Move _skip and _parse_number functions to C extension module

pyproject.toml

@@ -1,6 +1,6 @@
 [build-system]
-requires = ["uv_build>=0.9,<0.10"]
-build-backend = "uv_build"
+requires = ["setuptools>=61", "wheel"]
+build-backend = "setuptools.build_meta"
 
 [project]
 name = "foamlib"
@@ -119,3 +119,6 @@ extend-ignore = [
 
 [tool.ruff.lint.pydocstyle]
 convention = "pep257"
+
+[tool.setuptools.packages.find]
+where = ["src"]

setup.py

@@ -0,0 +1,13 @@
+"""Setup script for building C extension modules."""
+
+from setuptools import Extension, setup
+
+# Define the C extension module
+skip_ext = Extension(
+    "foamlib._files._parsing._skip_ext",
+    sources=["src/foamlib/_files/_parsing/_skip_ext.c"],
+)
+
+setup(
+    ext_modules=[skip_ext],
+)

src/foamlib/_files/_parsing/_parser.py

@@ -29,6 +29,9 @@
     SubDict,
     Tensor,
 )
+from ._skip_ext import ParseError as _CParseError
+from ._skip_ext import _parse_number as _parse_number_c
+from ._skip_ext import _skip
 from .exceptions import FoamFileDecodeError
 
 _DType = TypeVar("_DType", float, int)
@@ -46,21 +49,6 @@
 for c in b"0123456789._<>#$:+-*/|^%&=!":
     _IS_TOKEN_CONTINUATION[c] = True
 
-_IS_WHITESPACE = [False] * 256
-for c in b" \n\t\r\f\v":
-    _IS_WHITESPACE[c] = True
-
-_IS_WHITESPACE_NO_NEWLINE = _IS_WHITESPACE.copy()
-_IS_WHITESPACE_NO_NEWLINE[ord(b"\n")] = False
-
-_IS_POSSIBLE_FLOAT = [False] * 256
-for c in b"0123456789.-+eEinfnatyINFNATY":
-    _IS_POSSIBLE_FLOAT[c] = True
-
-_IS_POSSIBLE_INTEGER = [False] * 256
-for c in b"0123456789-+":
-    _IS_POSSIBLE_INTEGER[c] = True
-
 _COMMENTS = re.compile(rb"(?:(?:/\*(?:[^*]|\*(?!/))*\*/)|(?://(?:\\\n|[^\n])*))+")
 _SKIP = re.compile(rb"(?:\s+|" + _COMMENTS.pattern + rb")+")
 _POSSIBLE_FLOAT = re.compile(rb"[0-9.\-+einfatyEINFATY]+", re.ASCII)
@@ -78,52 +66,6 @@ def make_fatal(self) -> FoamFileDecodeError:
         return FoamFileDecodeError(self._contents, self.pos, expected=self._expected)
 
 
-def _skip(
-    contents: bytes | bytearray,
-    pos: int,
-    *,
-    newline_ok: bool = True,
-) -> int:
-    is_whitespace = _IS_WHITESPACE if newline_ok else _IS_WHITESPACE_NO_NEWLINE
-
-    with contextlib.suppress(IndexError):
-        while True:
-            while is_whitespace[contents[pos]]:
-                pos += 1
-
-            next1 = contents[pos]
-            next2 = contents[pos + 1]
-
-            if next1 == ord("/") and next2 == ord("/"):
-                pos += 2
-                while True:
-                    if contents[pos] == ord("\n"):
-                        if newline_ok:
-                            pos += 1
-                        break
-                    if contents[pos] == ord("\\"):
-                        with contextlib.suppress(IndexError):
-                            if contents[pos + 1] == ord("\n"):
-                                pos += 1
-                        pos += 1
-                        continue
-                    pos += 1
-                continue
-
-            if next1 == ord("/") and next2 == ord("*"):
-                if (pos := contents.find(b"*/", pos + 2)) == -1:
-                    raise FoamFileDecodeError(
-                        contents,
-                        len(contents),
-                        expected="*/",
-                    )
-                pos += 2
-                continue
-            break
-
-    return pos
-
-
 def _expect(contents: bytes | bytearray, pos: int, expected: bytes | bytearray) -> int:
     length = len(expected)
     if contents[pos : pos + length] != expected:
@@ -216,31 +158,13 @@ def _parse_number(
     *,
     target: type[int] | type[float] | type[int | float] = int | float,
 ) -> tuple[int | float, int]:
-    is_numeric = _IS_POSSIBLE_INTEGER if target is int else _IS_POSSIBLE_FLOAT
-    end = pos
-    with contextlib.suppress(IndexError):
-        while is_numeric[contents[end]]:
-            end += 1
-
-        if _IS_TOKEN_CONTINUATION[contents[end]]:
-            raise ParseError(contents, pos, expected="number")
-
-    if pos == end:
-        raise ParseError(contents, pos, expected="number")
-
-    chars = contents[pos:end]
-    if target is not float:
-        try:
-            return int(chars), end
-        except ValueError as e:
-            if target is int:
-                raise ParseError(contents, pos, expected="integer") from e
+    """Parse a number from contents and convert C ParseError to Python ParseError."""
     try:
-        return float(chars), end
-    except ValueError as e:
-        if target is float:
-            raise ParseError(contents, pos, expected="float") from e
-        raise ParseError(contents, pos, expected="number") from e
+        return _parse_number_c(contents, pos, target=target)
+    except _CParseError as e:
+        # Convert C ParseError to Python ParseError
+        # The C exception has attributes _contents, pos, and _expected
+        raise ParseError(e._contents, e.pos, expected=e._expected) from None
 
 
 class _ASCIINumericListParser(Generic[_DType, _ElShape]):