ryanpeach · ryanpeach · Nov 3, 2024 · Nov 3, 2024 · Nov 3, 2024 · Nov 3, 2024
diff --git a/Cargo.toml b/Cargo.toml
@@ -11,13 +11,17 @@ debug = true
 opt-level = 3
 debug = false
 
+[build-dependencies]
+cc="*"
+
 [dependencies]
 bon = "2.3.0"
-cached = "0.53.1"
 clap = { version = "4.5.16", features = ["derive"] }
+derive_more = { version = "1.0.0", features = ["full"] }
 env_logger = "0.11.5"
 fuzzy-matcher = "0.3.7"
 getset = "0.1.3"
+hashbrown = "0.15.0"
 indicatif = "0.17.8"
 itertools = "0.13.0"
 lazy_static = "1.5.0"
@@ -29,6 +33,8 @@ serde = { version = "1.0.208", features = ["derive"] }
 serde_yaml = "0.9.34"
 thiserror = "1.0.63"
 toml = "0.8.19"
+tree-sitter = "0.23.2"
+tree-sitter-md = { version = "0.3.2", features = ["parser"] }
 walkdir = "2.5.0"
 
 [dev-dependencies]
@@ -72,3 +78,4 @@ too_many_lines = "allow"
 missing_errors_doc = "allow"
 similar_names = "allow"
 missing_panics_doc = "allow"
+module_name_repetitions = "allow"
diff --git a/Justfile b/Justfile
@@ -0,0 +1,3 @@
+# Runs tests and supresses the progress bar
+test:
+    RUNNING_TESTS=true cargo test
diff --git a/src/config.rs b/src/config.rs
@@ -7,7 +7,8 @@ use crate::{
         content::wikilink::Alias,
         name::{Filename, FilenameLowercase},
     },
-    sed::{ReplacePair, ReplacePairError},
+    rules::ErrorCode,
+    sed::{ReplacePair, ReplacePairCompilationError},
 };
 use bon::Builder;
 use clap::Parser;
@@ -17,13 +18,15 @@ use toml;
 
 /// Errors derived from config file reads
 #[derive(thiserror::Error, Debug)]
-pub enum Error {
+pub enum NewConfigError {
     #[error("The config file at {path} does not exist")]
     FileDoesNotExistError { path: PathBuf },
     #[error("Failed to read the config file")]
     FileDoesNotReadError(#[from] io::Error),
     #[error("The config file does not have expected values")]
     FileDoesNotParseError(#[from] toml::de::Error),
+    #[error("ReplacePair compilation error")]
+    ReplacePairCompilationError(#[from] ReplacePairCompilationError),
 }
 
 /// Config which contains both the cli and the config file
@@ -39,9 +42,6 @@ pub struct Config {
     /// See [`self::cli::Config::boundary_pattern`]
     #[builder(default=r"___".to_owned())]
     pub boundary_pattern: String,
-    /// See [`self::cli::Config::wikilink_pattern`]
-    #[builder(default=r"\[\[([A-Za-z0-9_/\- ]+)]]|(?:(\s|^))#([A-Za-z0-9_]+)".to_owned())]
-    pub wikilink_pattern: String,
     /// See [`self::cli::Config::filename_spacing_pattern`]
     #[builder(default=r"-|_|\s".to_owned())]
     pub filename_spacing_pattern: String,
@@ -50,13 +50,13 @@ pub struct Config {
     pub filename_match_threshold: i64,
     /// See [`self::cli::Config::exclude`]
     #[builder(default=vec![])]
-    pub exclude: Vec<String>,
+    pub exclude: Vec<ErrorCode>,
     /// See [`self::file::Config::filename_to_alias`]
-    #[builder(default=Ok(ReplacePair::new(r"___", r"/").expect("Constant")))]
-    pub filename_to_alias: Result<ReplacePair<Filename, Alias>, ReplacePairError>,
+    #[builder(default=ReplacePair::new(r"___", r"/").expect("Constant"))]
+    pub filename_to_alias: ReplacePair<Filename, Alias>,
     /// See [`self::file::Config::alias_to_filename`]
-    #[builder(default=Ok(ReplacePair::new(r"/", r"___").expect("Constant")))]
-    pub alias_to_filename: Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairError>,
+    #[builder(default=ReplacePair::new(r"/", r"___").expect("Constant"))]
+    pub alias_to_filename: ReplacePair<Alias, FilenameLowercase>,
 }
 
 /// Things which implement the partial config trait
@@ -67,33 +67,41 @@ pub trait Partial {
     fn directories(&self) -> Option<Vec<PathBuf>>;
     fn ngram_size(&self) -> Option<usize>;
     fn boundary_pattern(&self) -> Option<String>;
-    fn wikilink_pattern(&self) -> Option<String>;
     fn filename_spacing_pattern(&self) -> Option<String>;
     fn filename_match_threshold(&self) -> Option<i64>;
-    fn exclude(&self) -> Option<Vec<String>>;
-    fn filename_to_alias(&self) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairError>>;
+    fn exclude(&self) -> Option<Vec<ErrorCode>>;
+    fn filename_to_alias(
+        &self,
+    ) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairCompilationError>>;
     fn alias_to_filename(
         &self,
-    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairError>>;
+    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairCompilationError>>;
 }
 
 /// Now we implement a combine function for patrial configs which
 /// iterates over the partials and if they have a Some field they use that field in the final
 /// config.
 ///
 /// Note: This makes last elements in the input slice first priority
-fn combine_partials(partials: &[&dyn Partial]) -> Config {
-    Config::builder()
+fn combine_partials(partials: &[&dyn Partial]) -> Result<Config, ReplacePairCompilationError> {
+    Ok(Config::builder()
         .maybe_directories(partials.iter().find_map(|p| p.directories()))
         .maybe_ngram_size(partials.iter().find_map(|p| p.ngram_size()))
         .maybe_boundary_pattern(partials.iter().find_map(|p| p.boundary_pattern()))
-        .maybe_wikilink_pattern(partials.iter().find_map(|p| p.wikilink_pattern()))
         .maybe_filename_spacing_pattern(partials.iter().find_map(|p| p.filename_spacing_pattern()))
         .maybe_filename_match_threshold(partials.iter().find_map(|p| p.filename_match_threshold()))
         .maybe_exclude(partials.iter().find_map(|p| p.exclude()))
-        .maybe_filename_to_alias(partials.iter().find_map(|p| p.filename_to_alias()))
-        .maybe_alias_to_filename(partials.iter().find_map(|p| p.alias_to_filename()))
-        .build()
+        .maybe_filename_to_alias(match partials.iter().find_map(|p| p.filename_to_alias()) {
+            Some(Ok(pair)) => Some(pair),
+            Some(Err(e)) => return Err(e),
+            None => None,
+        })
+        .maybe_alias_to_filename(match partials.iter().find_map(|p| p.alias_to_filename()) {
+            Some(Ok(pair)) => Some(pair),
+            Some(Err(e)) => return Err(e),
+            None => None,
+        })
+        .build())
 }
 
 impl Config {
@@ -105,7 +113,7 @@ impl Config {
     /// - [`Error::FileDoesNotParseError`] - Config file does not parse from toml into the
     ///     expected format
     ///
-    pub fn new() -> Result<Self, Error> {
+    pub fn new() -> Result<Self, NewConfigError> {
         let cli = cli::Config::parse();
 
         // If the config file doesn't exist, and it's not the default, error out
@@ -115,12 +123,12 @@ impl Config {
                 Err(report) => Err(report)?,
             }
         } else {
-            Err(Error::FileDoesNotExistError {
+            Err(NewConfigError::FileDoesNotExistError {
                 path: cli.config_path.clone(),
             })?
         };
 
         // CLI has priority over file by being last
-        Ok(combine_partials(&[&file, &cli]))
+        combine_partials(&[&file, &cli]).map_err(derive_more::Into::into)
     }
 }
diff --git a/src/config/cli.rs b/src/config/cli.rs
@@ -6,7 +6,8 @@ use crate::{
         content::wikilink::Alias,
         name::{Filename, FilenameLowercase},
     },
-    sed::{ReplacePair, ReplacePairError},
+    rules::ErrorCode,
+    sed::{ReplacePair, ReplacePairCompilationError},
 };
 
 use super::Partial;
@@ -33,11 +34,6 @@ pub(super) struct Config {
     #[clap(short = 'b', long = "bound")]
     pub boundary_pattern: Option<String>,
 
-    /// Regex pattern for wikilinks
-    /// Capture group 0 is skipped to enable lookbehind
-    #[clap(short = 'w', long = "wikilink")]
-    pub wikilink_pattern: Option<String>,
-
     /// Regex pattern to split filenames on, like ___ or /
     #[clap(short = 's', long = "space")]
     pub filename_spacing_pattern: Option<String>,
@@ -67,29 +63,28 @@ impl Partial for Config {
     fn boundary_pattern(&self) -> Option<String> {
         self.boundary_pattern.clone()
     }
-    fn wikilink_pattern(&self) -> Option<String> {
-        self.wikilink_pattern.clone()
-    }
     fn filename_spacing_pattern(&self) -> Option<String> {
         self.filename_spacing_pattern.clone()
     }
     fn filename_match_threshold(&self) -> Option<i64> {
         self.filename_match_threshold
     }
-    fn exclude(&self) -> Option<Vec<String>> {
+    fn exclude(&self) -> Option<Vec<ErrorCode>> {
         let out = self.exclude.clone();
         if out.is_empty() {
             None
         } else {
-            Some(out)
+            Some(out.into_iter().map(ErrorCode::new).collect())
         }
     }
-    fn filename_to_alias(&self) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairError>> {
+    fn filename_to_alias(
+        &self,
+    ) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairCompilationError>> {
         None
     }
     fn alias_to_filename(
         &self,
-    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairError>> {
+    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairCompilationError>> {
         None
     }
 }
diff --git a/src/config/file.rs b/src/config/file.rs
@@ -7,10 +7,11 @@ use crate::{
         content::wikilink::Alias,
         name::{Filename, FilenameLowercase},
     },
-    sed::{ReplacePair, ReplacePairError},
+    rules::ErrorCode,
+    sed::{ReplacePair, ReplacePairCompilationError},
 };
 
-use super::{Error, Partial};
+use super::{NewConfigError, Partial};
 
 #[derive(Serialize, Deserialize, Debug, Default)]
 pub(super) struct Config {
@@ -54,9 +55,10 @@ pub(super) struct Config {
 }
 
 impl Config {
-    pub fn new(path: &Path) -> Result<Self, Error> {
-        let contents = std::fs::read_to_string(path).map_err(Error::FileDoesNotReadError)?;
-        toml::from_str(&contents).map_err(Error::FileDoesNotParseError)
+    pub fn new(path: &Path) -> Result<Self, NewConfigError> {
+        let contents =
+            std::fs::read_to_string(path).map_err(NewConfigError::FileDoesNotReadError)?;
+        toml::from_str(&contents).map_err(NewConfigError::FileDoesNotParseError)
     }
 }
 
@@ -78,10 +80,6 @@ impl Partial for Config {
         self.boundary_pattern.clone()
     }
 
-    fn wikilink_pattern(&self) -> Option<String> {
-        self.wikilink_pattern.clone()
-    }
-
     fn filename_spacing_pattern(&self) -> Option<String> {
         self.filename_spacing_pattern.clone()
     }
@@ -90,42 +88,44 @@ impl Partial for Config {
         self.filename_match_threshold
     }
 
-    fn exclude(&self) -> Option<Vec<String>> {
+    fn exclude(&self) -> Option<Vec<ErrorCode>> {
         let out = self.exclude.clone();
         if out.is_empty() {
             None
         } else {
-            Some(out)
+            Some(out.into_iter().map(ErrorCode::new).collect())
         }
     }
 
     fn alias_to_filename(
         &self,
-    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairError>> {
+    ) -> Option<Result<ReplacePair<Alias, FilenameLowercase>, ReplacePairCompilationError>> {
         let (to, from) = self.alias_to_filename.clone();
         match (to.is_empty(), from.is_empty()) {
             (true, true) => None,
             (false, false) => Some(ReplacePair::new(&to, &from)),
-            (true, false) => Some(Err(ReplacePairError::ToError(regex::Error::Syntax(
-                "To is empty".to_string(),
-            )))),
-            (false, true) => Some(Err(ReplacePairError::FromError(regex::Error::Syntax(
-                "From is empty".to_string(),
-            )))),
+            (true, false) => Some(Err(ReplacePairCompilationError::ToError(
+                regex::Error::Syntax("To is empty".to_string()),
+            ))),
+            (false, true) => Some(Err(ReplacePairCompilationError::FromError(
+                regex::Error::Syntax("From is empty".to_string()),
+            ))),
         }
     }
 
-    fn filename_to_alias(&self) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairError>> {
+    fn filename_to_alias(
+        &self,
+    ) -> Option<Result<ReplacePair<Filename, Alias>, ReplacePairCompilationError>> {
         let (to, from) = self.alias_to_filename.clone();
         match (to.is_empty(), from.is_empty()) {
             (true, true) => None,
             (false, false) => Some(ReplacePair::new(&to, &from)),
-            (true, false) => Some(Err(ReplacePairError::ToError(regex::Error::Syntax(
-                "To is empty".to_string(),
-            )))),
-            (false, true) => Some(Err(ReplacePairError::FromError(regex::Error::Syntax(
-                "From is empty".to_string(),
-            )))),
+            (true, false) => Some(Err(ReplacePairCompilationError::ToError(
+                regex::Error::Syntax("To is empty".to_string()),
+            ))),
+            (false, true) => Some(Err(ReplacePairCompilationError::FromError(
+                regex::Error::Syntax("From is empty".to_string()),
+            ))),
         }
     }
 }
diff --git a/src/file.rs b/src/file.rs
@@ -6,13 +6,11 @@ use thiserror::Error;
 
 use std;
 
-use crate::sed::RegexError;
-
 pub mod content;
 pub mod name;
 
 /// Walk the directories and get just the files
-pub fn get_files(dirs: Vec<PathBuf>) -> Vec<PathBuf> {
+pub fn get_files(dirs: &Vec<PathBuf>) -> Vec<PathBuf> {
     let mut out = Vec::new();
     for path in dirs {
         let walk = WalkDir::new(path);
@@ -24,17 +22,3 @@ pub fn get_files(dirs: Vec<PathBuf>) -> Vec<PathBuf> {
     }
     out
 }
-
-/// A bunch of bad things can happen while you're reading files,
-/// This covers most of them.
-#[derive(Debug, Error)]
-pub enum Error {
-    #[error("Error reading the file.")]
-    IoError(#[from] std::io::Error),
-    #[error("Error parsing the yaml based on expected template.")]
-    SerdeError(#[from] serde_yaml::Error),
-    #[error("Found duplicate property {0} in file contents")]
-    DuplicateProperty(String),
-    #[error("Regex error: {0}")]
-    RegexError(#[from] RegexError),
-}