Commit | Line | Data |
---|---|---|
1adfc4f4 AT |
1 | /* |
2 | * (c) 2019 Aaron Taylor <ataylor at subgeniuskitty dot com> | |
3 | * All rights reserved. | |
4 | */ | |
5 | ||
6 | #include <stdio.h> | |
7 | #include <stdlib.h> | |
8 | #include <unistd.h> | |
9 | #include <string.h> | |
10 | #include <errno.h> | |
11 | #include <stdint.h> | |
12 | #include <sys/select.h> | |
13 | #include <getopt.h> | |
14 | ||
15 | #define VERSION 1 | |
16 | ||
17 | #define STACKSIZE 1024 | |
18 | #define HEAPSIZE 1024 | |
19 | #define RETURNSTACKSIZE 1024 | |
20 | ||
21 | void | |
22 | print_usage(char ** argv) | |
23 | { | |
bf43fa3f | 24 | printf( "VVhitespace Interpreter v%d (www.subgeniuskitty.com)\n" |
1adfc4f4 AT |
25 | "Usage: %s -i <file>\n" |
26 | " -h Help (prints this message)\n" | |
bf43fa3f | 27 | " -i <file> Specify a VVhitespace source file to interpret.\n" |
1adfc4f4 AT |
28 | , VERSION, argv[0] |
29 | ); | |
30 | } | |
31 | ||
32 | int | |
33 | stdin_empty(void) | |
34 | { | |
35 | fd_set read_fds; | |
36 | FD_ZERO(&read_fds); | |
37 | FD_SET(STDIN_FILENO, &read_fds); | |
38 | ||
39 | struct timeval timeout; | |
40 | timeout.tv_sec = 0; | |
41 | timeout.tv_usec = 0; | |
42 | ||
43 | int retval = select(1, &read_fds, NULL, NULL, &timeout); | |
44 | /* retval could be -1. Ignoring that for now. */ | |
45 | if (retval > 0) return 0; | |
46 | return 1; | |
47 | } | |
48 | ||
49 | void | |
50 | ws_die(size_t * pc, char * msg) | |
51 | { | |
1adfc4f4 AT |
52 | printf("SIM_ERROR @ PC %lu: %s\n", *pc, msg); |
53 | fflush(stdout); | |
54 | exit(EXIT_FAILURE); | |
55 | } | |
56 | ||
57 | void | |
58 | stack_push(int32_t ** sp, int32_t word) | |
59 | { | |
60 | *((*sp)++) = word; | |
61 | } | |
62 | ||
63 | int32_t | |
64 | stack_pop(int32_t ** sp) | |
65 | { | |
66 | return *(--(*sp)); | |
67 | } | |
68 | ||
69 | int32_t | |
70 | stack_peek(int32_t ** sp, size_t offset) | |
71 | /* offset=0 peeks TOS, offset=1 peeks NOS, etc. */ | |
72 | { | |
73 | return *((*sp)-offset-1); | |
74 | } | |
75 | ||
76 | uint8_t | |
77 | next_code_byte(uint8_t * code, size_t * pc) | |
78 | { | |
79 | return code[(*pc)++]; | |
80 | } | |
81 | ||
9686c901 AT |
82 | /* |
83 | * In addition to returning the parsed label, this function advances the PC to | |
84 | * the next instruction. | |
85 | */ | |
1adfc4f4 AT |
86 | uint16_t |
87 | parse_label(uint8_t * code, size_t * pc) | |
88 | { | |
89 | uint16_t label = 0; | |
90 | uint8_t c; | |
91 | while ((c = code[(*pc)++]) != '\n') { | |
92 | label = label << 1; | |
93 | if (c == ' ') label++; | |
94 | } | |
9686c901 AT |
95 | // TODO: Where should I handle attempts to access an unitialized label? |
96 | // For now, leave it undefined in a nasal demon sense. | |
1adfc4f4 AT |
97 | return label; |
98 | } | |
99 | ||
bf43fa3f AT |
100 | void |
101 | populate_labels(uint32_t * labels, uint8_t * code, size_t code_size) | |
102 | { | |
9686c901 AT |
103 | size_t cp = 0; |
104 | while (cp <= code_size) { | |
105 | if (code[cp] == '\v') { | |
106 | uint16_t temp_label = parse_label(code, &cp); | |
107 | labels[temp_label] = cp; | |
108 | } | |
109 | cp++; | |
110 | } | |
bf43fa3f AT |
111 | } |
112 | ||
1adfc4f4 AT |
113 | void |
114 | process_imp_stack(uint8_t * code, size_t * pc, int32_t ** sp) | |
115 | { | |
116 | switch (next_code_byte(code,pc)) { | |
117 | case ' ': | |
118 | /* Push number onto TOS. */ | |
119 | { | |
120 | /* First, pick off the sign */ | |
121 | int32_t sign = 0; | |
122 | switch (next_code_byte(code,pc)) { | |
123 | case ' ' : sign = 1; break; | |
124 | case '\t': sign = -1; break; | |
bf43fa3f | 125 | default : ws_die(pc, "expected sign"); break; |
1adfc4f4 AT |
126 | } |
127 | ||
128 | /* Now, construct the number and push to TOS. */ | |
129 | /* I'm assuming the numbers are read MSb first. */ | |
130 | int32_t temp, number = 0; | |
131 | while ((temp = next_code_byte(code,pc)) != '\n') { | |
bf43fa3f | 132 | if (temp == '\v') ws_die(pc, "non-binary digit in number"); |
1adfc4f4 AT |
133 | number <<= 1; |
134 | if (temp == '\t') number++; | |
135 | } | |
136 | stack_push(sp, number*sign); | |
137 | } | |
138 | break; | |
139 | case '\n': | |
140 | /* Stack sub-command */ | |
141 | { | |
142 | switch (next_code_byte(code,pc)) { | |
143 | /* Duplicate the TOS. */ | |
144 | case ' ': | |
145 | stack_push(sp, stack_peek(sp,0)); | |
146 | break; | |
147 | /* Swap TOS and NOS. */ | |
148 | case '\t': | |
149 | { | |
150 | int32_t t1 = stack_pop(sp); | |
151 | int32_t t2 = stack_pop(sp); | |
152 | stack_push(sp, t1); | |
153 | stack_push(sp, t2); | |
154 | } | |
155 | break; | |
156 | /* Discard TOS. */ | |
157 | case '\n': | |
158 | stack_pop(sp); | |
159 | break; | |
bf43fa3f AT |
160 | default: |
161 | ws_die(pc, "malformed stack IMP"); | |
162 | break; | |
1adfc4f4 AT |
163 | } |
164 | } | |
165 | break; | |
bf43fa3f | 166 | default: ws_die(pc, "malformed stack IMP"); break; |
1adfc4f4 AT |
167 | } |
168 | } | |
169 | ||
170 | void | |
171 | process_imp_arithmetic(uint8_t * code, size_t * pc, int32_t ** sp) | |
172 | { | |
173 | int32_t temp; | |
174 | switch (next_code_byte(code,pc)) { | |
175 | case ' ': | |
176 | { | |
177 | switch (next_code_byte(code,pc)) { | |
178 | case ' ': | |
179 | /* Addition */ | |
180 | stack_push(sp, stack_pop(sp)+stack_pop(sp)); | |
181 | break; | |
182 | case '\t': | |
183 | /* Subtraction */ | |
184 | temp = stack_pop(sp); | |
185 | stack_push(sp, stack_pop(sp)-temp); | |
186 | break; | |
187 | case '\n': | |
188 | /* Multiplication */ | |
189 | stack_push(sp, stack_pop(sp)*stack_pop(sp)); | |
190 | break; | |
bf43fa3f AT |
191 | default: |
192 | ws_die(pc, "malformed arithmetic IMP"); | |
193 | break; | |
1adfc4f4 AT |
194 | } |
195 | } | |
196 | break; | |
197 | case '\t': | |
198 | { | |
199 | switch (next_code_byte(code,pc)) { | |
200 | case ' ': | |
201 | /* Division */ | |
202 | temp = stack_pop(sp); | |
203 | stack_push(sp, stack_pop(sp)/temp); | |
204 | break; | |
205 | case '\t': | |
206 | /* Modulo */ | |
207 | temp = stack_pop(sp); | |
208 | stack_push(sp, stack_pop(sp)%temp); | |
209 | break; | |
bf43fa3f | 210 | default: ws_die(pc, "malformed arithmetic IMP"); break; |
1adfc4f4 AT |
211 | } |
212 | } | |
213 | break; | |
bf43fa3f | 214 | default: ws_die(pc, "malformed arithmetic IMP"); break; |
1adfc4f4 AT |
215 | } |
216 | } | |
217 | ||
218 | void | |
219 | process_imp_flowcontrol(uint8_t * code, size_t * pc, int32_t ** sp, uint32_t * labels, | |
220 | uint32_t ** rsp) | |
221 | { | |
222 | switch (next_code_byte(code,pc)) { | |
223 | case '\n': | |
224 | /* Technically another LF is required but we ignore it. */ | |
1adfc4f4 AT |
225 | fflush(stdout); |
226 | exit(EXIT_SUCCESS); | |
227 | case ' ': | |
228 | { | |
229 | switch (next_code_byte(code,pc)) { | |
230 | case ' ': | |
231 | /* Mark a location in the program. */ | |
9686c901 AT |
232 | if (next_code_byte(code,pc) != '\v') ws_die(pc,"expected vtab, " |
233 | "perhaps a whitespace program, rather than vvhitespace?"); | |
234 | /* Jump to next instruction since labels were parsed during startup. */ | |
235 | parse_label( code, pc); | |
1adfc4f4 AT |
236 | break; |
237 | case '\t': | |
238 | /* Call a subroutine. */ | |
239 | *((*rsp)++) = *pc; | |
240 | *pc = labels[parse_label(code, pc)]; | |
241 | break; | |
242 | case '\n': | |
243 | /* Jump unconditionally to a label. */ | |
244 | *pc = labels[parse_label(code, pc)]; | |
245 | break; | |
bf43fa3f AT |
246 | default: |
247 | ws_die(pc, "malformed flow control IMP"); | |
248 | break; | |
1adfc4f4 AT |
249 | } |
250 | } | |
251 | break; | |
252 | case '\t': | |
253 | { | |
254 | switch (next_code_byte(code,pc)) { | |
255 | case ' ': | |
256 | /* Jump to a label if TOS == 0 */ | |
257 | if (stack_peek(sp,0) == 0) *pc = labels[parse_label(code, pc)]; | |
258 | break; | |
259 | case '\t': | |
260 | /* Jump to a label if TOS < 0. */ | |
261 | if (stack_peek(sp,0) < 0) *pc = labels[parse_label(code, pc)]; | |
262 | break; | |
263 | case '\n': | |
264 | /* Return from subroutine. */ | |
265 | *pc = *(--(*rsp)); | |
266 | break; | |
bf43fa3f AT |
267 | default: |
268 | ws_die(pc, "malformed flow control IMP"); | |
269 | break; | |
1adfc4f4 AT |
270 | } |
271 | } | |
272 | break; | |
bf43fa3f AT |
273 | default: |
274 | ws_die(pc, "malformed flow control IMP"); | |
275 | break; | |
1adfc4f4 AT |
276 | } |
277 | } | |
278 | ||
279 | void | |
280 | process_imp_heap(uint8_t * code, size_t * pc, int32_t ** sp, int32_t ** hp) | |
281 | { | |
282 | switch (next_code_byte(code,pc)) { | |
283 | case ' ' : /* Store to heap */ *(*hp + *((*sp)-1)) = **sp; *sp -= 2; break; | |
284 | case '\t': /* Retrieve from heap */ **sp = *(*hp + **sp); break; | |
9686c901 | 285 | default : ws_die(pc, "malformed heap IMP"); break; |
1adfc4f4 AT |
286 | } |
287 | } | |
288 | ||
289 | void | |
290 | process_imp_io(uint8_t * code, size_t * pc, int32_t ** sp, int32_t ** hp) | |
291 | { | |
292 | switch (next_code_byte(code,pc)) { | |
293 | case ' ': | |
294 | /* Output */ | |
295 | { | |
296 | switch (next_code_byte(code,pc)) { | |
297 | case ' ' : /* Output character from TOS */ printf("%c", stack_pop(sp)); break; | |
298 | case '\t': /* Output number from TOS */ printf("%d", stack_pop(sp)); break; | |
9686c901 | 299 | default : ws_die(pc, "malformed output IMP"); break; |
1adfc4f4 AT |
300 | } |
301 | fflush(stdout); | |
302 | } | |
303 | break; | |
304 | case '\t': | |
305 | /* Input */ | |
306 | { | |
307 | while (stdin_empty()) continue; | |
308 | char c = getchar(); | |
309 | switch (next_code_byte(code,pc)) { | |
310 | case '\t': /* Input digit */ c -= '0'; /* fallthrough */ | |
311 | case ' ' : /* Input character */ *(*hp + *((*sp)--)) = c; break; | |
9686c901 | 312 | default : ws_die(pc, "malformed input IMP"); break; |
1adfc4f4 AT |
313 | } |
314 | } | |
315 | break; | |
bf43fa3f | 316 | default: ws_die(pc, "malformed i/o IMP"); break; |
1adfc4f4 AT |
317 | } |
318 | } | |
319 | ||
1adfc4f4 AT |
320 | int |
321 | main(int argc, char ** argv) | |
322 | { | |
323 | /* | |
324 | * Process command line arguments | |
325 | */ | |
326 | int c; | |
327 | FILE * input = NULL; | |
328 | while ((c = getopt(argc,argv,"i:h")) != -1) { | |
329 | switch (c) { | |
330 | case 'i': | |
331 | if ((input = fopen(optarg, "r")) == NULL) { | |
332 | fprintf(stderr, "ERROR: %s: %s\n", optarg, strerror(errno)); | |
333 | } | |
334 | break; | |
335 | case 'h': | |
336 | print_usage(argv); | |
337 | exit(EXIT_SUCCESS); | |
338 | break; | |
339 | default: | |
340 | break; | |
341 | } | |
342 | } | |
343 | if (input == NULL) { | |
bf43fa3f | 344 | fprintf(stderr, "ERROR: Must specify a VVhitespace source file with -f flag.\n"); |
1adfc4f4 AT |
345 | print_usage(argv); |
346 | exit(EXIT_FAILURE); | |
347 | } | |
348 | ||
349 | /* | |
bf43fa3f | 350 | * Read just the VVhitespace source code into memory. |
1adfc4f4 AT |
351 | * We will use the array indices as addresses for the virtual PC when jumping to labels. |
352 | */ | |
353 | size_t ws_code_size = 0; | |
354 | uint8_t temp_byte; | |
355 | while (fread(&temp_byte, 1, 1, input)) { | |
bf43fa3f AT |
356 | if (temp_byte == ' ' || temp_byte == '\t' || temp_byte == '\n' || temp_byte == '\v') { |
357 | ws_code_size++; | |
358 | } | |
1adfc4f4 AT |
359 | } |
360 | rewind(input); | |
361 | uint8_t * ws_code_space = malloc(ws_code_size); | |
362 | ws_code_size = 0; | |
363 | while (fread(&temp_byte, 1, 1, input)) { | |
bf43fa3f AT |
364 | if (temp_byte == ' ' || temp_byte == '\t' || temp_byte == '\n' || temp_byte == '\v') { |
365 | ws_code_space[ws_code_size++] = temp_byte; | |
366 | } | |
1adfc4f4 AT |
367 | } |
368 | fclose(input); | |
369 | ||
370 | /* | |
371 | * Setup a stack and heap. | |
372 | * Assume a 32-bit word size. | |
373 | */ | |
bf43fa3f | 374 | // TODO: Make everything 64-bit. |
1adfc4f4 AT |
375 | int32_t * hp = malloc(HEAPSIZE*4); |
376 | int32_t * sp = malloc(STACKSIZE*4); | |
377 | ||
378 | /* | |
379 | * Setup the return stack and the label array. | |
380 | */ | |
381 | uint32_t * rsp = malloc(RETURNSTACKSIZE*4); | |
bf43fa3f AT |
382 | uint32_t labels[65536] = {0}; |
383 | populate_labels(labels, ws_code_space, ws_code_size); | |
1adfc4f4 AT |
384 | |
385 | /* | |
386 | * Main Loop | |
387 | */ | |
388 | ||
389 | size_t pc = 0; /* Virtual program counter. Operates in the ws_code_space[] address space. */ | |
390 | while (1) { | |
391 | if (pc >= ws_code_size) { | |
bf43fa3f AT |
392 | fprintf(stderr, "SIM_ERROR: PC Overrun\n Requested PC: %lu\n Max Address: %lu\n", |
393 | pc, ws_code_size-1); | |
1adfc4f4 AT |
394 | exit(EXIT_FAILURE); |
395 | } | |
9686c901 AT |
396 | // TODO: Have the SIGTERM signal handler and normal term point return the value |
397 | // on TOS so I can do rudimentary automated tests. | |
1adfc4f4 AT |
398 | |
399 | /* Decode the IMPs */ | |
400 | switch (ws_code_space[pc++]) { | |
401 | case ' ': | |
402 | /* Stack Manipulation */ | |
403 | process_imp_stack(ws_code_space, &pc, &sp); | |
404 | break; | |
405 | case '\n': | |
406 | /* Flow Control */ | |
407 | process_imp_flowcontrol(ws_code_space, &pc, &sp, labels, &rsp); | |
408 | break; | |
409 | case '\t': | |
410 | /* Arithmetic, Heap Access, or I/O */ | |
411 | { | |
412 | switch (ws_code_space[pc++]) { | |
413 | case ' ': | |
414 | /* Arithmetic */ | |
415 | process_imp_arithmetic(ws_code_space, &pc, &sp); | |
416 | break; | |
417 | case '\t': | |
418 | /* Heap Access */ | |
419 | process_imp_heap(ws_code_space, &pc, &sp, &hp); | |
420 | break; | |
421 | case '\n': | |
422 | /* I/O */ | |
423 | process_imp_io(ws_code_space, &pc, &sp, &hp); | |
424 | break; | |
425 | } | |
426 | } | |
427 | break; | |
9686c901 | 428 | default: ws_die(&pc, "unexpected VTab"); break; |
1adfc4f4 AT |
429 | } |
430 | } | |
431 | ||
432 | printf("\n"); | |
433 | printf("Program executed.\n"); | |
434 | ||
435 | exit(EXIT_SUCCESS); | |
436 | } |