feat: added support for fancy regex (#292)

Author: ErikBjare

Cargo.lock

@@ -12,7 +12,7 @@ serde_json = "1.0"
 chrono = { version = "0.4", features = ["serde"] }
 plex = "0.2"
 log = "0.4"
-regex = "1.5"
+fancy-regex = "0.10.0"
 aw-datastore = { path = "../aw-datastore" }
 aw-models = { path = "../aw-models" }
 aw-transform = { path = "../aw-transform" }

aw-query/Cargo.toml

@@ -444,7 +444,7 @@ mod qfunctions {
         Ok(DataType::List(filtered_tagged_events))
     }
 
-    use regex::RegexBuilder;
+    use fancy_regex::RegexBuilder;
 
     pub fn filter_keyvals_regex(
         args: Vec<DataType>,

aw-query/src/functions.rs

@@ -10,7 +10,7 @@ edition = "2021"
 serde = "1.0"
 rocket = "0.5.0-rc.1" # TODO: Remove this dependency
 serde_json = "1.0"
-regex = "1.5"
+fancy-regex = "0.10.0"
 log = "0.4"
 chrono = { version = "0.4", features = ["serde"] }
 aw-models = { path = "../aw-models" }

aw-transform/Cargo.toml

@@ -2,7 +2,7 @@
 ///
 /// Based on code in aw_research: https://github.com/ActivityWatch/aw-research/blob/master/aw_research/classify.py
 use aw_models::Event;
-use regex::{Regex, RegexBuilder};
+use fancy_regex::Regex;
 
 pub enum Rule {
     None,
@@ -27,10 +27,16 @@ pub struct RegexRule {
 }
 
 impl RegexRule {
-    pub fn new(regex_str: &str, ignore_case: bool) -> Result<RegexRule, regex::Error> {
-        let mut regex_builder = RegexBuilder::new(regex_str);
-        regex_builder.case_insensitive(ignore_case);
-        let regex = regex_builder.build()?;
+    pub fn new(regex_str: &str, ignore_case: bool) -> Result<RegexRule, fancy_regex::Error> {
+        // can't use `RegexBuilder::case_insensitive` because it's not supported by fancy_regex,
+        // so we need to prefix with `(?i)` to make it case insensitive.
+        let regex = if ignore_case {
+            let regex_str = format!("(?i){}", regex_str);
+            Regex::new(&regex_str)?
+        } else {
+            Regex::new(regex_str)?
+        };
+
         Ok(RegexRule { regex })
     }
 }
@@ -46,7 +52,7 @@ impl RuleTrait for RegexRule {
             .data
             .values()
             .filter(|val| val.is_string())
-            .any(|val| self.regex.is_match(val.as_str().unwrap()))
+            .any(|val| self.regex.is_match(val.as_str().unwrap()).unwrap())
     }
 }
 
@@ -139,6 +145,18 @@ fn test_rule() {
     assert_eq!(rule_none.matches(&e_match), false);
 }
 
+#[test]
+fn test_rule_lookahead() {
+    // Originally requested by a user here, to match aw-server-python: https://canary.discord.com/channels/755040852727955476/755334543891759194/994291987878522961
+    let mut e_match = Event::default();
+    e_match
+        .data
+        .insert("test".into(), serde_json::json!("testing lookahead"));
+
+    let rule_from_regex = Rule::from(Regex::new("testing (?!lookahead)").unwrap());
+    assert_eq!(rule_from_regex.matches(&e_match), false);
+}
+
 #[test]
 fn test_categorize() {
     let mut e = Event::default();

aw-transform/src/classify.rs

@@ -1,4 +1,4 @@
-use regex::Regex;
+use fancy_regex::Regex;
 use serde_json::value::Value;
 
 use aw_models::Event;
@@ -42,7 +42,7 @@ pub fn filter_keyvals_regex(mut events: Vec<Event>, key: &str, regex: &Regex) ->
 
     for event in events.drain(..) {
         if let Some(v) = event.data.get(key) {
-            if regex.is_match(v.as_str().unwrap()) {
+            if regex.is_match(v.as_str().unwrap()).unwrap() {
                 filtered_events.push(event);
             }
         }
@@ -78,9 +78,8 @@ pub fn exclude_keyvals(mut events: Vec<Event>, key: &str, vals: &[Value]) -> Vec
 mod tests {
     use std::str::FromStr;
 
-    use chrono::DateTime;
-    use chrono::Duration;
-    use regex::RegexBuilder;
+    use chrono::{DateTime, Duration};
+    use fancy_regex::RegexBuilder;
     use serde_json::json;
 
     use aw_models::Event;