Skip to main content

provenant/utils/
language.rs

1use content_inspector::{ContentType, inspect};
2use std::path::Path;
3
4fn is_utf8_text(content_type: ContentType) -> bool {
5    content_type == ContentType::UTF_8 || content_type == ContentType::UTF_8_BOM
6}
7
8pub fn detect_language(path: &Path, content: &[u8]) -> String {
9    if content.len() > 32 && !is_utf8_text(inspect(content)) {
10        return "Binary".to_string();
11    }
12
13    // Check for shebang in script files
14    if content.len() > 2 && content[0] == b'#' && content[1] == b'!' {
15        let shebang_end = content
16            .iter()
17            .position(|&b| b == b'\n')
18            .unwrap_or(content.len());
19        let shebang = String::from_utf8_lossy(&content[0..shebang_end]);
20
21        if shebang.contains("python") {
22            return "Python".to_string();
23        } else if shebang.contains("node") {
24            return "JavaScript".to_string();
25        } else if shebang.contains("ruby") {
26            return "Ruby".to_string();
27        } else if shebang.contains("perl") {
28            return "Perl".to_string();
29        } else if shebang.contains("php") {
30            return "PHP".to_string();
31        } else if shebang.contains("bash") || shebang.contains("sh") {
32            return "Shell".to_string();
33        }
34    }
35
36    // Check file extension
37    if let Some(extension) = path.extension().and_then(|e| e.to_str()) {
38        match extension.to_lowercase().as_str() {
39            "rs" => return "Rust".to_string(),
40            "py" => return "Python".to_string(),
41            "js" => return "JavaScript".to_string(),
42            "ts" => return "TypeScript".to_string(),
43            "html" | "htm" => return "HTML".to_string(),
44            "css" => return "CSS".to_string(),
45            "c" => return "C".to_string(),
46            "cpp" | "cc" | "cxx" => return "C++".to_string(),
47            "h" => return "C".to_string(),
48            "hpp" => return "C++".to_string(),
49            "s" => return "GAS".to_string(),
50            "java" => return "Java".to_string(),
51            "go" => return "Go".to_string(),
52            "rb" => return "Ruby".to_string(),
53            "php" => return "PHP".to_string(),
54            "pl" => return "Perl".to_string(),
55            "swift" => return "Swift".to_string(),
56            "md" | "markdown" => return "Markdown".to_string(),
57            "json" => return "JSON".to_string(),
58            "xml" => return "XML".to_string(),
59            "yml" | "yaml" => return "YAML".to_string(),
60            "sql" => return "SQL".to_string(),
61            "sh" | "bash" => return "Shell".to_string(),
62            "kt" | "kts" => return "Kotlin".to_string(),
63            "dart" => return "Dart".to_string(),
64            "scala" => return "Scala".to_string(),
65            "cs" => return "C#".to_string(),
66            "fs" => return "F#".to_string(),
67            "r" => return "R".to_string(),
68            "lua" => return "Lua".to_string(),
69            "jl" => return "Julia".to_string(),
70            "ex" | "exs" => return "Elixir".to_string(),
71            "clj" => return "Clojure".to_string(),
72            "hs" => return "Haskell".to_string(),
73            "erl" => return "Erlang".to_string(),
74            "sc" => return "SuperCollider".to_string(),
75            "tex" => return "TeX".to_string(),
76            _ => {}
77        }
78    }
79
80    // Check file name for special cases
81    let file_name = path
82        .file_name()
83        .and_then(|n| n.to_str())
84        .map(|s| s.to_lowercase())
85        .unwrap_or_default();
86
87    if matches!(
88        file_name.as_str(),
89        "dockerfile" | "containerfile" | "containerfile.core"
90    ) {
91        return "Dockerfile".to_string();
92    } else if file_name == "makefile" {
93        return "Makefile".to_string();
94    } else if file_name == "gemfile" || file_name == "rakefile" {
95        return "Ruby".to_string();
96    }
97
98    if is_utf8_text(inspect(content)) {
99        let text_sample = String::from_utf8_lossy(&content[..std::cmp::min(content.len(), 1000)]);
100
101        if text_sample.contains("<?php") {
102            return "PHP".to_string();
103        } else if text_sample.contains("<html") || text_sample.contains("<!DOCTYPE html") {
104            return "HTML".to_string();
105        } else if text_sample.contains("import React") || text_sample.contains("import {") {
106            return "JavaScript/TypeScript".to_string();
107        } else if text_sample.contains("def ") && text_sample.contains(":") {
108            return "Python".to_string();
109        } else if text_sample.contains("package ")
110            && text_sample.contains("import ")
111            && text_sample.contains("{")
112        {
113            return "Go".to_string();
114        }
115
116        return "Text".to_string();
117    }
118
119    "Unknown".to_string()
120}
121
122#[cfg(test)]
123mod tests {
124    use super::detect_language;
125    use std::path::Path;
126
127    #[test]
128    fn detect_language_supports_containerfile_names() {
129        assert_eq!(
130            detect_language(Path::new("Containerfile"), b"FROM scratch\n"),
131            "Dockerfile"
132        );
133        assert_eq!(
134            detect_language(Path::new("containerfile.core"), b"FROM scratch\n"),
135            "Dockerfile"
136        );
137    }
138
139    #[test]
140    fn detect_language_maps_c_headers_to_c() {
141        assert_eq!(detect_language(Path::new("zlib.h"), b"/* header */\n"), "C");
142    }
143
144    #[test]
145    fn detect_language_maps_uppercase_s_to_gas() {
146        assert_eq!(detect_language(Path::new("gvmat64.S"), b"; asm\n"), "GAS");
147    }
148}