Unwanted zeros in AWK output

Question

Problem explanation

Using bash on Ubuntu, I read some data packets continuously over the serial (it's connected to a wireless base-station) using a java prog and pipe it through cut and awk to get a result. I use unbuffer to see the output in real-time (otherwise the output waits till the buffer is full).

Syntax of the command that I issue on the terminal:

java prog | unbuffer -p process1 | unbuffer -p process2

Actual command I issue:

java net.tinyos.tools.Listen -comm serial@/dev/ttyUSB0:telosb |
unbuffer -p cut -c49-50,52-53,55-56,58-59,60,61-62,64-65,67-68,70-71,72,73-74,76-77,79-80,82-83 |
unbuffer -p awk -F " " '{printf("%d %5.4f %5.4f\n",("0x"$1)*1,("0x"$2)*2*0.0005714,("0x"$3)/612500)}'

My output:

43301 1.9370 5.0124
0 0.0000 0.0000
43303 1.9279 5.0071
0 0.0000 0.0000
43305 1.9428 5.0125
0 0.0000 0.0000
43306 1.9428 5.0072
0 0.0000 0.0000
. 
.
.

Line of zeros aren't real but spurious. If I execute only the java prog (i.e., before cut) or java prog + process1 (i.e., before AWK), they don't appear. They appear in the output only after process2, after using awk.

What I want

I wanna find out why the zeros are there in my output and how.

Appreciate your help. Thanks.

---

Supplementary data

One of the received data packets looks like this (output after java prog). All are hex (ungrouped at this stage):

 00 FF FF 00 01 1C 22 95 00 00 00 00 00 00 00 00 00 00 25 83 00 00 06 95 00 2E BF 60 00 2E C0 BB 00 00 00 00

Above, it looks like all numbers are in groups of two. But there are number clusters within -- hence why I cut them in such a way that I regroup them accordingly. Output after cut is as follows:

Output after java prog + process1:

0000A995 00000697 002EBE00
0000A998 0000069F 002EE4FB
0000A99C 000006A4 002EBF9F
0000B421 000006A4 002ECC6B
 .
 .
 .

NOTE: The two outputs are not their own conversions. They were taken at different instances, just to provide an example only.

Answer 1

So, I answer my question inspired by the answer I got + the great comments and help tidbits.

Problem

Using awk. No gawk installed.

Command:

 java net.tinyos.tools.Listen -comm serial@/dev/ttyUSB0:telosb |
 unbuffer -p cut -c49-50,52-53,55-56,58-59,60,61-62,64-65,67-68,70-71,72,73-74,76-77,79-80,82-83|
 unbuffer -p awk -F " " '{printf("%d\t%5.4f\t%5.4f\t\n","0x"$1,("0x"$2)*2*0.0005714,("0x"$3)/612500)}'

Output:

 2723   1.9233  5.0164  
 0  0.0000  0.0000  
 2726   1.9268  5.0164  
 0  0.0000  0.0000  
 2730   1.9428  5.0073  
 0  0.0000  0.0000  
 2732   1.9348  4.9866  
 0  0.0000  0.0000
 .
 .
 .

Zeros were undesirable.

Solution

Installed gawk. Using it, I changed the command as in the following:

Command:

 java net.tinyos.tools.Listen -comm serial@/dev/ttyUSB0:telosb |
 unbuffer -p cut -c49-50,52-53,55-56,58-59,60,61-62,64-65,67-68,70-71,72,73-74,76-77,79-80,82-83|
 gawk -F " " --non-decimal-data '{printf("%d\t%5.4f\t%5.4f\t\n","0x"$1,("0x"$2)*2*0.0005714,("0x"$3)/612500)}'

Output:

 2723   1.9233  5.0164  
 2726   1.9268  5.0164  
 2730   1.9428  5.0073  
 2732   1.9348  4.9866  
 .
 .
 .

Zeros gone. Only changes were gawk instead of awk, addition of --non-decimal-data option, and removal of unbuffer -p.

Note: When I leave unbuffer -p on, the zeros reappear. Food for thought. Perhaps, someone can attend to that later.

Answer 2

You are using printf("%d \n","0x"$1), that is printf expecting a digit to be the value of Ox$1. Hence, when it is not a digit, awk prints a 0.

Probably you want to use %s instead, so that a string will be printed.

See an example:

$ awk 'BEGIN {printf("%d \n","0xAA")}'
0 
$ awk 'BEGIN {printf("%s \n","0xAA")}'
0xAA 

You can see a full list of Format control letters of printf in the awk manual: http://www.gnu.org/software/gawk/manual/gawk.html#Control-Letters.

Interesting reading related to hexadecimal values in awk and gawk: http://www.delorie.com/gnu/docs/gawk/gawk_77.html