jscpd-rs 0.1.0

package/src/files.rs

@@ -0,0 +1,25 @@
+mod discovery;
+mod gitignore;
+mod paths;
+mod shebang;
+
+#[cfg(test)]
+mod tests;
+
+pub use discovery::discover;
+pub(crate) use gitignore::collect_cwd_gitignore_patterns;
+
+/// Source content prepared by the caller for in-memory detection.
+///
+/// Use this with `detect_source_files` when an integration already owns file
+/// contents or wants to check generated snippets without reading from disk.
+#[derive(Clone, Debug)]
+pub struct SourceFile {
+    /// Stable source identifier used in clone reports, usually a file path.
+    pub source_id: String,
+    /// Source format name, for example `javascript`, `typescript`, `rust`, or
+    /// another value returned by `get_supported_formats`.
+    pub format: String,
+    /// Full source text.
+    pub content: String,
+}

package/src/formats.rs

@@ -0,0 +1,570 @@
+use std::path::Path;
+
+use crate::cli::FormatMappings;
+
+// Generated by scripts/sync-formats.mjs from jscpd/packages/tokenizer/dist/index.mjs.
+// Upstream getFormatByFile() uses first-writer-wins extension mappings and no
+// built-in filename mappings; extensionless names are supported through
+// --formats-names only.
+
+pub const SUPPORTED_FORMATS: &[&str] = &[
+    "abap",
+    "actionscript",
+    "ada",
+    "apacheconf",
+    "apl",
+    "applescript",
+    "arduino",
+    "arff",
+    "asciidoc",
+    "asm6502",
+    "aspnet",
+    "autohotkey",
+    "autoit",
+    "bash",
+    "basic",
+    "batch",
+    "bison",
+    "brainfuck",
+    "bro",
+    "c",
+    "c-header",
+    "clike",
+    "clojure",
+    "coffeescript",
+    "comments",
+    "cpp",
+    "cpp-header",
+    "crystal",
+    "csharp",
+    "csp",
+    "css-extras",
+    "css",
+    "d",
+    "dart",
+    "diff",
+    "django",
+    "docker",
+    "eiffel",
+    "elixir",
+    "elm",
+    "erb",
+    "erlang",
+    "flow",
+    "fortran",
+    "fsharp",
+    "gdscript",
+    "gedcom",
+    "gherkin",
+    "git",
+    "glsl",
+    "go",
+    "graphql",
+    "groovy",
+    "haml",
+    "handlebars",
+    "haskell",
+    "haxe",
+    "hpkp",
+    "hsts",
+    "http",
+    "ichigojam",
+    "icon",
+    "inform7",
+    "ini",
+    "io",
+    "j",
+    "java",
+    "javascript",
+    "jolie",
+    "json",
+    "jsx",
+    "julia",
+    "keymap",
+    "kotlin",
+    "latex",
+    "less",
+    "liquid",
+    "lisp",
+    "livescript",
+    "lolcode",
+    "lua",
+    "makefile",
+    "markdown",
+    "markup",
+    "matlab",
+    "mel",
+    "mizar",
+    "monkey",
+    "n4js",
+    "nasm",
+    "nginx",
+    "nim",
+    "nix",
+    "nsis",
+    "objectivec",
+    "ocaml",
+    "opencl",
+    "oz",
+    "parigp",
+    "pascal",
+    "perl",
+    "php",
+    "plsql",
+    "powershell",
+    "processing",
+    "prolog",
+    "properties",
+    "protobuf",
+    "pug",
+    "puppet",
+    "pure",
+    "python",
+    "q",
+    "qore",
+    "r",
+    "reason",
+    "renpy",
+    "rest",
+    "rip",
+    "roboconf",
+    "ruby",
+    "rust",
+    "sas",
+    "sass",
+    "scala",
+    "scheme",
+    "scss",
+    "svelte",
+    "smalltalk",
+    "smarty",
+    "soy",
+    "sql",
+    "stylus",
+    "swift",
+    "tap",
+    "tcl",
+    "textile",
+    "tsx",
+    "tt2",
+    "twig",
+    "typescript",
+    "txt",
+    "vbnet",
+    "velocity",
+    "verilog",
+    "vhdl",
+    "vim",
+    "visual-basic",
+    "astro",
+    "vue",
+    "wasm",
+    "wiki",
+    "xeora",
+    "xojo",
+    "xquery",
+    "yaml",
+    "abnf",
+    "agda",
+    "antlr4",
+    "apex",
+    "aql",
+    "armasm",
+    "awk",
+    "bicep",
+    "bnf",
+    "cfscript",
+    "cfml",
+    "cmake",
+    "cobol",
+    "csv",
+    "cypher",
+    "dhall",
+    "dns-zone-file",
+    "dot",
+    "ebnf",
+    "editorconfig",
+    "excel-formula",
+    "factor",
+    "ftl",
+    "gcode",
+    "gettext",
+    "gml",
+    "go-module",
+    "hcl",
+    "hlsl",
+    "idris",
+    "ignore",
+    "jq",
+    "json5",
+    "kusto",
+    "lilypond",
+    "linker-script",
+    "llvm",
+    "log",
+    "mermaid",
+    "mongodb",
+    "n1ql",
+    "odin",
+    "openqasm",
+    "plant-uml",
+    "powerquery",
+    "promql",
+    "purescript",
+    "qsharp",
+    "racket",
+    "regex",
+    "rego",
+    "rescript",
+    "robotframework",
+    "shell-session",
+    "smali",
+    "solidity",
+    "sparql",
+    "stata",
+    "toml",
+    "turtle",
+    "typoscript",
+    "unrealscript",
+    "uri",
+    "vala",
+    "wgsl",
+    "wolfram",
+    "zig",
+];
+
+const EXTENSION_FORMATS: &[(&str, &str)] = &[
+    ("as", "actionscript"),
+    ("ada", "ada"),
+    ("apl", "apl"),
+    ("asp", "aspnet"),
+    ("aspx", "aspnet"),
+    ("sh", "bash"),
+    ("ksh", "bash"),
+    ("bash", "bash"),
+    ("bas", "basic"),
+    ("b", "brainfuck"),
+    ("bf", "brainfuck"),
+    ("c", "c"),
+    ("z80", "c"),
+    ("h", "c-header"),
+    ("cljs", "clojure"),
+    ("clj", "clojure"),
+    ("cljc", "clojure"),
+    ("cljx", "clojure"),
+    ("edn", "clojure"),
+    ("coffee", "coffeescript"),
+    ("cpp", "cpp"),
+    ("c++", "cpp"),
+    ("cc", "cpp"),
+    ("cxx", "cpp"),
+    ("hpp", "cpp-header"),
+    ("h++", "cpp-header"),
+    ("hh", "cpp-header"),
+    ("hxx", "cpp-header"),
+    ("cr", "crystal"),
+    ("cs", "csharp"),
+    ("css", "css"),
+    ("gss", "css"),
+    ("d", "d"),
+    ("dart", "dart"),
+    ("diff", "diff"),
+    ("patch", "diff"),
+    ("e", "eiffel"),
+    ("elm", "elm"),
+    ("erl", "erlang"),
+    ("erlang", "erlang"),
+    ("f", "fortran"),
+    ("for", "fortran"),
+    ("f77", "fortran"),
+    ("f90", "fortran"),
+    ("fs", "fsharp"),
+    ("gd", "gdscript"),
+    ("feature", "gherkin"),
+    ("go", "go"),
+    ("graphql", "graphql"),
+    ("groovy", "groovy"),
+    ("gradle", "groovy"),
+    ("haml", "haml"),
+    ("hb", "handlebars"),
+    ("hbs", "handlebars"),
+    ("handlebars", "handlebars"),
+    ("hs", "haskell"),
+    ("lhs", "haskell"),
+    ("hx", "haxe"),
+    ("hxml", "haxe"),
+    ("ini", "ini"),
+    ("java", "java"),
+    ("js", "javascript"),
+    ("es", "javascript"),
+    ("es6", "javascript"),
+    ("mjs", "javascript"),
+    ("cjs", "javascript"),
+    ("json", "json"),
+    ("map", "json"),
+    ("jsonld", "json"),
+    ("jsx", "jsx"),
+    ("jl", "julia"),
+    ("kt", "kotlin"),
+    ("kts", "kotlin"),
+    ("tex", "latex"),
+    ("less", "less"),
+    ("cl", "lisp"),
+    ("lisp", "lisp"),
+    ("el", "lisp"),
+    ("ls", "livescript"),
+    ("lua", "lua"),
+    ("md", "markdown"),
+    ("markdown", "markdown"),
+    ("mkd", "markdown"),
+    ("html", "markup"),
+    ("htm", "markup"),
+    ("xml", "markup"),
+    ("xsl", "markup"),
+    ("xslt", "markup"),
+    ("svg", "markup"),
+    ("ejs", "markup"),
+    ("jsp", "markup"),
+    ("nsh", "nsis"),
+    ("nsi", "nsis"),
+    ("m", "objectivec"),
+    ("mm", "objectivec"),
+    ("ocaml", "ocaml"),
+    ("ml", "ocaml"),
+    ("mli", "ocaml"),
+    ("mll", "ocaml"),
+    ("mly", "ocaml"),
+    ("oz", "oz"),
+    ("pas", "pascal"),
+    ("p", "pascal"),
+    ("pl", "perl"),
+    ("pm", "perl"),
+    ("php", "php"),
+    ("phtml", "php"),
+    ("plsql", "plsql"),
+    ("ps1", "powershell"),
+    ("psd1", "powershell"),
+    ("psm1", "powershell"),
+    ("pro", "prolog"),
+    ("properties", "properties"),
+    ("proto", "protobuf"),
+    ("pug", "pug"),
+    ("jade", "pug"),
+    ("pp", "puppet"),
+    ("puppet", "puppet"),
+    ("py", "python"),
+    ("pyx", "python"),
+    ("pxd", "python"),
+    ("pxi", "python"),
+    ("q", "q"),
+    ("r", "r"),
+    ("R", "r"),
+    ("rb", "ruby"),
+    ("rs", "rust"),
+    ("sas", "sas"),
+    ("sass", "sass"),
+    ("scala", "scala"),
+    ("scm", "scheme"),
+    ("ss", "scheme"),
+    ("scss", "scss"),
+    ("svelte", "svelte"),
+    ("st", "smalltalk"),
+    ("smarty", "smarty"),
+    ("tpl", "smarty"),
+    ("soy", "soy"),
+    ("sql", "sql"),
+    ("cql", "sql"),
+    ("styl", "stylus"),
+    ("stylus", "stylus"),
+    ("swift", "swift"),
+    ("tap", "tap"),
+    ("tcl", "tcl"),
+    ("textile", "textile"),
+    ("tsx", "tsx"),
+    ("tt2", "tt2"),
+    ("twig", "twig"),
+    ("ts", "typescript"),
+    ("mts", "typescript"),
+    ("cts", "typescript"),
+    ("txt", "txt"),
+    ("vb", "vbnet"),
+    ("vtl", "velocity"),
+    ("v", "verilog"),
+    ("vhd", "vhdl"),
+    ("vhdl", "vhdl"),
+    ("astro", "astro"),
+    ("vue", "vue"),
+    ("xy", "xquery"),
+    ("xquery", "xquery"),
+    ("yaml", "yaml"),
+    ("yml", "yaml"),
+    ("g4", "antlr4"),
+    ("cls", "apex"),
+    ("trigger", "apex"),
+    ("apex", "apex"),
+    ("awk", "awk"),
+    ("bicep", "bicep"),
+    ("cfc", "cfscript"),
+    ("cfm", "cfml"),
+    ("cmake", "cmake"),
+    ("csv", "csv"),
+    ("dot", "dot"),
+    ("gv", "dot"),
+    ("xlsx", "excel-formula"),
+    ("xls", "excel-formula"),
+    ("po", "gettext"),
+    ("tf", "hcl"),
+    ("hcl", "hcl"),
+    ("idr", "idris"),
+    ("gitignore", "ignore"),
+    ("json5", "json5"),
+    ("ly", "lilypond"),
+    ("ld", "linker-script"),
+    ("ll", "llvm"),
+    ("log", "log"),
+    ("qasm", "openqasm"),
+    ("puml", "plant-uml"),
+    ("plantuml", "plant-uml"),
+    ("pq", "powerquery"),
+    ("purs", "purescript"),
+    ("qs", "qsharp"),
+    ("rkt", "racket"),
+    ("res", "rescript"),
+    ("robot", "robotframework"),
+    ("sol", "solidity"),
+    ("rq", "sparql"),
+    ("toml", "toml"),
+    ("ttl", "turtle"),
+    ("uc", "unrealscript"),
+    ("wgsl", "wgsl"),
+    ("wl", "wolfram"),
+    ("nb", "wolfram"),
+    ("zig", "zig"),
+];
+
+pub fn format_for_path<'a>(
+    path: &Path,
+    formats_exts: &'a FormatMappings,
+    formats_names: &'a FormatMappings,
+) -> Option<&'a str> {
+    let file_name = path.file_name()?.to_string_lossy();
+    if !formats_names.is_empty()
+        && let Some(format) = formats_names.find_format_for_value(&file_name)
+    {
+        return Some(format);
+    }
+
+    let ext = path.extension()?.to_string_lossy();
+    if !formats_exts.is_empty() {
+        return formats_exts.find_format_for_value(&ext);
+    }
+
+    EXTENSION_FORMATS
+        .iter()
+        .find_map(|(candidate, format)| (*candidate == ext).then_some(*format))
+}
+
+pub fn supported_formats() -> Vec<&'static str> {
+    SUPPORTED_FORMATS.to_vec()
+}
+
+#[cfg(test)]
+mod tests {
+    use std::path::Path;
+
+    use crate::cli::FormatMappings;
+
+    #[test]
+    fn maps_module_typescript_extensions_like_upstream() {
+        let formats_exts = FormatMappings::default();
+        let formats_names = FormatMappings::default();
+
+        assert_eq!(
+            super::format_for_path(Path::new("index.mts"), &formats_exts, &formats_names),
+            Some("typescript")
+        );
+        assert_eq!(
+            super::format_for_path(Path::new("index.cts"), &formats_exts, &formats_names),
+            Some("typescript")
+        );
+    }
+
+    #[test]
+    fn maps_javascript_module_extensions_like_upstream() {
+        let formats_exts = FormatMappings::default();
+        let formats_names = FormatMappings::default();
+
+        assert_eq!(
+            super::format_for_path(Path::new("index.es"), &formats_exts, &formats_names),
+            Some("javascript")
+        );
+        assert_eq!(
+            super::format_for_path(Path::new("index.es6"), &formats_exts, &formats_names),
+            Some("javascript")
+        );
+    }
+
+    #[test]
+    fn syncs_supported_format_surface_from_upstream() {
+        assert_eq!(super::SUPPORTED_FORMATS.len(), 223);
+        assert_eq!(super::EXTENSION_FORMATS.len(), 206);
+        assert_eq!(super::supported_formats().first(), Some(&"abap"));
+        assert!(super::supported_formats().contains(&"typescript"));
+        assert!(super::supported_formats().contains(&"excel-formula"));
+    }
+
+    #[test]
+    fn maps_long_tail_extensions_like_upstream() {
+        let formats_exts = FormatMappings::default();
+        let formats_names = FormatMappings::default();
+
+        assert_eq!(
+            super::format_for_path(Path::new("schema.g4"), &formats_exts, &formats_names),
+            Some("antlr4")
+        );
+        assert_eq!(
+            super::format_for_path(Path::new("component.bicep"), &formats_exts, &formats_names),
+            Some("bicep")
+        );
+        assert_eq!(
+            super::format_for_path(Path::new("formula.xlsx"), &formats_exts, &formats_names),
+            Some("excel-formula")
+        );
+    }
+
+    #[test]
+    fn maps_header_extensions_like_upstream() {
+        let formats_exts = FormatMappings::default();
+        let formats_names = FormatMappings::default();
+
+        assert_eq!(
+            super::format_for_path(Path::new("foo.h"), &formats_exts, &formats_names),
+            Some("c-header")
+        );
+        assert_eq!(
+            super::format_for_path(Path::new("foo.hpp"), &formats_exts, &formats_names),
+            Some("cpp-header")
+        );
+    }
+
+    #[test]
+    fn extensionless_names_require_formats_names_like_upstream() {
+        let formats_exts = FormatMappings::default();
+        let formats_names = FormatMappings::default();
+
+        assert_eq!(
+            super::format_for_path(Path::new("Makefile"), &formats_exts, &formats_names),
+            None
+        );
+
+        let formats_names = FormatMappings::from_pairs(vec![(
+            "makefile".to_string(),
+            vec!["Makefile".to_string(), "GNUmakefile".to_string()],
+        )]);
+
+        assert_eq!(
+            super::format_for_path(Path::new("Makefile"), &formats_exts, &formats_names),
+            Some("makefile")
+        );
+    }
+}