Add substr short-ciruit for regexp_like

apache · Dec 6, 2020 · 960406e · 960406e
1 parent 42a8736
commit 960406e
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 2 deletions.
diff --git a/cpp/src/gandiva/regexp_matches_holder.cc b/cpp/src/gandiva/regexp_matches_holder.cc
@@ -25,6 +25,7 @@ namespace gandiva {
 
 RE2 RegexpMatchesHolder::starts_with_regex_(R"(\^([\w\s]+)(\.\*)?)");
 RE2 RegexpMatchesHolder::ends_with_regex_(R"((\.\*)?([\w\s]+)\$)");
+RE2 RegexpMatchesHolder::is_substr_regex_(R"((\w|\s)*)");
 
 // Short-circuit pattern matches for the two common sub cases :
 // - starts_with and ends_with.
@@ -45,6 +46,11 @@ const FunctionNode RegexpMatchesHolder::TryOptimize(const FunctionNode& node) {
  std::make_shared<LiteralNode>(literal_type, LiteralHolder(substr), false);
  return FunctionNode("ends_with", {node.children().at(0), suffix_node},
  node.return_type());
+ } else if (RE2::FullMatch(pattern, is_substr_regex_)) {
+ auto substr_node =
+ std::make_shared<LiteralNode>(literal_type, LiteralHolder(pattern), false);
+ return FunctionNode("is_substr", {node.children().at(0), substr_node},
+ node.return_type());
  }
  }
 

diff --git a/cpp/src/gandiva/regexp_matches_holder.h b/cpp/src/gandiva/regexp_matches_holder.h
@@ -55,6 +55,7 @@ class GANDIVA_EXPORT RegexpMatchesHolder : public LikeHolder {
 
  static RE2 starts_with_regex_; // pre-compiled pattern for matching starts_with
  static RE2 ends_with_regex_; // pre-compiled pattern for matching ends_with
+ static RE2 is_substr_regex_; // pre-compiled pattern for matching is_substr
 };
 } // namespace gandiva
 

diff --git a/cpp/src/gandiva/regexp_matches_holder_test.cc b/cpp/src/gandiva/regexp_matches_holder_test.cc
@@ -172,6 +172,11 @@ TEST_F(TestRegexpMatchesHolder, TestOptimise) {
  EXPECT_EQ(fnode.descriptor()->name(), "ends_with");
  EXPECT_EQ(fnode.ToString(), "bool ends_with((string) in, (const string) xyz)");
 
+ // optimise for 'is_substr'
+ fnode = RegexpMatchesHolder::TryOptimize(BuildRegexpMatches("xyz"));
+ EXPECT_EQ(fnode.descriptor()->name(), "is_substr");
+ EXPECT_EQ(fnode.ToString(), "bool is_substr((string) in, (const string) xyz)");
+
  // no optimisation for others.
  fnode = RegexpMatchesHolder::TryOptimize(BuildRegexpMatches("^xyz$"));
  EXPECT_EQ(fnode.descriptor()->name(), "regexp_matches");

diff --git a/cpp/src/gandiva/sql_like_holder.cc b/cpp/src/gandiva/sql_like_holder.cc
@@ -46,8 +46,8 @@ const FunctionNode SQLLikeHolder::TryOptimize(const FunctionNode& node) {
  auto suffix = pattern.substr(2); // skip .*
  auto suffix_node =
  std::make_shared<LiteralNode>(literal_type, LiteralHolder(suffix), false);
-  return FunctionNode("ends_with", {node.children().at(0), suffix_node},
-  node.return_type());
+ return FunctionNode("ends_with", {node.children().at(0), suffix_node},
+ node.return_type());
  } else if (RE2::FullMatch(pattern, is_substr_regex_)) {
  auto substr =
  pattern.substr(2, pattern.length() - 4); // trim starting and ending .*