Merge pull request #157 from epage/json

Polish typos

crates/typos/src/checks.rs

@@ -112,19 +112,12 @@ impl ParseIdentifiers {
             return Ok(typos_found);
         }
 
-        let buffer = std::fs::read(path)
-            .map_err(|e| crate::ErrorKind::IoError.into_error().with_source(e))?;
-        if !explicit && !self.binary {
-            let content_type = content_inspector::inspect(&buffer);
-            if content_type.is_binary()
-                // HACK: We only support UTF-8 at the moment
-                || (content_type != content_inspector::ContentType::UTF_8_BOM
-                    && content_type != content_inspector::ContentType::UTF_8)
-            {
-                let msg = report::BinaryFile { path };
-                reporter.report(msg.into());
-                return Ok(typos_found);
-            }
+        let buffer = read_file(path)?;
+        let (buffer, content_type) = massage_data(buffer)?;
+        if !explicit && !self.binary && content_type.is_binary() {
+            let msg = report::BinaryFile { path };
+            reporter.report(msg.into());
+            return Ok(typos_found);
         }
 
         for line in buffer.lines() {
@@ -188,19 +181,12 @@ impl ParseWords {
             return Ok(typos_found);
         }
 
-        let buffer = std::fs::read(path)
-            .map_err(|e| crate::ErrorKind::IoError.into_error().with_source(e))?;
-        if !explicit && !self.binary {
-            let content_type = content_inspector::inspect(&buffer);
-            // HACK: We only support UTF-8 at the moment
-            if content_type.is_binary()
-                || (content_type != content_inspector::ContentType::UTF_8_BOM
-                    && content_type != content_inspector::ContentType::UTF_8)
-            {
-                let msg = report::BinaryFile { path };
-                reporter.report(msg.into());
-                return Ok(typos_found);
-            }
+        let buffer = read_file(path)?;
+        let (buffer, content_type) = massage_data(buffer)?;
+        if !explicit && !self.binary && content_type.is_binary() {
+            let msg = report::BinaryFile { path };
+            reporter.report(msg.into());
+            return Ok(typos_found);
         }
 
         for line in buffer.lines() {
@@ -249,7 +235,7 @@ impl Checks {
                 Some(Status::Valid) => {}
                 Some(corrections) => {
                     let byte_offset = ident.offset();
-                    let msg = report::PathCorrection {
+                    let msg = report::PathTypo {
                         path,
                         byte_offset,
                         typo: ident.token(),
@@ -263,7 +249,7 @@ impl Checks {
                             Some(Status::Valid) => {}
                             Some(corrections) => {
                                 let byte_offset = word.offset();
-                                let msg = report::PathCorrection {
+                                let msg = report::PathTypo {
                                     path,
                                     byte_offset,
                                     typo: word.token(),
@@ -295,20 +281,12 @@ impl Checks {
             return Ok(typos_found);
         }
 
-        let buffer = std::fs::read(path)
-            .map_err(|e| crate::ErrorKind::IoError.into_error().with_source(e))?;
-        if !explicit && !self.binary {
-            let content_type = content_inspector::inspect(&buffer);
-            // HACK: We only support UTF-8 at the moment
-            if content_type.is_binary()
-                || (content_type != content_inspector::ContentType::UTF_8_BOM
-                    && content_type != content_inspector::ContentType::UTF_8)
-            {
-                // HACK: we don't support alternative encodings atm
-                let msg = report::BinaryFile { path };
-                reporter.report(msg.into());
-                return Ok(typos_found);
-            }
+        let buffer = read_file(path)?;
+        let (buffer, content_type) = massage_data(buffer)?;
+        if !explicit && !self.binary && content_type.is_binary() {
+            let msg = report::BinaryFile { path };
+            reporter.report(msg.into());
+            return Ok(typos_found);
         }
 
         for (line_idx, line) in buffer.lines().enumerate() {
@@ -318,7 +296,7 @@ impl Checks {
                     Some(Status::Valid) => {}
                     Some(corrections) => {
                         let byte_offset = ident.offset();
-                        let msg = report::Correction {
+                        let msg = report::FileTypo {
                             path,
                             line,
                             line_num,
@@ -334,7 +312,7 @@ impl Checks {
                                 Some(Status::Valid) => {}
                                 Some(corrections) => {
                                     let byte_offset = word.offset();
-                                    let msg = report::Correction {
+                                    let msg = report::FileTypo {
                                         path,
                                         line,
                                         line_num,
@@ -355,3 +333,22 @@ impl Checks {
         Ok(typos_found)
     }
 }
+
+fn read_file(path: &std::path::Path) -> Result<Vec<u8>, crate::Error> {
+    std::fs::read(path).map_err(|e| crate::ErrorKind::IoError.into_error().with_source(e))
+}
+
+fn massage_data(
+    buffer: Vec<u8>,
+) -> Result<(Vec<u8>, content_inspector::ContentType), crate::Error> {
+    let mut content_type = content_inspector::inspect(&buffer);
+
+    // HACK: We only support UTF-8 at the moment
+    if content_type != content_inspector::ContentType::UTF_8_BOM
+        && content_type != content_inspector::ContentType::UTF_8
+    {
+        content_type = content_inspector::ContentType::BINARY;
+    }
+
+    Ok((buffer, content_type))
+}

crates/typos/src/report.rs

@@ -8,8 +8,8 @@ use std::io::{self, Write};
 #[non_exhaustive]
 pub enum Message<'m> {
     BinaryFile(BinaryFile<'m>),
-    Correction(Correction<'m>),
-    PathCorrection(PathCorrection<'m>),
+    FileTypo(FileTypo<'m>),
+    PathTypo(PathTypo<'m>),
     File(File<'m>),
     Parse(Parse<'m>),
     PathError(PathError<'m>),
@@ -20,8 +20,8 @@ impl<'m> Message<'m> {
     pub fn is_correction(&self) -> bool {
         match self {
             Message::BinaryFile(_) => false,
-            Message::Correction(c) => c.corrections.is_correction(),
-            Message::PathCorrection(c) => c.corrections.is_correction(),
+            Message::FileTypo(c) => c.corrections.is_correction(),
+            Message::PathTypo(c) => c.corrections.is_correction(),
             Message::File(_) => false,
             Message::Parse(_) => false,
             Message::PathError(_) => false,
@@ -32,8 +32,8 @@ impl<'m> Message<'m> {
     pub fn is_error(&self) -> bool {
         match self {
             Message::BinaryFile(_) => false,
-            Message::Correction(_) => false,
-            Message::PathCorrection(_) => false,
+            Message::FileTypo(_) => false,
+            Message::PathTypo(_) => false,
             Message::File(_) => false,
             Message::Parse(_) => false,
             Message::PathError(_) => true,
@@ -51,7 +51,7 @@ pub struct BinaryFile<'m> {
 
 #[derive(Clone, Debug, serde::Serialize, derive_setters::Setters)]
 #[non_exhaustive]
-pub struct Correction<'m> {
+pub struct FileTypo<'m> {
     pub path: &'m std::path::Path,
     #[serde(skip)]
     pub line: &'m [u8],
@@ -61,7 +61,7 @@ pub struct Correction<'m> {
     pub corrections: crate::Status<'m>,
 }
 
-impl<'m> Default for Correction<'m> {
+impl<'m> Default for FileTypo<'m> {
     fn default() -> Self {
         Self {
             path: std::path::Path::new("-"),
@@ -76,14 +76,14 @@ impl<'m> Default for Correction<'m> {
 
 #[derive(Clone, Debug, serde::Serialize, derive_setters::Setters)]
 #[non_exhaustive]
-pub struct PathCorrection<'m> {
+pub struct PathTypo<'m> {
     pub path: &'m std::path::Path,
     pub byte_offset: usize,
     pub typo: &'m str,
     pub corrections: crate::Status<'m>,
 }
 
-impl<'m> Default for PathCorrection<'m> {
+impl<'m> Default for PathTypo<'m> {
     fn default() -> Self {
         Self {
             path: std::path::Path::new("-"),
@@ -195,7 +195,7 @@ impl Report for PrintBrief {
             Message::BinaryFile(msg) => {
                 log::info!("{}", msg);
             }
-            Message::Correction(msg) => match &msg.corrections {
+            Message::FileTypo(msg) => match &msg.corrections {
                 crate::Status::Valid => {}
                 crate::Status::Invalid => {
                     println!(
@@ -217,7 +217,7 @@ impl Report for PrintBrief {
                     );
                 }
             },
-            Message::PathCorrection(msg) => match &msg.corrections {
+            Message::PathTypo(msg) => match &msg.corrections {
                 crate::Status::Valid => {}
                 crate::Status::Invalid => {
                     println!("{}: {} is disallowed", msg.path.display(), msg.typo,);
@@ -257,8 +257,8 @@ impl Report for PrintLong {
             Message::BinaryFile(msg) => {
                 log::info!("{}", msg);
             }
-            Message::Correction(msg) => print_long_correction(msg),
-            Message::PathCorrection(msg) => match &msg.corrections {
+            Message::FileTypo(msg) => print_long_correction(msg),
+            Message::PathTypo(msg) => match &msg.corrections {
                 crate::Status::Valid => {}
                 crate::Status::Invalid => {
                     println!(
@@ -293,7 +293,7 @@ impl Report for PrintLong {
     }
 }
 
-fn print_long_correction(msg: &Correction) {
+fn print_long_correction(msg: &FileTypo) {
     let line_num = msg.line_num.to_string();
     let line_indent: String = itertools::repeat_n(" ", line_num.len()).collect();
 

docs/reference.md

@@ -8,14 +8,13 @@ Configuration is read from the following (in precedence order)
 
 - Command line arguments
 - File specified via `--config PATH`
-- Search parents of specified file / directory for `typos.toml`
+- Search parents of specified file / directory for one of `typos.toml`, `_typos.toml`, or `.typos.toml`
 
 ### Config Fields
 
 | Field                  | Argument          | Format | Description |
 |------------------------|-------------------|--------|-------------|
 | files.binary           | --binary          | bool   | Check binary files as text |
-| files.ignore-patterns  |                   | list of strings | Typos-specific ignore globs (gitignore syntax) |
 | files.ignore-hidden    | --hidden          | bool   | Skip hidden files and directories. |
 | files.ignore-files     | --ignore          | bool   | Respect ignore files. |
 | files.ignore-dot       | --ignore-dot      | bool   | Respect .ignore files. |

src/config.rs

@@ -17,16 +17,6 @@ pub trait WalkSource {
         None
     }
 
-    /// The root for `ignore_patterns`
-    fn ignore_root(&self) -> Option<&std::path::Path> {
-        None
-    }
-
-    /// Ignore the specified patterns (gitignore syntax)
-    fn ignore_patterns(&self) -> Option<&[String]> {
-        None
-    }
-
     /// Skip hidden files and directories.
     fn ignore_hidden(&self) -> Option<bool> {
         None
@@ -120,9 +110,7 @@ impl Config {
         let mut file = std::fs::File::open(path)?;
         let mut s = String::new();
         file.read_to_string(&mut s)?;
-        let mut c = Self::from_toml(&s)?;
-        c.files.ignore_root = path.parent().map(|p| p.to_owned());
-        Ok(c)
+        Self::from_toml(&s)
     }
 
     pub fn from_toml(data: &str) -> Result<Self, anyhow::Error> {
@@ -131,7 +119,7 @@ impl Config {
     }
 
     pub fn derive(cwd: &std::path::Path) -> Result<Self, anyhow::Error> {
-        if let Some(path) = find_project_file(cwd.to_owned(), "typos.toml") {
+        if let Some(path) = find_project_file(cwd, &["typos.toml", "_typos.toml", ".typos.toml"]) {
             Self::from_file(&path)
         } else {
             Ok(Default::default())
@@ -152,6 +140,10 @@ impl ConfigSource for Config {
     fn walk(&self) -> Option<&dyn WalkSource> {
         Some(&self.files)
     }
+
+    fn default(&self) -> Option<&dyn FileSource> {
+        Some(&self.default)
+    }
 }
 
 #[derive(Debug, Clone, Default, serde::Serialize, serde::Deserialize)]
@@ -159,9 +151,6 @@ impl ConfigSource for Config {
 #[serde(rename_all = "kebab-case")]
 pub struct Walk {
     pub binary: Option<bool>,
-    #[serde(skip)]
-    pub ignore_root: Option<std::path::PathBuf>,
-    pub ignore_patterns: Option<Vec<String>>,
     pub ignore_hidden: Option<bool>,
     pub ignore_files: Option<bool>,
     pub ignore_dot: Option<bool>,
@@ -175,10 +164,6 @@ impl Walk {
         if let Some(source) = source.binary() {
             self.binary = Some(source);
         }
-        if let (Some(root), Some(source)) = (source.ignore_root(), source.ignore_patterns()) {
-            self.ignore_root = Some(root.to_owned());
-            self.ignore_patterns = Some(source.to_owned());
-        }
         if let Some(source) = source.ignore_hidden() {
             self.ignore_hidden = Some(source);
         }
@@ -208,14 +193,6 @@ impl Walk {
         self.binary.unwrap_or(false)
     }
 
-    pub fn ignore_root(&self) -> Option<&std::path::Path> {
-        self.ignore_root.as_deref()
-    }
-
-    pub fn ignore_patterns(&self) -> Option<&[String]> {
-        self.ignore_patterns.as_deref()
-    }
-
     pub fn ignore_hidden(&self) -> bool {
         self.ignore_hidden.unwrap_or(true)
     }
@@ -245,14 +222,6 @@ impl WalkSource for Walk {
         self.binary
     }
 
-    fn ignore_root(&self) -> Option<&std::path::Path> {
-        self.ignore_root.as_deref()
-    }
-
-    fn ignore_patterns(&self) -> Option<&[String]> {
-        self.ignore_patterns.as_deref()
-    }
-
     fn ignore_hidden(&self) -> Option<bool> {
         self.ignore_hidden
     }
@@ -431,18 +400,17 @@ impl FileSource for FileConfig {
     }
 }
 
-fn find_project_file(dir: std::path::PathBuf, name: &str) -> Option<std::path::PathBuf> {
-    let mut file_path = dir;
-    file_path.push(name);
-    while !file_path.exists() {
-        file_path.pop(); // filename
-        let hit_bottom = !file_path.pop();
-        if hit_bottom {
-            return None;
+fn find_project_file(dir: &std::path::Path, names: &[&str]) -> Option<std::path::PathBuf> {
+    for ancestor in dir.ancestors() {
+        let mut file_path = ancestor.join("placeholder");
+        for name in names {
+            file_path.set_file_name(name);
+            if file_path.exists() {
+                return Some(file_path);
+            }
         }
-        file_path.push(name);
     }
-    Some(file_path)
+    None
 }
 
 #[derive(Debug, Copy, Clone, serde::Serialize, serde::Deserialize)]

src/main.rs

@@ -14,7 +14,13 @@ mod dict;
 mod replace;
 
 fn main() {
-    let code = run().unwrap();
+    let code = match run() {
+        Ok(code) => code,
+        Err(err) => {
+            eprintln!("{}", err);
+            1
+        }
+    };
     std::process::exit(code);
 }
 
@@ -78,16 +84,6 @@ fn run() -> Result<i32, anyhow::Error> {
             .git_ignore(config.files.ignore_vcs())
             .git_exclude(config.files.ignore_vcs())
             .parents(config.files.ignore_parent());
-        if let (Some(root), Some(patterns)) =
-            (config.files.ignore_root(), config.files.ignore_patterns())
-        {
-            let mut overrides = ignore::overrides::OverrideBuilder::new(root);
-            for pattern in patterns {
-                overrides.add(pattern)?;
-            }
-            let overrides = overrides.build()?;
-            walk.overrides(overrides);
-        }
 
         let mut reporter = args.format.reporter();
         let replace_reporter = replace::Replace::new(reporter);

src/replace.rs

@@ -57,7 +57,7 @@ impl<'r> Replace<'r> {
 impl<'r> typos::report::Report for Replace<'r> {
     fn report(&self, msg: typos::report::Message<'_>) -> bool {
         match msg {
-            typos::report::Message::Correction(msg) => match msg.corrections {
+            typos::report::Message::FileTypo(msg) => match msg.corrections {
                 typos::Status::Corrections(corrections) if corrections.len() == 1 => {
                     let path = msg.path.to_owned();
                     let line_num = msg.line_num;
@@ -73,11 +73,9 @@ impl<'r> typos::report::Report for Replace<'r> {
                     content.push(correction);
                     false
                 }
-                _ => self
-                    .reporter
-                    .report(typos::report::Message::Correction(msg)),
+                _ => self.reporter.report(typos::report::Message::FileTypo(msg)),
             },
-            typos::report::Message::PathCorrection(msg) => match msg.corrections {
+            typos::report::Message::PathTypo(msg) => match msg.corrections {
                 typos::Status::Corrections(corrections) if corrections.len() == 1 => {
                     let path = msg.path.to_owned();
                     let correction =
@@ -87,9 +85,7 @@ impl<'r> typos::report::Report for Replace<'r> {
                     content.push(correction);
                     false
                 }
-                _ => self
-                    .reporter
-                    .report(typos::report::Message::PathCorrection(msg)),
+                _ => self.reporter.report(typos::report::Message::PathTypo(msg)),
             },
             _ => self.reporter.report(msg),
         }
@@ -208,7 +204,7 @@ mod test {
         let primary = typos::report::PrintSilent;
         let replace = Replace::new(&primary);
         replace.report(
-            typos::report::Correction::default()
+            typos::report::FileTypo::default()
                 .path(input_file.path())
                 .line(b"1 foo 2\n3 4 5")
                 .line_num(1)
@@ -233,7 +229,7 @@ mod test {
         let primary = typos::report::PrintSilent;
         let replace = Replace::new(&primary);
         replace.report(
-            typos::report::PathCorrection::default()
+            typos::report::PathTypo::default()
                 .path(input_file.path())
                 .byte_offset(0)
                 .typo("foo")