Merge pull request #913 from sabbellasri/regex-fancy-fallback

Add fancy regex fallback for advanced filters

Author: dalance

Cargo.lock

@@ -54,6 +54,7 @@ tokio         = { version = "1.52", optional = true, features = ["rt"] }
 toml          = "1.1"
 unicode-width = "0.2"
 regex         = "1.12"
+fancy-regex   = "0.16"
 
 [build-dependencies]
 anyhow        = "1.0"

Cargo.toml

@@ -3,6 +3,7 @@ mod columns;
 mod config;
 mod opt;
 mod process;
+mod search_regex;
 mod style;
 mod term_info;
 mod util;

src/main.rs

@@ -0,0 +1,60 @@
+use anyhow::Error;
+
+pub enum SearchRegex {
+    Fast(regex::Regex),
+    Fancy(fancy_regex::Regex),
+}
+
+impl SearchRegex {
+    pub fn new(pattern: &str, ignore_case: bool) -> Result<Self, Error> {
+        if let Ok(regex) = regex::RegexBuilder::new(pattern)
+            .case_insensitive(ignore_case)
+            .build()
+        {
+            return Ok(Self::Fast(regex));
+        }
+
+        let regex = fancy_regex::RegexBuilder::new(pattern)
+            .case_insensitive(ignore_case)
+            .build()?;
+        Ok(Self::Fancy(regex))
+    }
+
+    pub fn is_match(&self, text: &str) -> Result<bool, Error> {
+        match self {
+            Self::Fast(regex) => Ok(regex.is_match(text)),
+            Self::Fancy(regex) => Ok(regex.is_match(text)?),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::SearchRegex;
+
+    #[test]
+    fn uses_fast_engine_for_standard_regex() {
+        let regex = SearchRegex::new("proc.*", false).unwrap();
+
+        assert!(matches!(regex, SearchRegex::Fast(_)));
+        assert!(regex.is_match("procs").unwrap());
+    }
+
+    #[test]
+    fn falls_back_to_fancy_engine_for_lookahead() {
+        let regex = SearchRegex::new("proc(?=s)", false).unwrap();
+
+        assert!(matches!(regex, SearchRegex::Fancy(_)));
+        assert!(regex.is_match("procs").unwrap());
+        assert!(!regex.is_match("procx").unwrap());
+    }
+
+    #[test]
+    fn falls_back_to_fancy_engine_for_negative_lookahead() {
+        let regex = SearchRegex::new("proc(?!x)", false).unwrap();
+
+        assert!(matches!(regex, SearchRegex::Fancy(_)));
+        assert!(regex.is_match("procs").unwrap());
+        assert!(!regex.is_match("procx").unwrap());
+    }
+}

src/search_regex.rs

@@ -4,6 +4,7 @@ use crate::columns::*;
 use crate::config::*;
 use crate::opt::{ArgColorMode, ArgPagerMode};
 use crate::process::collect_proc;
+use crate::search_regex::SearchRegex;
 use crate::style::{apply_color, apply_style, color_to_column_style};
 use crate::term_info::TermInfo;
 use crate::util::{
@@ -13,7 +14,6 @@ use crate::util::{
 use anyhow::{Error, bail};
 #[cfg(not(target_os = "windows"))]
 use pager::Pager;
-use regex::RegexBuilder;
 use std::collections::HashMap;
 use std::time::Duration;
 
@@ -52,10 +52,8 @@ impl View {
 
         // Adding the sort column to inserts if not already present
         match (&opt.sorta, &opt.sortd) {
-            (_, Some(col)) | (Some(col), _) => {
-                if !opt.insert.contains(col) {
-                    opt.insert.push(col.clone());
-                }
+            (_, Some(col)) | (Some(col), _) if !opt.insert.contains(col) => {
+                opt.insert.push(col.clone());
             }
             _ => {}
         }
@@ -263,9 +261,7 @@ impl View {
                 ConfigSearchCase::Insensitive => true,
                 ConfigSearchCase::Sensitive => false,
             };
-            let regex = RegexBuilder::new(pattern)
-                .case_insensitive(ignore_case)
-                .build()?;
+            let regex = SearchRegex::new(pattern, ignore_case)?;
             Some(regex)
         } else {
             None
@@ -316,7 +312,7 @@ impl View {
             } else if opt.keyword.is_empty() {
                 true
             } else if let Some(regex) = &regex {
-                View::search_regex(*pid, cols_searchable.as_slice(), regex)
+                View::search_regex(*pid, cols_searchable.as_slice(), regex)?
             } else {
                 View::search(
                     *pid,
@@ -714,8 +710,13 @@ impl View {
         }
     }
 
-    fn search_regex(pid: i32, cols: &[&dyn Column], regex: &regex::Regex) -> bool {
-        cols.iter().any(|c| regex.is_match(&c.display_json(pid)))
+    fn search_regex(pid: i32, cols: &[&dyn Column], regex: &SearchRegex) -> Result<bool, Error> {
+        for c in cols {
+            if regex.is_match(&c.display_json(pid))? {
+                return Ok(true);
+            }
+        }
+        Ok(false)
     }
 
     #[cfg(not(any(target_os = "windows", any(target_os = "linux", target_os = "android"))))]

src/view.rs

@@ -1,12 +1,12 @@
 use crate::Opt;
 use crate::config::*;
+use crate::search_regex::SearchRegex;
 use crate::term_info::TermInfo;
 use crate::util::{get_theme, has_regex_syntax};
 use crate::view::View;
 use anyhow::Error;
 use chrono::offset::Local;
 use getch::Getch;
-use regex::RegexBuilder;
 use std::collections::HashMap;
 use std::sync::mpsc::{Receiver, Sender, channel};
 use std::thread;
@@ -221,10 +221,7 @@ impl Watcher {
                                                 ConfigSearchCase::Insensitive => true,
                                                 ConfigSearchCase::Sensitive => false,
                                             };
-                                            match RegexBuilder::new(&candidate)
-                                                .case_insensitive(ignore_case)
-                                                .build()
-                                            {
+                                            match SearchRegex::new(&candidate, ignore_case) {
                                                 Ok(_) => {
                                                     opt.keyword = vec![candidate];
                                                     regex_error = None;