Implement regexp extension

COMPAT.md

@@ -500,3 +500,13 @@ UUID's in Limbo are `blobs` by default.
 | uuid7_timestamp_ms(X) | Yes    | Convert a UUID v7 to milliseconds since epoch                 |
 | uuid_str(X)           | Yes    | Convert a valid UUID to string                                |
 | uuid_blob(X)          | Yes    | Convert a valid UUID to blob                                  |
+
+### REGEXP
+
+| Function                                       | Status | Comment | 
+|------------------------------------------------|--------|---------|
+| regexp(pattern, source)                        | Yes    |         |
+| regexp_like(source, pattern)                   | Yes    |         |
+| regexp_substr(source, pattern)                 | Yes    |         |
+| regexp_capture(source, pattern[, n])           | No     |         |
+| regexp_replace(source, pattern, replacement)   | No     |         |

Cargo.lock

@@ -1274,6 +1274,15 @@ dependencies = [
  "syn 2.0.96",
 ]
 
+[[package]]
+name = "limbo_regexp"
+version = "0.0.12"
+dependencies = [
+ "limbo_ext",
+ "log",
+ "regex",
+]
+
 [[package]]
 name = "limbo_sim"
 version = "0.0.12"

Cargo.toml

@@ -11,6 +11,7 @@ members = [
     "core",
     "extensions/core",
     "extensions/uuid",
+    "extensions/regexp",
     "macros",
     "simulator",
     "sqlite3",

extensions/regexp/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+name = "limbo_regexp"
+version.workspace = true
+authors.workspace = true
+edition.workspace = true
+license.workspace = true
+repository.workspace = true
+
+[lib]
+crate-type = ["cdylib", "lib"]
+
+
+[dependencies]
+limbo_ext = { path = "../core"}
+regex = "1.11.1"
+log = "0.4.20"

extensions/regexp/src/lib.rs

@@ -0,0 +1,65 @@
+use limbo_ext::{export_scalar, register_extension, register_scalar_functions, Value, ValueType};
+use regex::Regex;
+
+register_extension! {
+    scalars: {
+        "regexp" => regexp,
+        "regexp_like" => regexp_like,
+        "regexp_substr" => regexp_substr,
+    },
+}
+
+#[export_scalar]
+#[args(2)]
+fn regexp(args: &[Value]) -> Value {
+    regex(&args[0], &args[1])
+}
+
+#[export_scalar]
+#[args(2)]
+fn regexp_like(args: &[Value]) -> Value {
+    regex(&args[1], &args[0])
+}
+
+fn regex(pattern: &Value, haystack: &Value) -> Value {
+    match (pattern.value_type(), haystack.value_type()) {
+        (ValueType::Text, ValueType::Text) => {
+            let Some(pattern) = pattern.to_text() else {
+                return Value::null();
+            };
+            let Some(haystack) = haystack.to_text() else {
+                return Value::null();
+            };
+            let re = match Regex::new(&pattern) {
+                Ok(re) => re,
+                Err(_) => return Value::null(),
+            };
+            Value::from_integer(re.is_match(&haystack) as i64)
+        }
+        _ => Value::null(),
+    }
+}
+
+#[export_scalar]
+#[args(2)]
+fn regexp_substr(args: &[Value]) -> Value {
+    return match (args[0].value_type(), args[1].value_type()) {
+        (ValueType::Text, ValueType::Text) => {
+            let Some(haystack) = &args[0].to_text() else {
+                return Value::null();
+            };
+            let Some(pattern) = &args[1].to_text() else {
+                return Value::null();
+            };
+            let re = match Regex::new(pattern) {
+                Ok(re) => re,
+                Err(_) => return Value::null(),
+            };
+            match re.find(haystack) {
+                Some(mat) => Value::from_text(mat.as_str().to_string()),
+                None => Value::null(),
+            }
+        }
+        _ => Value::null(),
+    };
+}

testing/extensions.py

@@ -77,6 +77,11 @@ def run_test(pipe, sql, validator=None):
             raise Exception("Validation failed")
     print("Test PASSED")
 
+def validate_true(result):
+    return result == "1"
+
+def validate_false(result):
+    return result == "0"
 
 def validate_blob(result):
     # HACK: blobs are difficult to test because the shell
@@ -100,33 +105,54 @@ def assert_now_unixtime(result):
 def assert_specific_time(result):
     return result == "1736720789"
 
-
-def main():
+def test_uuid(pipe):
     specific_time = "01945ca0-3189-76c0-9a8f-caf310fc8b8e"
     extension_path = "./target/debug/liblimbo_uuid.so"
+
+    # before extension loads, assert no function
+    run_test(pipe, "SELECT uuid4();", returns_null)
+    run_test(pipe, "SELECT uuid4_str();", returns_null)
+    run_test(pipe, f".load {extension_path}", returns_null)
+    print(f"Extension {extension_path} loaded successfully.")
+    run_test(pipe, "SELECT hex(uuid4());", validate_blob)
+    run_test(pipe, "SELECT uuid4_str();", validate_string_uuid)
+    run_test(pipe, "SELECT hex(uuid7());", validate_blob)
+    run_test(
+        pipe,
+        "SELECT uuid7_timestamp_ms(uuid7()) / 1000;",
+    )
+    run_test(pipe, "SELECT uuid7_str();", validate_string_uuid)
+    run_test(pipe, "SELECT uuid_str(uuid7());", validate_string_uuid)
+    run_test(pipe, "SELECT hex(uuid_blob(uuid7_str()));", validate_blob)
+    run_test(pipe, "SELECT uuid_str(uuid_blob(uuid7_str()));", validate_string_uuid)
+    run_test(
+        pipe,
+        f"SELECT uuid7_timestamp_ms('{specific_time}') / 1000;",
+        assert_specific_time,
+    )
+
+def test_regexp(pipe):
+    extension_path = "./target/debug/liblimbo_regexp.so"
+
+    # before extension loads, assert no function
+    run_test(pipe, "SELECT regexp('a.c', 'abc');", returns_null)
+    run_test(pipe, f".load {extension_path}", returns_null)
+    print(f"Extension {extension_path} loaded successfully.")
+    run_test(pipe, "SELECT regexp('a.c', 'abc');", validate_true)
+    run_test(pipe, "SELECT regexp('a.c', 'ac');", validate_false)
+    run_test(pipe, "SELECT regexp('[0-9]+', 'the year is 2021');", validate_true)
+    run_test(pipe, "SELECT regexp('[0-9]+', 'the year is unknow');", validate_false)
+    run_test(pipe, "SELECT regexp_like('the year is 2021', '[0-9]+');", validate_true)
+    run_test(pipe, "SELECT regexp_like('the year is unknow', '[0-9]+');", validate_false)
+    run_test(pipe, "SELECT regexp_substr('the year is 2021', '[0-9]+') = '2021';", validate_true)
+    run_test(pipe, "SELECT regexp_substr('the year is unknow', '[0-9]+');", returns_null)
+
+    
+def main():
     pipe = init_limbo()
     try:
-        # before extension loads, assert no function
-        run_test(pipe, "SELECT uuid4();", returns_null)
-        run_test(pipe, "SELECT uuid4_str();", returns_null)
-        run_test(pipe, f".load {extension_path}", returns_null)
-        print("Extension loaded successfully.")
-        run_test(pipe, "SELECT hex(uuid4());", validate_blob)
-        run_test(pipe, "SELECT uuid4_str();", validate_string_uuid)
-        run_test(pipe, "SELECT hex(uuid7());", validate_blob)
-        run_test(
-            pipe,
-            "SELECT uuid7_timestamp_ms(uuid7()) / 1000;",
-        )
-        run_test(pipe, "SELECT uuid7_str();", validate_string_uuid)
-        run_test(pipe, "SELECT uuid_str(uuid7());", validate_string_uuid)
-        run_test(pipe, "SELECT hex(uuid_blob(uuid7_str()));", validate_blob)
-        run_test(pipe, "SELECT uuid_str(uuid_blob(uuid7_str()));", validate_string_uuid)
-        run_test(
-            pipe,
-            f"SELECT uuid7_timestamp_ms('{specific_time}') / 1000;",
-            assert_specific_time,
-        )
+        test_regexp(pipe)
+        test_uuid(pipe)
     except Exception as e:
         print(f"Test FAILED: {e}")
         pipe.terminate()