Parsing extra metadata

GNU Radio
1

Hello,

I am trying to add some extra information to the standard metadata
file header.

Following the information posted on
http://gnuradio.org/doc/doxygen/page_metadata.html , I am able to write
and read the standard metadata. I have attempted to add extra
metadata with the following lines in my flowgraph:

            # add some extra meta data
key = pmt.pmt_intern("dummy")
val = pmt.pmt_from_long(77)
extras = pmt.pmt_make_dict()
extras = pmt.pmt_dict_add(extras, key, val)
extras_str = pmt.pmt_serialize_str(extras)

            nItems = 1024*size
self.blocks_file_meta_sink_0 =

blocks.file_meta_sink(gr.sizeof_float*1,“./meta_test.dat”,
samp_rate, 1, blocks.GR_FILE_FLOAT, False, nItems, extras_str,
True)
self.blocks_file_meta_sink_0.set_unbuffered(False)

I think that the writing is going OK, since the length of the header
increases by a reasonable amount and a Unix hexdump indicates that the
“dummy” data is present. However, I am stuck when it comes to getting
at the extra data using the parse_file_metadata functions. In
particular, when I try to read the data using:

    cHeader = c[0:185]   # character array read from file
pmt_dict = pmt.pmt_deserialize_str(cHeader)
p = parse_file_metadata.parse_header(pmt_dict,False)
p_extra = parse_file_metadata.parse_extra_dict(pmt_dict,info,True)

The first parse_file_metadata line successfully provides the standard
metadata, but the second parse_file_metadata does not provide the extra
“dummy” data. I suspect that it has to do with my poor understanding
of what should go in the info argument.

Any help would be greatly appreciated.

Sincerely,
Dan Marlow

2

On Tue, Aug 19, 2014 at 3:13 PM, Daniel M. [email protected]
wrote:

            # add some extra meta data

self.blocks_file_meta_sink_0.set_unbuffered(False)
p_extra = parse_file_metadata.parse_extra_dict(pmt_dict,info,True)

Dan,

What happens when you run gr_read_metadata on the file that’s created?
Does
that show your extra header info?

Tom

3

Hi Tom,

Thanks for looking at this. The answer is “yes,” as can be seen from
the output below. The rx_freq field is not something that I added.
I did add the “dumm” field, however.

Sincerely,
Dan Marlow

marlow@phy-gr2:~/pulsar/DAQ$ gr_read_file_metadata meta_test.dat.hdr
HEADER 0
Version Number: 0
Sample Rate: 4000000.00 sps
Seconds: 0.184295
Item size: 4
Data Type: float (5)
Complex? False
Header Length: 189 bytes
Extra Length: 40
Extra Header? True
Size of Data: 65536 bytes
16384 items

Extra Header:
rx_freq:
1.2e+09
dumm:
1408561917746092

4

On Thu, Aug 21, 2014 at 6:16 PM, Daniel M. [email protected]
wrote:

HEADER 0
16384 items

Extra Header:
rx_freq:
1.2e+09
dumm:
1408561917746092

Ok, so that’s good. Looks like the basic concept is working fine, then.
The
rx_freq field is coming from the USRP. The UHD sends tags with
information
about the settings, which get stored in the metadata files. If you
change
the frequency, it’ll generate another tag for you.

It looks like you should look into the gr_read_file_meatadata app to see
how it’s using the parse_metadata. That might clue you into what’s going
wrong in your project.

Tom