Merge pull request #652 from epage/type

feat(cli): '--file-types' flag to help debug detection issues

Cargo.lock

@@ -196,12 +196,12 @@ dependencies = [
 
 [[package]]
 name = "clap"
-version = "4.0.32"
+version = "4.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a7db700bc935f9e43e88d00b0850dae18a63773cfbec6d8e070fccf7fef89a39"
+checksum = "aa91278560fc226a5d9d736cc21e485ff9aad47d26b8ffe1f54cba868b684b9f"
 dependencies = [
  "bitflags",
- "clap_derive 4.0.21",
+ "clap_derive 4.1.0",
  "clap_lex 0.3.0",
  "is-terminal",
  "once_cell",
@@ -215,7 +215,7 @@ version = "2.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "23e2b6c3dcdb73299f48ae05b294da14e2f560b3ed2c09e742269eb1b22af231"
 dependencies = [
- "clap 4.0.32",
+ "clap 4.1.0",
  "log",
 ]
 
@@ -234,9 +234,9 @@ dependencies = [
 
 [[package]]
 name = "clap_derive"
-version = "4.0.21"
+version = "4.1.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0177313f9f02afc995627906bbd8967e2be069f5261954222dac78290c2b9014"
+checksum = "684a277d672e91966334af371f1a7b5833f9aa00b07c84e92fbce95e00208ce8"
 dependencies = [
  "heck",
  "proc-macro-error",
@@ -303,7 +303,7 @@ version = "0.0.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "64a9572308414bfdbda25e41cdd4a15a90e68d93f2d5a66e3ff6e1cdc856f923"
 dependencies = [
- "clap 4.0.32",
+ "clap 4.1.0",
  "concolor",
 ]
 
@@ -1619,7 +1619,7 @@ dependencies = [
  "assert_fs",
  "atty",
  "bstr 1.1.0",
- "clap 4.0.32",
+ "clap 4.1.0",
  "clap-verbosity-flag",
  "concolor",
  "concolor-clap",

Cargo.toml

@@ -65,7 +65,7 @@ typos-dict = { version = "^0.9", path = "crates/typos-dict", optional = true }
 typos-vars = { version = "^0.8", path = "crates/typos-vars", optional = true }
 unicase = "2.6"
 anyhow = "1.0"
-clap = { version = "4.0.32", features = ["derive"] }
+clap = { version = "4.1.0", features = ["derive"] }
 clap-verbosity-flag = "2.0"
 ignore = "0.4"
 serde = { version = "1.0", features = ["derive"] }

src/bin/typos-cli/args.rs

@@ -46,51 +46,55 @@ impl Default for Format {
     )]
 #[command(group = clap::ArgGroup::new("mode").multiple(false))]
 pub(crate) struct Args {
-    #[arg(default_value = ".")]
     /// Paths to check with `-` for stdin
+    #[arg(default_value = ".")]
     pub(crate) path: Vec<std::path::PathBuf>,
 
-    #[arg(short = 'c', long = "config")]
     /// Custom config file
+    #[arg(short = 'c', long = "config")]
     pub(crate) custom_config: Option<std::path::PathBuf>,
 
-    #[arg(long)]
     /// Ignore implicit configuration files.
+    #[arg(long)]
     pub(crate) isolated: bool,
 
-    #[arg(long, group = "mode")]
     /// Print a diff of what would change
+    #[arg(long, group = "mode")]
     pub(crate) diff: bool,
 
-    #[arg(long, short = 'w', group = "mode")]
     /// Write fixes out
+    #[arg(long, short = 'w', group = "mode")]
     pub(crate) write_changes: bool,
 
-    #[arg(long, group = "mode")]
     /// Debug: Print each file that would be spellchecked.
+    #[arg(long, group = "mode")]
     pub(crate) files: bool,
 
+    /// Debug: Print each file's type
     #[arg(long, group = "mode")]
+    pub(crate) file_types: bool,
+
     /// Debug: Print each identifier that would be spellchecked.
+    #[arg(long, group = "mode")]
     pub(crate) identifiers: bool,
 
-    #[arg(long, group = "mode")]
     /// Debug: Print each word that would be spellchecked.
+    #[arg(long, group = "mode")]
     pub(crate) words: bool,
 
-    #[arg(long, group = "mode")]
     /// Write the current configuration to file with `-` for stdout
+    #[arg(long, group = "mode")]
     pub(crate) dump_config: Option<std::path::PathBuf>,
 
-    #[arg(long, group = "mode")]
     /// Show all supported file types.
+    #[arg(long, group = "mode")]
     pub(crate) type_list: bool,
 
     #[arg(long, value_enum, ignore_case = true, default_value("long"))]
     pub(crate) format: Format,
 
-    #[arg(short = 'j', long = "threads", default_value = "0")]
     /// The approximate number of threads to use.
+    #[arg(short = 'j', long = "threads", default_value = "0")]
     pub(crate) threads: usize,
 
     #[command(flatten)]
@@ -106,28 +110,28 @@ pub(crate) struct Args {
 #[derive(Debug, Clone, clap::Args)]
 #[command(rename_all = "kebab-case")]
 pub(crate) struct FileArgs {
-    #[arg(long, overrides_with("no_binary"))]
     /// Search binary files.
+    #[arg(long, overrides_with("no_binary"))]
     binary: bool,
     #[arg(long, overrides_with("binary"), hide = true)]
     no_binary: bool,
 
-    #[arg(long, overrides_with("check_filenames"))]
     /// Skip verifying spelling in file names.
+    #[arg(long, overrides_with("check_filenames"))]
     no_check_filenames: bool,
     #[arg(long, overrides_with("no_check_filenames"), hide = true)]
     check_filenames: bool,
 
-    #[arg(long, overrides_with("check_files"))]
     /// Skip verifying spelling in files.
+    #[arg(long, overrides_with("check_files"))]
     no_check_files: bool,
     #[arg(long, overrides_with("no_check_files"), hide = true)]
     check_files: bool,
 
     #[arg(long, overrides_with("no_unicode"), hide = true)]
     unicode: bool,
-    #[arg(long, overrides_with("unicode"))]
     /// Only allow ASCII characters in identifiers
+    #[arg(long, overrides_with("unicode"))]
     no_unicode: bool,
 
     #[arg(long)]
@@ -173,11 +177,11 @@ impl FileArgs {
 }
 
 #[derive(Debug, clap::Args)]
-#[clap(rename_all = "kebab-case")]
+#[command(rename_all = "kebab-case")]
 pub(crate) struct ConfigArgs {
-    #[clap(flatten)]
+    #[command(flatten)]
     walk: WalkArgs,
-    #[clap(flatten)]
+    #[command(flatten)]
     overrides: FileArgs,
 }
 
@@ -192,46 +196,46 @@ impl ConfigArgs {
 }
 
 #[derive(Debug, clap::Args)]
-#[clap(rename_all = "kebab-case")]
+#[command(rename_all = "kebab-case")]
 pub(crate) struct WalkArgs {
-    #[clap(long, name = "GLOB")]
     /// Ignore files & directories matching the glob.
+    #[arg(long, value_name = "GLOB")]
     exclude: Vec<String>,
 
-    #[clap(long, overrides_with("no_hidden"))]
     /// Search hidden files and directories.
+    #[arg(long, overrides_with("no_hidden"))]
     hidden: bool,
-    #[clap(long, overrides_with("hidden"), hide = true)]
+    #[arg(long, overrides_with("hidden"), hide = true)]
     no_hidden: bool,
 
-    #[clap(long, overrides_with("ignore"))]
     /// Don't respect ignore files.
+    #[arg(long, overrides_with("ignore"))]
     no_ignore: bool,
-    #[clap(long, overrides_with("no_ignore"), hide = true)]
+    #[arg(long, overrides_with("no_ignore"), hide = true)]
     ignore: bool,
 
-    #[clap(long, overrides_with("ignore_dot"))]
     /// Don't respect .ignore files.
+    #[arg(long, overrides_with("ignore_dot"))]
     no_ignore_dot: bool,
-    #[clap(long, overrides_with("no_ignore_dot"), hide = true)]
+    #[arg(long, overrides_with("no_ignore_dot"), hide = true)]
     ignore_dot: bool,
 
-    #[clap(long, overrides_with("ignore_global"))]
     /// Don't respect global ignore files.
+    #[arg(long, overrides_with("ignore_global"))]
     no_ignore_global: bool,
-    #[clap(long, overrides_with("no_ignore_global"), hide = true)]
+    #[arg(long, overrides_with("no_ignore_global"), hide = true)]
     ignore_global: bool,
 
-    #[clap(long, overrides_with("ignore_parent"))]
     /// Don't respect ignore files in parent directories.
+    #[arg(long, overrides_with("ignore_parent"))]
     no_ignore_parent: bool,
-    #[clap(long, overrides_with("no_ignore_parent"), hide = true)]
+    #[arg(long, overrides_with("no_ignore_parent"), hide = true)]
     ignore_parent: bool,
 
-    #[clap(long, overrides_with("ignore_vcs"))]
     /// Don't respect ignore files in vcs directories.
+    #[arg(long, overrides_with("ignore_vcs"))]
     no_ignore_vcs: bool,
-    #[clap(long, overrides_with("no_ignore_vcs"), hide = true)]
+    #[arg(long, overrides_with("no_ignore_vcs"), hide = true)]
     ignore_vcs: bool,
 }
 

src/bin/typos-cli/main.rs

@@ -227,6 +227,8 @@ fn run_checks(
 
         let selected_checks: &dyn typos_cli::file::FileChecker = if args.files {
             &typos_cli::file::FoundFiles
+        } else if args.file_types {
+            &typos_cli::file::FileTypes
         } else if args.identifiers {
             &typos_cli::file::Identifiers
         } else if args.words {

src/bin/typos-cli/report.rs

@@ -89,6 +89,14 @@ impl Report for PrintBrief {
                 log::info!("{}", msg);
             }
             Message::Typo(msg) => print_brief_correction(msg, self.stdout_palette)?,
+            Message::FileType(msg) => {
+                writeln!(
+                    io::stdout(),
+                    "{}:{}",
+                    msg.path.display(),
+                    msg.file_type.unwrap_or("-")
+                )?;
+            }
             Message::File(msg) => {
                 writeln!(io::stdout(), "{}", msg.path.display())?;
             }
@@ -116,6 +124,14 @@ impl Report for PrintLong {
                 log::info!("{}", msg);
             }
             Message::Typo(msg) => print_long_correction(msg, self.stdout_palette)?,
+            Message::FileType(msg) => {
+                writeln!(
+                    io::stdout(),
+                    "{}:{}",
+                    msg.path.display(),
+                    msg.file_type.unwrap_or("-")
+                )?;
+            }
             Message::File(msg) => {
                 writeln!(io::stdout(), "{}", msg.path.display())?;
             }

src/file.rs

@@ -352,6 +352,36 @@ impl FileChecker for Words {
     }
 }
 
+#[derive(Debug, Clone, Copy)]
+pub struct FileTypes;
+
+impl FileChecker for FileTypes {
+    fn check_file(
+        &self,
+        path: &std::path::Path,
+        explicit: bool,
+        policy: &crate::policy::Policy,
+        reporter: &dyn report::Report,
+    ) -> Result<(), std::io::Error> {
+        // Check `policy.binary` first so we can easily check performance of walking vs reading
+        if policy.binary {
+            let msg = report::FileType::new(path, policy.file_type);
+            reporter.report(msg.into())?;
+        } else {
+            let (_buffer, content_type) = read_file(path, reporter)?;
+            if !explicit && content_type.is_binary() {
+                let msg = report::BinaryFile { path };
+                reporter.report(msg.into())?;
+            } else {
+                let msg = report::FileType::new(path, policy.file_type);
+                reporter.report(msg.into())?;
+            }
+        }
+
+        Ok(())
+    }
+}
+
 #[derive(Debug, Clone, Copy)]
 pub struct FoundFiles;
 

src/policy.rs

@@ -91,10 +91,11 @@ impl<'s> ConfigEngine<'s> {
     pub fn policy(&self, path: &std::path::Path) -> Policy<'_, '_> {
         debug_assert!(path.is_absolute(), "{} is not absolute", path.display());
         let dir = self.get_dir(path).expect("`walk()` should be called first");
-        let file_config = dir.get_file_config(path);
+        let (file_type, file_config) = dir.get_file_config(path);
         Policy {
             check_filenames: file_config.check_filenames,
             check_files: file_config.check_files,
+            file_type,
             binary: file_config.binary,
             tokenizer: self.get_tokenizer(&file_config),
             dict: self.get_dict(&file_config),
@@ -299,10 +300,11 @@ struct DirConfig {
 }
 
 impl DirConfig {
-    fn get_file_config(&self, path: &std::path::Path) -> FileConfig {
+    fn get_file_config(&self, path: &std::path::Path) -> (Option<&str>, FileConfig) {
         let name = self.type_matcher.file_matched(path);
 
-        name.and_then(|name| {
+        let config = name
+            .and_then(|name| {
                 log::debug!("{}: `{}` policy", path.display(), name);
                 self.types.get(name).copied()
             })
@@ -313,7 +315,8 @@ impl DirConfig {
                     name.unwrap_or("<unknown>")
                 );
                 self.default
-        })
+            });
+        (name, config)
     }
 }
 
@@ -331,6 +334,7 @@ struct FileConfig {
 pub struct Policy<'t, 'd> {
     pub check_filenames: bool,
     pub check_files: bool,
+    pub file_type: Option<&'d str>,
     pub binary: bool,
     pub tokenizer: &'t typos::tokens::Tokenizer,
     pub dict: &'d dyn typos::Dictionary,
@@ -351,6 +355,7 @@ impl<'t, 'd> Default for Policy<'t, 'd> {
         Self {
             check_filenames: true,
             check_files: true,
+            file_type: None,
             binary: false,
             tokenizer: &DEFAULT_TOKENIZER,
             dict: &DEFAULT_DICT,

src/report.rs

@@ -13,6 +13,7 @@ pub trait Report: Send + Sync {
 pub enum Message<'m> {
     BinaryFile(BinaryFile<'m>),
     Typo(Typo<'m>),
+    FileType(FileType<'m>),
     File(File<'m>),
     Parse(Parse<'m>),
     Error(Error<'m>),
@@ -23,6 +24,7 @@ impl<'m> Message<'m> {
         match self {
             Message::BinaryFile(_) => false,
             Message::Typo(c) => c.corrections.is_correction(),
+            Message::FileType(_) => false,
             Message::File(_) => false,
             Message::Parse(_) => false,
             Message::Error(_) => false,
@@ -33,6 +35,7 @@ impl<'m> Message<'m> {
         match self {
             Message::BinaryFile(_) => false,
             Message::Typo(_) => false,
+            Message::FileType(_) => false,
             Message::File(_) => false,
             Message::Parse(_) => false,
             Message::Error(_) => true,
@@ -144,6 +147,28 @@ pub enum ParseKind {
     Word,
 }
 
+#[derive(Clone, Debug, serde::Serialize, derive_setters::Setters)]
+#[non_exhaustive]
+pub struct FileType<'m> {
+    pub path: &'m std::path::Path,
+    pub file_type: Option<&'m str>,
+}
+
+impl<'m> FileType<'m> {
+    pub fn new(path: &'m std::path::Path, file_type: Option<&'m str>) -> Self {
+        Self { path, file_type }
+    }
+}
+
+impl<'m> Default for FileType<'m> {
+    fn default() -> Self {
+        Self {
+            path: std::path::Path::new("-"),
+            file_type: None,
+        }
+    }
+}
+
 #[derive(Clone, Debug, serde::Serialize, derive_setters::Setters)]
 #[non_exhaustive]
 pub struct File<'m> {