Now we take the file paths directly

.pre-commit-hooks.yaml

@@ -1,17 +1,17 @@
--   id: mdlinker
-    name: Markdown Linker
-    description: This hook checks that all markdown links are accounted for.
-    entry: mdlinker
-    language: rust
-    pass_filenames: false
--   id: enforce-ascii
-    name: Enforce ASCII Compliance
-    description: Detects and replaces non-ASCII characters in specified files.
-    entry: ./bin/enforce-ascii
-    language: script
-    types: [text]
-    pass_filenames: true
-    always_run: false
-    args:
+- id: mdlinker
+  name: Markdown Linker
+  description: This hook checks that all markdown links are accounted for.
+  entry: mdlinker
+  language: rust
+  pass_filenames: false
+- id: enforce-ascii
+  name: Enforce ASCII Compliance
+  description: Detects and replaces non-ASCII characters in specified files.
+  entry: ./bin/enforce-ascii
+  language: script
+  types: [text]
+  pass_filenames: true
+  always_run: false
+  args:
     - --fix
-    additional_dependencies: []
+  additional_dependencies: []

rust-toolchain.toml

@@ -1,3 +1,3 @@
 [toolchain]
-channel = "nightly"
+channel = "1.91.0"
 components = ["clippy", "rustfmt"]

src/config.rs

@@ -37,13 +37,16 @@ pub enum NewConfigError {
 /// Used to reconcile the two
 #[derive(Builder)]
 pub struct Config {
+    #[allow(clippy::struct_field_names)]
     file_config: file::Config,
+    #[allow(clippy::struct_field_names)]
     cli_config: cli::Config,
-    /// See [`self::cli::Config::pages_directory`]
-    pub pages_directory: PathBuf,
-    /// See [`self::cli::Config::other_directories`]
+    /// See [`self::cli::Config::files`]
     #[builder(default=vec![])]
-    pub other_directories: Vec<PathBuf>,
+    pub files: Vec<PathBuf>,
+    /// See [`self::cli::Config::root_directory`]
+    #[builder(default=PathBuf::from("."))]
+    pub new_files_directory: PathBuf,
     /// See [`self::cli::Config::ngram_size`]
     #[builder(default = 2)]
     pub ngram_size: usize,
@@ -84,8 +87,8 @@ pub struct Config {
 /// these can be unioned with one another
 /// and then we can use that to create the final config
 pub trait Partial {
-    fn pages_directory(&self) -> Option<PathBuf>;
-    fn other_directories(&self) -> Option<Vec<PathBuf>>;
+    fn files(&self) -> Option<Vec<PathBuf>>;
+    fn new_files_directory(&self) -> Option<PathBuf>;
     fn ngram_size(&self) -> Option<usize>;
     fn boundary_pattern(&self) -> Option<String>;
     fn filename_spacing_pattern(&self) -> Option<String>;
@@ -162,18 +165,12 @@ fn combine_partials(
         })
         .maybe_fix(cli_config.fix().or(file_config.fix()))
         .maybe_allow_dirty(cli_config.allow_dirty().or(file_config.allow_dirty()))
-        .pages_directory(
+        .files(
             cli_config
-                .pages_directory()
-                .or(file_config.pages_directory())
+                .files()
+                .or(file_config.files())
                 .expect("A default is set"),
         )
-        .maybe_other_directories(Some(
-            cli_config
-                .other_directories()
-                .or(file_config.other_directories())
-                .expect("A default is set"),
-        ))
         .maybe_ignore_word_pairs(
             cli_config
                 .ignore_word_pairs()
@@ -192,9 +189,9 @@ impl Config {
     ///
     /// # Errors
     ///
-    /// - [`Error::FileDoesNotExistError`] - Config file does not exist
-    /// - [`Error::FileDoesNotParseError`] - Config file does not parse from toml into the
-    ///     expected format
+    ///  - [`Error::FileDoesNotExistError`] - Config file does not exist
+    ///  - [`Error::FileDoesNotParseError`] - Config file does not parse from toml into the
+    ///    expected format
     ///
     pub fn new() -> Result<Self, NewConfigError> {
         let cli = cli::Config::parse();
@@ -227,11 +224,8 @@ impl Config {
     /// Legacy directories function
     /// Gets all the directories into one vec
     #[must_use]
-    pub fn directories(&self) -> Vec<PathBuf> {
-        let mut out = Vec::new();
-        out.push(self.pages_directory.clone());
-        out.extend(self.other_directories.clone());
-        out
+    pub fn files(&self) -> &[PathBuf] {
+        &self.files
     }
 
     pub fn add_report_to_ignore(&mut self, report: &impl ReportTrait) {

src/config/cli.rs

@@ -15,14 +15,14 @@ use super::Partial;
 #[derive(Parser, Default, Clone)]
 #[command(version, about, long_about = None)]
 pub struct Config {
-    /// The pages directory is the directory where pages are named for their alias
-    /// and where new pages should be created when running --fix
-    #[clap(short = 'p', long = "pages")]
-    pub pages_directory: Option<PathBuf>,
+    /// Globs or paths to relevant files
+    #[clap()]
+    pub files: Vec<PathBuf>,
 
-    /// Other directories to search in
-    #[clap(short = 'd', long = "dir")]
-    pub other_directories: Vec<PathBuf>,
+    /// A location to store new files in created by --fix
+    /// for the [`super::rules::broken_wikilink::BrokenWikilink`] rule
+    #[clap(short = 'n', long = "newf")]
+    pub new_files_directory: Option<PathBuf>,
 
     /// Path to a configuration file
     #[clap(short = 'c', long = "config", default_value = "mdlinker.toml")]
@@ -67,17 +67,16 @@ pub struct Config {
 }
 
 impl Partial for Config {
-    fn pages_directory(&self) -> Option<PathBuf> {
-        self.pages_directory.clone()
-    }
-    fn other_directories(&self) -> Option<Vec<PathBuf>> {
-        let out = self.other_directories.clone();
-        if out.is_empty() {
+    fn files(&self) -> Option<Vec<PathBuf>> {
+        if self.files.is_empty() {
             None
         } else {
-            Some(out)
+            Some(self.files.clone())
         }
     }
+    fn new_files_directory(&self) -> Option<PathBuf> {
+        self.new_files_directory.clone()
+    }
     fn ngram_size(&self) -> Option<usize> {
         self.ngram_size
     }

src/config/file.rs

@@ -15,11 +15,13 @@ use super::{Config as MasterConfig, NewConfigError, Partial};
 
 #[derive(Serialize, Deserialize, Debug, Default, Clone)]
 pub struct Config {
-    /// See [`super::cli::Config::pages_directory`]
-    pub pages_directory: PathBuf,
+    /// See [`super::cli::Config::files`]
+    #[serde(default)]
+    pub files: Option<Vec<String>>,
 
-    /// See [`super::cli::Config::other_directories`]
-    pub other_directories: Vec<PathBuf>,
+    /// See [`super::cli::Config::new_files_directory`]
+    #[serde(default)]
+    pub new_files_directory: Option<PathBuf>,
 
     /// See [`super::cli::Config::ngram_size`]
     #[serde(default)]
@@ -68,8 +70,14 @@ impl Config {
 impl From<MasterConfig> for Config {
     fn from(value: MasterConfig) -> Self {
         Self {
-            pages_directory: value.pages_directory,
-            other_directories: value.other_directories,
+            files: Some(
+                value
+                    .files
+                    .into_iter()
+                    .map(|file| file.to_string_lossy().to_string())
+                    .collect(),
+            ),
+            new_files_directory: Some(value.new_files_directory),
             ngram_size: Some(value.ngram_size),
             boundary_pattern: Some(value.boundary_pattern),
             filename_spacing_pattern: Some(value.filename_spacing_pattern),
@@ -83,18 +91,56 @@ impl From<MasterConfig> for Config {
 }
 
 impl Partial for Config {
-    fn pages_directory(&self) -> Option<PathBuf> {
-        Some(self.pages_directory.clone())
-    }
-    fn other_directories(&self) -> Option<Vec<PathBuf>> {
-        let out = self.other_directories.clone();
+    fn files(&self) -> Option<Vec<PathBuf>> {
+        let mut out = Vec::new();
+        match &self.files {
+            None => return None,
+            Some(files) if files.is_empty() => return None,
+            Some(files) => {
+                for file in files {
+                    if file.contains('*') {
+                        let pattern = glob::Pattern::new(file);
+                        match pattern {
+                            Ok(_) => {
+                                let globs = glob::glob(file);
+                                match globs {
+                                    Ok(globs) => {
+                                        for glob in globs {
+                                            match glob {
+                                                Ok(path) => out.push(path),
+                                                Err(e) => {
+                                                    eprintln!(
+                                                        "Error processing glob '{file}': {e}",
+                                                    );
+                                                }
+                                            }
+                                        }
+                                    }
+                                    Err(e) => {
+                                        eprintln!("Error parsing glob pattern '{file}': {e}");
+                                    }
+                                }
+                            }
+                            Err(e) => {
+                                eprintln!("Error parsing glob pattern '{file}': {e}");
+                                return None;
+                            }
+                        }
+                    }
+                }
+            }
+        }
         if out.is_empty() {
             None
         } else {
             Some(out)
         }
     }
 
+    fn new_files_directory(&self) -> Option<PathBuf> {
+        self.new_files_directory.clone()
+    }
+
     fn ngram_size(&self) -> Option<usize> {
         self.ngram_size
     }

src/file.rs

@@ -1,24 +1,4 @@
-use std::path::PathBuf;
-
-use walkdir::WalkDir;
-
 use thiserror::Error;
 
-use std;
-
 pub mod content;
 pub mod name;
-
-/// Walk the directories and get just the files
-pub fn get_files(dirs: &Vec<PathBuf>) -> Vec<PathBuf> {
-    let mut out = Vec::new();
-    for path in dirs {
-        let walk = WalkDir::new(path);
-        for entry in walk.into_iter().filter_map(Result::ok) {
-            if entry.file_type().is_file() {
-                out.push(entry.into_path());
-            }
-        }
-    }
-    out
-}

src/lib.rs

@@ -8,7 +8,7 @@ pub mod sed;
 pub mod visitor;
 
 use console::{style, Emoji};
-use file::{get_files, name::ngrams};
+use file::name::ngrams;
 use indicatif::ProgressBar;
 use miette::{Diagnostic, Result};
 use ngrams::CalculateError;
@@ -114,6 +114,7 @@ fn is_repo_dirty(repo: &Repository) -> Result<bool, Error> {
 }
 
 /// Runs [`check`] in a loop until no more fixes can be made
+#[allow(clippy::result_large_err)]
 fn fix(config: &config::Config) -> Result<OutputReport, OutputErrors> {
     // Check if the git repo is dirty
     match git2::Repository::open_from_env() {
@@ -145,7 +146,7 @@ fn fix(config: &config::Config) -> Result<OutputReport, OutputErrors> {
             style("[1/3]").bold().dim(),
             CHECK
         );
-    };
+    }
 
     let mut output_report = check(config)?;
 
@@ -191,7 +192,7 @@ fn fix(config: &config::Config) -> Result<OutputReport, OutputErrors> {
                 style("[3/3]").bold().dim(),
                 CHECK_AGAIN
             );
-        };
+        }
         output_report = check(config)?;
     } else if env::var("RUNNING_TESTS").is_err() {
         println!(
@@ -204,12 +205,13 @@ fn fix(config: &config::Config) -> Result<OutputReport, OutputErrors> {
     Ok(output_report)
 }
 
+#[allow(clippy::result_large_err)]
 fn check(config: &config::Config) -> Result<OutputReport, OutputErrors> {
     // Compile our regex patterns
     let boundary_regex = regex::Regex::new(&config.boundary_pattern)?;
     let filename_spacing_regex = regex::Regex::new(&config.filename_spacing_pattern)?;
 
-    let all_files = get_files(&config.directories());
+    let all_files = config.files().to_vec();
     let file_ngrams = ngrams(
         &all_files,
         config.ngram_size,
@@ -329,6 +331,7 @@ fn check(config: &config::Config) -> Result<OutputReport, OutputErrors> {
 ///
 /// Basically if this library fails, this returns an Err
 /// but if this library runs, even if it finds linting violations, this returns an Ok
+#[allow(clippy::result_large_err)]
 pub fn lib(config: &config::Config) -> Result<OutputReport, OutputErrors> {
     if config.fix {
         fix(config)

src/rules.rs

@@ -80,7 +80,6 @@ fn dedupe_by_code<T: ReportTrait + PartialOrd>(mut this: Vec<T>) -> Vec<T> {
 
 /// Used for filtering out items that start with the exclude code
 impl<T: ReportTrait + PartialOrd> VecHasIdExtensions<T> for Vec<T> {
-    #[must_use]
     fn finalize(self, excludes: &[ErrorCode]) -> Self {
         dedupe_by_code(filter_by_excludes(self, excludes))
     }

src/rules/broken_wikilink.rs

@@ -57,7 +57,7 @@ impl ReportTrait for BrokenWikilink {
             self.src.name()
         );
         let filename = format!("{}.md", FilenameLowercase::from_alias(&self.alias, config));
-        let path = config.pages_directory.join(filename);
+        let path = config.new_files_directory.join(filename);
         std::fs::write(path.clone(), "").map_err(|source| FixError::IOError {
             source,
             backtrace: Backtrace::force_capture(),

src/rules/unlinked_text.rs

@@ -56,7 +56,7 @@ impl ReportTrait for UnlinkedText {
     /// TODO: Be able to handle this in parallel with other reports
     fn fix(&self, _config: &Config) -> Result<Option<()>, FixError> {
         let file = self.src.name().to_owned();
-        trace!("Fixing unlinked text: {:?}", file);
+        trace!("Fixing unlinked text: {file:?}");
         let mut source = std::fs::read_to_string(&file).map_err(|src| FixError::IOError {
             source: src,
             file: file.clone(),

src/visitor.rs

@@ -58,7 +58,7 @@ pub trait Visitor {
     /// WARNING: Don't overwrite this, its already written for you.
     /// Implement [`Self::_finalize_file`] instead
     fn finalize_file(&mut self, source: &str, path: &Path) -> Result<(), FinalizeError> {
-        trace!("{:?} finalizing file {:?}", self.name(), path);
+        trace!("{:?} finalizing file {:?}", self.name(), path.display());
         #[allow(clippy::used_underscore_items)]
         self._finalize_file(source, path)
     }
@@ -118,7 +118,7 @@ pub enum ParseError {
 /// Parse the source code and visit all the nodes using tree-sitter
 #[allow(clippy::result_large_err)]
 pub fn parse(path: &PathBuf, visitors: Vec<Rc<RefCell<dyn Visitor>>>) -> Result<(), ParseError> {
-    debug!("Parsing file {:?}", path);
+    debug!("Parsing file {:?}", path.display());
     let source = std::fs::read_to_string(path).map_err(|source| ParseError::IoError {
         file: path.clone(),
         source,