Skip to content

About

AIS toolkit scaffold in development for vessel tracks, activity, and maritime traffic data. No processing pipelines are implemented here yet.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

17 Commits

Folders and files

Repository files navigation

AIS Toolkit

Ink panorama of a coastal archipelago with vessels, AIS signal paths, and a shore antenna

Independently installable, species-neutral normalization and contracts for future source-to-daily-H3 processing. Phase 2 adds offline local ingestion to private native Parquet using bounded batches and SQLite spill. Development/tests use synthetic fixtures only. Phase 3 adds bounded local estimated trajectories and direct R6/R7 contributions from verified ingestion evidence. Scientific defaults and interpolation accuracy remain unqualified. Live acquisition, final daily products, regional qualification and consumer integration remain deferred.

python -m venv .venv
source .venv/bin/activate
python -m pip install -e '.[dev]'
ais inspect-catalog
ais inspect-metrics
ais validate-csv tests/fixtures/current_synthetic.csv --era '2025+'
ais validate-csv tests/fixtures/legacy_synthetic.csv --era 2018-2024
pytest -q
ruff check .
ruff format --check .
python -m build --outdir dist-current
python scripts/wheel_smoke.py dist-current

When network access is unavailable, an environment with installed build/setuptools and pytest tools can use python -m build --no-isolation and python scripts/wheel_smoke.py dist-current --offline-test-tools --wheelhouse /chosen/wheels. The wheelhouse must contain compatible declared Arrow/Zstandard/H3/Shapely (and transitive NumPy) dependency wheels. Only pytest tool modules are copied into the temporary runner; the package and its runtime dependencies install into a fresh environment from wheels with no index access.

validate-csv inspects a bounded, already-provided uncompressed CSV; its default cap is 10,000 rows. Exact field sets are required, independent of column order. Exit 0 means nonempty contract inspection passed; 1 means rejected rows, conflicts or empty input; 2 means file/schema/cap failure with quarantine status. It does not certify source rights, coverage or scientific suitability. The default rejects naive timestamps; explicitly select --naive-time-policy dictionary_utc only when the pinned source dictionary applies.

See offline ingestion for deliberate input/output paths, receipts, halos, carried barriers, atomic completion, limits and runnable synthetic example. Use ais ingest-local job.json /chosen/private/output --carried /chosen/previous/run only with explicitly provisioned local inputs; the carried directory is optional for the first run.

See local estimated trajectories for explicit gap/speed/coast/ projection settings, direct H3 clipping, UTC/core ownership, unresolved identity episodes and numerical limits. ais estimate-local job.json /chosen/private/output emits private contributions and accepted/rejected ledgers. To run a fabricated example in a NEW directory:

python scripts/synthetic_trajectories.py /chosen/new-synthetic-workspace
ais estimate-local /chosen/new-synthetic-workspace/trajectory-job.json /chosen/private-estimates

The script constructs fabricated Phase 2 evidence and a job; the CLI runs estimation. Its synthetic thresholds, coordinates and fake producer SHA are not study defaults.

See foundation contracts for schemas, metric semantics, source verification, unresolved choices and deferred work. Real AIS, vessel-level intermediates and model inputs stay private and outside tracked source. Code can be public; AIS presence does not measure people aboard, whale presence, noise or disturbance.

About

AIS toolkit scaffold in development for vessel tracks, activity, and maritime traffic data. No processing pipelines are implemented here yet.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages