telemetry-cli

Command-line tools for binary telemetry, built for Unix pipelines:

• pick decodes fields from fixed-length binary records: integers, floats, complex numbers, characters and bit fields, in any byte order (manual).
• recs extracts records from a byte stream by their start (and end) markers (manual).
• recl estimates the record length of a binary file (manual).
• tgen generates synthetic telemetry from a command file describing each record (manual).

$ printf '\x00\x00\x00\x01\xff\xfeHi\x00\x00\x00\x02\x00\x07ok' > demo.bin
$ pick demo.bin 8 -q order=big u s 2S
1 -2 Hi
2 7 ok
$ pick demo.bin 8 -q order=big -n u0x w2b
0 00 00 00 01  1111 1111 1111 1110
1 00 00 00 02  0000 0000 0000 0111

Together: tgen makes a stream, recs pulls the frames out of it, recl finds their length, and pick decodes them. See Using the tools together.

Install

Requires Python 3.10 or later. The package is telemetry-cli on PyPI, and is best installed as an isolated command-line tool:

pipx install telemetry-cli
# or
uv tool install telemetry-cli

Either installs the commands pick, recs, recl and tgen. You can also pip install telemetry-cli in a virtual environment.

History

These are Python ports of tools written at JPL. pick and recl began together as C++ tools, for a project whose telemetry arrived with no documentation of its structure: recl to find the length of the records, and pick to take them apart and work out what was in them. (A bit reminiscent of the film Contact.) Both were later rewritten in Perl (pick by 2000), and tgen (2003) and recs joined them.

The command lines are the same, and so is the output, apart from fixes listed in the CHANGELOG: chiefly pick's byte-order handling on little-endian machines, and recl's measurement. tgen's command files are now Python; tgen --convert translates the old Perl ones. The Perl originals, with their history, are in legacy/.

Development

python -m venv .venv && .venv/bin/pip install -e '.[dev]'
.venv/bin/pytest

The tests compare each tool with its Perl original. tests/golden/ holds the Perl's output for large tables of command lines (about 2,200 for pick, 700 for recs), and the tests check that the Python version reproduces it byte for byte. Where it deliberately differs, the case names the CHANGELOG entry that explains why. With perl installed, further tests run random command lines through both versions and compare them (HYPOTHESIS_PROFILE=deep pytest for a longer search), and python tools/make_golden.py --check confirms the saved output is current. The examples in the documentation run as tests too.