Function ParseNameTokenHelper

be/src/runtime/datetime-parser-common.cc:219–265

Helper function used by ParseMonthNameToken() and ParseDayNameToken(). Parses 'token_start' string and returns true if it finds a valid short or normal name. The valid name prefixes are stored as keys in 'prefix_to_suffix' map. The valid name prefixes are expected to be equal to the corresponding valid short names. The valid name suffixes and the corresponding int IDs are stored as values in 'pref

// name is found using 'prefix_to_suffix'. If parsing is successful, '*token_end' is
// adjusted to point to the character right after the end of the name.
bool ParseNameTokenHelper(
    bool is_short_name, int short_name_len, int max_name_len,
    const char* token_start, const char** token_end,
    bool is_strict_matching,
    const unordered_map<string, pair<string, int>>& prefix_to_suffix,
    int* ret_val) {
  DCHECK(token_start != nullptr);
  DCHECK(ret_val != nullptr);
  DCHECK(token_end != nullptr && *token_end != nullptr);
  DCHECK(token_start <= *token_end);
  int token_len = *token_end - token_start;
  if (token_len < short_name_len) return false;

  string buff(token_start, token_len);
  boost::to_lower(buff);
  const string& prefix = buff.substr(0, short_name_len);
  const auto it = prefix_to_suffix.find(prefix);
  if (UNLIKELY(it == prefix_to_suffix.end())) return false;
  if (is_short_name) {
    DCHECK(token_len == short_name_len);
    *ret_val = it->second.second;
    return true;
  }

  if (is_strict_matching) {
    DCHECK(buff.length() == max_name_len);
    trim_right_if(buff, is_any_of(" "));
  }

  // Check if the remaining characters match.
  const string& expected_suffix = it->second.first;
  if (buff.length() - short_name_len < expected_suffix.length()) return false;
  const char* actual_suffix = buff.c_str() + short_name_len;
  if (strncmp(actual_suffix, expected_suffix.c_str(), expected_suffix.length()) != 0) {
    return false;
  }
  *ret_val = it->second.second;

  // If the end of the name token wasn't identified because it wasn't followed by a
  // separator then the end of the name token has to be adjusted.
  if (prefix.length() + expected_suffix.length() < buff.length()) {
    if (is_strict_matching) return false;
    *token_end = token_start + prefix.length() + expected_suffix.length();
  }

  return true;
}

}