Move _skip and _parse_number functions to C extension module

[Copilot]

Moves the _skip and _parse_number functions from _parser.py to a new C extension module _skip_ext.c to make room for future parser enhancements.

Changes

• Build system: Switched from uv_build to setuptools to support C extensions

  • Added setup.py defining the extension module
  • Updated pyproject.toml with setuptools configuration

• C extension (_skip_ext.c): 400+ line implementation handling:

  • _skip function: Whitespace and comment skipping
    • Whitespace skipping (space, tab, newline, etc.)
    • Single-line comments (//) with backslash line continuation
    • Multi-line comments (/* */) with proper error on unclosed blocks
    • newline_ok parameter to control newline handling
  • _parse_number function: Number parsing
    • Integer parsing
    • Float parsing (including scientific notation, infinity, NaN)
    • Type-specific parsing (int, float, or int|float)
    • Proper error handling with ParseError
  • Direct buffer usage (no copies) for memory efficiency
  • Proper reference counting and module cleanup
  • Lookup tables for efficient character classification

• Parser module: Replaced Python implementations with imports from C extension

  • _skip: Direct import from C extension (44 lines removed)
  • _parse_number: Wrapper that calls C version and converts exceptions (32 lines removed)
  • Removed unused lookup tables (_IS_POSSIBLE_FLOAT, _IS_POSSIBLE_INTEGER)

Both C functions maintain identical behavior to the Python versions. All 68 parsing tests pass.

# Before: Python implementations in _parser.py
def _skip(contents: bytes | bytearray, pos: int, *, newline_ok: bool = True) -> int:
    # 44 lines of Python logic
    ...

def _parse_number(contents: bytes | bytearray, pos: int, *, target: type[int] | type[float] | type[int | float] = int | float) -> tuple[int | float, int]:
    # 32 lines of Python logic
    ...

# After: Imports from C extension
from ._skip_ext import _skip
from ._skip_ext import _parse_number as _parse_number_c

def _parse_number(...):
    # Wrapper that converts C ParseError to Python ParseError
    ...

Original prompt

Move the _skip function from the parser module to a new C extension module, to make room for

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 84.59%. Comparing base (5fbadb9) to head (67a7661).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #740      +/-   ##
==========================================
- Coverage   84.67%   84.59%   -0.08%     
==========================================
  Files          27       27              
  Lines        3315     3259      -56     
==========================================
- Hits         2807     2757      -50     
+ Misses        508      502       -6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[gerlero]

@copilot also translate the _parse_number function to C

[Copilot]

@copilot also translate the _parse_number function to C

Done! Added _parse_number to the C extension in commit 67a7661.

The implementation includes:

• Support for int, float, and int|float target types
• Scientific notation and special float values
• Proper error handling with ParseError
• All 68 parsing tests pass

The Python wrapper converts the C ParseError to Python ParseError for compatibility with existing code.