File
Blob: src/shared/web/syntax.ts
| 1 | // Syntax highlighting helpers for Highlight.js |
| 2 | |
| 3 | /** |
| 4 | * Infer Highlight.js language slug from a filename/path. |
| 5 | * Returns null when unknown. |
| 6 | */ |
| 7 | export function inferHljsLang(fileName: string): string | null { |
| 8 | const name = fileName || ""; |
| 9 | const lower = name.toLowerCase(); |
| 10 | // Special filenames |
| 11 | if (lower === "makefile") return "makefile"; |
| 12 | if (lower === "dockerfile" || lower.endsWith("/dockerfile")) return "dockerfile"; |
| 13 | |
| 14 | const lastDot = lower.lastIndexOf("."); |
| 15 | const ext = lastDot >= 0 ? lower.slice(lastDot + 1) : ""; |
| 16 | |
| 17 | switch (ext) { |
| 18 | case "ts": |
| 19 | case "tsx": |
| 20 | return "typescript"; |
| 21 | case "js": |
| 22 | case "mjs": |
| 23 | case "cjs": |
| 24 | case "jsx": |
| 25 | return "javascript"; |
| 26 | case "json": |
| 27 | return "json"; |
| 28 | case "go": |
| 29 | return "go"; |
| 30 | case "py": |
| 31 | return "python"; |
| 32 | case "rb": |
| 33 | return "ruby"; |
| 34 | case "php": |
| 35 | return "php"; |
| 36 | case "java": |
| 37 | return "java"; |
| 38 | case "c": |
| 39 | return "c"; |
| 40 | case "cc": |
| 41 | case "cpp": |
| 42 | case "cxx": |
| 43 | case "hpp": |
| 44 | return "cpp"; |
| 45 | case "cs": |
| 46 | return "csharp"; |
| 47 | case "rs": |
| 48 | return "rust"; |
| 49 | case "kt": |
| 50 | return "kotlin"; |
| 51 | case "swift": |
| 52 | return "swift"; |
| 53 | case "dart": |
| 54 | return "dart"; |
| 55 | case "sql": |
| 56 | return "sql"; |
| 57 | case "sh": |
| 58 | case "bash": |
| 59 | case "zsh": |
| 60 | return "bash"; // highlight.js alias for sh |
| 61 | case "ps1": |
| 62 | case "psm1": |
| 63 | return "powershell"; |
| 64 | case "less": |
| 65 | return "less"; |
| 66 | case "lua": |
| 67 | return "lua"; |
| 68 | case "yml": |
| 69 | case "yaml": |
| 70 | return "yaml"; |
| 71 | case "toml": |
| 72 | return "toml"; |
| 73 | case "proto": |
| 74 | return "protobuf"; |
| 75 | case "gradle": |
| 76 | return "groovy"; |
| 77 | case "mm": |
| 78 | return "objectivec"; |
| 79 | case "pl": |
| 80 | return "perl"; |
| 81 | case "h": |
| 82 | return "cpp"; |
| 83 | case "ini": |
| 84 | case "cfg": |
| 85 | case "conf": |
| 86 | return "ini"; |
| 87 | case "md": |
| 88 | case "markdown": |
| 89 | return "markdown"; |
| 90 | case "html": |
| 91 | case "htm": |
| 92 | case "xml": |
| 93 | case "svg": |
| 94 | return "xml"; // highlight.js uses xml for HTML/XML/SVG |
| 95 | case "css": |
| 96 | return "css"; |
| 97 | case "scss": |
| 98 | case "sass": |
| 99 | return "scss"; |
| 100 | case "diff": |
| 101 | return "diff"; |
| 102 | default: |
| 103 | return null; |
| 104 | } |
| 105 | } |
| 106 | |
| 107 | /** |
| 108 | * Default small set of languages useful for README/markdown pages. |
| 109 | * Keep this list short to honor "load only when needed" while covering common fences. |
| 110 | */ |
| 111 | export function getMarkdownHighlightLangs(): string[] { |
| 112 | return ["typescript", "javascript", "json", "bash", "yaml", "diff", "go", "python"]; |
| 113 | } |
| 114 | |
| 115 | /** |
| 116 | * Infer language from a shebang line (e.g., #!/usr/bin/env python3) |
| 117 | */ |
| 118 | function inferFromShebang(text: string): string | null { |
| 119 | const firstLine = (text || "").split(/\r?\n/, 1)[0]; |
| 120 | if (!firstLine || !firstLine.startsWith("#!")) return null; |
| 121 | const lower = firstLine.toLowerCase(); |
| 122 | if (lower.includes("python")) return "python"; |
| 123 | if (lower.includes("node") || lower.includes("deno")) return "javascript"; |
| 124 | if (lower.includes("bash") || lower.includes("sh") || lower.includes("zsh")) return "bash"; |
| 125 | if (lower.includes("ruby")) return "ruby"; |
| 126 | if (lower.includes("ts-node")) return "typescript"; |
| 127 | return null; |
| 128 | } |
| 129 | |
| 130 | /** |
| 131 | * Smarter inference: use file extension first, then shebang fallback. |
| 132 | */ |
| 133 | export function inferHljsLangSmart(fileName: string, text?: string): string | null { |
| 134 | const byExt = inferHljsLang(fileName); |
| 135 | if (byExt) return byExt; |
| 136 | if (text) { |
| 137 | const byShebang = inferFromShebang(text); |
| 138 | if (byShebang) return byShebang; |
| 139 | } |
| 140 | return null; |
| 141 | } |
| 142 | |
| 143 | /** |
| 144 | * Languages to load for a blob using smart inference. |
| 145 | */ |
| 146 | export function getHighlightLangsForBlobSmart(fileName: string, text?: string): string[] { |
| 147 | const lang = inferHljsLangSmart(fileName, text); |
| 148 | const normalized = normalizeLangForCdn(lang); |
| 149 | return normalized ? [normalized] : []; |
| 150 | } |
| 151 | |
| 152 | // Supported language slugs on our pinned CDN build. Keep this conservative. |
| 153 | const SUPPORTED_LANGS = new Set<string>([ |
| 154 | "typescript", |
| 155 | "javascript", |
| 156 | "json", |
| 157 | "bash", |
| 158 | "yaml", |
| 159 | "diff", |
| 160 | "go", |
| 161 | "python", |
| 162 | "ini", |
| 163 | "xml", |
| 164 | "css", |
| 165 | "scss", |
| 166 | "less", |
| 167 | "java", |
| 168 | "c", |
| 169 | "cpp", |
| 170 | "csharp", |
| 171 | "rust", |
| 172 | "kotlin", |
| 173 | "swift", |
| 174 | "php", |
| 175 | "ruby", |
| 176 | "sql", |
| 177 | "powershell", |
| 178 | "dockerfile", |
| 179 | "makefile", |
| 180 | "nginx", |
| 181 | "objectivec", |
| 182 | "perl", |
| 183 | "protobuf", |
| 184 | "groovy", |
| 185 | "lua", |
| 186 | ]); |
| 187 | |
| 188 | /** |
| 189 | * Normalize an inferred language slug to a CDN-available one. |
| 190 | * - Map missing or alias languages to closest available grammar (e.g., toml -> ini) |
| 191 | */ |
| 192 | export function normalizeLangForCdn(lang: string | null): string | null { |
| 193 | if (!lang) return null; |
| 194 | let slug = lang; |
| 195 | // Aliases or fallbacks |
| 196 | if (slug === "toml") slug = "ini"; // close enough for highlighting |
| 197 | // Ensure it's in our allowlist |
| 198 | return SUPPORTED_LANGS.has(slug) ? slug : null; |
| 199 | } |