| 32 | enum ParserState { PARSER_DEFAULT, PARSER_IN_QUOTE, PARSER_IN_REGEX, PARSER_IN_EXPANSION, PARSER_IN_BRACE }; |
| 33 | |
| 34 | bool |
| 35 | Parser::parse_line(const std::string &original_line) |
| 36 | { |
| 37 | std::string line = original_line; |
| 38 | ParserState state = PARSER_DEFAULT; |
| 39 | bool extracting_token = false; |
| 40 | off_t cur_token_start = 0; |
| 41 | size_t cur_token_length = 0; |
| 42 | |
| 43 | for (size_t i = 0; i < line.size(); ++i) { |
| 44 | if ((state == PARSER_DEFAULT) && (std::isspace(line[i]) || ((line[i] == '=')))) { |
| 45 | if (extracting_token) { |
| 46 | cur_token_length = i - cur_token_start; |
| 47 | if (cur_token_length > 0) { |
| 48 | _tokens.push_back(line.substr(cur_token_start, cur_token_length)); |
| 49 | } |
| 50 | extracting_token = false; |
| 51 | state = PARSER_DEFAULT; |
| 52 | } else if (!std::isspace(line[i])) { |
| 53 | // we got a standalone =, > or < |
| 54 | _tokens.push_back(std::string(1, line[i])); |
| 55 | } |
| 56 | } else if ((state != PARSER_IN_QUOTE) && (line[i] == '/')) { |
| 57 | // Deal with regexes, nothing gets escaped / quoted in here |
| 58 | if ((state != PARSER_IN_REGEX) && !extracting_token) { |
| 59 | state = PARSER_IN_REGEX; |
| 60 | extracting_token = true; |
| 61 | cur_token_start = i; |
| 62 | } else if ((state == PARSER_IN_REGEX) && extracting_token && (line[i - 1] != '\\')) { |
| 63 | cur_token_length = i - cur_token_start + 1; |
| 64 | _tokens.push_back(line.substr(cur_token_start, cur_token_length)); |
| 65 | state = PARSER_DEFAULT; |
| 66 | extracting_token = false; |
| 67 | } |
| 68 | } else if ((state != PARSER_IN_REGEX) && (line[i] == '\\')) { |
| 69 | // Escaping - convert escape sequences to control characters |
| 70 | if (!extracting_token) { |
| 71 | extracting_token = true; |
| 72 | cur_token_start = i; |
| 73 | } |
| 74 | // Check if next character forms an escape sequence we want to convert |
| 75 | if (i + 1 < line.size()) { |
| 76 | constexpr std::string_view controls{"trn", 3}; |
| 77 | constexpr char mapped_ctrls[] = "\t\r\n"; |
| 78 | |
| 79 | if (auto pos = controls.find(line[i + 1]); pos != std::string_view::npos) { |
| 80 | line[i] = mapped_ctrls[pos]; |
| 81 | line.erase(i + 1, 1); |
| 82 | } else { |
| 83 | line.erase(i, 1); |
| 84 | } |
| 85 | } else { |
| 86 | line.erase(i, 1); // Backslash at end of line |
| 87 | } |
| 88 | } else if ((state != PARSER_IN_REGEX) && (line[i] == '"')) { |
| 89 | if ((state != PARSER_IN_QUOTE) && !extracting_token) { |
| 90 | state = PARSER_IN_QUOTE; |
| 91 | extracting_token = true; |