Home" ** 16,
        .shell = "She.LllL" ** 8,
    }, User{
        .uid = 1002,
        .gid = 1002,
        .name = "svc-bar",
        .gecos = "",
        .home = "/",
        .shell = "/",
    } };
    for (users) |user|
        try PackedUserTest.packTo(&buf, user, math.maxInt(u64), TestShellIndex{});

    var i: u29 = 0;
    var it1 = PackedUserTest.iterator(buf.items, testShell);
    while (try it1.next()) |user| : (i += 1) {
        try testing.expectEqual(users[i].uid, user.uid());
        try testing.expectEqual(users[i].gid, user.gid());
        try testing.expectEqual(
            @as(u64, math.maxInt(u64)),
            user.additionalGidsOffset(),
        );
        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(testShell));
    }
    try testing.expectEqual(users.len, i);
}

test "PackedUser.maxSize()" {
    // TODO(motiejus) try using a slice that points to an array in stack.
    // As of writing, I am getting a stack smashing error.
    var buf = try ArrayList(u8).initCapacity(
        testing.allocator,
        PackedUserHash.maxSize(),
    );
    defer buf.deinit();

    const largeUser = User{
        .uid = math.maxInt(u32),
        .gid = math.maxInt(u32),
        .name = "Name" ** 8, // 32
        .gecos = "Gecos" ** 51, // 255
        .home = "Home" ** 16, // 64
        .shell = "She.LllL" ** 8, // 64
    };
    try PackedUserTest.packTo(&buf, largeUser, math.maxInt(u29), TestShellIndex{});
    try testing.expectEqual(PackedUserTest.maxSize(), buf.items.len);
}

test "User.clone" {
    var allocator = testing.allocator;
    const user = User{
        .uid = 1000,
        .gid = 1000,
        .name = "vidmantas",
        .gecos = "Vidmantas Kaminskas",
        .home = "/home/vidmantas",
        .shell = "/bin/bash",
    };
    var user2 = try user.clone(allocator);
    defer user2.deinit(allocator);

    try testing.expectEqualStrings(user.shell, "/bin/bash");
}