Add Rust module with _skip function implementation

Co-authored-by: gerlero <[email protected]>

Author: Copilot

Cargo.lock

@@ -0,0 +1,12 @@
+[package]
+name = "foamlib"
+version = "1.5.4"
+edition = "2021"
+
+[lib]
+name = "foamlib_rust"
+crate-type = ["cdylib"]
+path = "rust_src/lib.rs"
+
+[dependencies]
+pyo3 = { version = "0.23", features = ["extension-module"] }

Cargo.toml

@@ -1,6 +1,6 @@
 [build-system]
-requires = ["uv_build>=0.9,<0.10"]
-build-backend = "uv_build"
+requires = ["maturin>=1.0,<2.0"]
+build-backend = "maturin"
 
 [project]
 name = "foamlib"
@@ -119,3 +119,7 @@ extend-ignore = [
 
 [tool.ruff.lint.pydocstyle]
 convention = "pep257"
+
+[tool.maturin]
+python-source = "src"
+module-name = "foamlib.foamlib_rust"

pyproject.toml

@@ -0,0 +1,135 @@
+use pyo3::prelude::*;
+
+// Whitespace lookup table
+static IS_WHITESPACE: [bool; 256] = {
+    let mut table = [false; 256];
+    table[b' ' as usize] = true;
+    table[b'\n' as usize] = true;
+    table[b'\t' as usize] = true;
+    table[b'\r' as usize] = true;
+    table[b'\x0C' as usize] = true; // \f
+    table[b'\x0B' as usize] = true; // \v
+    table
+};
+
+// Whitespace without newline lookup table
+static IS_WHITESPACE_NO_NEWLINE: [bool; 256] = {
+    let mut table = [false; 256];
+    table[b' ' as usize] = true;
+    table[b'\t' as usize] = true;
+    table[b'\r' as usize] = true;
+    table[b'\x0C' as usize] = true; // \f
+    table[b'\x0B' as usize] = true; // \v
+    table
+};
+
+/// Skip whitespace and comments in OpenFOAM files
+/// 
+/// This function skips over whitespace and comments (both // and /* */ style).
+/// It handles line continuations with backslash in line comments.
+/// 
+/// # Arguments
+/// * `contents` - The file contents as bytes or bytearray
+/// * `pos` - Current position in the file
+/// * `newline_ok` - Whether newlines should be skipped (default: True)
+/// 
+/// # Returns
+/// The new position after skipping whitespace and comments
+#[pyfunction]
+#[pyo3(signature = (contents, pos, *, newline_ok=true))]
+fn skip(contents: &Bound<'_, pyo3::types::PyAny>, mut pos: usize, newline_ok: bool) -> PyResult<usize> {
+    // Extract bytes from either bytes or bytearray
+    let contents = if let Ok(bytes) = contents.downcast::<pyo3::types::PyBytes>() {
+        bytes.as_bytes()
+    } else if let Ok(bytearray) = contents.downcast::<pyo3::types::PyByteArray>() {
+        unsafe { bytearray.as_bytes() }
+    } else {
+        return Err(PyErr::new::<pyo3::exceptions::PyTypeError, _>(
+            "contents must be bytes or bytearray"
+        ));
+    };
+    let is_whitespace = if newline_ok {
+        &IS_WHITESPACE
+    } else {
+        &IS_WHITESPACE_NO_NEWLINE
+    };
+    
+    loop {
+        // Skip whitespace
+        while pos < contents.len() && is_whitespace[contents[pos] as usize] {
+            pos += 1;
+        }
+        
+        // Check if we're at the end of content
+        if pos >= contents.len() {
+            break;
+        }
+        
+        // Check for comments
+        if pos + 1 < contents.len() {
+            let next1 = contents[pos];
+            let next2 = contents[pos + 1];
+            
+            // Single-line comment //
+            if next1 == b'/' && next2 == b'/' {
+                pos += 2;
+                loop {
+                    if pos >= contents.len() {
+                        break;
+                    }
+                    
+                    if contents[pos] == b'\n' {
+                        if newline_ok {
+                            pos += 1;
+                        }
+                        break;
+                    }
+                    
+                    // Handle line continuation
+                    if contents[pos] == b'\\' && pos + 1 < contents.len() && contents[pos + 1] == b'\n' {
+                        pos += 1;
+                    }
+                    
+                    pos += 1;
+                }
+                continue;
+            }
+            
+            // Multi-line comment /* */
+            if next1 == b'/' && next2 == b'*' {
+                pos += 2;
+                
+                // Find the closing */
+                let mut found = false;
+                while pos + 1 < contents.len() {
+                    if contents[pos] == b'*' && contents[pos + 1] == b'/' {
+                        pos += 2;
+                        found = true;
+                        break;
+                    }
+                    pos += 1;
+                }
+                
+                if !found {
+                    return Err(PyErr::new::<pyo3::exceptions::PyValueError, _>(
+                        format!("Unterminated comment at position {}", pos)
+                    ));
+                }
+                
+                continue;
+            }
+        }
+        
+        // No more whitespace or comments
+        break;
+    }
+    
+    Ok(pos)
+}
+
+/// A Python module implemented in Rust for performance-critical parsing operations.
+#[pymodule]
+fn foamlib_rust(m: &Bound<'_, PyModule>) -> PyResult<()> {
+    m.add_function(wrap_pyfunction!(skip, m)?)?;
+    Ok(())
+}

rust_src/lib.rs

@@ -13,6 +13,13 @@
     from typing_extensions import Unpack, assert_never
 
 import numpy as np
+
+# Import the Rust implementation of _skip
+try:
+    from foamlib.foamlib_rust import skip as _skip_rust
+    _USE_RUST_SKIP = True
+except ImportError:
+    _USE_RUST_SKIP = False
 from multicollections import MultiDict
 
 from ...typing import (
@@ -84,6 +91,19 @@ def _skip(
     *,
     newline_ok: bool = True,
 ) -> int:
+    # Use Rust implementation if available
+    if _USE_RUST_SKIP:
+        try:
+            return _skip_rust(contents, pos, newline_ok=newline_ok)
+        except ValueError as e:
+            # Convert ValueError from Rust to FoamFileDecodeError
+            raise FoamFileDecodeError(
+                contents,
+                len(contents),
+                expected="*/",
+            ) from e
+    
+    # Fallback to Python implementation
     is_whitespace = _IS_WHITESPACE if newline_ok else _IS_WHITESPACE_NO_NEWLINE
 
     with contextlib.suppress(IndexError):