It's very easy for a short packet to contain a very high number of
nested arrays (a packet with N bytes can be N-1 nested arrays), which
could cause stack overflow problems in devices using
cbor_value_map_find_value, so ensure that we won't crash.

The actual value needs to be chosen by the implementor depending on the
stack size of their devices and the expected usage.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

The encoder only supports two error conditions: out of memory (need more
buffer) and illegal simple types. It's possible for the encoder to
encode an array or map with a size the decoder can't parse (UINT32_MAX),
but that's not an error condition.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This method is recursive and O(n) on the number of elements in the
payload.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This is the first convenience API. As such, this begins by skipping
tagged values and it will notnot assert on the wrong types.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

The next commit will introduce a function to compare a CBOR text string
with a const char * without extracting it. Since the iteration code is
pretty much the same, just replacing memcpy with memcmp, we can reuse
the same code. That makes it more maintainable in the future too.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

I tried to use mostly valid tag + data combinations in the test, even
though the parser does and will not validate the combinations unless
asked for it (in a function that doesn't exist yet).

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Again a silly mistake in sign-extension of (char)0xff to -1.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Wrong direction of the operator → we thought we overflowed when adding 0
to 0...

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This commit also starts to add unit tests.

The API, as tested, works. The simplereader.c example is probably broken
and will be fixed later.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This version compiles but does not work yet.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This version doesn't work yet.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>

This code was originally part of a contribution to QtCore, see:
     https://codereview.qt-project.org/107465
     https://codereview.qt-project.org/107466



This version does not compile nor work yet.

Signed-off-by: Thiago Macieira <thiago.macieira@intel.com>