From 62f10571f335c19c069927eb6e6fab5851ee80b1 Mon Sep 17 00:00:00 2001 From: Ed Page Date: Fri, 13 Jan 2023 20:41:35 -0600 Subject: [PATCH 1/4] style(cli): Adjust macro/comment style --- src/bin/typos-cli/args.rs | 44 +++++++++++++++++++-------------------- 1 file changed, 22 insertions(+), 22 deletions(-) diff --git a/src/bin/typos-cli/args.rs b/src/bin/typos-cli/args.rs index 3d812e4..1857b39 100644 --- a/src/bin/typos-cli/args.rs +++ b/src/bin/typos-cli/args.rs @@ -46,51 +46,51 @@ 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, - #[arg(short = 'c', long = "config")] /// Custom config file + #[arg(short = 'c', long = "config")] pub(crate) custom_config: Option, - #[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, - #[arg(long, group = "mode")] /// 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, - #[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 +106,28 @@ 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)] @@ -194,42 +194,42 @@ impl ConfigArgs { #[derive(Debug, clap::Args)] #[clap(rename_all = "kebab-case")] pub(crate) struct WalkArgs { - #[clap(long, name = "GLOB")] /// Ignore files & directories matching the glob. + #[clap(long, name = "GLOB")] exclude: Vec, - #[clap(long, overrides_with("no_hidden"))] /// Search hidden files and directories. + #[clap(long, overrides_with("no_hidden"))] hidden: bool, #[clap(long, overrides_with("hidden"), hide = true)] no_hidden: bool, - #[clap(long, overrides_with("ignore"))] /// Don't respect ignore files. + #[clap(long, overrides_with("ignore"))] no_ignore: bool, #[clap(long, overrides_with("no_ignore"), hide = true)] ignore: bool, - #[clap(long, overrides_with("ignore_dot"))] /// Don't respect .ignore files. + #[clap(long, overrides_with("ignore_dot"))] no_ignore_dot: bool, #[clap(long, overrides_with("no_ignore_dot"), hide = true)] ignore_dot: bool, - #[clap(long, overrides_with("ignore_global"))] /// Don't respect global ignore files. + #[clap(long, overrides_with("ignore_global"))] no_ignore_global: bool, #[clap(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. + #[clap(long, overrides_with("ignore_parent"))] no_ignore_parent: bool, #[clap(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. + #[clap(long, overrides_with("ignore_vcs"))] no_ignore_vcs: bool, #[clap(long, overrides_with("no_ignore_vcs"), hide = true)] ignore_vcs: bool, From 4b6e9c3c1bbb98fb297028243df7a116f7384faa Mon Sep 17 00:00:00 2001 From: Ed Page Date: Fri, 13 Jan 2023 20:43:18 -0600 Subject: [PATCH 2/4] style(cli): Resolve clap deprecations --- src/bin/typos-cli/args.rs | 34 +++++++++++++++++----------------- 1 file changed, 17 insertions(+), 17 deletions(-) diff --git a/src/bin/typos-cli/args.rs b/src/bin/typos-cli/args.rs index 1857b39..ba4ea3e 100644 --- a/src/bin/typos-cli/args.rs +++ b/src/bin/typos-cli/args.rs @@ -173,11 +173,11 @@ 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 +192,46 @@ impl ConfigArgs { } #[derive(Debug, clap::Args)] -#[clap(rename_all = "kebab-case")] +#[command(rename_all = "kebab-case")] pub(crate) struct WalkArgs { /// Ignore files & directories matching the glob. - #[clap(long, name = "GLOB")] + #[arg(long, value_name = "GLOB")] exclude: Vec, /// Search hidden files and directories. - #[clap(long, overrides_with("no_hidden"))] + #[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, /// Don't respect ignore files. - #[clap(long, overrides_with("ignore"))] + #[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, /// Don't respect .ignore files. - #[clap(long, overrides_with("ignore_dot"))] + #[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, /// Don't respect global ignore files. - #[clap(long, overrides_with("ignore_global"))] + #[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, /// Don't respect ignore files in parent directories. - #[clap(long, overrides_with("ignore_parent"))] + #[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, /// Don't respect ignore files in vcs directories. - #[clap(long, overrides_with("ignore_vcs"))] + #[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, } From 222da3d81af3d6bcf2ad60e1a033b2942a7c093b Mon Sep 17 00:00:00 2001 From: Ed Page Date: Fri, 13 Jan 2023 20:44:33 -0600 Subject: [PATCH 3/4] chore: Upgrade to clap v4 --- Cargo.lock | 16 ++++++++-------- Cargo.toml | 2 +- 2 files changed, 9 insertions(+), 9 deletions(-) diff --git a/Cargo.lock b/Cargo.lock index 7335465..379b938 100644 --- a/Cargo.lock +++ b/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", diff --git a/Cargo.toml b/Cargo.toml index 7475c11..e346c62 100644 --- a/Cargo.toml +++ b/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"] } From b92a8c1eea8d4e865800b5894ccc7da1f98e0bd0 Mon Sep 17 00:00:00 2001 From: Ed Page Date: Fri, 13 Jan 2023 20:59:49 -0600 Subject: [PATCH 4/4] feat(cli): '--file-types' flag to help debug detection issues Like #567 --- src/bin/typos-cli/args.rs | 4 ++++ src/bin/typos-cli/main.rs | 2 ++ src/bin/typos-cli/report.rs | 16 ++++++++++++++++ src/file.rs | 30 ++++++++++++++++++++++++++++++ src/policy.rs | 33 +++++++++++++++++++-------------- src/report.rs | 25 +++++++++++++++++++++++++ 6 files changed, 96 insertions(+), 14 deletions(-) diff --git a/src/bin/typos-cli/args.rs b/src/bin/typos-cli/args.rs index ba4ea3e..8261486 100644 --- a/src/bin/typos-cli/args.rs +++ b/src/bin/typos-cli/args.rs @@ -70,6 +70,10 @@ pub(crate) struct Args { #[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, diff --git a/src/bin/typos-cli/main.rs b/src/bin/typos-cli/main.rs index f0bc0af..b9aeb91 100644 --- a/src/bin/typos-cli/main.rs +++ b/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 { diff --git a/src/bin/typos-cli/report.rs b/src/bin/typos-cli/report.rs index 5a8b438..7074bf7 100644 --- a/src/bin/typos-cli/report.rs +++ b/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())?; } diff --git a/src/file.rs b/src/file.rs index 887a9f9..f727f79 100644 --- a/src/file.rs +++ b/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; diff --git a/src/policy.rs b/src/policy.rs index 42948db..fd4b71f 100644 --- a/src/policy.rs +++ b/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), @@ -299,21 +300,23 @@ 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| { - log::debug!("{}: `{}` policy", path.display(), name); - self.types.get(name).copied() - }) - .unwrap_or_else(|| { - log::debug!( - "{}: default policy for `{}` file type", - path.display(), - name.unwrap_or("") - ); - self.default - }) + let config = name + .and_then(|name| { + log::debug!("{}: `{}` policy", path.display(), name); + self.types.get(name).copied() + }) + .unwrap_or_else(|| { + log::debug!( + "{}: default policy for `{}` file type", + path.display(), + name.unwrap_or("") + ); + 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, diff --git a/src/report.rs b/src/report.rs index 871accb..e11ac52 100644 --- a/src/report.rs +++ b/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> {