implemented decoding indefinite array

[nagrawal63]

This has the indefinite array implemented in it. Please let me know if there are any issues.

[bergzand]

Please rebase on latest master as this fixes the current CI issues in this PR.

[bergzand]

Also, please remove any commented code from your commits

[benpicco]

@Nishchay-sopho ping 😉

[nagrawal63]

Yeah sure, I totally forgot about it 😅 . Will do the needful asap

[benpicco]

I would have expected to find the implementation in src/decoder.c, not in the test case 😉

[nagrawal63]

The latest commit resolves the issue #20 and also resolves indefinite map support.

Testing :
Use the function below to encode an indefinite array

static void _encode(nanocbor_encoder_t *enc)
{
    nanocbor_fmt_array_indefinite(enc);
    nanocbor_fmt_bool(enc, true);
    nanocbor_fmt_bool(enc, false);
    nanocbor_fmt_uint(enc, UINT32_MAX);
    nanocbor_fmt_int(enc, INT32_MIN);
    nanocbor_fmt_map(enc, 4);
    nanocbor_fmt_uint(enc, 8);
    nanocbor_fmt_int(enc, 30);
    nanocbor_fmt_int(enc, -30);
    nanocbor_fmt_int(enc, 500);
    nanocbor_fmt_int(enc, -500);
    nanocbor_put_tstr(enc, "this is a long string");
    nanocbor_fmt_float(enc, 0.34);
    nanocbor_put_bstr(enc, (uint8_t*)"bytez", sizeof("bytez"));
    nanocbor_fmt_null(enc);
    nanocbor_fmt_end_indefinite(enc);
}

On calling decode on this encoded buffer we get the following output:

[
  True,
  False,
  4294967295,
  -2147483648,
  {
    8: 30,
    -30: 500,
    -500: "this is a long string",
    Unsupported float: "0x62, 0x79, 0x74, 0x65, 0x7a, 0x00, ",
  },
  NULL,
],

[benpicco]

You'll still need to do a git rebase master and solve the merge conflict :)

[benpicco]

Lint has still some issues, most of them should be fixed if you rebase.

Maybe then @bergzand can take a look at this 😉

[nagrawal63]

The lint check is giving a weird error:

/tmp/cirrus-ci-build/src/decoder.c:282:33: error: use of a signed integer operand with a binary bitwise operator [hicpp-signed-bitwise,-warnings-as-errors]
    if (_value_match_exact(it, (type << NANOCBOR_TYPE_OFFSET | NANOCBOR_SIZE_INDEFINITE)) == 1) {

More on it here

All the variables in boolean operation type, NANOCBOR_TYPE_OFFSET and NANOCBOR_SIZE_INDEFINITE are unsigned integers.

Could you kindly let me know why this is happening?

[benpicco]

Suggested change

	if (_value_match_exact(it, (type << NANOCBOR_TYPE_OFFSET | NANOCBOR_SIZE_INDEFINITE)) == 1) {
	if (_value_match_exact(it, (type << NANOCBOR_TYPE_OFFSET) | NANOCBOR_SIZE_INDEFINITE) == 1) {

Does that help? (it makes it at least easier to parse for humans)

[nagrawal63]

lint is still not satisfied by this change

[benpicco]

hmm I don't know either 😟 - maybe try casting it to (unsigned)

[nagrawal63]

Now lint is happy. This is so weird. The type variable was already uint8_t. 😕

Let me know if this works for you, I will then squash the commits to make them ready for merge.

[benpicco]

Sure, it's up to @bergzand to merge this though

[bergzand]

Tested this locally, code looks good! Thanks for fixing