Lexer.cpp 9.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371
  1. /*
  2. * Copyright (c) 2023, Sam Atkins <atkinssj@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "Lexer.h"
  7. #include <AK/CharacterTypes.h>
  8. #include <AK/Debug.h>
  9. #include <AK/Format.h>
  10. #include <AK/ScopeLogger.h>
  11. namespace CMake {
  12. static bool is_valid_identifier_initial_char(char c)
  13. {
  14. return is_ascii_alpha(c) || c == '_';
  15. }
  16. static bool is_valid_identifier_char(char c)
  17. {
  18. return is_ascii_alphanumeric(c) || c == '_';
  19. }
  20. ErrorOr<Vector<Token>> Lexer::lex(StringView input)
  21. {
  22. Lexer lexer { input };
  23. return lexer.lex_file();
  24. }
  25. Lexer::Lexer(StringView input)
  26. : GenericLexer(input)
  27. {
  28. }
  29. ErrorOr<Vector<Token>> Lexer::lex_file()
  30. {
  31. m_tokens.clear_with_capacity();
  32. while (!is_eof()) {
  33. consume_whitespace_or_comments();
  34. if (is_eof())
  35. break;
  36. if (is_valid_identifier_initial_char(peek())) {
  37. consume_command_invocation();
  38. } else {
  39. consume_garbage();
  40. }
  41. }
  42. return m_tokens;
  43. }
  44. void Lexer::skip_whitespace()
  45. {
  46. while (!is_eof()) {
  47. if (next_is('\n')) {
  48. next_line();
  49. continue;
  50. }
  51. auto consumed = consume_while([&](char c) {
  52. return c == ' ' || c == '\t';
  53. });
  54. if (consumed.is_empty())
  55. break;
  56. }
  57. }
  58. void Lexer::consume_whitespace_or_comments()
  59. {
  60. ScopeLogger<CMAKE_DEBUG> log;
  61. while (!is_eof()) {
  62. skip_whitespace();
  63. if (next_is('#')) {
  64. consume_comment();
  65. } else {
  66. break;
  67. }
  68. }
  69. }
  70. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#command-invocations
  71. void Lexer::consume_command_invocation()
  72. {
  73. ScopeLogger<CMAKE_DEBUG> log;
  74. auto identifier_start = position();
  75. auto identifier = consume_while(is_valid_identifier_char);
  76. auto control_keyword = control_keyword_from_string(identifier);
  77. if (control_keyword.has_value()) {
  78. emit_token(Token::Type::ControlKeyword, identifier, identifier_start, position(), control_keyword.release_value());
  79. } else {
  80. emit_token(Token::Type::Identifier, identifier, identifier_start, position());
  81. }
  82. consume_whitespace_or_comments();
  83. if (next_is('('))
  84. consume_open_paren();
  85. consume_arguments();
  86. if (next_is(')'))
  87. consume_close_paren();
  88. }
  89. void Lexer::consume_arguments()
  90. {
  91. ScopeLogger<CMAKE_DEBUG> log;
  92. while (!is_eof()) {
  93. consume_whitespace_or_comments();
  94. if (next_is('(')) {
  95. consume_open_paren();
  96. consume_whitespace_or_comments();
  97. consume_arguments();
  98. consume_whitespace_or_comments();
  99. if (next_is(')'))
  100. consume_close_paren();
  101. continue;
  102. }
  103. if (next_is(')'))
  104. return;
  105. consume_argument();
  106. }
  107. }
  108. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#command-arguments
  109. void Lexer::consume_argument()
  110. {
  111. ScopeLogger<CMAKE_DEBUG> log;
  112. consume_whitespace_or_comments();
  113. if (next_is('[')) {
  114. consume_bracket_argument();
  115. return;
  116. }
  117. if (next_is('"')) {
  118. consume_quoted_argument();
  119. return;
  120. }
  121. consume_unquoted_argument();
  122. }
  123. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#bracket-argument
  124. void Lexer::consume_bracket_argument()
  125. {
  126. ScopeLogger<CMAKE_DEBUG> log;
  127. auto start = position();
  128. auto value = read_bracket_argument();
  129. emit_token(Token::Type::BracketArgument, value, start, position());
  130. }
  131. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#quoted-argument
  132. void Lexer::consume_quoted_argument()
  133. {
  134. ScopeLogger<CMAKE_DEBUG> log;
  135. auto start = position();
  136. auto start_offset = tell();
  137. VERIFY(consume_specific('"'));
  138. while (!is_eof()) {
  139. if (next_is('"')) {
  140. ignore();
  141. break;
  142. }
  143. if (next_is("\\\""sv)) {
  144. ignore(2);
  145. continue;
  146. }
  147. if (next_is('\n')) {
  148. next_line();
  149. continue;
  150. }
  151. ignore();
  152. }
  153. auto whole_token = m_input.substring_view(start_offset, tell() - start_offset);
  154. auto value = whole_token.substring_view(1, whole_token.length() - 2);
  155. auto variable_references = parse_variable_references_from_argument(whole_token, start);
  156. emit_token(Token::Type::QuotedArgument, value, start, position(), {}, move(variable_references));
  157. }
  158. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#unquoted-argument
  159. void Lexer::consume_unquoted_argument()
  160. {
  161. ScopeLogger<CMAKE_DEBUG> log;
  162. auto start_offset = tell();
  163. auto start = position();
  164. while (!is_eof()) {
  165. if (next_is('\\')) {
  166. consume_escaped_character('\\');
  167. continue;
  168. }
  169. auto consumed = consume_until([](char c) { return is_ascii_space(c) || "()#\"\\'"sv.contains(c); });
  170. if (consumed.is_empty())
  171. break;
  172. // FIXME: `unquoted_legacy`
  173. }
  174. auto value = m_input.substring_view(start_offset, tell() - start_offset);
  175. auto variable_references = parse_variable_references_from_argument(value, start);
  176. emit_token(Token::Type::UnquotedArgument, value, start, position(), {}, move(variable_references));
  177. }
  178. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#comments
  179. void Lexer::consume_comment()
  180. {
  181. ScopeLogger<CMAKE_DEBUG> log;
  182. auto start = position();
  183. VERIFY(consume_specific('#'));
  184. if (next_is('[')) {
  185. // Bracket comment
  186. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#bracket-comment
  187. auto comment = read_bracket_argument();
  188. emit_token(Token::Type::BracketComment, comment, start, position());
  189. return;
  190. }
  191. // Line comment
  192. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#line-comment
  193. auto comment = consume_until('\n');
  194. emit_token(Token::Type::LineComment, comment, start, position());
  195. }
  196. void Lexer::consume_open_paren()
  197. {
  198. auto start = position();
  199. VERIFY(consume_specific('('));
  200. emit_token(Token::Type::OpenParen, "("sv, start, position());
  201. }
  202. void Lexer::consume_close_paren()
  203. {
  204. auto start = position();
  205. VERIFY(consume_specific(')'));
  206. emit_token(Token::Type::CloseParen, ")"sv, start, position());
  207. }
  208. void Lexer::consume_garbage()
  209. {
  210. ScopeLogger<CMAKE_DEBUG> log;
  211. auto start = position();
  212. auto contents = consume_until(is_ascii_space);
  213. if (!contents.is_empty())
  214. emit_token(Token::Type::Garbage, contents, start, position());
  215. }
  216. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#bracket-argument
  217. // Used by both bracket arguments and bracket comments.
  218. StringView Lexer::read_bracket_argument()
  219. {
  220. VERIFY(consume_specific('['));
  221. auto leading_equals_signs = consume_while([](char c) { return c == '='; });
  222. consume_specific('[');
  223. auto start = tell();
  224. auto end = start;
  225. while (!is_eof()) {
  226. // Read everything until we see `]={len}]`.
  227. ignore_until(']');
  228. end = tell();
  229. ignore();
  230. if (next_is(leading_equals_signs))
  231. ignore(leading_equals_signs.length());
  232. if (consume_specific(']'))
  233. break;
  234. }
  235. return m_input.substring_view(start, end - start);
  236. }
  237. // https://cmake.org/cmake/help/latest/manual/cmake-language.7.html#variable-references
  238. Vector<VariableReference> Lexer::parse_variable_references_from_argument(StringView argument_value, Position argument_start)
  239. {
  240. auto position = argument_start;
  241. GenericLexer lexer { argument_value };
  242. Vector<VariableReference> variable_references;
  243. while (!lexer.is_eof()) {
  244. if (lexer.next_is('\n')) {
  245. lexer.ignore();
  246. position.column = 0;
  247. position.line++;
  248. continue;
  249. }
  250. if (lexer.next_is('\\')) {
  251. lexer.ignore();
  252. if (lexer.next_is('\n')) {
  253. lexer.ignore();
  254. position.column = 0;
  255. position.line++;
  256. continue;
  257. }
  258. lexer.ignore();
  259. position.column += 2;
  260. }
  261. if (lexer.next_is('$')) {
  262. auto start = position;
  263. lexer.ignore();
  264. position.column++;
  265. if (lexer.next_is("ENV{"sv)) {
  266. lexer.ignore(4);
  267. position.column += 4;
  268. } else if (lexer.next_is('{')) {
  269. lexer.ignore();
  270. position.column++;
  271. } else {
  272. auto skipped = lexer.consume_until(is_any_of("$ \n"sv));
  273. position.column += skipped.length();
  274. continue;
  275. }
  276. auto variable_name = lexer.consume_until(is_any_of("} \n"sv));
  277. position.column += variable_name.length();
  278. if (lexer.next_is('}')) {
  279. lexer.ignore();
  280. position.column++;
  281. variable_references.empend(variable_name, start, position);
  282. }
  283. continue;
  284. }
  285. lexer.ignore();
  286. position.column++;
  287. }
  288. return variable_references;
  289. }
  290. Position Lexer::position() const
  291. {
  292. return Position {
  293. .line = m_line,
  294. .column = tell() - m_string_offset_after_previous_newline,
  295. };
  296. }
  297. void Lexer::next_line()
  298. {
  299. VERIFY(consume_specific('\n'));
  300. m_string_offset_after_previous_newline = tell();
  301. m_line++;
  302. }
  303. void Lexer::emit_token(Token::Type type, StringView value, Position start, Position end, Optional<ControlKeywordType> control_keyword, Vector<VariableReference> variable_references)
  304. {
  305. dbgln_if(CMAKE_DEBUG, "Emitting {} token: `{}` ({}:{} to {}:{})", to_string(type), value, start.line, start.column, end.line, end.column);
  306. m_tokens.empend(type, value, start, end, move(control_keyword), move(variable_references));
  307. }
  308. }