use RE2::QuoteMeta to escape special chars

chrisiou · Dec 19, 2023 · 5173e1e · 5173e1e
1 parent bb5e3ea
commit 5173e1e
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 37 deletions.
diff --git a/src/core_functions/scalar/string/functions.json b/src/core_functions/scalar/string/functions.json
@@ -308,7 +308,7 @@
     {
         "name": "regexp_escape",
         "parameters": "string",
-        "description": "Return the input string with all the regular expression metacharacters escaped",
+        "description": "Escapes all potentially meaningful regexp characters in the input string",
         "example": "regexp_escape('https://duckdb.org')",
         "struct": "RegexpEscapeFun",
         "type": "scalar_function"

diff --git a/src/core_functions/scalar/string/regexp_escape.cpp b/src/core_functions/scalar/string/regexp_escape.cpp
@@ -1,31 +1,18 @@
 #include "duckdb/core_functions/scalar/string_functions.hpp"
-#include "duckdb/common/string_util.hpp"
+#include "re2/re2.h"
 
 namespace duckdb {
 
-static string_t escape(const string_t &str, vector<char> &escaped_pattern) {
-	auto input_str = str.GetData();
-	auto size_str = str.GetSize();
-
-	escaped_pattern.clear(); // reuse the buffer
-	// note: reserving double the size to account for escaping ('\\') as an average case
-	// to have half of the characters in the input string are special
-	escaped_pattern.reserve(2 * size_str);
-	string special_chars = "()[]{}?*+-|^$\\.&~#";
-	for (idx_t i = 0; i < size_str; ++i) {
-		char ch = input_str[i];
-		if (special_chars.find(ch) != std::string::npos) {
-			escaped_pattern.push_back('\\'); // escape the special character
-		}
-		escaped_pattern.push_back(ch);
+struct EscapeOperator {
+	template <class INPUT_TYPE, class RESULT_TYPE>
+	static RESULT_TYPE Operation(INPUT_TYPE &input, Vector &result) {
+		auto escaped_pattern = RE2::QuoteMeta(input.GetString());
+		return StringVector::AddString(result, escaped_pattern);
 	}
-	return string_t(escaped_pattern.data(), escaped_pattern.size());
-}
+};
 
 static void RegexpEscapeFunction(DataChunk &args, ExpressionState &state, Vector &result) {
-	vector<char> escaped_pattern;
-	auto input_str = args.GetValue(0, 0);
-	result.Reference(escape(input_str.ToString(), escaped_pattern));
+	UnaryExecutor::ExecuteString<string_t, string_t, EscapeOperator>(args.data[0], result, args.size());
 }
 
 ScalarFunction RegexpEscapeFun::GetFunction() {

diff --git a/src/include/duckdb/core_functions/scalar/string_functions.hpp b/src/include/duckdb/core_functions/scalar/string_functions.hpp
@@ -483,7 +483,7 @@ struct ToBaseFun {
 struct RegexpEscapeFun {
 	static constexpr const char *Name = "regexp_escape";
 	static constexpr const char *Parameters = "string";
-	static constexpr const char *Description = "Return the input string with all the regular expression metacharacters escaped";
+	static constexpr const char *Description = "Escapes all potentially meaningful regexp characters in the input string";
 	static constexpr const char *Example = "regexp_escape('https://duckdb.org')";
 
 	static ScalarFunction GetFunction();

diff --git a/test/sql/function/string/regex_escape.test b/test/sql/function/string/regex_escape.test
@@ -9,9 +9,9 @@ PRAGMA enable_verification
 query T
 SELECT regexp_escape('https://duckdb.org');
 ----
-https://duckdb\.org
+https\:\/\/duckdb\.org
 
-# no special chars in the input
+# no special chars
 query T
 SELECT regexp_escape('abc');
 ----
@@ -23,27 +23,22 @@ SELECT regexp_escape('a.b');
 ----
 a\.b
 
-query T
-SELECT regexp_escape('()[]{}*+?|^$');
-----
-\(\)\[\]\{\}\*\+\?\|\^\$
-
 query T
 SELECT regexp_escape('a.b[c]*');
 ----
 a\.b\[c\]\*
 
 # empty string
 query T
-SELECT regexp_escape('""');
+SELECT regexp_escape('\n');
 ----
-""
+\\n
 
 # whitespaces
 query T
 SELECT regexp_escape('a b c');
 ----
-a b c
+a\ b\ c
 
 query T
 SELECT regexp_escape('line1\nline2');
@@ -52,24 +47,35 @@ line1\\nline2
 
 # case sensitive
 query T
-SELECT regexp_escape('CaseSensitive');
+SELECT regexp_escape('CaseSensitivE');
 ----
-CaseSensitive
+CaseSensitivE
 
 # unicode character
 query T
 SELECT regexp_escape('@');
 ----
-@
+\@
 
 # backslashes
 query T
 SELECT regexp_escape('path\to\wonderland');
 ----
 path\\to\\wonderland
 
-# escape all special characters
+# more special characters
 query T
 SELECT regexp_escape('$()*+.?[\]^{|}-');
 ----
 \$\(\)\*\+\.\?\[\\\]\^\{\|\}\-
+
+query I
+CREATE TABLE tbl (a VARCHAR);
+INSERT INTO tbl VALUES('cde');
+INSERT INTO tbl VALUES('-cd+e');
+INSERT INTO tbl VALUES('[^!$');
+SELECT regexp_escape(a) FROM tbl;
+----
+cde
+\-cd\+e
+\[\^\!\$