use std::path::Path; pub fn detect_language(filename: &Path, contents: &str) -> &'static str { detect_from_extension(filename) .or_else(|| detect_from_shebang(contents)) .unwrap_or("") } fn detect_from_extension(filename: &Path) -> Option<&'static str> { let ext = filename .extension() .and_then(|e| e.to_str()) .map(|e| e.to_ascii_lowercase()); let ext_str = ext.as_deref(); match ext_str { Some("rs") => Some("rust"), Some("py") => Some("python"), Some("json") => Some("json"), Some("toml") => Some("toml"), Some("yaml") | Some("yml") => Some("yaml"), Some("js") | Some("jsx") => Some("javascript"), Some("ts") | Some("tsx") => Some("typescript"), _ => None, } } fn detect_from_shebang(contents: &str) -> Option<&'static str> { if let Some(first_line) = contents.lines().next() && first_line.starts_with("#!") { if first_line.contains("python") { return Some("python"); } if first_line.contains("bash") || first_line.contains("/sh") || first_line.contains(" sh") { return Some("bash"); } } None } #[cfg(test)] mod tests { use std::path::Path; use super::detect_language; #[test] fn non_shebang_line_does_not_trigger_detection() { let contents = "this mentions bash but is not a shebang"; assert_eq!(detect_language(Path::new("file"), contents), ""); } #[test] fn detects_python_from_shebang() { let contents = "#!/usr/bin/python3\nprint('hi')"; assert_eq!(detect_language(Path::new("file"), contents), "python"); } #[test] fn sh_shebang_maps_to_bash_for_llm() { let contents = "#!/bin/sh\necho hi"; assert_eq!(detect_language(Path::new("file"), contents), "bash"); } #[test] fn env_shebang_is_detected() { let contents = "#!/usr/bin/env sh\necho hi"; assert_eq!(detect_language(Path::new("file"), contents), "bash"); } #[test] fn extension_overrides_shebang() { let contents = "#!/bin/bash\nprint('hello')"; assert_eq!(detect_language(Path::new("main.py"), contents), "python"); } #[test] fn javascript_family_extensions_map_correctly() { let cases = [ ("file.js", "javascript"), ("file.jsx", "javascript"), ("file.ts", "typescript"), ("file.tsx", "typescript"), ]; for (filename, expected) in cases { assert_eq!(detect_language(Path::new(filename), ""), expected); } } #[test] fn json_derived_extensions_map_correctly() { let cases = [ ("file.json", "json"), ("file.yml", "yaml"), ("file.yaml", "yaml"), ]; for (filename, expected) in cases { assert_eq!(detect_language(Path::new(filename), ""), expected); } } #[test] fn toml_extension_is_detected() { assert_eq!(detect_language(Path::new("Cargo.toml"), ""), "toml"); } }