strict_cpp.re 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575
  1. /*=============================================================================
  2. Boost.Wave: A Standard compliant C++ preprocessor library
  3. Copyright (c) 2001 Daniel C. Nuffer
  4. Copyright (c) 2001-2011 Hartmut Kaiser.
  5. Distributed under the Boost Software License, Version 1.0. (See accompanying
  6. file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
  7. This is a strict lexer conforming to the Standard as close as possible.
  8. It does not allow the '$' to be part of identifiers. If you need the '$'
  9. character in identifiers please include the lexer definition provided
  10. in the cpp.re file.
  11. TODO:
  12. handle errors better.
  13. =============================================================================*/
  14. /*!re2c
  15. re2c:indent:string = " ";
  16. any = [\t\v\f\r\n\040-\377];
  17. anyctrl = [\001-\037];
  18. OctalDigit = [0-7];
  19. Digit = [0-9];
  20. HexDigit = [a-fA-F0-9];
  21. Integer = (("0" [xX] HexDigit+) | ("0" OctalDigit*) | ([1-9] Digit*));
  22. ExponentStart = [Ee] [+-];
  23. ExponentPart = [Ee] [+-]? Digit+;
  24. FractionalConstant = (Digit* "." Digit+) | (Digit+ ".");
  25. FloatingSuffix = [fF] [lL]? | [lL] [fF]?;
  26. IntegerSuffix = [uU] [lL]? | [lL] [uU]?;
  27. LongIntegerSuffix = [uU] ([lL] [lL]) | ([lL] [lL]) [uU]?;
  28. Backslash = [\\] | "??/";
  29. EscapeSequence = Backslash ([abfnrtv?'"] | Backslash | "x" HexDigit+ | OctalDigit OctalDigit? OctalDigit?);
  30. HexQuad = HexDigit HexDigit HexDigit HexDigit;
  31. UniversalChar = Backslash ("u" HexQuad | "U" HexQuad HexQuad);
  32. Newline = "\r\n" | "\n" | "\r";
  33. PPSpace = ([ \t\f\v]|("/*"(any\[*]|Newline|("*"+(any\[*/]|Newline)))*"*"+"/"))*;
  34. Pound = "#" | "??=" | "%:";
  35. NonDigit = [a-zA-Z_] | UniversalChar;
  36. */
  37. /*!re2c
  38. "/*" { goto ccomment; }
  39. "//" { goto cppcomment; }
  40. "."? Digit { goto pp_number; }
  41. "alignas" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_ALIGNAS : T_IDENTIFIER); }
  42. "alignof" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_ALIGNOF : T_IDENTIFIER); }
  43. "asm" { BOOST_WAVE_RET(T_ASM); }
  44. "auto" { BOOST_WAVE_RET(T_AUTO); }
  45. "bool" { BOOST_WAVE_RET(T_BOOL); }
  46. "break" { BOOST_WAVE_RET(T_BREAK); }
  47. "case" { BOOST_WAVE_RET(T_CASE); }
  48. "catch" { BOOST_WAVE_RET(T_CATCH); }
  49. "char" { BOOST_WAVE_RET(T_CHAR); }
  50. "char16_t" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_CHAR16_T : T_IDENTIFIER); }
  51. "char32_t" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_CHAR32_T : T_IDENTIFIER); }
  52. "class" { BOOST_WAVE_RET(T_CLASS); }
  53. "const" { BOOST_WAVE_RET(T_CONST); }
  54. "constexpr" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_CONSTEXPR : T_IDENTIFIER); }
  55. "const_cast" { BOOST_WAVE_RET(T_CONSTCAST); }
  56. "continue" { BOOST_WAVE_RET(T_CONTINUE); }
  57. "decltype" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_DECLTYPE : T_IDENTIFIER); }
  58. "default" { BOOST_WAVE_RET(T_DEFAULT); }
  59. "delete" { BOOST_WAVE_RET(T_DELETE); }
  60. "do" { BOOST_WAVE_RET(T_DO); }
  61. "double" { BOOST_WAVE_RET(T_DOUBLE); }
  62. "dynamic_cast" { BOOST_WAVE_RET(T_DYNAMICCAST); }
  63. "else" { BOOST_WAVE_RET(T_ELSE); }
  64. "enum" { BOOST_WAVE_RET(T_ENUM); }
  65. "explicit" { BOOST_WAVE_RET(T_EXPLICIT); }
  66. "export" { BOOST_WAVE_RET(T_EXPORT); }
  67. "extern" { BOOST_WAVE_RET(T_EXTERN); }
  68. "false" { BOOST_WAVE_RET(T_FALSE); }
  69. "float" { BOOST_WAVE_RET(T_FLOAT); }
  70. "for" { BOOST_WAVE_RET(T_FOR); }
  71. "friend" { BOOST_WAVE_RET(T_FRIEND); }
  72. "goto" { BOOST_WAVE_RET(T_GOTO); }
  73. "if" { BOOST_WAVE_RET(T_IF); }
  74. "import" { BOOST_WAVE_RET(s->enable_import_keyword ? T_IMPORT : T_IDENTIFIER); }
  75. "inline" { BOOST_WAVE_RET(T_INLINE); }
  76. "int" { BOOST_WAVE_RET(T_INT); }
  77. "long" { BOOST_WAVE_RET(T_LONG); }
  78. "mutable" { BOOST_WAVE_RET(T_MUTABLE); }
  79. "namespace" { BOOST_WAVE_RET(T_NAMESPACE); }
  80. "new" { BOOST_WAVE_RET(T_NEW); }
  81. "noexcept" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_NOEXCEPT : T_IDENTIFIER); }
  82. "nullptr" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_NULLPTR : T_IDENTIFIER); }
  83. "operator" { BOOST_WAVE_RET(T_OPERATOR); }
  84. "private" { BOOST_WAVE_RET(T_PRIVATE); }
  85. "protected" { BOOST_WAVE_RET(T_PROTECTED); }
  86. "public" { BOOST_WAVE_RET(T_PUBLIC); }
  87. "register" { BOOST_WAVE_RET(T_REGISTER); }
  88. "reinterpret_cast" { BOOST_WAVE_RET(T_REINTERPRETCAST); }
  89. "return" { BOOST_WAVE_RET(T_RETURN); }
  90. "short" { BOOST_WAVE_RET(T_SHORT); }
  91. "signed" { BOOST_WAVE_RET(T_SIGNED); }
  92. "sizeof" { BOOST_WAVE_RET(T_SIZEOF); }
  93. "static" { BOOST_WAVE_RET(T_STATIC); }
  94. "static_cast" { BOOST_WAVE_RET(T_STATICCAST); }
  95. "static_assert" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_STATICASSERT : T_IDENTIFIER); }
  96. "struct" { BOOST_WAVE_RET(T_STRUCT); }
  97. "switch" { BOOST_WAVE_RET(T_SWITCH); }
  98. "template" { BOOST_WAVE_RET(T_TEMPLATE); }
  99. "this" { BOOST_WAVE_RET(T_THIS); }
  100. "thread_local" { BOOST_WAVE_RET(s->act_in_cpp0x_mode ? T_THREADLOCAL : T_IDENTIFIER); }
  101. "throw" { BOOST_WAVE_RET(T_THROW); }
  102. "true" { BOOST_WAVE_RET(T_TRUE); }
  103. "try" { BOOST_WAVE_RET(T_TRY); }
  104. "typedef" { BOOST_WAVE_RET(T_TYPEDEF); }
  105. "typeid" { BOOST_WAVE_RET(T_TYPEID); }
  106. "typename" { BOOST_WAVE_RET(T_TYPENAME); }
  107. "union" { BOOST_WAVE_RET(T_UNION); }
  108. "unsigned" { BOOST_WAVE_RET(T_UNSIGNED); }
  109. "using" { BOOST_WAVE_RET(T_USING); }
  110. "virtual" { BOOST_WAVE_RET(T_VIRTUAL); }
  111. "void" { BOOST_WAVE_RET(T_VOID); }
  112. "volatile" { BOOST_WAVE_RET(T_VOLATILE); }
  113. "wchar_t" { BOOST_WAVE_RET(T_WCHART); }
  114. "while" { BOOST_WAVE_RET(T_WHILE); }
  115. "__int8" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_INT8 : T_IDENTIFIER); }
  116. "__int16" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_INT16 : T_IDENTIFIER); }
  117. "__int32" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_INT32 : T_IDENTIFIER); }
  118. "__int64" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_INT64 : T_IDENTIFIER); }
  119. "_"? "_based" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_BASED : T_IDENTIFIER); }
  120. "_"? "_declspec" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_DECLSPEC : T_IDENTIFIER); }
  121. "_"? "_cdecl" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_CDECL : T_IDENTIFIER); }
  122. "_"? "_fastcall" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_FASTCALL : T_IDENTIFIER); }
  123. "_"? "_stdcall" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_STDCALL : T_IDENTIFIER); }
  124. "__try" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_TRY : T_IDENTIFIER); }
  125. "__except" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_EXCEPT : T_IDENTIFIER); }
  126. "__finally" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_FINALLY : T_IDENTIFIER); }
  127. "__leave" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_LEAVE : T_IDENTIFIER); }
  128. "_"? "_inline" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_INLINE : T_IDENTIFIER); }
  129. "_"? "_asm" { BOOST_WAVE_RET(s->enable_ms_extensions ? T_MSEXT_ASM : T_IDENTIFIER); }
  130. "{" { BOOST_WAVE_RET(T_LEFTBRACE); }
  131. "??<" { BOOST_WAVE_RET(T_LEFTBRACE_TRIGRAPH); }
  132. "<%" { BOOST_WAVE_RET(T_LEFTBRACE_ALT); }
  133. "}" { BOOST_WAVE_RET(T_RIGHTBRACE); }
  134. "??>" { BOOST_WAVE_RET(T_RIGHTBRACE_TRIGRAPH); }
  135. "%>" { BOOST_WAVE_RET(T_RIGHTBRACE_ALT); }
  136. "[" { BOOST_WAVE_RET(T_LEFTBRACKET); }
  137. "??(" { BOOST_WAVE_RET(T_LEFTBRACKET_TRIGRAPH); }
  138. "<:" { BOOST_WAVE_RET(T_LEFTBRACKET_ALT); }
  139. "]" { BOOST_WAVE_RET(T_RIGHTBRACKET); }
  140. "??)" { BOOST_WAVE_RET(T_RIGHTBRACKET_TRIGRAPH); }
  141. ":>" { BOOST_WAVE_RET(T_RIGHTBRACKET_ALT); }
  142. "#" { BOOST_WAVE_RET(T_POUND); }
  143. "%:" { BOOST_WAVE_RET(T_POUND_ALT); }
  144. "??=" { BOOST_WAVE_RET(T_POUND_TRIGRAPH); }
  145. "##" { BOOST_WAVE_RET(T_POUND_POUND); }
  146. "#??=" { BOOST_WAVE_RET(T_POUND_POUND_TRIGRAPH); }
  147. "??=#" { BOOST_WAVE_RET(T_POUND_POUND_TRIGRAPH); }
  148. "??=??=" { BOOST_WAVE_RET(T_POUND_POUND_TRIGRAPH); }
  149. "%:%:" { BOOST_WAVE_RET(T_POUND_POUND_ALT); }
  150. "(" { BOOST_WAVE_RET(T_LEFTPAREN); }
  151. ")" { BOOST_WAVE_RET(T_RIGHTPAREN); }
  152. ";" { BOOST_WAVE_RET(T_SEMICOLON); }
  153. ":" { BOOST_WAVE_RET(T_COLON); }
  154. "..." { BOOST_WAVE_RET(T_ELLIPSIS); }
  155. "?" { BOOST_WAVE_RET(T_QUESTION_MARK); }
  156. "::"
  157. {
  158. if (s->act_in_c99_mode) {
  159. --YYCURSOR;
  160. BOOST_WAVE_RET(T_COLON);
  161. }
  162. else {
  163. BOOST_WAVE_RET(T_COLON_COLON);
  164. }
  165. }
  166. "." { BOOST_WAVE_RET(T_DOT); }
  167. ".*"
  168. {
  169. if (s->act_in_c99_mode) {
  170. --YYCURSOR;
  171. BOOST_WAVE_RET(T_DOT);
  172. }
  173. else {
  174. BOOST_WAVE_RET(T_DOTSTAR);
  175. }
  176. }
  177. "+" { BOOST_WAVE_RET(T_PLUS); }
  178. "-" { BOOST_WAVE_RET(T_MINUS); }
  179. "*" { BOOST_WAVE_RET(T_STAR); }
  180. "/" { BOOST_WAVE_RET(T_DIVIDE); }
  181. "%" { BOOST_WAVE_RET(T_PERCENT); }
  182. "^" { BOOST_WAVE_RET(T_XOR); }
  183. "??'" { BOOST_WAVE_RET(T_XOR_TRIGRAPH); }
  184. "xor" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_XOR_ALT); }
  185. "&" { BOOST_WAVE_RET(T_AND); }
  186. "bitand" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_AND_ALT); }
  187. "|" { BOOST_WAVE_RET(T_OR); }
  188. "bitor" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_OR_ALT); }
  189. "??!" { BOOST_WAVE_RET(T_OR_TRIGRAPH); }
  190. "~" { BOOST_WAVE_RET(T_COMPL); }
  191. "??-" { BOOST_WAVE_RET(T_COMPL_TRIGRAPH); }
  192. "compl" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_COMPL_ALT); }
  193. "!" { BOOST_WAVE_RET(T_NOT); }
  194. "not" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_NOT_ALT); }
  195. "=" { BOOST_WAVE_RET(T_ASSIGN); }
  196. "<" { BOOST_WAVE_RET(T_LESS); }
  197. ">" { BOOST_WAVE_RET(T_GREATER); }
  198. "+=" { BOOST_WAVE_RET(T_PLUSASSIGN); }
  199. "-=" { BOOST_WAVE_RET(T_MINUSASSIGN); }
  200. "*=" { BOOST_WAVE_RET(T_STARASSIGN); }
  201. "/=" { BOOST_WAVE_RET(T_DIVIDEASSIGN); }
  202. "%=" { BOOST_WAVE_RET(T_PERCENTASSIGN); }
  203. "^=" { BOOST_WAVE_RET(T_XORASSIGN); }
  204. "xor_eq" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_XORASSIGN_ALT); }
  205. "??'=" { BOOST_WAVE_RET(T_XORASSIGN_TRIGRAPH); }
  206. "&=" { BOOST_WAVE_RET(T_ANDASSIGN); }
  207. "and_eq" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_ANDASSIGN_ALT); }
  208. "|=" { BOOST_WAVE_RET(T_ORASSIGN); }
  209. "or_eq" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_ORASSIGN_ALT); }
  210. "??!=" { BOOST_WAVE_RET(T_ORASSIGN_TRIGRAPH); }
  211. "<<" { BOOST_WAVE_RET(T_SHIFTLEFT); }
  212. ">>" { BOOST_WAVE_RET(T_SHIFTRIGHT); }
  213. ">>=" { BOOST_WAVE_RET(T_SHIFTRIGHTASSIGN); }
  214. "<<=" { BOOST_WAVE_RET(T_SHIFTLEFTASSIGN); }
  215. "==" { BOOST_WAVE_RET(T_EQUAL); }
  216. "!=" { BOOST_WAVE_RET(T_NOTEQUAL); }
  217. "not_eq" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_NOTEQUAL_ALT); }
  218. "<=" { BOOST_WAVE_RET(T_LESSEQUAL); }
  219. ">=" { BOOST_WAVE_RET(T_GREATEREQUAL); }
  220. "&&" { BOOST_WAVE_RET(T_ANDAND); }
  221. "and" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_ANDAND_ALT); }
  222. "||" { BOOST_WAVE_RET(T_OROR); }
  223. "??!|" { BOOST_WAVE_RET(T_OROR_TRIGRAPH); }
  224. "|??!" { BOOST_WAVE_RET(T_OROR_TRIGRAPH); }
  225. "or" { BOOST_WAVE_RET(s->act_in_c99_mode ? T_IDENTIFIER : T_OROR_ALT); }
  226. "??!??!" { BOOST_WAVE_RET(T_OROR_TRIGRAPH); }
  227. "++" { BOOST_WAVE_RET(T_PLUSPLUS); }
  228. "--" { BOOST_WAVE_RET(T_MINUSMINUS); }
  229. "," { BOOST_WAVE_RET(T_COMMA); }
  230. "->*"
  231. {
  232. if (s->act_in_c99_mode) {
  233. --YYCURSOR;
  234. BOOST_WAVE_RET(T_ARROW);
  235. }
  236. else {
  237. BOOST_WAVE_RET(T_ARROWSTAR);
  238. }
  239. }
  240. "->" { BOOST_WAVE_RET(T_ARROW); }
  241. "??/" { BOOST_WAVE_RET(T_ANY_TRIGRAPH); }
  242. "L"? (['] (EscapeSequence | UniversalChar | any\[\n\r\\'])+ ['])
  243. { BOOST_WAVE_RET(T_CHARLIT); }
  244. "L"? (["] (EscapeSequence | UniversalChar | any\[\n\r\\"])* ["])
  245. { BOOST_WAVE_RET(T_STRINGLIT); }
  246. "L"? "R" ["]
  247. {
  248. if (s->act_in_cpp0x_mode)
  249. {
  250. rawstringdelim = "";
  251. goto extrawstringlit;
  252. }
  253. --YYCURSOR;
  254. BOOST_WAVE_RET(T_IDENTIFIER);
  255. }
  256. [uU] [']
  257. {
  258. if (s->act_in_cpp0x_mode)
  259. goto extcharlit;
  260. --YYCURSOR;
  261. BOOST_WAVE_RET(T_IDENTIFIER);
  262. }
  263. ([uU] | "u8") ["]
  264. {
  265. if (s->act_in_cpp0x_mode)
  266. goto extstringlit;
  267. --YYCURSOR;
  268. BOOST_WAVE_RET(T_IDENTIFIER);
  269. }
  270. ([uU] | "u8") "R" ["]
  271. {
  272. if (s->act_in_cpp0x_mode)
  273. {
  274. rawstringdelim = "";
  275. goto extrawstringlit;
  276. }
  277. --YYCURSOR;
  278. BOOST_WAVE_RET(T_IDENTIFIER);
  279. }
  280. ([a-zA-Z_] | UniversalChar) ([a-zA-Z_0-9] | UniversalChar)*
  281. { BOOST_WAVE_RET(T_IDENTIFIER); }
  282. Pound PPSpace ( "include" | "include_next") PPSpace "<" (any\[\n\r>])+ ">"
  283. { BOOST_WAVE_RET(T_PP_HHEADER); }
  284. Pound PPSpace ( "include" | "include_next") PPSpace "\"" (any\[\n\r"])+ "\""
  285. { BOOST_WAVE_RET(T_PP_QHEADER); }
  286. Pound PPSpace ( "include" | "include_next") PPSpace
  287. { BOOST_WAVE_RET(T_PP_INCLUDE); }
  288. Pound PPSpace "if" { BOOST_WAVE_RET(T_PP_IF); }
  289. Pound PPSpace "ifdef" { BOOST_WAVE_RET(T_PP_IFDEF); }
  290. Pound PPSpace "ifndef" { BOOST_WAVE_RET(T_PP_IFNDEF); }
  291. Pound PPSpace "else" { BOOST_WAVE_RET(T_PP_ELSE); }
  292. Pound PPSpace "elif" { BOOST_WAVE_RET(T_PP_ELIF); }
  293. Pound PPSpace "endif" { BOOST_WAVE_RET(T_PP_ENDIF); }
  294. Pound PPSpace "define" { BOOST_WAVE_RET(T_PP_DEFINE); }
  295. Pound PPSpace "undef" { BOOST_WAVE_RET(T_PP_UNDEF); }
  296. Pound PPSpace "line" { BOOST_WAVE_RET(T_PP_LINE); }
  297. Pound PPSpace "error" { BOOST_WAVE_RET(T_PP_ERROR); }
  298. Pound PPSpace "pragma" { BOOST_WAVE_RET(T_PP_PRAGMA); }
  299. Pound PPSpace "warning" { BOOST_WAVE_RET(T_PP_WARNING); }
  300. Pound PPSpace "region" { BOOST_WAVE_RET(T_MSEXT_PP_REGION); }
  301. Pound PPSpace "endregion" { BOOST_WAVE_RET(T_MSEXT_PP_ENDREGION); }
  302. [ \t\v\f]+
  303. { BOOST_WAVE_RET(T_SPACE); }
  304. Newline
  305. {
  306. s->line++;
  307. cursor.column = 1;
  308. BOOST_WAVE_RET(T_NEWLINE);
  309. }
  310. "\000"
  311. {
  312. if (s->eof && cursor != s->eof)
  313. {
  314. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  315. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  316. "invalid character '\\000' in input stream");
  317. }
  318. BOOST_WAVE_RET(T_EOF);
  319. }
  320. any { BOOST_WAVE_RET(TOKEN_FROM_ID(*s->tok, UnknownTokenType)); }
  321. anyctrl
  322. {
  323. // flag the error
  324. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  325. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  326. "invalid character '\\%03o' in input stream", *--YYCURSOR);
  327. }
  328. */
  329. ccomment:
  330. /*!re2c
  331. "*/" { BOOST_WAVE_RET(T_CCOMMENT); }
  332. Newline
  333. {
  334. /*if(cursor == s->eof) BOOST_WAVE_RET(T_EOF);*/
  335. /*s->tok = cursor; */
  336. s->line += count_backslash_newlines(s, cursor) +1;
  337. cursor.column = 1;
  338. goto ccomment;
  339. }
  340. any { goto ccomment; }
  341. "\000"
  342. {
  343. if(cursor == s->eof)
  344. {
  345. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  346. (*s->error_proc)(s, lexing_exception::generic_lexing_warning,
  347. "Unterminated 'C' style comment");
  348. }
  349. else
  350. {
  351. --YYCURSOR; // next call returns T_EOF
  352. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  353. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  354. "invalid character: '\\000' in input stream");
  355. }
  356. }
  357. anyctrl
  358. {
  359. // flag the error
  360. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  361. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  362. "invalid character '\\%03o' in input stream", *--YYCURSOR);
  363. }
  364. */
  365. cppcomment:
  366. /*!re2c
  367. Newline
  368. {
  369. /*if(cursor == s->eof) BOOST_WAVE_RET(T_EOF); */
  370. /*s->tok = cursor; */
  371. s->line++;
  372. cursor.column = 1;
  373. BOOST_WAVE_RET(T_CPPCOMMENT);
  374. }
  375. any { goto cppcomment; }
  376. "\000"
  377. {
  378. if (s->eof && cursor != s->eof)
  379. {
  380. --YYCURSOR; // next call returns T_EOF
  381. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  382. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  383. "invalid character '\\000' in input stream");
  384. }
  385. --YYCURSOR; // next call returns T_EOF
  386. if (!s->single_line_only)
  387. {
  388. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  389. (*s->error_proc)(s, lexing_exception::generic_lexing_warning,
  390. "Unterminated 'C++' style comment");
  391. }
  392. BOOST_WAVE_RET(T_CPPCOMMENT);
  393. }
  394. anyctrl
  395. {
  396. // flag the error
  397. BOOST_WAVE_UPDATE_CURSOR(); // adjust the input cursor
  398. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  399. "invalid character '\\%03o' in input stream", *--YYCURSOR);
  400. }
  401. */
  402. /* this subscanner is called whenever a pp_number has been started */
  403. pp_number:
  404. {
  405. cursor = uchar_wrapper(s->tok = s->cur, s->column = s->curr_column);
  406. marker = uchar_wrapper(s->ptr);
  407. limit = uchar_wrapper(s->lim);
  408. if (s->detect_pp_numbers) {
  409. /*!re2c
  410. "."? Digit (Digit | NonDigit | ExponentStart | ".")*
  411. { BOOST_WAVE_RET(T_PP_NUMBER); }
  412. */
  413. }
  414. else {
  415. /*!re2c
  416. ((FractionalConstant ExponentPart?) | (Digit+ ExponentPart)) FloatingSuffix?
  417. { BOOST_WAVE_RET(T_FLOATLIT); }
  418. Integer { goto integer_suffix; }
  419. */
  420. }
  421. }
  422. /* this subscanner is called, whenever an Integer was recognized */
  423. integer_suffix:
  424. {
  425. if (s->enable_ms_extensions) {
  426. /*!re2c
  427. LongIntegerSuffix | "i64"
  428. { BOOST_WAVE_RET(T_LONGINTLIT); }
  429. IntegerSuffix?
  430. { BOOST_WAVE_RET(T_INTLIT); }
  431. */
  432. }
  433. else {
  434. /*!re2c
  435. LongIntegerSuffix
  436. { BOOST_WAVE_RET(T_LONGINTLIT); }
  437. IntegerSuffix?
  438. { BOOST_WAVE_RET(T_INTLIT); }
  439. */
  440. }
  441. }
  442. /* this subscanner is invoked for C++0x extended character literals */
  443. extcharlit:
  444. {
  445. /*!re2c
  446. ((EscapeSequence | UniversalChar | any\[\n\r\\']) ['])
  447. { BOOST_WAVE_RET(T_CHARLIT); }
  448. any
  449. { BOOST_WAVE_RET(TOKEN_FROM_ID(*s->tok, UnknownTokenType)); }
  450. */
  451. }
  452. /* this subscanner is invoked for C++0x extended character string literals */
  453. extstringlit:
  454. {
  455. /*!re2c
  456. ((EscapeSequence | UniversalChar | any\[\n\r\\"])* ["])
  457. { BOOST_WAVE_RET(T_STRINGLIT); }
  458. any
  459. { BOOST_WAVE_RET(TOKEN_FROM_ID(*s->tok, UnknownTokenType)); }
  460. */
  461. }
  462. extrawstringlit:
  463. {
  464. // we have consumed the double quote but not the lparen
  465. // at this point we may see a delimiter
  466. /*!re2c
  467. * {
  468. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  469. "Invalid character in raw string delimiter ('%c')", yych);
  470. }
  471. // delimiters are any character but parentheses, backslash, and whitespace
  472. any\[()\\\t\v\f\r\n]
  473. {
  474. rawstringdelim += yych;
  475. if (rawstringdelim.size() > 16)
  476. {
  477. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  478. "Raw string delimiter of excessive length (\"%s\") in input stream",
  479. rawstringdelim.c_str());
  480. }
  481. goto extrawstringlit;
  482. }
  483. "("
  484. {
  485. rawstringdelim = ")" + rawstringdelim;
  486. goto extrawstringbody;
  487. }
  488. */
  489. }
  490. extrawstringbody:
  491. {
  492. /*!re2c
  493. * {
  494. (*s->error_proc)(s, lexing_exception::generic_lexing_error,
  495. "Invalid character in raw string body ('%c')", yych);
  496. }
  497. Newline
  498. {
  499. s->line += count_backslash_newlines(s, cursor) +1;
  500. cursor.column = 1;
  501. goto extrawstringbody;
  502. }
  503. (EscapeSequence | UniversalChar | any\["])
  504. {
  505. goto extrawstringbody;
  506. }
  507. ["]
  508. {
  509. // check to see if we have completed a delimiter
  510. if (string_type((char *)(YYCURSOR - rawstringdelim.size() - 1),
  511. (char *)(YYCURSOR - 1)) == rawstringdelim)
  512. {
  513. BOOST_WAVE_RET(T_RAWSTRINGLIT);
  514. } else {
  515. goto extrawstringbody;
  516. }
  517. }
  518. */
  519. }