pdf-odin-llm/filter_runlength_test.odin

package pdf

import "core:mem/virtual"
import "core:slice"
import "core:testing"

@(private = "file")
Rl_Case :: struct {
    name: string,
    in_:  []byte,
    want: []byte,
}

@(private = "file")
check_rl :: proc(t: ^testing.T, c: Rl_Case) {
    out, err := run_length_decode(c.in_)
    defer delete(out)
    testing.expectf(t, err == .None, "%s: err = %v", c.name, err)
    testing.expectf(t, slice.equal(out, c.want), "%s: got %v, want %v", c.name, out, c.want)
}

@(test)
test_run_length_literals :: proc(t: ^testing.T) {
    cases := []Rl_Case{
        {"one byte",   {0, 'A', 128},                {'A'}},
        {"three bytes", {2, 'a', 'b', 'c', 128},     {'a', 'b', 'c'}},
        {"two runs",   {1, 'a', 'b', 0, 'c', 128},   {'a', 'b', 'c'}},
        {"nul literal", {1, 0x00, 0x80, 128},        {0x00, 0x80}},
        // 128 only means EOD in a length position; as data it is just a byte.
        {"128 as data", {0, 128, 128},               {128}},
    }
    for c in cases {
        check_rl(t, c)
    }
}

@(test)
test_run_length_repeats :: proc(t: ^testing.T) {
    cases := []Rl_Case{
        // 257-255 = 2 is the shortest repeat the encoding can express.
        {"two",   {255, 'x', 128},        {'x', 'x'}},
        {"three", {254, 'x', 128},        {'x', 'x', 'x'}},
        {"four",  {253, 0xFF, 128},       {0xFF, 0xFF, 0xFF, 0xFF}},
        {"mixed", {1, 'a', 'b', 254, 'c', 0, 'd', 128},
                  {'a', 'b', 'c', 'c', 'c', 'd'}},
    }
    for c in cases {
        check_rl(t, c)
    }
}

@(test)
test_run_length_boundaries :: proc(t: ^testing.T) {
    // The longest literal run (127 -> 128 bytes) and the longest repeat
    // (129 -> 128 copies), the two lengths an off-by-one lands on.
    enc := make([dynamic]byte)
    defer delete(enc)
    want := make([dynamic]byte)
    defer delete(want)

    append(&enc, 127)
    for i in 0 ..< 128 {
        append(&enc, byte(i))
        append(&want, byte(i))
    }
    append(&enc, 129, 'Z')
    for _ in 0 ..< 128 {
        append(&want, 'Z')
    }
    append(&enc, 128)

    check_rl(t, {"max runs", enc[:], want[:]})
}

@(test)
test_run_length_degenerate :: proc(t: ^testing.T) {
    cases := []Rl_Case{
        {"empty",           {},                     {}},
        {"eod immediately", {128},                  {}},
        {"eod then junk",   {128, 5, 'a', 'b'},     {}},
        // No EOD marker: the data simply ends.
        {"missing eod",     {1, 'a', 'b'},          {'a', 'b'}},
        {"repeat, no eod",  {254, 'q'},             {'q', 'q', 'q'}},
    }
    for c in cases {
        check_rl(t, c)
    }
}

@(test)
test_run_length_truncated :: proc(t: ^testing.T) {
    cases := []Rl_Case{
        // A literal run promising more bytes than remain keeps what is there.
        {"short literal",   {5, 'a', 'b'},          {'a', 'b'}},
        {"literal, none",   {5},                    {}},
        // A repeat with no byte to repeat produces nothing and ends.
        {"repeat, no byte", {250},                  {}},
        {"tail after good", {1, 'a', 'b', 9, 'c'},  {'a', 'b', 'c'}},
    }
    for c in cases {
        check_rl(t, c)
    }
}

@(test)
test_run_length_truncated_everywhere :: proc(t: ^testing.T) {
    // Every prefix of a well-formed stream must decode a prefix of the
    // plaintext without crashing.
    enc := []byte{2, 'a', 'b', 'c', 254, 'z', 0, 'Q', 129, '!', 128}
    full := "abczzzQ"
    for cut in 0 ..= len(enc) {
        out, err := run_length_decode(enc[:cut])
        testing.expectf(t, err == .None, "cut %d: err = %v", cut, err)
        n := min(len(out), len(full))
        testing.expectf(t, slice.equal(out[:n], transmute([]byte)full[:n]),
            "cut %d: got %v", cut, out)
        // Nothing past the '!' run can appear, so the length is bounded.
        testing.expectf(t, len(out) <= len(full) + 128, "cut %d: %d bytes", cut, len(out))
        delete(out)
    }
}

@(test)
test_run_length_all_lengths :: proc(t: ^testing.T) {
    // Sweep every length byte as the first byte of a stream, with one byte of
    // payload behind it, and assert the decoded length is exactly what the
    // encoding promises. This is the table the filter is.
    for length in 0 ..= 255 {
        buf := [3]byte{byte(length), 'p', 128}
        out, err := run_length_decode(buf[:])
        want := 0
        switch {
        case length == 128:
            want = 0
        case length < 128:
            // "length+1 literal bytes", of which only 2 are present here
            // ('p' and the 128 that would otherwise have been EOD).
            want = min(length + 1, 2)
        case:
            want = 257 - length
        }
        testing.expectf(t, err == .None, "length %d: err = %v", length, err)
        testing.expectf(t, len(out) == want, "length %d: got %d bytes, want %d",
            length, len(out), want)
        delete(out)
    }
}

@(test)
test_run_length_cross_check :: proc(t: ^testing.T) {
    // Vectors from an independent RunLength encoder over random bytes drawn
    // from a small alphabet, so literal and repeat runs interleave.
    cases := []struct{name, in_, want: string}{
        {"n=1",  "\x00\xFF\x80", "\xFF"},
        {"n=3",  "\xFF\x00\x00a\x80", "\x00\x00a"},
        {"n=9",  "\x05\x00\xFFb\x00\xFF\x00\xFFc\x00a\x80", "\x00\xFFb\x00\xFF\x00cca"},
        {"n=40",
         "\x00b\xFF\x00\x02a\xFFa\xFF\x00\x09cbcab\xFFcbc\x00\xFFa\x07\x00bacab\xFFa\xFFc\x09\x00a\xFFca\x00ba\x00b\x80",
         "b\x00\x00a\xFFa\x00\x00cbcab\xFFcbc\x00aa\x00bacab\xFFacc\x00a\xFFca\x00ba\x00b"},
    }
    for c in cases {
        check_rl(t, {c.name, transmute([]byte)c.in_, transmute([]byte)c.want})
    }
}

@(private = "file")
rl_fuzz_next :: proc(state: ^u64) -> u32 {
    state^ = state^ * 6364136223846793005 + 1442695040888963407
    return u32(state^ >> 33)
}

@(test)
test_run_length_fuzz :: proc(t: ^testing.T) {
    // Random bytes decode to at most 128 bytes per input byte and must never
    // crash or hang, whatever the length bytes happen to say.
    state: u64 = 0xBADF00D
    buf: [256]byte
    for round in 0 ..< 400 {
        n := int(rl_fuzz_next(&state) % u32(len(buf) + 1))
        for i in 0 ..< n {
            buf[i] = byte(rl_fuzz_next(&state))
        }
        out, err := run_length_decode(buf[:n])
        testing.expectf(t, err == .None, "round %d: err %v", round, err)
        testing.expectf(t, len(out) <= 128 * n, "round %d: grew to %d from %d",
            round, len(out), n)
        delete(out)
    }
}

@(test)
test_run_length_honours_allocator :: proc(t: ^testing.T) {
    arena: virtual.Arena
    testing.expect(t, virtual.arena_init_growing(&arena) == nil)
    defer virtual.arena_destroy(&arena)

    out, err := run_length_decode({2, 'a', 'b', 'c', 128}, virtual.arena_allocator(&arena))
    testing.expect(t, err == .None)
    testing.expect(t, slice.equal(out, transmute([]byte)string("abc")))
}