feat(expr): switch to fancy-regex crate & update the original versi…

…on (#12329)

Co-authored-by: xzhseh <[email protected]>

e2e_test/batch/basic/func.slt.part

@@ -570,6 +570,30 @@ select regexp_replace('💩💩💩💩💩foo🤔️bar亲爱的😭baz这不
 ----
 💩💩💩💩💩foo🤔️bar亲爱的😭这是🥵爱情❤️‍🔥
 
+# Positive Lookahead
+query T
+select regexp_replace('foobarbaz', 'a(?=r)', 'X');
+----
+foobXrbaz
+
+# Negative Lookahead
+query T
+select regexp_replace('chocolate', 'o(?!c)', 'X');
+----
+chocXlate
+
+# Positive Lookbehind
+query T
+select regexp_replace('foobarXaz', '(?<=X)a', 'X');
+----
+foobarXXz
+
+# Negative Lookbehind
+query T
+select regexp_replace('foobarXaz', '(?<!X)a', 'X');
+----
+foobXrXaz
+
 query T
 select regexp_count('ABCABCAXYaxy', 'A.');
 ----

src/expr/Cargo.toml

@@ -28,6 +28,7 @@ ctor = "0.2"
 downcast-rs = "1.2"
 easy-ext = "1"
 either = "1"
+fancy-regex = "0.11.0"
 futures = { version = "0.3", default-features = false, features = ["alloc"] }
 futures-async-stream = { workspace = true }
 futures-util = "0.3"

src/expr/src/table_function/regexp_matches.rs

@@ -36,6 +36,7 @@ fn regexp_matches<'a>(
         // ignored in PostgreSQL's behavior.
         let skip_flag = regex.regex.captures_len() > 1;
         let list = capture
+            .unwrap()
             .iter()
             .skip(if skip_flag { 1 } else { 0 })
             .map(|mat| mat.map(|m| m.as_str().into()))

src/expr/src/vector_op/regexp.rs

@@ -16,7 +16,7 @@
 
 use std::str::FromStr;
 
-use regex::{Regex, RegexBuilder};
+use fancy_regex::{Regex, RegexBuilder};
 use risingwave_common::array::ListValue;
 use risingwave_expr_macro::function;
 
@@ -32,10 +32,17 @@ pub struct RegexpContext {
 impl RegexpContext {
     fn new(pattern: &str, flags: &str, replacement: &str) -> Result<Self> {
         let options = RegexpOptions::from_str(flags)?;
+
+        let origin = if options.case_insensitive {
+            format!("(?i:{})", pattern)
+        } else {
+            pattern.to_string()
+        };
+
         Ok(Self {
-            regex: RegexBuilder::new(pattern)
-                .case_insensitive(options.case_insensitive)
-                .build()?,
+            regex: RegexBuilder::new(&origin)
+                .build()
+                .map_err(|e| ExprError::Parse(e.to_string().into()))?,
             global: options.global,
             replacement: make_replacement(replacement),
         })
@@ -142,7 +149,7 @@ fn regexp_match(text: &str, regex: &RegexpContext) -> Option<ListValue> {
     // If there are multiple captures, then the first one is the whole match, and should be
     // ignored in PostgreSQL's behavior.
     let skip_first = regex.regex.captures_len() > 1;
-    let capture = regex.regex.captures(text)?;
+    let capture = regex.regex.captures(text).unwrap()?;
     let list = capture
         .iter()
         .skip(if skip_first { 1 } else { 0 })
@@ -190,7 +197,7 @@ fn regexp_count(text: &str, start: i32, regex: &RegexpContext) -> Result<i32> {
     };
 
     let mut count = 0;
-    while let Some(captures) = regex.regex.captures(&text[start..]) {
+    while let Ok(Some(captures)) = regex.regex.captures(&text[start..]) {
         count += 1;
         start += captures.get(0).unwrap().end();
     }
@@ -297,7 +304,7 @@ fn regexp_replace(
             let mut ret = text[..search_start].to_string();
 
             // Begin the actual replace logic
-            while let Some(capture) = ctx.regex.captures(&text[search_start..]) {
+            while let Ok(Some(capture)) = ctx.regex.captures(&text[search_start..]) {
                 let match_start = capture.get(0).unwrap().start();
                 let match_end = capture.get(0).unwrap().end();
 
@@ -344,7 +351,7 @@ fn regexp_replace(
                 let mut count = 1;
                 // The absolute index for the start of searching
                 let mut search_start = start;
-                while let Some(capture) = ctx.regex.captures(&text[search_start..]) {
+                while let Ok(Some(capture)) = ctx.regex.captures(&text[search_start..]) {
                     // Get the current start & end index
                     let match_start = capture.get(0).unwrap().start();
                     let match_end = capture.get(0).unwrap().end();
@@ -378,7 +385,7 @@ fn regexp_replace(
             if let Some(n) = n {
                 // Replace only the N-th match
                 let mut count = 1;
-                while let Some(capture) = ctx.regex.captures(&text[start..]) {
+                while let Ok(Some(capture)) = ctx.regex.captures(&text[start..]) {
                     if count == n {
                         // We've reached the pattern to replace
                         let match_start = capture.get(0).unwrap().start();
@@ -406,12 +413,13 @@ fn regexp_replace(
                 }
             } else {
                 // `N` is not specified
-                if ctx.regex.captures(&text[start..]).is_none() {
+                if let Ok(None) = ctx.regex.captures(&text[start..]) {
                     // No match
                     return Ok(text.into());
                 }
+
                 // Otherwise replace the source text
-                if let Some(capture) = ctx.regex.captures(&text[start..]) {
+                if let Ok(Some(capture)) = ctx.regex.captures(&text[start..]) {
                     let match_start = capture.get(0).unwrap().start();
                     let match_end = capture.get(0).unwrap().end();
 

src/workspace-hack/Cargo.toml

@@ -24,6 +24,7 @@ aws-credential-types = { version = "0.55", default-features = false, features =
 aws-sdk-s3 = { version = "0.28", features = ["native-tls"] }
 aws-smithy-client = { version = "0.55", default-features = false, features = ["native-tls", "rustls"] }
 base64 = { version = "0.21", features = ["alloc"] }
+bit-vec = { version = "0.6" }
 bitflags = { version = "2", default-features = false, features = ["std"] }
 byteorder = { version = "1", features = ["i128"] }
 bytes = { version = "1", features = ["serde"] }
@@ -121,6 +122,7 @@ aws-credential-types = { version = "0.55", default-features = false, features =
 aws-sdk-s3 = { version = "0.28", features = ["native-tls"] }
 aws-smithy-client = { version = "0.55", default-features = false, features = ["native-tls", "rustls"] }
 base64 = { version = "0.21", features = ["alloc"] }
+bit-vec = { version = "0.6" }
 bitflags = { version = "2", default-features = false, features = ["std"] }
 byteorder = { version = "1", features = ["i128"] }
 bytes = { version = "1", features = ["serde"] }