implemented regexp_matches function

fixed linting issues

fixed check style issues

Fixed some names

Remove extra whitespace

Add substr short-ciruit for regexp_like

Refactor to share logic between like and rlike

Fix style issues

Fix segfault

Fix SqlLikePatternToPcre to use partial matching

Fix style issues and warning

Fix formatting; ran the docker image and used clang-format

cpp/src/gandiva/expr_decomposer.cc

@@ -53,7 +53,10 @@ Status ExprDecomposer::Visit(const FieldNode& node) {
 // time.
 const FunctionNode ExprDecomposer::TryOptimize(const FunctionNode& node) {
  if (node.descriptor()->name() == "like") {
- return LikeHolder::TryOptimize(node);
+ return SQLLikeHolder::TryOptimize(node);
+ } else if (node.descriptor()->name() == "regexp_matches" ||
+ node.descriptor()->name() == "regexp_like") {
+ return RegexpMatchesHolder::TryOptimize(node);
  } else {
  return node;
  }

cpp/src/gandiva/function_holder_registry.h

@@ -23,7 +23,6 @@
 #include <unordered_map>
 
 #include "arrow/status.h"
-
 #include "gandiva/function_holder.h"
 #include "gandiva/like_holder.h"
 #include "gandiva/node.h"
@@ -61,8 +60,10 @@ class FunctionHolderRegistry {
  private:
  static map_type& makers() {
  static map_type maker_map = {
- {"like", LAMBDA_MAKER(LikeHolder)},
- {"ilike", LAMBDA_MAKER(LikeHolder)},
+ {"like", LAMBDA_MAKER(SQLLikeHolder)},
+ {"ilike", LAMBDA_MAKER(SQLLikeHolder)},
+ {"regexp_matches", LAMBDA_MAKER(RegexpMatchesHolder)},
+ {"regexp_like", LAMBDA_MAKER(RegexpMatchesHolder)},
  {"to_date", LAMBDA_MAKER(ToDateHolder)},
  {"random", LAMBDA_MAKER(RandomGeneratorHolder)},
  {"rand", LAMBDA_MAKER(RandomGeneratorHolder)},

cpp/src/gandiva/function_registry_string.cc

@@ -146,6 +146,10 @@ std::vector<NativeFunction> GetStringFunctionRegistry() {
  kResultNullIfNull, "gdv_fn_ilike_utf8_utf8",
  NativeFunction::kNeedsFunctionHolder),
 
+ NativeFunction("regexp_matches", {"regexp_like"}, DataTypeVector{utf8(), utf8()},
+ boolean(), kResultNullIfNull, "gdv_fn_regexp_matches_utf8_utf8",
+ NativeFunction::kNeedsFunctionHolder),
+
  NativeFunction("ltrim", {}, DataTypeVector{utf8(), utf8()}, utf8(),
  kResultNullIfNull, "ltrim_utf8_utf8", NativeFunction::kNeedsContext),
 

cpp/src/gandiva/gdv_function_stubs.cc

@@ -42,21 +42,25 @@ extern "C" {
 
 bool gdv_fn_like_utf8_utf8(int64_t ptr, const char* data, int data_len,
  const char* pattern, int pattern_len) {
- gandiva::LikeHolder* holder = reinterpret_cast<gandiva::LikeHolder*>(ptr);
+ gandiva::RegexpMatchesHolder* holder =
+ reinterpret_cast<gandiva::RegexpMatchesHolder*>(ptr);
  return (*holder)(std::string(data, data_len));
 }
 
 bool gdv_fn_like_utf8_utf8_utf8(int64_t ptr, const char* data, int data_len,
  const char* pattern, int pattern_len,
  const char* escape_char, int escape_char_len) {
- gandiva::LikeHolder* holder = reinterpret_cast<gandiva::LikeHolder*>(ptr);
- return (*holder)(std::string(data, data_len));
+ return gdv_fn_like_utf8_utf8(ptr, data, data_len, pattern, pattern_len);
 }
 
 bool gdv_fn_ilike_utf8_utf8(int64_t ptr, const char* data, int data_len,
  const char* pattern, int pattern_len) {
- gandiva::LikeHolder* holder = reinterpret_cast<gandiva::LikeHolder*>(ptr);
- return (*holder)(std::string(data, data_len));
+ return gdv_fn_like_utf8_utf8(ptr, data, data_len, pattern, pattern_len);
+}
+
+bool gdv_fn_regexp_matches_utf8_utf8(int64_t ptr, const char* data, int data_len,
+ const char* pattern, int pattern_len) {
+ return gdv_fn_like_utf8_utf8(ptr, data, data_len, pattern, pattern_len);
 }
 
 double gdv_fn_random(int64_t ptr) {
@@ -884,6 +888,17 @@ void ExportedStubFunctions::AddMappings(Engine* engine) const {
  types->i1_type() /*return_type*/, args,
  reinterpret_cast<void*>(gdv_fn_ilike_utf8_utf8));
 
+ // gdv_fn_regexp_matches_utf8_utf8
+ args = {types->i64_type(), // int64_t ptr
+ types->i8_ptr_type(), // const char* data
+ types->i32_type(), // int data_len
+ types->i8_ptr_type(), // const char* pattern
+ types->i32_type()}; // int pattern_len
+
+ engine->AddGlobalMappingForFunc(
+ "gdv_fn_regexp_matches_utf8_utf8", types->i1_type() /*return_type*/, args,
+ reinterpret_cast<void*>(gdv_fn_regexp_matches_utf8_utf8));
+
  // gdv_fn_to_date_utf8_utf8
  args = {types->i64_type(), // int64_t execution_context
  types->i64_type(), // int64_t holder_ptr

cpp/src/gandiva/gdv_function_stubs.h

@@ -64,6 +64,9 @@ bool gdv_fn_like_utf8_utf8_utf8(int64_t ptr, const char* data, int data_len,
 bool gdv_fn_ilike_utf8_utf8(int64_t ptr, const char* data, int data_len,
  const char* pattern, int pattern_len);
 
+bool gdv_fn_regexp_matches_utf8_utf8(int64_t ptr, const char* data, int data_len,
+ const char* pattern, int pattern_len);
+
 int64_t gdv_fn_to_date_utf8_utf8_int32(int64_t context, int64_t ptr, const char* data,
  int data_len, bool in1_validity,
  const char* pattern, int pattern_len,