chore: add --stdin-paths and -stdin-paths0

Author: dpc

Cargo.lock

@@ -78,6 +78,7 @@ colorchoice-clap = "1.0.3"
 serde_regex = "1.1.0"
 regex = "1.9.6"
 encoding_rs = "0.8.33"
+os_str_bytes = "6.5.1"
 
 [dev-dependencies]
 assert_fs = "1.0"

crates/typos-cli/Cargo.toml

@@ -39,9 +39,17 @@ impl Format {
 #[command(group = clap::ArgGroup::new("mode").multiple(false))]
 pub(crate) struct Args {
     /// Paths to check with `-` for stdin
-    #[arg(default_value = ".")]
+    #[arg(default_value = ".", group = "source")]
     pub(crate) path: Vec<std::path::PathBuf>,
 
+    /// Read the list of newline separated paths from stdin
+    #[arg(long, group = "source")]
+    pub(crate) stdin_paths: bool,
+
+    /// Read the list of '\0' separated paths from stdin
+    #[arg(long, group = "source")]
+    pub(crate) stdin_paths0: bool,
+
     /// The approximate number of threads to use.
     #[arg(short = 'j', long = "threads", default_value = "0")]
     pub(crate) threads: usize,
@@ -110,6 +118,12 @@ pub(crate) struct Args {
     pub(crate) verbose: clap_verbosity_flag::Verbosity,
 }
 
+impl Args {
+    pub fn is_using_stdin_paths(&self) -> bool {
+        self.stdin_paths || self.stdin_paths0
+    }
+}
+
 #[derive(Debug, Clone, clap::Args)]
 #[command(rename_all = "kebab-case")]
 pub(crate) struct FileArgs {

crates/typos-cli/src/bin/typos-cli/args.rs

@@ -1,4 +1,6 @@
-use std::io::Write;
+use std::ffi::OsString;
+use std::io::{Read as _, Write as _};
+use std::path::PathBuf;
 
 use clap::Parser;
 
@@ -167,8 +169,29 @@ fn run_checks(args: &args::Args) -> proc_exit::ExitResult {
 
     let mut typos_found = false;
     let mut errors_found = false;
-    for path in args.path.iter() {
+
+    let stdin_paths = if args.stdin_paths {
+        Some(
+            std::io::stdin()
+                .lines()
+                .map(|res| res.map(PathBuf::from))
+                .collect::<Result<_, _>>()
+                .with_code(proc_exit::sysexits::IO_ERR)?,
+        )
+    } else if args.stdin_paths0 {
+        Some(unix_read_paths_from_stdin()?)
+    } else {
+        None
+    };
+
+    // Note: stdin_paths and args.path are mutually exclusive, enforced by clap
+    for path in stdin_paths.as_ref().unwrap_or(&args.path) {
+        // Note paths are passed through stdin, `-` is treated like a normal path
         let cwd = if path == std::path::Path::new("-") {
+            if args.is_using_stdin_paths() {
+                return Err(proc_exit::sysexits::USAGE_ERR
+                    .with_message("Can't use `-` (stdin) while using stdin provided paths"));
+            };
             global_cwd.clone()
         } else if path.is_file() {
             let mut cwd = path
@@ -293,6 +316,48 @@ fn run_checks(args: &args::Args) -> proc_exit::ExitResult {
     }
 }
 
+#[cfg(target_family = "unix")]
+fn os_str_from_bytes(bytes: Vec<u8>) -> Result<OsString, proc_exit::Exit> {
+    use std::os::unix::ffi::OsStringExt;
+
+    Ok(OsString::from_vec(bytes))
+}
+
+#[cfg(not(target_family = "unix"))]
+fn os_str_from_bytes(bytes: Vec<u8>) -> Result<OsString, proc_exit::Exit> {
+    use os_str_bytes::RawOsString;
+
+    Ok(RawOsString::from_raw_vec(bytes).map_err(|_e| {
+        proc_exit::sysexits::USAGE_ERR
+            .with_message("Invalid encoding")?
+            .into_os_string()
+    }))
+}
+
+fn unix_read_paths_from_stdin() -> Result<Vec<PathBuf>, proc_exit::Exit> {
+    let mut buf = Vec::new();
+    std::io::stdin()
+        .read_to_end(&mut buf)
+        .with_code(proc_exit::sysexits::IO_ERR)?;
+
+    let (mut paths, rest) = buf.into_iter().try_fold(
+        (vec![], vec![]),
+        |(mut paths, mut cur_path): (Vec<PathBuf>, Vec<u8>), byte| {
+            if byte == 0 {
+                paths.push(PathBuf::from(os_str_from_bytes(cur_path)?));
+                Ok((paths, vec![]))
+            } else {
+                cur_path.push(byte);
+                Ok((paths, cur_path))
+            }
+        },
+    )?;
+    if !rest.is_empty() {
+        paths.push(PathBuf::from(os_str_from_bytes(rest)?));
+    }
+    Ok(paths)
+}
+
 fn init_logging(level: Option<log::Level>) {
     if let Some(level) = level {
         let mut builder = env_logger::Builder::new();

crates/typos-cli/src/bin/typos-cli/main.rs

@@ -1,5 +1,7 @@
 #[test]
 #[cfg(feature = "dict")]
 fn cli_tests() {
-    trycmd::TestCases::new().case("tests/cmd/*.toml");
+    trycmd::TestCases::new()
+        .case("tests/cmd/*.toml")
+        .case("tests/cmd/*.trycmd");
 }

crates/typos-cli/tests/cli_tests.rs

@@ -9,6 +9,8 @@ Arguments:
   [PATH]...  Paths to check with `-` for stdin [default: .]
 
 Options:
+      --stdin-paths        Read the list of newline separated paths from stdin
+      --stdin-paths0       Read the list of '/0' separated paths from stdin
   -j, --threads <THREADS>  The approximate number of threads to use [default: 0]
       --force-exclude      Respect excluded files even for paths passed explicitly
   -h, --help               Print help

crates/typos-cli/tests/cmd/help.toml

@@ -0,0 +1,8 @@
+[files]
+extend-exclude = ["_typos.toml"]
+
+[default.extend-identifiers]
+hello = "goodbye"
+
+[type.fail]
+extend-glob = ["*.fail"]

crates/typos-cli/tests/cmd/stdin-paths.in/_typos.toml

@@ -0,0 +1 @@
+hello

crates/typos-cli/tests/cmd/stdin-paths.in/a.fail

@@ -0,0 +1 @@
+hello

crates/typos-cli/tests/cmd/stdin-paths.in/b.fail

@@ -0,0 +1 @@
+hello