Commit | Line | Data |
---|---|---|
f76eb4c0 KB |
1 | /* |
2 | * Copyright (c) 1982 Regents of the University of California | |
3 | */ | |
4 | #ifndef lint | |
5 | static char sccsid[] = "@(#)asscan2.c 4.14 7/6/83"; | |
6 | #endif not lint | |
7 | ||
8 | #include "asscanl.h" | |
9 | ||
10 | static inttoktype oval = NL; | |
11 | #define ASINBUFSIZ 4096 | |
12 | char inbufunget[8]; | |
13 | char inbuffer[ASINBUFSIZ]; | |
14 | char *Ginbufptr = inbuffer; | |
15 | int Ginbufcnt = 0; | |
16 | int scannerhadeof; | |
17 | ||
18 | fillinbuffer() | |
19 | { | |
20 | int nread; | |
21 | int goal; | |
22 | int got; | |
23 | ||
24 | nread = 0; | |
25 | if (scannerhadeof == 0){ | |
26 | goal = sizeof(inbuffer); | |
27 | do { | |
28 | got = read(stdin->_file, inbuffer + nread, goal); | |
29 | if (got == 0) | |
30 | scannerhadeof = 1; | |
31 | if (got <= 0) | |
32 | break; | |
33 | nread += got; | |
34 | goal -= got; | |
35 | } while (goal); | |
36 | } else { | |
37 | scannerhadeof = 0; | |
38 | } | |
39 | /* | |
40 | * getchar assumes that Ginbufcnt and Ginbufptr | |
41 | * are adjusted as if one character has been removed | |
42 | * from the input. | |
43 | */ | |
44 | if (nread == 0){ | |
45 | inbuffer[0] = EOFCHAR; | |
46 | nread = 1; | |
47 | } | |
48 | Ginbufcnt = nread - 1; | |
49 | Ginbufptr = inbuffer + 1; | |
50 | } | |
51 | ||
52 | scan_dot_s(bufferbox) | |
53 | struct tokbufdesc *bufferbox; | |
54 | { | |
55 | reg char *inbufptr; | |
56 | reg int inbufcnt; | |
57 | reg int ryylval; /* local copy of lexical value */ | |
58 | extern int yylval; /* global copy of lexical value */ | |
59 | reg int val; /* the value returned */ | |
60 | int i; /* simple counter */ | |
61 | reg char *rcp; | |
62 | int ch; /* treated as a character */ | |
63 | int ch1; /* shadow value */ | |
64 | struct symtab *op; | |
65 | reg ptrall bufptr; /* where to stuff tokens */ | |
66 | ptrall bufub; /* where not to stuff tokens */ | |
67 | long intval; /* value of int */ | |
68 | int linescrossed; /* when doing strings and comments */ | |
69 | u_char opstruct; | |
70 | reg int strlg; /* the length of a string */ | |
71 | ||
72 | (bytetoktype *)bufptr = (bytetoktype *) & (bufferbox->toks[0]); | |
73 | (bytetoktype *)bufub = &(bufferbox->toks[AVAILTOKS]); | |
74 | ||
75 | MEMTOREGBUF; | |
76 | if (newfflag){ | |
77 | newfflag = 0; | |
78 | ryylval = (int)savestr(newfname, strlen(newfname)+1, STR_BOTH); | |
79 | ||
80 | ptoken(bufptr, IFILE); | |
81 | ptoken(bufptr, STRING); | |
82 | pptr(bufptr, ryylval); | |
83 | ||
84 | ptoken(bufptr, ILINENO); | |
85 | ptoken(bufptr, INT); | |
86 | pint(bufptr, 1); | |
87 | } | |
88 | ||
89 | while (bufptr < bufub){ | |
90 | loop: | |
91 | switch(ryylval = (type+1)[ch = getchar()]) { | |
92 | case SCANEOF: | |
93 | endoffile: ; | |
94 | inbufptr = 0; | |
95 | ptoken(bufptr, PARSEEOF); | |
96 | goto done; | |
97 | ||
98 | case DIV: /*process C style comments*/ | |
99 | if ( (ch = getchar()) == '*') { /*comment prelude*/ | |
100 | int incomment; | |
101 | linescrossed = 0; | |
102 | incomment = 1; | |
103 | ch = getchar(); /*skip over the * */ | |
104 | while(incomment){ | |
105 | switch(ch){ | |
106 | case '*': | |
107 | ch = getchar(); | |
108 | incomment = (ch != '/'); | |
109 | break; | |
110 | case '\n': | |
111 | scanlineno++; | |
112 | linescrossed++; | |
113 | ch = getchar(); | |
114 | break; | |
115 | case EOFCHAR: | |
116 | goto endoffile; | |
117 | default: | |
118 | ch = getchar(); | |
119 | break; | |
120 | } | |
121 | } | |
122 | val = ILINESKIP; | |
123 | ryylval = linescrossed; | |
124 | goto ret; | |
125 | } else { /*just an ordinary DIV*/ | |
126 | ungetc(ch); | |
127 | val = ryylval = DIV; | |
128 | goto ret; | |
129 | } | |
130 | case SH: | |
131 | if (oval == NL){ | |
132 | /* | |
133 | * Attempt to recognize a C preprocessor | |
134 | * style comment '^#[ \t]*[0-9]*[ \t]*".*" | |
135 | */ | |
136 | ch = getchar(); /*bump the #*/ | |
137 | while (INCHARSET(ch, SPACE)) | |
138 | ch = getchar();/*bump white */ | |
139 | if (INCHARSET(ch, DIGIT)){ | |
140 | intval = 0; | |
141 | while(INCHARSET(ch, DIGIT)){ | |
142 | intval = intval*10 + ch - '0'; | |
143 | ch = getchar(); | |
144 | } | |
145 | while (INCHARSET(ch, SPACE)) | |
146 | ch = getchar(); | |
147 | if (ch == '"' || ch == '\n'){ | |
148 | ptoken(bufptr, ILINENO); | |
149 | ptoken(bufptr, INT); | |
150 | pint(bufptr, intval - 1); | |
151 | if (ch == '"') | |
152 | { | |
153 | ptoken(bufptr, IFILE); | |
154 | /* | |
155 | * The '"' has already been | |
156 | * munched | |
157 | * | |
158 | * eatstr will not eat | |
159 | * the trailing \n, so | |
160 | * it is given to the parser | |
161 | * and counted. | |
162 | */ | |
163 | goto eatstr; | |
164 | } | |
165 | } | |
166 | } | |
167 | } | |
168 | /* | |
169 | * Well, its just an ordinary decadent comment | |
170 | */ | |
171 | while ((ch != '\n') && (ch != EOFCHAR)) | |
172 | ch = getchar(); | |
173 | if (ch == EOFCHAR) | |
174 | goto endoffile; | |
175 | val = ryylval = oval = NL; | |
176 | scanlineno++; | |
177 | goto ret; | |
178 | ||
179 | case NL: | |
180 | scanlineno++; | |
181 | val = ryylval; | |
182 | goto ret; | |
183 | ||
184 | case SP: | |
185 | oval = SP; /*invalidate ^# meta comments*/ | |
186 | goto loop; | |
187 | ||
188 | case REGOP: /* % , could be used as modulo, or register*/ | |
189 | ch = getchar(); | |
190 | if (INCHARSET(ch, DIGIT)){ | |
191 | ryylval = ch-'0'; | |
192 | if (ch=='1') { | |
193 | if (INCHARSET( (ch = getchar()), REGDIGIT)) | |
194 | ryylval = 10+ch-'0'; | |
195 | else | |
196 | ungetc(ch); | |
197 | } | |
198 | /* | |
199 | * God only knows what the original author | |
200 | * wanted this undocumented feature to | |
201 | * do. | |
202 | * %5++ is really r7 | |
203 | */ | |
204 | while(INCHARSET( (ch = getchar()), SIGN)) { | |
205 | if (ch=='+') | |
206 | ryylval++; | |
207 | else | |
208 | ryylval--; | |
209 | } | |
210 | ungetc(ch); | |
211 | val = REG; | |
212 | } else { | |
213 | ungetc(ch); | |
214 | val = REGOP; | |
215 | } | |
216 | goto ret; | |
217 | ||
218 | case ALPH: | |
219 | ch1 = ch; | |
220 | if (INCHARSET(ch, SZSPECBEGIN)){ | |
221 | if( (ch = getchar()) == '`' || ch == '^'){ | |
222 | ch1 |= 0100; /*convert to lower*/ | |
223 | switch(ch1){ | |
224 | case 'b': ryylval = 1; break; | |
225 | case 'w': ryylval = 2; break; | |
226 | case 'l': ryylval = 4; break; | |
227 | default: ryylval = d124; break; | |
228 | } | |
229 | val = SIZESPEC; | |
230 | goto ret; | |
231 | } else { | |
232 | ungetc(ch); | |
233 | ch = ch1; /*restore first character*/ | |
234 | } | |
235 | } | |
236 | rcp = yytext; | |
237 | do { | |
238 | if (rcp < &yytext[NCPName]) | |
239 | *rcp++ = ch; | |
240 | } while (INCHARSET ( (ch = getchar()), ALPHA | DIGIT)); | |
241 | *rcp = '\0'; | |
242 | while (INCHARSET(ch, SPACE)) | |
243 | ch = getchar(); | |
244 | ungetc(ch); | |
245 | ||
246 | switch((op = *lookup(1))->s_tag){ | |
247 | case 0: | |
248 | case LABELID: | |
249 | /* | |
250 | * Its a name... (Labels are subsets of name) | |
251 | */ | |
252 | ryylval = (int)op; | |
253 | val = NAME; | |
254 | break; | |
255 | case INST0: | |
256 | case INSTn: | |
257 | case IJXXX: | |
258 | opstruct = ( (struct instab *)op)->i_opcode; | |
259 | val = op->s_tag; | |
260 | break; | |
261 | default: | |
262 | ryylval = ( (struct instab *)op)->i_opcode; | |
263 | val = op->s_tag; | |
264 | break; | |
265 | } | |
266 | goto ret; | |
267 | ||
268 | case DIG: | |
269 | /* | |
270 | * restore local inbufptr and inbufcnt | |
271 | */ | |
272 | REGTOMEMBUF; | |
273 | val = number(ch); | |
274 | MEMTOREGBUF; | |
275 | /* | |
276 | * yylval or yybignum has been stuffed as a side | |
277 | * effect to number(); get the global yylval | |
278 | * into our fast local copy in case it was an INT. | |
279 | */ | |
280 | ryylval = yylval; | |
281 | goto ret; | |
282 | ||
283 | case LSH: | |
284 | case RSH: | |
285 | /* | |
286 | * We allow the C style operators | |
287 | * << and >>, as well as < and > | |
288 | */ | |
289 | if ( (ch1 = getchar()) != ch) | |
290 | ungetc(ch1); | |
291 | val = ryylval; | |
292 | goto ret; | |
293 | ||
294 | case MINUS: | |
295 | if ( (ch = getchar()) =='(') | |
296 | ryylval=val=MP; | |
297 | else { | |
298 | ungetc(ch); | |
299 | val=MINUS; | |
300 | } | |
301 | goto ret; | |
302 | ||
303 | case SQ: | |
304 | if ((ryylval = getchar()) == '\n') | |
305 | scanlineno++; /*not entirely correct*/ | |
306 | val = INT; | |
307 | goto ret; | |
308 | ||
309 | case DQ: | |
310 | eatstr: | |
311 | linescrossed = 0; | |
312 | for (strlg = 0; /*VOID*/; strlg++){ | |
313 | switch(ch = getchar()){ | |
314 | case '"': | |
315 | goto tailDQ; | |
316 | default: | |
317 | stuff: | |
318 | putc(ch, strfile); | |
319 | break; | |
320 | case '\n': | |
321 | yywarning("New line in a string constant"); | |
322 | scanlineno++; | |
323 | linescrossed++; | |
324 | ch = getchar(); | |
325 | switch(ch){ | |
326 | case EOFCHAR: | |
327 | putc('\n', strfile); | |
328 | ungetc(EOFCHAR); | |
329 | goto tailDQ; | |
330 | default: | |
331 | ungetc(ch); | |
332 | ch = '\n'; | |
333 | goto stuff; | |
334 | } | |
335 | break; | |
336 | ||
337 | case '\\': | |
338 | ch = getchar(); /*skip the '\\'*/ | |
339 | if ( INCHARSET(ch, BSESCAPE)){ | |
340 | switch (ch){ | |
341 | case 'b': ch = '\b'; goto stuff; | |
342 | case 'f': ch = '\f'; goto stuff; | |
343 | case 'n': ch = '\n'; goto stuff; | |
344 | case 'r': ch = '\r'; goto stuff; | |
345 | case 't': ch = '\t'; goto stuff; | |
346 | } | |
347 | } | |
348 | if ( !(INCHARSET(ch, OCTDIGIT)) ) | |
349 | goto stuff; | |
350 | i = 0; | |
351 | intval = 0; | |
352 | while ( (i < 3) && (INCHARSET(ch, OCTDIGIT))){ | |
353 | i++; | |
354 | intval <<= 3; | |
355 | intval += ch - '0'; | |
356 | ch = getchar(); | |
357 | } | |
358 | ungetc(ch); | |
359 | ch = (char)intval; | |
360 | goto stuff; | |
361 | } | |
362 | } | |
363 | tailDQ: ; | |
364 | /* | |
365 | * account for any lines that were crossed | |
366 | */ | |
367 | if (linescrossed){ | |
368 | ptoken(bufptr, ILINESKIP); | |
369 | pint(bufptr, linescrossed); | |
370 | } | |
371 | /* | |
372 | * Cheat: append a trailing null to the string | |
373 | * and then adjust the string length to ignore | |
374 | * the trailing null. If any STRING client requires | |
375 | * the trailing null, the client can just change STRLEN | |
376 | */ | |
377 | putc(0, strfile); | |
378 | ryylval = (int)savestr((char *)0, strlg + 1, STR_FILE); | |
379 | val = STRING; | |
380 | ((struct strdesc *)ryylval)->sd_strlen -= 1; | |
381 | goto ret; | |
382 | ||
383 | case BADCHAR: | |
384 | linescrossed = lineno; | |
385 | lineno = scanlineno; | |
386 | yyerror("Illegal character mapped: %d, char read:(octal) %o", | |
387 | ryylval, ch); | |
388 | lineno = linescrossed; | |
389 | val = BADCHAR; | |
390 | goto ret; | |
391 | ||
392 | default: | |
393 | val = ryylval; | |
394 | goto ret; | |
395 | } /*end of the switch*/ | |
396 | /* | |
397 | * here with one token, so stuff it | |
398 | */ | |
399 | ret: | |
400 | oval = val; | |
401 | ptoken(bufptr, val); | |
402 | switch(val){ | |
403 | case ILINESKIP: | |
404 | pint(bufptr, ryylval); | |
405 | break; | |
406 | case SIZESPEC: | |
407 | pchar(bufptr, ryylval); | |
408 | break; | |
409 | case BFINT: plong(bufptr, ryylval); | |
410 | break; | |
411 | case INT: plong(bufptr, ryylval); | |
412 | break; | |
413 | case BIGNUM: pnumber(bufptr, yybignum); | |
414 | break; | |
415 | case STRING: pptr(bufptr, (int)(char *)ryylval); | |
416 | break; | |
417 | case NAME: pptr(bufptr, (int)(struct symtab *)ryylval); | |
418 | break; | |
419 | case REG: pchar(bufptr, ryylval); | |
420 | break; | |
421 | case INST0: | |
422 | case INSTn: | |
423 | popcode(bufptr, opstruct); | |
424 | break; | |
425 | case IJXXX: | |
426 | popcode(bufptr, opstruct); | |
427 | pptr(bufptr, (int)(struct symtab *)symalloc()); | |
428 | break; | |
429 | case ISTAB: | |
430 | case ISTABSTR: | |
431 | case ISTABNONE: | |
432 | case ISTABDOT: | |
433 | case IALIGN: | |
434 | pptr(bufptr, (int)(struct symtab *)symalloc()); | |
435 | break; | |
436 | /* | |
437 | * default: | |
438 | */ | |
439 | } | |
440 | builtval: ; | |
441 | } /*end of the while to stuff the buffer*/ | |
442 | done: | |
443 | bufferbox->tok_count = (bytetoktype *)bufptr - &(bufferbox->toks[0]); | |
444 | /* | |
445 | * This is a real kludge: | |
446 | * | |
447 | * We put the last token in the buffer to be a MINUS | |
448 | * symbol. This last token will never be picked up | |
449 | * in the normal way, but can be looked at during | |
450 | * a peekahead look that the short circuit expression | |
451 | * evaluator uses to see if an expression is complicated. | |
452 | * | |
453 | * Consider the following situation: | |
454 | * | |
455 | * .word 45 + 47 | |
456 | * buffer 1 | buffer 0 | |
457 | * the peekahead would want to look across the buffer, | |
458 | * but will look in the buffer end zone, see the minus, and | |
459 | * fail. | |
460 | */ | |
461 | ptoken(bufptr, MINUS); | |
462 | REGTOMEMBUF; | |
463 | } |