feat!: DSV swim-meet file parser

[nilskntl]

Initial implementation: a standalone parser for DSV swim-meet files (.dsv5–.dsv8, .dsv8z) into a Pydantic schema, exposed as a library, a CLI and an optional FastAPI service.

Design

• One declarative element table (spec/elements.py): every element is an ElementSpec with its attributes in file order; version and Listart deltas are versions=/file_types= markers, not branches. Transcribed from the official Format 6/7/8 documents; Format 5 derived from the Format 6 change log.
• Four-stage reader (core/): decoding (ZIP, encoding sniffing) → lexer → binder → parser. The reader never raises on file content — bad values become None plus a Diagnostic.
• Swim times as integer milliseconds, enums as speaking values, no database, no state.

Included

• dsv-parser CLI (parse, check, spec) and FastAPI service as an optional extra
• Docker/compose setup, CI, release-please, docs under docs/
• 157 tests on synthetic fixtures (~98% coverage); validated against nine real EasyWk files in Format 6/7 with zero errors and warnings
• Fixes from a first self-review round: overflow/encrypted-zip handling, strict swim-time validation, future-version clamping, U+2028-safe line splitting, duplicate-FORMAT diagnostic

🤖 Generated with Claude Code

https://claude.ai/code/session_01SV14g59Zd2sS1UGZxJVfhB