[Support] Add \g<ref> backreferences in Regex::sub() (#67220)

The existing format of backreferences, `\<ref>`, does not allow digits to be placed directly after the reference because they are included in the reference number. The new format solves this problem by adding explicit delimiters.

Author: igorkudrin

llvm/include/llvm/Support/Regex.h

@@ -85,8 +85,9 @@ namespace llvm {
  std::string *Error = nullptr) const;
 
  /// sub - Return the result of replacing the first match of the regex in
- /// \p String with the \p Repl string. Backreferences like "\0" in the
- /// replacement string are replaced with the appropriate match substring.
+ /// \p String with the \p Repl string. Backreferences like "\0" and "\g<1>"
+ /// in the replacement string are replaced with the appropriate match
+ /// substring.
  ///
  /// Note that the replacement string has backslash escaping performed on
  /// it. Invalid backreferences are ignored (replaced by empty strings).

llvm/lib/Support/Regex.cpp

@@ -163,6 +163,25 @@ std::string Regex::sub(StringRef Repl, StringRef String,
 
  // FIXME: We should have a StringExtras function for mapping C99 escapes.
  switch (Repl[0]) {
+
+ // Backreference with the "\g<ref>" syntax
+ case 'g':
+ if (Repl.size() >= 4 && Repl[1] == '<') {
+ size_t End = Repl.find('>');
+ StringRef Ref = Repl.slice(2, End);
+ unsigned RefValue;
+ if (End != StringRef::npos && !Ref.getAsInteger(10, RefValue)) {
+ Repl = Repl.substr(End + 1);
+ if (RefValue < Matches.size())
+ Res += Matches[RefValue];
+ else if (Error && Error->empty())
+ *Error =
+ ("invalid backreference string 'g<" + Twine(Ref) + ">'").str();
+ break;
+ }
+ }
+ [[fallthrough]];
+
  // Treat all unrecognized characters as self-quoting.
  default:
  Res += Repl[0];

llvm/unittests/Support/RegexTest.cpp

@@ -127,6 +127,34 @@ TEST_F(RegexTest, Substitution) {
 
  EXPECT_EQ("aber", Regex("a[0-9]+b").sub("a\\100b", "a1234ber", &Error));
  EXPECT_EQ(Error, "invalid backreference string '100'");
+
+ EXPECT_EQ("012345", Regex("a([0-9]+).*").sub("0\\g<1>5", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0a1234ber5",
+ Regex("a([0-9]+).*").sub("0\\g<0>5", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0A5", Regex("a(.)(.)(.)(.)(.)(.)(.)(.)(.)(.).*")
+ .sub("0\\g<10>5", "a123456789Aber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0g<-1>5",
+ Regex("a([0-9]+).*").sub("0\\g<-1>5", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0g<15", Regex("a([0-9]+).*").sub("0\\g<15", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0g<>15", Regex("a([0-9]+).*").sub("0\\g<>15", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("0g<3e>1",
+ Regex("a([0-9]+).*").sub("0\\g<3e>1", "a1234ber", &Error));
+ EXPECT_EQ("", Error);
+
+ EXPECT_EQ("aber", Regex("a([0-9]+)b").sub("a\\g<100>b", "a1234ber", &Error));
+ EXPECT_EQ(Error, "invalid backreference string 'g<100>'");
 }
 
 TEST_F(RegexTest, IsLiteralERE) {