Skip to content

CLAUDE.md

Latest commit

 

History

History
116 lines (88 loc) · 4.39 KB

CLAUDE.md

File metadata and controls

116 lines (88 loc) · 4.39 KB
Raw

CLAUDE.md — Python Module

This file provides guidance to Claude Code (claude.ai/code) when working with the Python module.

Prerequisites

The Python module depends on the C++ shared library (libtsfile). Build C++ first:

# From repo root
./mvnw clean package -P with-cpp -DskipTests

Build Commands

# Build everything via Maven (builds C++ then Python)
./mvnw clean verify -P with-python

# Or build directly (after C++ is built)
cd python
python setup.py build_ext --inplace

# Run tests
pytest tests/

# Run a single test
pytest tests/test_write_and_read.py -k "test_name"

# ASan/LSan coverage of the bindings (Linux; run from the repo root; see
# SANITIZER.md). TSFILE_SANITIZER is required or test_sanitizer_ownership.py
# skips every case, and the runtime must be preloaded or the instrumented
# extensions abort on import.
./mvnw -P with-python -Dtsfile.sanitizer=address -Dbuild.type=RelWithDebInfo \
  -Dbuild.test=OFF clean test-compile
TSFILE_SANITIZER=address \
ASAN_OPTIONS="detect_leaks=0:abort_on_error=1:halt_on_error=1" \
LSAN_OPTIONS="suppressions=$PWD/python/tests/sanitizer/lsan_suppressions.txt" \
LD_PRELOAD="$(gcc -print-file-name=libasan.so):$(gcc -print-file-name=libstdc++.so.6)" \
  python/venv/bin/python -m pytest python/tests/test_sanitizer_ownership.py -v

# Format code
cd ../ && ./mvnw spotless:apply   # Uses Black 26.3.1

Source Structure

python/tsfile/
├── __init__.py              # Public API exports
├── constants.py             # Enums: TSDataType, TSEncoding, Compressor, ColumnCategory
├── schema.py                # TimeseriesSchema, DeviceSchema, TableSchema, ColumnSchema
├── tablet.py                # Tablet class for batch columnar data
├── field.py / row_record.py # Row-based data structures
├── tag_filter.py            # Tag filtering DSL for queries
├── tsfile_table_writer.py   # High-level table writer API
├── utils.py                 # to_dataframe(), dataframe_to_tsfile()
├── exceptions.py            # Custom exception types
│
├── tsfile_py_cpp.pyx        # Cython core: Python ↔ C++ type conversion
├── tsfile_reader.pyx        # Cython reader wrapper
├── tsfile_writer.pyx        # Cython writer wrapper
├── tsfile_cpp.pxd           # C++ interface declarations (cwrapper)
│
└── dataset/                 # High-level multi-file API
    ├── dataframe.py         # TsFileDataFrame (pandas-like interface)
    ├── timeseries.py        # Timeseries, AlignedTimeseries
    ├── reader.py            # Multi-file reader coordination
    └── merge.py             # K-way merge for overlapping time ranges

Architecture Notes

  • Layered design: C++ core → Cython bridge (*.pyx) → Pure Python API
  • setup.py copies pre-built C++ headers and shared libraries from ../cpp/target/build/
  • Cython binds to cwrapper/tsfile_cwrapper.h from the C++ module
  • C error codes are mapped to Python exceptions via check_error() in tsfile_py_cpp.pyx
  • Requires Python 3.9+, numpy >= 2.0, pandas >= 2.0, pyarrow >= 16.0

Code Style

  • Formatter: Black 26.3.1 (via Spotless Maven plugin)

Testing

  • Framework: pytest
  • Tests in python/tests/, 19 test files covering write, read, filtering, DataFrame conversion, and Arrow format

License Header

Every new file must include the Apache License 2.0 header at the top. For Python files, use # line comments. For .pyx/.pxd files, also use # comments. See any existing file for the exact wording.

Git Commit

  • Do NOT add Co-Authored-By trailer to commit messages.