Method RegexpExtract

be/src/exprs/string-functions-ir.cc:1094–1125

}

StringVal StringFunctions::RegexpExtract(FunctionContext* context, const StringVal& str,
    const StringVal& pattern, const BigIntVal& index) {
  if (str.is_null || pattern.is_null || index.is_null) return StringVal::null();
  if (index.val < 0) return StringVal();

  re2::RE2* re = reinterpret_cast<re2::RE2*>(
      context->GetFunctionState(FunctionContext::THREAD_LOCAL));
  scoped_ptr<re2::RE2> scoped_re; // destroys re if we have to locally compile it
  if (re == NULL) {
    DCHECK(!context->IsArgConstant(1));
    string error_str;
    re = CompileRegex(pattern, &error_str, StringVal::null());
    if (re == NULL) {
      context->AddWarning(error_str.c_str());
      return StringVal::null();
    }
    scoped_re.reset(re);
  }

  re2::StringPiece str_sp(reinterpret_cast<char*>(str.ptr), str.len);
  int max_matches = 1 + re->NumberOfCapturingGroups();
  if (index.val >= max_matches) return StringVal();
  // Use a vector because clang complains about non-POD varlen arrays
  // TODO: fix this
  vector<re2::StringPiece> matches(max_matches);
  bool success =
      re->Match(str_sp, 0, str.len, re2::RE2::UNANCHORED, matches.data(), max_matches);
  if (!success) return StringVal();
  // matches[0] is the whole string, matches[1] the first group, etc.
  const re2::StringPiece& match = matches[index.val];
  return AnyValUtil::FromBuffer(context, match.data(), match.size());
}

StringVal StringFunctions::RegexpReplace(FunctionContext* context, const StringVal& str,
    const StringVal& pattern, const StringVal& replace) {