turbonss/src/cmph.zig

const std = @import("std");
const Allocator = std.mem.Allocator;
const bdz = @import("bdz.zig");

const c = @cImport({
    @cInclude("cmph.h");
});

// pack packs cmph hashes for the given input and returns a slice ("cmph pack
// minus first 4 bytes") for further storage. The slice must be freed by the
// caller.
const packErr = Allocator.Error || error{Overflow};
pub fn pack(allocator: Allocator, input: [][*:0]const u8) packErr![]const u8 {
    var cvector = @ptrCast([*c][*c]u8, input.ptr);
    const len = try std.math.cast(c_uint, input.len);
    var source = c.cmph_io_vector_adapter(cvector, len);
    defer c.cmph_io_vector_adapter_destroy(source);
    var config: *c.cmph_config_t = c.cmph_config_new(source) orelse return error.OutOfMemory;
    c.cmph_config_set_algo(config, c.CMPH_BDZ);
    c.cmph_config_set_b(config, 7);
    var hash: *c.cmph_t = c.cmph_new(config) orelse return error.OutOfMemory;
    c.cmph_config_destroy(config);

    const size = c.cmph_packed_size(hash);
    var buf = try allocator.alloc(u8, size);
    c.cmph_pack(hash, &buf[0]);
    c.cmph_destroy(hash);
    return buf[4..];
}

const testing = std.testing;

const items = .{
    "aaaaaaaaaa",
    "bbbbbbbbbb",
    "cccccccccc",
    "dddddddddd",
    "eeeeeeeeee",
    "ffffffffff",
    "gggggggggg",
    "hhhhhhhhhh",
    "iiiiiiiiii",
    "jjjjjjjjjj",
};
const items_len = items.len;

fn samplePack(allocator: Allocator) ![]const u8 {
    var vector = std.ArrayList([*:0]const u8).init(allocator);
    defer vector.deinit();
    try vector.appendSlice(&items);
    return pack(allocator, vector.items);
}

test "basic pack/unpack" {
    const buf = try samplePack(testing.allocator);
    defer testing.allocator.free(buf);
    try testing.expect(buf.len < 100);

    var used: [items_len]bool = undefined;

    inline for (items) |elem| {
        const hashed = try bdz.search(buf, elem);
        used[hashed] = true;
    }

    for (used) |item| {
        try testing.expect(item);
    }
}