turbonss/src/sections.zig

const std = @import("std");
const fmt = std.fmt;
const math = std.math;
const sort = std.sort;
const unicode = std.unicode;
const Allocator = std.mem.Allocator;
const ArrayListUnmanaged = std.ArrayListUnmanaged;
const ArrayList = std.ArrayList;
const MultiArrayList = std.MultiArrayList;
const StringHashMap = std.StringHashMap;
const AutoHashMap = std.AutoHashMap;
const BufSet = std.BufSet;

const pad = @import("padding.zig");
const compress = @import("compress.zig");
const userImport = @import("user.zig");
const groupImport = @import("group.zig");
const cmph = @import("cmph.zig");
const bdz = @import("bdz.zig");
const User = userImport.User;
const Group = groupImport.Group;

const Corpus = struct {
    arena: std.heap.ArenaAllocator,

    // sorted by name, by unicode codepoint
    users: []User,
    // sorted by gid
    groups: []Group,

    // convenience users and groups by column
    usersMulti: MultiArrayList(User),
    groupsMulti: MultiArrayList(Group),

    // pointing to `users` and `groups` slices above.
    name2user: StringHashMap(*const User),
    uid2user: AutoHashMap(u32, *const User),
    name2group: StringHashMap(*const Group),
    gid2group: AutoHashMap(u32, *const Group),
    groupname2users: StringHashMap([]*const User),
    username2groups: StringHashMap([]*const Group),

    pub const initErr = Allocator.Error ||
        error{InvalidUtf8} ||
        error{Duplicate} ||
        error{NotFound};

    pub fn init(
        baseAllocator: Allocator,
        usersConst: []const User,
        groupsConst: []const Group,
    ) initErr!Corpus {
        var arena = std.heap.ArenaAllocator.init(baseAllocator);
        var allocator = arena.allocator();

        var users = try allocator.alloc(User, usersConst.len);
        var groups = try allocator.alloc(Group, groupsConst.len);
        for (usersConst) |*user, i|
            users[i] = try user.clone(allocator);
        for (groupsConst) |*group, i|
            groups[i] = try group.clone(allocator);

        sort.sort(User, users, {}, cmpUser);
        sort.sort(Group, groups, {}, cmpGroup);

        var usersMulti = MultiArrayList(User){};
        try usersMulti.ensureTotalCapacity(allocator, users.len);
        for (users) |user|
            usersMulti.appendAssumeCapacity(user);
        var groupsMulti = MultiArrayList(Group){};
        try groupsMulti.ensureTotalCapacity(allocator, groups.len);
        for (groups) |group|
            groupsMulti.appendAssumeCapacity(group);

        var name2user = StringHashMap(*const User).init(allocator);
        var uid2user = AutoHashMap(u32, *const User).init(allocator);
        var name2group = StringHashMap(*const Group).init(allocator);
        var gid2group = AutoHashMap(u32, *const Group).init(allocator);
        for (users) |*user| {
            var res1 = try name2user.getOrPut(user.name);
            if (res1.found_existing)
                return error.Duplicate;
            res1.value_ptr.* = user;

            var res2 = try uid2user.getOrPut(user.uid);
            if (res2.found_existing)
                return error.Duplicate;

            res2.value_ptr.* = user;
        }

        for (groups) |*group| {
            var res1 = try name2group.getOrPut(group.name);
            if (res1.found_existing)
                return error.Duplicate;
            res1.value_ptr.* = group;
            var res2 = try gid2group.getOrPut(group.gid);
            if (res2.found_existing)
                return error.Duplicate;
            res2.value_ptr.* = group;
        }

        var groupname2users = StringHashMap([]*const User).init(allocator);

        // uses baseAllocator, because it will be freed before
        // returning from this function. This keeps the arena clean.
        var username2groups = StringHashMap(
            ArrayListUnmanaged(*const Group),
        ).init(baseAllocator);
        for (groups) |*group| {
            var members = try allocator.alloc(*const User, group.members.count());
            members.len = 0;

            var it = group.members.iterator();
            while (it.next()) |memberName| {
                if (name2user.get(memberName.*)) |user| {
                    members.len += 1;
                    members[members.len - 1] = user;
                } else {
                    return error.NotFound;
                }

                var groupsOfMember = try username2groups.getOrPut(memberName.*);
                if (!groupsOfMember.found_existing)
                    groupsOfMember.value_ptr.* = ArrayListUnmanaged(*const Group){};
                try groupsOfMember.value_ptr.*.append(allocator, group);
            }

            var result = try groupname2users.getOrPut(group.name);
            if (result.found_existing)
                return error.Duplicate;
            result.value_ptr.* = members;
        }

        var it1 = groupname2users.valueIterator();
        while (it1.next()) |groupUsers| {
            sort.sort(*const User, groupUsers.*, {}, cmpUserPtr);
        }

        var it2 = username2groups.valueIterator();
        while (it2.next()) |userGroups|
            sort.sort(*const Group, userGroups.items, {}, cmpGroupPtr);

        var username2groups_final = StringHashMap([]*const Group).init(allocator);
        for (users) |user| {
            const userGroups = username2groups.get(user.name).?.toOwnedSlice(allocator);
            try username2groups_final.put(user.name, userGroups);
        }
        username2groups.deinit();

        return Corpus{
            .arena = arena,
            .users = users,
            .groups = groups,
            .usersMulti = usersMulti,
            .groupsMulti = groupsMulti,
            .name2user = name2user,
            .uid2user = uid2user,
            .name2group = name2group,
            .gid2group = gid2group,
            .groupname2users = groupname2users,
            .username2groups = username2groups_final,
        };
    }

    pub fn deinit(self: *Corpus) void {
        self.arena.deinit();
        self.* = undefined;
    }
};

pub const Sections = struct {
    allocator: Allocator,
    corpus: *const Corpus,

    pub fn init(allocator: Allocator, corpus: *const Corpus) Sections {
        return Sections{
            .allocator = allocator,
            .corpus = corpus,
        };
    }

    pub const GroupMembers = struct {
        offsets: []const usize,
        bytes: []const u8,
    };

    const groupMembersErr = error{Overflow} || Allocator.Error;

    pub fn groupMembers(self: *const Sections) groupMembersErr!GroupMembers {
        var buf: [compress.maxVarintLen64]u8 = undefined;
        var offsets = ArrayListUnmanaged(usize).initCapacity(
            self.allocator,
            self.corpus.groups.len,
        );
        var bytes = ArrayList(u8).init(self.allocator);
        var offset: usize = 0;
        for (self.corpus.groups) |group, i| {
            offsets[i] = offset;
            const users = self.corpus.groupname2users.get(group.name).?;
            const len = try compress.putVarint(&buf, users.len);
            offset += len;
            try bytes.appendSlice(buf[0..len]);
            for (users) |user| {
                // TODO: offset into the User's record
                _ = user;
            }
        }
    }
};

// cmpUser compares two users for sorting. By username's utf8 codepoints, ascending.
fn cmpUser(_: void, a: User, b: User) bool {
    var utf8_a = (unicode.Utf8View.init(a.name) catch unreachable).iterator();
    var utf8_b = (unicode.Utf8View.init(b.name) catch unreachable).iterator();
    while (utf8_a.nextCodepoint()) |codepoint_a| {
        if (utf8_b.nextCodepoint()) |codepoint_b| {
            if (codepoint_a == codepoint_b) {
                continue;
            } else {
                return codepoint_a < codepoint_b;
            }
        }

        // a is a prefix of b. It is thus shorter.
        return false;
    }
    // b is a prefix of a
    return true;
}

fn cmpUserPtr(context: void, a: *const User, b: *const User) bool {
    return cmpUser(context, a.*, b.*);
}

fn cmpGroup(_: void, a: Group, b: Group) bool {
    return a.gid < b.gid;
}

fn cmpGroupPtr(context: void, a: *const Group, b: *const Group) bool {
    return cmpGroup(context, a.*, b.*);
}

const testing = std.testing;

fn testCorpus(allocator: Allocator) !Corpus {
    const users = [_]User{ User{
        .uid = 128,
        .gid = 128,
        .name = "vidmantas",
        .gecos = "Vidmantas Kaminskas",
        .home = "/home/vidmantas",
        .shell = "/bin/bash",
    }, 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 = "/",
    } };

    var members1 = try groupImport.someMembers(
        allocator,
        &[_][]const u8{"vidmantas"},
    );
    defer members1.deinit();

    var members2 = try groupImport.someMembers(
        allocator,
        &[_][]const u8{ "svc-bar", "vidmantas" },
    );
    defer members2.deinit();

    var members3 = try groupImport.someMembers(
        allocator,
        &[_][]const u8{ "svc-bar", "Name" ** 8, "vidmantas" },
    );
    defer members3.deinit();

    const groups = [_]Group{ Group{
        .gid = 128,
        .name = "vidmantas",
        .members = members1,
    }, Group{
        .gid = 9999,
        .name = "all",
        .members = members3,
    }, Group{
        .gid = 0,
        .name = "service-account",
        .members = members2,
    } };

    return try Corpus.init(allocator, users[0..], groups[0..]);
}

test "test corpus" {
    var corpus = try testCorpus(testing.allocator);
    defer corpus.deinit();

    try testing.expectEqualStrings(corpus.users[0].name, "Name" ** 8);
    try testing.expectEqualStrings(corpus.users[1].name, "svc-bar");
    try testing.expectEqualStrings(corpus.users[2].name, "vidmantas");

    try testing.expectEqual(corpus.name2user.get("404"), null);
    try testing.expectEqual(corpus.name2user.get("vidmantas").?.uid, 128);
    try testing.expectEqual(corpus.uid2user.get(42), null);
    try testing.expectEqual(corpus.uid2user.get(128).?.gid, 128);
    try testing.expectEqual(corpus.name2group.get("404"), null);
    try testing.expectEqual(corpus.name2group.get("vidmantas").?.gid, 128);
    try testing.expectEqual(corpus.gid2group.get(42), null);
    try testing.expectEqual(corpus.gid2group.get(128).?.gid, 128);

    const membersOfAll = corpus.groupname2users.get("all").?;
    try testing.expectEqualStrings(membersOfAll[0].name, "Name" ** 8);
    try testing.expectEqualStrings(membersOfAll[1].name, "svc-bar");
    try testing.expectEqualStrings(membersOfAll[2].name, "vidmantas");
    try testing.expectEqual(corpus.groupname2users.get("404"), null);

    const groupsOfVidmantas = corpus.username2groups.get("vidmantas").?;
    try testing.expectEqual(groupsOfVidmantas[0].gid, 0);
    try testing.expectEqual(groupsOfVidmantas[1].gid, 128);
    try testing.expectEqual(groupsOfVidmantas[2].gid, 9999);
    try testing.expectEqual(corpus.username2groups.get("404"), null);
}

test "pack gids" {
    const allocator = testing.allocator;
    var corpus = try testCorpus(allocator);
    defer corpus.deinit();

    const cmph_gid = try cmph.pack_u32(allocator, corpus.groupsMulti.items(.gid));
    defer allocator.free(cmph_gid);

    const k1 = bdz.search_u32(cmph_gid, 0);
    const k2 = bdz.search_u32(cmph_gid, 128);
    const k3 = bdz.search_u32(cmph_gid, 9999);
    var hashes = &[_]u32{ k1, k2, k3 };
    sort.sort(u32, hashes, {}, comptime sort.asc(u32));
    for (hashes) |hash, i|
        try testing.expectEqual(i, hash);
}

fn testUser(name: []const u8) User {
    var result = std.mem.zeroes(User);
    result.name = name;
    return result;
}

test "users compare function" {
    const a = testUser("a");
    const b = testUser("b");
    const bb = testUser("bb");
    try testing.expect(cmpUser({}, a, b));
    try testing.expect(!cmpUser({}, b, a));
    try testing.expect(cmpUser({}, a, bb));
    try testing.expect(!cmpUser({}, bb, a));
    try testing.expect(cmpUser({}, b, bb));
    try testing.expect(!cmpUser({}, bb, b));
}