1 | /*
|
---|
2 | * Cforall Version 1.0.0 Copyright (C) 2015 University of Waterloo
|
---|
3 | *
|
---|
4 | * The contents of this file are covered under the licence agreement in the
|
---|
5 | * file "LICENCE" distributed with Cforall.
|
---|
6 | *
|
---|
7 | * lex.ll --
|
---|
8 | *
|
---|
9 | * Author : Peter A. Buhr
|
---|
10 | * Created On : Sat Sep 22 08:58:10 2001
|
---|
11 | * Last Modified By : Peter A. Buhr
|
---|
12 | * Last Modified On : Sun Sep 10 22:29:15 2017
|
---|
13 | * Update Count : 620
|
---|
14 | */
|
---|
15 |
|
---|
16 | %option yylineno
|
---|
17 | %option noyywrap
|
---|
18 | %option nounput
|
---|
19 |
|
---|
20 | %{
|
---|
21 | // The lexer assumes the program has been preprocessed by cpp. Hence, all user level preprocessor directive have been
|
---|
22 | // performed and removed from the source. The only exceptions are preprocessor directives passed to the compiler (e.g.,
|
---|
23 | // line-number directives) and C/C++ style comments, which are ignored.
|
---|
24 |
|
---|
25 | //**************************** Includes and Defines ****************************
|
---|
26 |
|
---|
27 | unsigned int column = 0; // position of the end of the last token parsed
|
---|
28 | #define YY_USER_ACTION yylloc.first_line = yylineno; yylloc.first_column = column; column += yyleng; yylloc.last_column = column; yylloc.last_line = yylineno; yylloc.filename = yyfilename ? yyfilename : ""; // trigger before each matching rule's action
|
---|
29 |
|
---|
30 | #include <string>
|
---|
31 | #include <cstdio> // FILENAME_MAX
|
---|
32 | using namespace std;
|
---|
33 |
|
---|
34 | #include "ParseNode.h"
|
---|
35 | #include "TypedefTable.h"
|
---|
36 |
|
---|
37 | char *yyfilename;
|
---|
38 | string *strtext; // accumulate parts of character and string constant value
|
---|
39 |
|
---|
40 | #define RETURN_LOCN(x) yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x )
|
---|
41 | #define RETURN_VAL(x) yylval.tok.str = new string( yytext ); RETURN_LOCN( x )
|
---|
42 | #define RETURN_CHAR(x) yylval.tok.str = nullptr; RETURN_LOCN( x )
|
---|
43 | #define RETURN_STR(x) yylval.tok.str = strtext; RETURN_LOCN( x )
|
---|
44 |
|
---|
45 | #define WHITE_RETURN(x) // do nothing
|
---|
46 | #define NEWLINE_RETURN() column = 0; WHITE_RETURN( '\n' )
|
---|
47 | #define ASCIIOP_RETURN() RETURN_CHAR( (int)yytext[0] ) // single character operator
|
---|
48 | #define NAMEDOP_RETURN(x) RETURN_CHAR( x ) // multichar operator, with a name
|
---|
49 | #define NUMERIC_RETURN(x) rm_underscore(); RETURN_VAL( x ) // numeric constant
|
---|
50 | #define KEYWORD_RETURN(x) RETURN_CHAR( x ) // keyword
|
---|
51 | #define QKEYWORD_RETURN(x) typedefTable.isKind( yytext ); RETURN_VAL(x); // quasi-keyword
|
---|
52 | #define IDENTIFIER_RETURN() RETURN_VAL( typedefTable.isKind( yytext ) )
|
---|
53 | #define ATTRIBUTE_RETURN() RETURN_VAL( ATTR_IDENTIFIER )
|
---|
54 |
|
---|
55 | void rm_underscore() {
|
---|
56 | // Remove underscores in numeric constant by copying the non-underscore characters to the front of the string.
|
---|
57 | yyleng = 0;
|
---|
58 | for ( int i = 0; yytext[i] != '\0'; i += 1 ) {
|
---|
59 | if ( yytext[i] == '`' ) {
|
---|
60 | // copy user suffix
|
---|
61 | for ( ; yytext[i] != '\0'; i += 1 ) {
|
---|
62 | yytext[yyleng] = yytext[i];
|
---|
63 | yyleng += 1;
|
---|
64 | } // for
|
---|
65 | break;
|
---|
66 | } // if
|
---|
67 | if ( yytext[i] != '_' ) {
|
---|
68 | yytext[yyleng] = yytext[i];
|
---|
69 | yyleng += 1;
|
---|
70 | } // if
|
---|
71 | } // for
|
---|
72 | yytext[yyleng] = '\0';
|
---|
73 | }
|
---|
74 |
|
---|
75 | // Stop warning due to incorrectly generated flex code.
|
---|
76 | #pragma GCC diagnostic ignored "-Wsign-compare"
|
---|
77 | %}
|
---|
78 |
|
---|
79 | octal [0-7]
|
---|
80 | nonzero [1-9]
|
---|
81 | decimal [0-9]
|
---|
82 | hex [0-9a-fA-F]
|
---|
83 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2}))
|
---|
84 |
|
---|
85 | // identifier, GCC: $ in identifier
|
---|
86 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})*
|
---|
87 |
|
---|
88 | // attribute identifier, GCC: $ in identifier
|
---|
89 | attr_identifier "@"{identifier}
|
---|
90 |
|
---|
91 | user_suffix_opt ("`"{identifier})?
|
---|
92 |
|
---|
93 | // numeric constants, CFA: '_' in constant
|
---|
94 | hex_quad {hex}("_"?{hex}){3}
|
---|
95 | length ("ll"|"LL"|[lL])|("hh"|"HH"|[hH])
|
---|
96 | integer_suffix_opt ("_"?(([uU]({length}?[iI]?)|([iI]{length}))|([iI]({length}?[uU]?)|([uU]{length}))|({length}([iI]?[uU]?)|([uU][iI]))|[zZ]))?{user_suffix_opt}
|
---|
97 |
|
---|
98 | octal_digits ({octal})|({octal}({octal}|"_")*{octal})
|
---|
99 | octal_prefix "0""_"?
|
---|
100 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt}
|
---|
101 |
|
---|
102 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal})
|
---|
103 | decimal_constant {nonzero_digits}{integer_suffix_opt}
|
---|
104 |
|
---|
105 | hex_digits ({hex})|({hex}({hex}|"_")*{hex})
|
---|
106 | hex_prefix "0"[xX]"_"?
|
---|
107 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt}
|
---|
108 |
|
---|
109 | // GCC: D (double) and iI (imaginary) suffixes, and DL (long double)
|
---|
110 | exponent "_"?[eE]"_"?[+-]?{decimal_digits}
|
---|
111 | floating_suffix ([fFdDlL]?[iI]?)|([iI][lLfFdD])
|
---|
112 | floating_suffix_opt ("_"?({floating_suffix}|"DL"))?{user_suffix_opt}
|
---|
113 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal})
|
---|
114 | floating_decimal {decimal_digits}"."{exponent}?{floating_suffix_opt}
|
---|
115 | floating_fraction "."{decimal_digits}{exponent}?{floating_suffix_opt}
|
---|
116 | floating_constant ({decimal_digits}{exponent}{floating_suffix_opt})|({decimal_digits}{floating_fraction})
|
---|
117 |
|
---|
118 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits}
|
---|
119 | hex_floating_suffix_opt ("_"?({floating_suffix}))?{user_suffix_opt}
|
---|
120 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".")
|
---|
121 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt}
|
---|
122 |
|
---|
123 | // character escape sequence, GCC: \e => esc character
|
---|
124 | simple_escape "\\"[abefnrtv'"?\\]
|
---|
125 | // ' stop editor highlighting
|
---|
126 | octal_escape "\\"{octal}("_"?{octal}){0,2}
|
---|
127 | hex_escape "\\""x""_"?{hex_digits}
|
---|
128 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char}
|
---|
129 | cwide_prefix "L"|"U"|"u"
|
---|
130 | swide_prefix {cwide_prefix}|"u8"
|
---|
131 |
|
---|
132 | // display/white-space characters
|
---|
133 | h_tab [\011]
|
---|
134 | form_feed [\014]
|
---|
135 | v_tab [\013]
|
---|
136 | c_return [\015]
|
---|
137 | h_white [ ]|{h_tab}
|
---|
138 |
|
---|
139 | // overloadable operators
|
---|
140 | op_unary_only "~"|"!"
|
---|
141 | op_unary_binary "+"|"-"|"*"
|
---|
142 | op_unary_pre_post "++"|"--"
|
---|
143 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post}
|
---|
144 |
|
---|
145 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>="
|
---|
146 | op_binary_over {op_unary_binary}|{op_binary_only}
|
---|
147 | // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@="
|
---|
148 | // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over}
|
---|
149 |
|
---|
150 | %x COMMENT
|
---|
151 | %x BKQUOTE
|
---|
152 | %x QUOTE
|
---|
153 | %x STRING
|
---|
154 |
|
---|
155 | %%
|
---|
156 | /* line directives */
|
---|
157 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" {
|
---|
158 | /* " stop editor highlighting */
|
---|
159 | static char filename[FILENAME_MAX]; // temporarily store current source-file name
|
---|
160 | char *end_num;
|
---|
161 | char *begin_string, *end_string;
|
---|
162 | long lineno, length;
|
---|
163 | lineno = strtol( yytext + 1, &end_num, 0 );
|
---|
164 | begin_string = strchr( end_num, '"' );
|
---|
165 | if ( begin_string ) { // file name ?
|
---|
166 | end_string = strchr( begin_string + 1, '"' ); // look for ending delimiter
|
---|
167 | assert( end_string ); // closing quote ?
|
---|
168 | length = end_string - begin_string - 1; // file-name length without quotes or sentinel
|
---|
169 | assert( length < FILENAME_MAX ); // room for sentinel ?
|
---|
170 | memcpy( &filename, begin_string + 1, length ); // copy file name from yytext
|
---|
171 | filename[ length ] = '\0'; // terminate string with sentinel
|
---|
172 | //cout << "file " << filename << " line " << lineno << endl;
|
---|
173 | yylineno = lineno;
|
---|
174 | yyfilename = filename;
|
---|
175 | } // if
|
---|
176 | }
|
---|
177 |
|
---|
178 | /* ignore preprocessor directives (for now) */
|
---|
179 | ^{h_white}*"#"[^\n]*"\n" ;
|
---|
180 |
|
---|
181 | /* ignore C style comments (ALSO HANDLED BY CPP) */
|
---|
182 | "/*" { BEGIN COMMENT; }
|
---|
183 | <COMMENT>.|\n ;
|
---|
184 | <COMMENT>"*/" { BEGIN 0; }
|
---|
185 |
|
---|
186 | /* ignore C++ style comments (ALSO HANDLED BY CPP) */
|
---|
187 | "//"[^\n]*"\n" ;
|
---|
188 |
|
---|
189 | /* ignore whitespace */
|
---|
190 | {h_white}+ { WHITE_RETURN(' '); }
|
---|
191 | ({v_tab}|{c_return}|{form_feed})+ { WHITE_RETURN(' '); }
|
---|
192 | ({h_white}|{v_tab}|{c_return}|{form_feed})*"\n" { NEWLINE_RETURN(); }
|
---|
193 |
|
---|
194 | /* keywords */
|
---|
195 | _Alignas { KEYWORD_RETURN(ALIGNAS); } // C11
|
---|
196 | _Alignof { KEYWORD_RETURN(ALIGNOF); } // C11
|
---|
197 | __alignof { KEYWORD_RETURN(ALIGNOF); } // GCC
|
---|
198 | __alignof__ { KEYWORD_RETURN(ALIGNOF); } // GCC
|
---|
199 | asm { KEYWORD_RETURN(ASM); }
|
---|
200 | __asm { KEYWORD_RETURN(ASM); } // GCC
|
---|
201 | __asm__ { KEYWORD_RETURN(ASM); } // GCC
|
---|
202 | _At { KEYWORD_RETURN(AT); } // CFA
|
---|
203 | _Atomic { KEYWORD_RETURN(ATOMIC); } // C11
|
---|
204 | __attribute { KEYWORD_RETURN(ATTRIBUTE); } // GCC
|
---|
205 | __attribute__ { KEYWORD_RETURN(ATTRIBUTE); } // GCC
|
---|
206 | auto { KEYWORD_RETURN(AUTO); }
|
---|
207 | _Bool { KEYWORD_RETURN(BOOL); } // C99
|
---|
208 | break { KEYWORD_RETURN(BREAK); }
|
---|
209 | case { KEYWORD_RETURN(CASE); }
|
---|
210 | catch { KEYWORD_RETURN(CATCH); } // CFA
|
---|
211 | catchResume { KEYWORD_RETURN(CATCHRESUME); } // CFA
|
---|
212 | char { KEYWORD_RETURN(CHAR); }
|
---|
213 | choose { KEYWORD_RETURN(CHOOSE); } // CFA
|
---|
214 | _Complex { KEYWORD_RETURN(COMPLEX); } // C99
|
---|
215 | __complex { KEYWORD_RETURN(COMPLEX); } // GCC
|
---|
216 | __complex__ { KEYWORD_RETURN(COMPLEX); } // GCC
|
---|
217 | const { KEYWORD_RETURN(CONST); }
|
---|
218 | __const { KEYWORD_RETURN(CONST); } // GCC
|
---|
219 | __const__ { KEYWORD_RETURN(CONST); } // GCC
|
---|
220 | continue { KEYWORD_RETURN(CONTINUE); }
|
---|
221 | coroutine { KEYWORD_RETURN(COROUTINE); } // CFA
|
---|
222 | default { KEYWORD_RETURN(DEFAULT); }
|
---|
223 | disable { KEYWORD_RETURN(DISABLE); } // CFA
|
---|
224 | do { KEYWORD_RETURN(DO); }
|
---|
225 | double { KEYWORD_RETURN(DOUBLE); }
|
---|
226 | dtype { KEYWORD_RETURN(DTYPE); } // CFA
|
---|
227 | else { KEYWORD_RETURN(ELSE); }
|
---|
228 | enable { KEYWORD_RETURN(ENABLE); } // CFA
|
---|
229 | enum { KEYWORD_RETURN(ENUM); }
|
---|
230 | __extension__ { KEYWORD_RETURN(EXTENSION); } // GCC
|
---|
231 | extern { KEYWORD_RETURN(EXTERN); }
|
---|
232 | fallthrough { KEYWORD_RETURN(FALLTHRU); } // CFA
|
---|
233 | fallthru { KEYWORD_RETURN(FALLTHRU); } // CFA
|
---|
234 | finally { KEYWORD_RETURN(FINALLY); } // CFA
|
---|
235 | float { KEYWORD_RETURN(FLOAT); }
|
---|
236 | __float128 { KEYWORD_RETURN(FLOAT); } // GCC
|
---|
237 | for { KEYWORD_RETURN(FOR); }
|
---|
238 | forall { KEYWORD_RETURN(FORALL); } // CFA
|
---|
239 | fortran { KEYWORD_RETURN(FORTRAN); }
|
---|
240 | ftype { KEYWORD_RETURN(FTYPE); } // CFA
|
---|
241 | _Generic { KEYWORD_RETURN(GENERIC); } // C11
|
---|
242 | goto { KEYWORD_RETURN(GOTO); }
|
---|
243 | if { KEYWORD_RETURN(IF); }
|
---|
244 | _Imaginary { KEYWORD_RETURN(IMAGINARY); } // C99
|
---|
245 | __imag { KEYWORD_RETURN(IMAGINARY); } // GCC
|
---|
246 | __imag__ { KEYWORD_RETURN(IMAGINARY); } // GCC
|
---|
247 | inline { KEYWORD_RETURN(INLINE); } // C99
|
---|
248 | __inline { KEYWORD_RETURN(INLINE); } // GCC
|
---|
249 | __inline__ { KEYWORD_RETURN(INLINE); } // GCC
|
---|
250 | int { KEYWORD_RETURN(INT); }
|
---|
251 | __int128 { KEYWORD_RETURN(INT); } // GCC
|
---|
252 | __int128_t { KEYWORD_RETURN(INT); } // GCC
|
---|
253 | __label__ { KEYWORD_RETURN(LABEL); } // GCC
|
---|
254 | long { KEYWORD_RETURN(LONG); }
|
---|
255 | monitor { KEYWORD_RETURN(MONITOR); } // CFA
|
---|
256 | mutex { KEYWORD_RETURN(MUTEX); } // CFA
|
---|
257 | _Noreturn { KEYWORD_RETURN(NORETURN); } // C11
|
---|
258 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); } // GCC
|
---|
259 | one_t { NUMERIC_RETURN(ONE_T); } // CFA
|
---|
260 | otype { KEYWORD_RETURN(OTYPE); } // CFA
|
---|
261 | register { KEYWORD_RETURN(REGISTER); }
|
---|
262 | restrict { KEYWORD_RETURN(RESTRICT); } // C99
|
---|
263 | __restrict { KEYWORD_RETURN(RESTRICT); } // GCC
|
---|
264 | __restrict__ { KEYWORD_RETURN(RESTRICT); } // GCC
|
---|
265 | return { KEYWORD_RETURN(RETURN); }
|
---|
266 | short { KEYWORD_RETURN(SHORT); }
|
---|
267 | signed { KEYWORD_RETURN(SIGNED); }
|
---|
268 | __signed { KEYWORD_RETURN(SIGNED); } // GCC
|
---|
269 | __signed__ { KEYWORD_RETURN(SIGNED); } // GCC
|
---|
270 | sizeof { KEYWORD_RETURN(SIZEOF); }
|
---|
271 | static { KEYWORD_RETURN(STATIC); }
|
---|
272 | _Static_assert { KEYWORD_RETURN(STATICASSERT); } // C11
|
---|
273 | struct { KEYWORD_RETURN(STRUCT); }
|
---|
274 | switch { KEYWORD_RETURN(SWITCH); }
|
---|
275 | thread { KEYWORD_RETURN(THREAD); } // C11
|
---|
276 | _Thread_local { KEYWORD_RETURN(THREADLOCAL); } // C11
|
---|
277 | throw { KEYWORD_RETURN(THROW); } // CFA
|
---|
278 | throwResume { KEYWORD_RETURN(THROWRESUME); } // CFA
|
---|
279 | timeout { QKEYWORD_RETURN(TIMEOUT); } // CFA
|
---|
280 | trait { KEYWORD_RETURN(TRAIT); } // CFA
|
---|
281 | try { KEYWORD_RETURN(TRY); } // CFA
|
---|
282 | ttype { KEYWORD_RETURN(TTYPE); } // CFA
|
---|
283 | typedef { KEYWORD_RETURN(TYPEDEF); }
|
---|
284 | typeof { KEYWORD_RETURN(TYPEOF); } // GCC
|
---|
285 | __typeof { KEYWORD_RETURN(TYPEOF); } // GCC
|
---|
286 | __typeof__ { KEYWORD_RETURN(TYPEOF); } // GCC
|
---|
287 | __uint128_t { KEYWORD_RETURN(INT); } // GCC
|
---|
288 | union { KEYWORD_RETURN(UNION); }
|
---|
289 | unsigned { KEYWORD_RETURN(UNSIGNED); }
|
---|
290 | __builtin_va_list { KEYWORD_RETURN(VALIST); } // GCC
|
---|
291 | virtual { KEYWORD_RETURN(VIRTUAL); } // CFA
|
---|
292 | void { KEYWORD_RETURN(VOID); }
|
---|
293 | volatile { KEYWORD_RETURN(VOLATILE); }
|
---|
294 | __volatile { KEYWORD_RETURN(VOLATILE); } // GCC
|
---|
295 | __volatile__ { KEYWORD_RETURN(VOLATILE); } // GCC
|
---|
296 | waitfor { KEYWORD_RETURN(WAITFOR); }
|
---|
297 | or { QKEYWORD_RETURN(WOR); } // CFA
|
---|
298 | when { KEYWORD_RETURN(WHEN); }
|
---|
299 | while { KEYWORD_RETURN(WHILE); }
|
---|
300 | with { KEYWORD_RETURN(WITH); } // CFA
|
---|
301 | zero_t { NUMERIC_RETURN(ZERO_T); } // CFA
|
---|
302 |
|
---|
303 | /* identifier */
|
---|
304 | {identifier} { IDENTIFIER_RETURN(); }
|
---|
305 | {attr_identifier} { ATTRIBUTE_RETURN(); }
|
---|
306 | "`" { BEGIN BKQUOTE; }
|
---|
307 | <BKQUOTE>{identifier} { IDENTIFIER_RETURN(); }
|
---|
308 | <BKQUOTE>"`" { BEGIN 0; }
|
---|
309 |
|
---|
310 | /* numeric constants */
|
---|
311 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
---|
312 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
---|
313 | {hex_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
---|
314 | {floating_decimal} { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant
|
---|
315 | {floating_fraction} { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant
|
---|
316 | {floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
|
---|
317 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
|
---|
318 |
|
---|
319 | /* character constant, allows empty value */
|
---|
320 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); }
|
---|
321 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); }
|
---|
322 | <QUOTE>['\n]{user_suffix_opt} { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); }
|
---|
323 | /* ' stop editor highlighting */
|
---|
324 |
|
---|
325 | /* string constant */
|
---|
326 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); }
|
---|
327 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); }
|
---|
328 | <STRING>["\n]{user_suffix_opt} { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); }
|
---|
329 | /* " stop editor highlighting */
|
---|
330 |
|
---|
331 | /* common character/string constant */
|
---|
332 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); }
|
---|
333 | <QUOTE,STRING>"\\"{h_white}*"\n" {} // continuation (ALSO HANDLED BY CPP)
|
---|
334 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character
|
---|
335 |
|
---|
336 | /* punctuation */
|
---|
337 | "@" { ASCIIOP_RETURN(); }
|
---|
338 | "[" { ASCIIOP_RETURN(); }
|
---|
339 | "]" { ASCIIOP_RETURN(); }
|
---|
340 | "(" { ASCIIOP_RETURN(); }
|
---|
341 | ")" { ASCIIOP_RETURN(); }
|
---|
342 | "{" { ASCIIOP_RETURN(); }
|
---|
343 | "}" { ASCIIOP_RETURN(); }
|
---|
344 | "," { ASCIIOP_RETURN(); } // also operator
|
---|
345 | ":" { ASCIIOP_RETURN(); }
|
---|
346 | ";" { ASCIIOP_RETURN(); }
|
---|
347 | "." { ASCIIOP_RETURN(); } // also operator
|
---|
348 | "..." { NAMEDOP_RETURN(ELLIPSIS); }
|
---|
349 |
|
---|
350 | /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */
|
---|
351 | "<:" { RETURN_VAL('['); }
|
---|
352 | ":>" { RETURN_VAL(']'); }
|
---|
353 | "<%" { RETURN_VAL('{'); }
|
---|
354 | "%>" { RETURN_VAL('}'); }
|
---|
355 |
|
---|
356 | /* operators */
|
---|
357 | "!" { ASCIIOP_RETURN(); }
|
---|
358 | "+" { ASCIIOP_RETURN(); }
|
---|
359 | "-" { ASCIIOP_RETURN(); }
|
---|
360 | "*" { ASCIIOP_RETURN(); }
|
---|
361 | "\\" { ASCIIOP_RETURN(); } // CFA, exponentiation
|
---|
362 | "/" { ASCIIOP_RETURN(); }
|
---|
363 | "%" { ASCIIOP_RETURN(); }
|
---|
364 | "^" { ASCIIOP_RETURN(); }
|
---|
365 | "~" { ASCIIOP_RETURN(); }
|
---|
366 | "&" { ASCIIOP_RETURN(); }
|
---|
367 | "|" { ASCIIOP_RETURN(); }
|
---|
368 | "<" { ASCIIOP_RETURN(); }
|
---|
369 | ">" { ASCIIOP_RETURN(); }
|
---|
370 | "=" { ASCIIOP_RETURN(); }
|
---|
371 | "?" { ASCIIOP_RETURN(); }
|
---|
372 |
|
---|
373 | "++" { NAMEDOP_RETURN(ICR); }
|
---|
374 | "--" { NAMEDOP_RETURN(DECR); }
|
---|
375 | "==" { NAMEDOP_RETURN(EQ); }
|
---|
376 | "!=" { NAMEDOP_RETURN(NE); }
|
---|
377 | "<<" { NAMEDOP_RETURN(LS); }
|
---|
378 | ">>" { NAMEDOP_RETURN(RS); }
|
---|
379 | "<=" { NAMEDOP_RETURN(LE); }
|
---|
380 | ">=" { NAMEDOP_RETURN(GE); }
|
---|
381 | "&&" { NAMEDOP_RETURN(ANDAND); }
|
---|
382 | "||" { NAMEDOP_RETURN(OROR); }
|
---|
383 | "->" { NAMEDOP_RETURN(ARROW); }
|
---|
384 | "+=" { NAMEDOP_RETURN(PLUSassign); }
|
---|
385 | "-=" { NAMEDOP_RETURN(MINUSassign); }
|
---|
386 | "\\=" { NAMEDOP_RETURN(EXPassign); } // CFA, exponentiation
|
---|
387 | "*=" { NAMEDOP_RETURN(MULTassign); }
|
---|
388 | "/=" { NAMEDOP_RETURN(DIVassign); }
|
---|
389 | "%=" { NAMEDOP_RETURN(MODassign); }
|
---|
390 | "&=" { NAMEDOP_RETURN(ANDassign); }
|
---|
391 | "|=" { NAMEDOP_RETURN(ORassign); }
|
---|
392 | "^=" { NAMEDOP_RETURN(ERassign); }
|
---|
393 | "<<=" { NAMEDOP_RETURN(LSassign); }
|
---|
394 | ">>=" { NAMEDOP_RETURN(RSassign); }
|
---|
395 |
|
---|
396 | "@=" { NAMEDOP_RETURN(ATassign); } // CFA
|
---|
397 |
|
---|
398 | /* CFA, operator identifier */
|
---|
399 | {op_unary}"?" { IDENTIFIER_RETURN(); } // unary
|
---|
400 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); }
|
---|
401 | "^?{}" { IDENTIFIER_RETURN(); }
|
---|
402 | "?`"{identifier} { IDENTIFIER_RETURN(); } // unit operator
|
---|
403 | "?"{op_binary_over}"?" { IDENTIFIER_RETURN(); } // binary
|
---|
404 | /*
|
---|
405 | This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?" can be
|
---|
406 | lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier,
|
---|
407 | e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is
|
---|
408 | a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then
|
---|
409 | reparsing the trailing operator identifier. Otherwise a space is needed between the unary operator and operator
|
---|
410 | identifier to disambiguate this common case.
|
---|
411 |
|
---|
412 | A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers. The ambiguity
|
---|
413 | occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace
|
---|
414 | look-ahead for the routine-call parameter-list to disambiguate. However, the dereference operator must have a
|
---|
415 | parameter/argument to dereference *?(...). Hence, always interpreting the string *?() as * ?() does not preclude
|
---|
416 | any meaningful program.
|
---|
417 |
|
---|
418 | The remaining cases are with the increment/decrement operators and conditional expression:
|
---|
419 |
|
---|
420 | i++? ...(...);
|
---|
421 | i?++ ...(...);
|
---|
422 |
|
---|
423 | requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an
|
---|
424 | incorrect expression (juxtaposed identifiers). Therefore, it is necessary to disambiguate these cases with a
|
---|
425 | space:
|
---|
426 |
|
---|
427 | i++ ? i : 0;
|
---|
428 | i? ++i : 0;
|
---|
429 | */
|
---|
430 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") {
|
---|
431 | // 1 or 2 character unary operator ?
|
---|
432 | int i = yytext[1] == '?' ? 1 : 2;
|
---|
433 | yyless( i ); // put back characters up to first '?'
|
---|
434 | if ( i > 1 ) {
|
---|
435 | NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR );
|
---|
436 | } else {
|
---|
437 | ASCIIOP_RETURN();
|
---|
438 | } // if
|
---|
439 | }
|
---|
440 |
|
---|
441 | /* unknown character */
|
---|
442 | . { yyerror( "unknown character" ); }
|
---|
443 |
|
---|
444 | %%
|
---|
445 | // ----end of lexer----
|
---|
446 |
|
---|
447 | void yyerror( const char * errmsg ) {
|
---|
448 | cout << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1
|
---|
449 | << ": " << SemanticError::error_str() << errmsg << " at token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl;
|
---|
450 | }
|
---|
451 |
|
---|
452 | // Local Variables: //
|
---|
453 | // mode: c++ //
|
---|
454 | // tab-width: 4 //
|
---|
455 | // compile-command: "make install" //
|
---|
456 | // End: //
|
---|