Added more progress bars

Cargo.toml

@@ -12,13 +12,12 @@ opt-level = 3
 debug = false
 
 [dependencies]
+aho-corasick = "1.1.3"
 bon = "2.3.0"
-cached = "0.54.0"
 clap = { version = "4.5.16", features = ["derive"] }
 comrak = "0.29.0"
 derive_more = { version = "1.0.0", features = ["full"] }
 env_logger = "0.11.5"
-fancy-regex = "0.14.0"
 fuzzy-matcher = "0.3.7"
 getset = "0.1.3"
 git2 = "0.19.0"

src/lib.rs

@@ -7,9 +7,10 @@ pub mod rules;
 pub mod sed;
 pub mod visitor;
 
-use std::{backtrace::Backtrace, cell::RefCell, rc::Rc};
+use std::{backtrace::Backtrace, cell::RefCell, env, rc::Rc};
 
 use file::{get_files, name::ngrams};
+use indicatif::ProgressBar;
 use log::info;
 use miette::{Diagnostic, Result};
 use ngrams::MissingSubstringError;
@@ -179,7 +180,7 @@ fn check(config: &config::Config) -> Result<OutputReport, OutputErrors> {
 
     let mut reports: Vec<Report> = vec![];
 
-    // First pass
+    // Filename pass
     // Just over filenames
     // NOTE: Always use `filter_by_excludes` and `dedupe_by_code` on the reports
     let similar_filenames = SimilarFilename::calculate(
@@ -194,24 +195,44 @@ fn check(config: &config::Config) -> Result<OutputReport, OutputErrors> {
             .map(|x| Report::SimilarFilename(x.clone())),
     );
 
-    // Second pass
+    // First pass
     // This gives us metadata we need for all other rules from the content of files
     //  The duplicate alias visitor has to run first to get the table of aliases
+    let first_pass_bar: Option<ProgressBar> = if env::var("RUNNING_TESTS").is_ok() {
+        None
+    } else {
+        #[allow(clippy::cast_sign_loss)]
+        #[allow(clippy::cast_possible_truncation)]
+        Some(ProgressBar::new(all_files.len() as u64).with_prefix("First Pass"))
+    };
     let duplicate_alias_visitor = Rc::new(RefCell::new(DuplicateAliasVisitor::new(
         &all_files,
         &config.filename_to_alias,
     )));
     for file in &all_files {
         let visitors: Vec<Rc<RefCell<dyn Visitor>>> = vec![duplicate_alias_visitor.clone()];
         parse(file, visitors)?;
+        if let Some(bar) = &first_pass_bar {
+            bar.inc(1);
+        }
     }
     let mut duplicate_alias_visitor: DuplicateAliasVisitor =
         Rc::try_unwrap(duplicate_alias_visitor)
             .expect("parse is done")
             .into_inner();
     reports.extend(duplicate_alias_visitor.finalize(&config.exclude)?);
+    if let Some(bar) = &first_pass_bar {
+        bar.finish();
+    }
 
-    // Third Pass
+    // Second Pass
+    let second_pass_bar: Option<ProgressBar> = if env::var("RUNNING_TESTS").is_ok() {
+        None
+    } else {
+        #[allow(clippy::cast_sign_loss)]
+        #[allow(clippy::cast_possible_truncation)]
+        Some(ProgressBar::new(all_files.len() as u64).with_prefix("Second Pass"))
+    };
     let mut visitors: Vec<Rc<RefCell<dyn Visitor>>> = vec![];
     for rule in ThirdPassRule::iter() {
         visitors.push(match rule {
@@ -232,12 +253,18 @@ fn check(config: &config::Config) -> Result<OutputReport, OutputErrors> {
 
     for file in &all_files {
         parse(file, visitors.clone())?;
+        if let Some(bar) = &second_pass_bar {
+            bar.inc(1);
+        }
     }
 
     for visitor in visitors {
         let mut visitor_cell = (*visitor).borrow_mut();
         reports.extend(visitor_cell.finalize(&config.exclude)?);
     }
+    if let Some(bar) = &second_pass_bar {
+        bar.finish();
+    }
 
     Ok(OutputReport { reports })
 }

src/rules/similar_filename.rs

@@ -154,7 +154,7 @@ impl SimilarFilename {
         } else {
             #[allow(clippy::cast_sign_loss)]
             #[allow(clippy::cast_possible_truncation)]
-            Some(ProgressBar::new((n * (n + 1.0) / 2.0) as u64))
+            Some(ProgressBar::new((n * (n + 1.0) / 2.0) as u64).with_prefix("Crosschecking files"))
         };
         let matcher = SkimMatcherV2::default();
         let mut matches: Vec<SimilarFilename> = Vec::new();

src/rules/unlinked_text.rs

@@ -10,13 +10,12 @@ use crate::{
     sed::ReplacePair,
     visitor::{FinalizeError, VisitError, Visitor},
 };
+use aho_corasick::AhoCorasick;
 use bon::Builder;
-use cached::proc_macro::cached;
 use comrak::{
     arena_tree::Node,
     nodes::{Ast, NodeValue},
 };
-use fancy_regex::Regex;
 use hashbrown::HashMap;
 use miette::{Diagnostic, NamedSource, Result, SourceOffset, SourceSpan};
 use std::{
@@ -111,11 +110,45 @@ impl UnlinkedTextVisitor {
     }
 }
 
-#[cached]
-fn get_regex(alias: Alias) -> Regex {
-    // Compile the regex and cache it based on the alias
-    let pattern = format!(r"(?i)(?<![\w#]){alias}(?!\w)");
-    Regex::new(&pattern).expect("The regex is just case insensitive string search")
+/// Checks if the match at the given start and end indices is a whole word match.
+fn is_whole_word_match(text: &str, start: usize, end: usize) -> bool {
+    is_start_boundary(text, start) && is_end_boundary(text, end) && !is_start_hashtag(text, start)
+}
+
+/// Checks if the character before the start index is a word boundary.
+fn is_start_boundary(text: &str, start: usize) -> bool {
+    if start == 0 {
+        true
+    } else {
+        text[..start]
+            .chars()
+            .next_back()
+            .map_or(true, |c| !c.is_alphanumeric())
+    }
+}
+
+/// Checks if the character before the start index is a word boundary.
+fn is_start_hashtag(text: &str, start: usize) -> bool {
+    if start == 0 {
+        false
+    } else {
+        text[..start]
+            .chars()
+            .next_back()
+            .map_or(false, |c| c == '#')
+    }
+}
+
+/// Checks if the character after the end index is a word boundary.
+fn is_end_boundary(text: &str, end: usize) -> bool {
+    if end == text.len() {
+        true
+    } else {
+        text[end..]
+            .chars()
+            .next()
+            .map_or(true, |c| !c.is_alphanumeric())
+    }
 }
 
 impl Visitor for UnlinkedTextVisitor {
@@ -129,40 +162,47 @@ impl Visitor for UnlinkedTextVisitor {
         let sourcepos = data_ref.sourcepos;
         let parent = node.parent();
         if let NodeValue::Text(text) = data {
-            for alias in self.alias_table.keys() {
-                // Make sure neither the character before or after is a letter
-                // This makes sure you aren't matching a part of a word
-                // This should also handle tags
-                // Check the character before the match
-
-                let re = get_regex(alias.clone());
-                if let Ok(Some(found)) = re.find(text) {
-                    // Get our span
-                    let span = repair_span_due_to_frontmatter(
-                        SourceSpan::new(
-                            (SourceOffset::from_location(
-                                remove_frontmatter_from_source(source, node),
-                                sourcepos.start.line,
-                                sourcepos.start.column,
-                            )
-                            .offset()
-                                + found.start())
-                            .into(),
-                            alias.to_string().len(),
-                        ),
-                        node,
-                    );
-
-                    // Dont match inside wikilinks
-                    if let Some(parent) = parent {
-                        if let NodeValue::WikiLink(_) = parent.data.borrow().value {
-                            // If this is already in a link, skip it
-                            continue;
-                        }
+            let patterns: Vec<String> = self
+                .alias_table
+                .keys()
+                .map(std::string::ToString::to_string)
+                .collect();
+            let ac = AhoCorasick::builder()
+                .ascii_case_insensitive(true)
+                .build(&patterns)?;
+            // Make sure neither the character before or after is a letter
+            // This makes sure you aren't matching a part of a word
+            // This should also handle tags
+            // Check the character before the match
+            for found in ac.find_iter(text) {
+                if !is_whole_word_match(text, found.start(), found.end()) {
+                    continue;
+                }
+                let alias = Alias::new(&patterns[found.pattern().as_usize()]);
+                let span = repair_span_due_to_frontmatter(
+                    SourceSpan::new(
+                        (SourceOffset::from_location(
+                            remove_frontmatter_from_source(source, node),
+                            sourcepos.start.line,
+                            sourcepos.start.column,
+                        )
+                        .offset()
+                            + found.start())
+                        .into(),
+                        found.end() - found.start(),
+                    ),
+                    node,
+                );
+
+                // Dont match inside wikilinks
+                if let Some(parent) = parent {
+                    if let NodeValue::WikiLink(_) = parent.data.borrow().value {
+                        // If this is already in a link, skip it
+                        continue;
                     }
-
-                    self.new_unlinked_texts.push((alias.clone(), span));
                 }
+
+                self.new_unlinked_texts.push((alias, span));
             }
         }
         Ok(())

src/visitor.rs

@@ -22,6 +22,14 @@ pub enum VisitError {
         #[backtrace]
         backtrace: std::backtrace::Backtrace,
     },
+
+    #[error("Error making patterns from aliases")]
+    AhoBuildError {
+        #[from]
+        source: aho_corasick::BuildError,
+        #[backtrace]
+        backtrace: std::backtrace::Backtrace,
+    },
 }
 
 #[derive(Error, Debug)]