| 1 | //! Language detection for repositories by file-type counting. |
| 2 | |
| 3 | use std::{ |
| 4 | collections::BTreeMap, |
| 5 | path::Path, |
| 6 | }; |
| 7 | |
| 8 | use crate::error::Result; |
| 9 | |
| 10 | /// Language percentages for a repository. |
| 11 | #[derive(Clone, Debug)] |
| 12 | pub struct LanguageDetection { |
| 13 | /// Primary language by count, or empty if no files. |
| 14 | pub primary: String, |
| 15 | /// JSON array: [{"lang": "Rust", "percent": 75.5}, ...] |
| 16 | pub json: String, |
| 17 | } |
| 18 | |
| 19 | /// Map of file extensions to language names. |
| 20 | fn ext_to_language(ext: &str) -> Option<&'static str> { |
| 21 | match ext { |
| 22 | "rs" => Some("Rust"), |
| 23 | "go" => Some("Go"), |
| 24 | "py" => Some("Python"), |
| 25 | "js" | "jsx" => Some("JavaScript"), |
| 26 | "ts" | "tsx" => Some("TypeScript"), |
| 27 | "java" => Some("Java"), |
| 28 | "cpp" | "cc" | "cxx" | "c++" => Some("C++"), |
| 29 | "c" => Some("C"), |
| 30 | "h" | "hpp" => Some("C/C++"), |
| 31 | "cs" => Some("C#"), |
| 32 | "rb" => Some("Ruby"), |
| 33 | "php" => Some("PHP"), |
| 34 | "swift" => Some("Swift"), |
| 35 | "kt" => Some("Kotlin"), |
| 36 | "sh" | "bash" => Some("Shell"), |
| 37 | "yml" | "yaml" => Some("YAML"), |
| 38 | "json" => Some("JSON"), |
| 39 | "toml" => Some("TOML"), |
| 40 | "md" => Some("Markdown"), |
| 41 | "html" | "htm" => Some("HTML"), |
| 42 | "css" | "scss" | "sass" => Some("CSS"), |
| 43 | "sql" => Some("SQL"), |
| 44 | "xml" => Some("XML"), |
| 45 | _ => None, |
| 46 | } |
| 47 | } |
| 48 | |
| 49 | /// Detect languages in a repository by walking the file tree and counting by extension. |
| 50 | /// Returns primary language + JSON-encoded percentages. |
| 51 | pub async fn detect(repo_path: &Path) -> Result<LanguageDetection> { |
| 52 | let mut counts: BTreeMap<String, u64> = BTreeMap::new(); |
| 53 | let mut total = 0u64; |
| 54 | |
| 55 | // Walk the repository tree, skipping .git. |
| 56 | if let Ok(entries) = std::fs::read_dir(repo_path) { |
| 57 | walk_dir(&mut counts, &mut total, entries); |
| 58 | } |
| 59 | |
| 60 | let primary = if total == 0 { |
| 61 | String::new() |
| 62 | } else { |
| 63 | let mut max_lang = String::new(); |
| 64 | let mut max_count = 0u64; |
| 65 | for (lang, count) in &counts { |
| 66 | if *count > max_count { |
| 67 | max_count = *count; |
| 68 | max_lang = lang.clone(); |
| 69 | } |
| 70 | } |
| 71 | max_lang |
| 72 | }; |
| 73 | |
| 74 | let json = if counts.is_empty() { |
| 75 | "[]".to_string() |
| 76 | } else { |
| 77 | let mut entries: Vec<_> = counts.iter().collect(); |
| 78 | entries.sort_by(|a, b| b.1.cmp(a.1)); |
| 79 | |
| 80 | let percentages: Vec<String> = entries |
| 81 | .iter() |
| 82 | .map(|(lang, count)| { |
| 83 | let percent = (**count as f64 / total as f64) * 100.0; |
| 84 | format!(r#"{{"lang":"{}","percent":{:.1}}}"#, lang, percent) |
| 85 | }) |
| 86 | .collect(); |
| 87 | format!("[{}]", percentages.join(",")) |
| 88 | }; |
| 89 | |
| 90 | Ok(LanguageDetection { primary, json }) |
| 91 | } |
| 92 | |
| 93 | fn walk_dir(counts: &mut BTreeMap<String, u64>, total: &mut u64, entries: std::fs::ReadDir) { |
| 94 | for entry in entries.flatten() { |
| 95 | let Ok(metadata) = entry.metadata() else { |
| 96 | continue; |
| 97 | }; |
| 98 | |
| 99 | let path = entry.path(); |
| 100 | let file_name = entry.file_name(); |
| 101 | let name = file_name.to_string_lossy(); |
| 102 | |
| 103 | // Skip .git and hidden directories. |
| 104 | if name.starts_with('.') { |
| 105 | continue; |
| 106 | } |
| 107 | |
| 108 | if metadata.is_dir() { |
| 109 | if let Ok(subentries) = std::fs::read_dir(&path) { |
| 110 | walk_dir(counts, total, subentries); |
| 111 | } |
| 112 | } else if metadata.is_file() |
| 113 | && let Some(ext) = path.extension().and_then(|s| s.to_str()) |
| 114 | && let Some(lang) = ext_to_language(ext) |
| 115 | { |
| 116 | *counts.entry(lang.to_string()).or_insert(0) += 1; |
| 117 | *total += 1; |
| 118 | } |
| 119 | } |
| 120 | } |