- Replace [tool.poetry] with [project] section
- Use hatchling as build backend
- Add [dependency-groups] for dev dependencies
- Add duckdb>=1.0.0 to dependencies
- Generate uv.lock (remove poetry.lock)

Co-Authored-By: Warp <agent@warp.dev>

- Add query.py with QueryBuilder class for DuckDB data loading
- Add plot.py with Figure wrapper and TimeSeriesFigure stub
- Remove provider.py (drop provider abstraction)
- Remove data.py (replaced by DuckDB-based query.py)
- Rewrite __init__.py CLI to use new modules
- Add new CLI options: --where, --after, --before, --bucket, --json, --csv
- Add --format option for explicit input format
- Mark -F/--resample as deprecated (use -B/--bucket)
- Add --first, --last aggregation methods

Co-Authored-By: Warp <agent@warp.dev>

- Add bucket interval parsing (15min, 1h, 1d shorthand)
- Implement time_bucket() for time-series aggregation
- Implement --where SQL filtering
- Implement --after/--before datetime filters
- Implement -S/--scale datetime offset conversion
- Fix datetime precision handling (us/ns/ms)
- Add column caching in QueryBuilder
- Improve stdin format handling (CSV, JSON)
- Quote column names in SQL for safety

Co-Authored-By: Warp <agent@warp.dev>

- Port TimeGranularity enum and detect_time_granularity()
- Port generate_time_ticks() with adaptive tick generation
- Add TimeTickResult dataclass for tick information
- Implement _CustomTplotFigure with custom x-axis drawing
- Add TimeSeriesFigure wrapper with secondary label support
- Integrate TimeSeriesFigure in PlotApp for -T flag
- Use UTC consistently for timezone handling
- Add helper functions _epoch_to_datetime, _datetime_to_epoch

Time-series plots now show formatted HH:MM labels with date context.

Co-Authored-By: Warp <agent@warp.dev>

- Add --json and --csv output modes for data export
- Add pyarrow dependency for Parquet support
- Add comprehensive tests for QueryBuilder (test_query.py)
- Add CLI integration tests (test_cli.py)
- Update README with full usage documentation
- Document all new CLI options: --format, --where, --after, --before,
  -B/--bucket, aggregation methods, output modes
- Change install instruction to use 'uv tool install'

Co-Authored-By: Warp <agent@warp.dev>

- Add --by COLUMN CLI option to group data by a categorical column
- Implement PIVOT query in QueryBuilder to transform long-format to wide-format
- Handle NaN values in line plots (tplot can't render NaN coordinates)
- Works with and without time bucketing (-B/--bucket)

Example: plot data.csv -x timestamp -y value --by category -T -B 1h --mean

Co-Authored-By: Warp <agent@warp.dev>

When --count is used with -B/--bucket but no -y column specified,
uses COUNT(*) to count occurrences per time bucket. This eliminates
the need for awk injection of ',1' when plotting event frequency.

Example usage:
  # Count log events per hour
  grep 'opened session' logs | plot -x timestamp -T -B 1h --count

  # Count by category
  plot logs.csv -x timestamp -T -B 1h --count --by user

Co-Authored-By: Warp <agent@warp.dev>

Co-Authored-By: Warp <agent@warp.dev>

Wraps tplot.Figure.scatter with consistent interface.

Co-Authored-By: Oz <oz-agent@warp.dev>

Co-Authored-By: Oz <oz-agent@warp.dev>

Co-Authored-By: Oz <oz-agent@warp.dev>

Co-Authored-By: Oz <oz-agent@warp.dev>

Co-Authored-By: Oz <oz-agent@warp.dev>

Co-Authored-By: Oz <oz-agent@warp.dev>