1 | #include "adlib/lib.h" |
---|
2 | #include "adlib/map.h" |
---|
3 | #include "adlib/set.h" |
---|
4 | #include "pplex.h" |
---|
5 | |
---|
6 | #ifdef GCC |
---|
7 | #define noinline __attribute__((noinlinex)) |
---|
8 | #else |
---|
9 | #define noinline |
---|
10 | #endif |
---|
11 | |
---|
12 | typedef Map<Str *, Str *> Dict; |
---|
13 | |
---|
14 | static Token space, semicolon, eof, static_token, asterisk; |
---|
15 | static Str *decl_var; |
---|
16 | static Str *decl_extern_var; |
---|
17 | static Str *decl_static_var; |
---|
18 | static StrArr *init_list; |
---|
19 | static Dict *class_vars; |
---|
20 | static Dict *class_types; |
---|
21 | static StrSet *type_prefix_set; |
---|
22 | static Dict *namespaced; |
---|
23 | |
---|
24 | enum DeclType { |
---|
25 | ExternDecl, |
---|
26 | StaticDecl, |
---|
27 | NormalDecl |
---|
28 | }; |
---|
29 | |
---|
30 | INIT(DeclParser, { |
---|
31 | GCVar(space, Token(SymWS, S(" "))); |
---|
32 | GCVar(semicolon, Token(SymWS, S(";\n"))); |
---|
33 | GCVar(eof, Token(SymEOF, S(""))); |
---|
34 | GCVar(asterisk, Token(SymAst, S("*"))); |
---|
35 | GCVar(static_token, Token(SymIdent, S("static"))); |
---|
36 | GCVar(decl_var, S("__thread")); |
---|
37 | GCVar(decl_extern_var, S("extern __thread")); |
---|
38 | GCVar(decl_static_var, S("static __thread")); |
---|
39 | GCVar(init_list, A()); |
---|
40 | GCVar(class_vars, new Dict()); |
---|
41 | GCVar(class_types, new Dict()); |
---|
42 | GCVar(type_prefix_set, new StrSet()); |
---|
43 | type_prefix_set->add(S("class")); |
---|
44 | type_prefix_set->add(S("struct")); |
---|
45 | type_prefix_set->add(S("typedef")); |
---|
46 | // special treatment for some gfanlib variables for now. |
---|
47 | GCVar(namespaced, new Dict()); |
---|
48 | namespaced->add(S("MVMachineIntegerOverflow"), S("gfan")); |
---|
49 | namespaced->add(S("lpSolver"), S("gfan")); |
---|
50 | }); |
---|
51 | |
---|
52 | struct State { |
---|
53 | Int pos, marker; |
---|
54 | }; |
---|
55 | |
---|
56 | class Parser : public GC { |
---|
57 | public: |
---|
58 | TokenList *input, *output; |
---|
59 | SourceFile *source; |
---|
60 | Int pos, marker; |
---|
61 | Int init_count; |
---|
62 | noinline Parser(SourceFile *_source) { |
---|
63 | source = _source; |
---|
64 | input = _source->tokens; |
---|
65 | if (input->len() == 0 || input->last().sym != SymEOF) |
---|
66 | input->add(eof); |
---|
67 | output = new TokenList(); |
---|
68 | pos = 0; |
---|
69 | marker = 0; |
---|
70 | } |
---|
71 | bool c_source() { |
---|
72 | return source->filename->ends_with(".c"); |
---|
73 | } |
---|
74 | void skip_until(Word64 syms) { |
---|
75 | syms |= BIT(SymEOF); |
---|
76 | while (!TEST(syms, input->at(pos).sym)) { |
---|
77 | pos++; |
---|
78 | } |
---|
79 | } |
---|
80 | void skip_while(Word64 syms) { |
---|
81 | while (TEST(syms, input->at(pos).sym)) { |
---|
82 | pos++; |
---|
83 | } |
---|
84 | } |
---|
85 | noinline Int find_back_until(Word64 syms) { |
---|
86 | Int p = pos; |
---|
87 | while (p > marker) { |
---|
88 | p--; |
---|
89 | if (TEST(syms, input->at(p).sym)) |
---|
90 | return p; |
---|
91 | } |
---|
92 | return p; |
---|
93 | } |
---|
94 | noinline void safe_skip_until(Word64 syms) { |
---|
95 | // handle parentheses and brackets properly. |
---|
96 | Int par_level = 0; |
---|
97 | Int brkt_level = 0; |
---|
98 | Int brace_level = 0; |
---|
99 | for (;;) { |
---|
100 | Symbol sym = input->at(pos).sym; |
---|
101 | switch (sym) { |
---|
102 | case SymEOF: |
---|
103 | return; |
---|
104 | case SymLPar: |
---|
105 | par_level++; |
---|
106 | break; |
---|
107 | case SymRPar: |
---|
108 | par_level--; |
---|
109 | break; |
---|
110 | case SymLBrkt: |
---|
111 | brkt_level++; |
---|
112 | break; |
---|
113 | case SymRBrkt: |
---|
114 | brkt_level--; |
---|
115 | break; |
---|
116 | case SymLBrace: |
---|
117 | brace_level++; |
---|
118 | break; |
---|
119 | case SymRBrace: |
---|
120 | brace_level--; |
---|
121 | break; |
---|
122 | default: |
---|
123 | if (par_level + brkt_level + brace_level > 0) |
---|
124 | break; |
---|
125 | if (TEST(syms, sym)) { |
---|
126 | return; |
---|
127 | } |
---|
128 | break; |
---|
129 | } |
---|
130 | advance(); |
---|
131 | } |
---|
132 | } |
---|
133 | void skipWhiteSpace() { |
---|
134 | while (BIT(input->at(pos).sym) & SymsWS) { |
---|
135 | pos++; |
---|
136 | } |
---|
137 | } |
---|
138 | void advance() { |
---|
139 | if (pos < input->len()) |
---|
140 | pos++; |
---|
141 | } |
---|
142 | Token &token(Int i) { |
---|
143 | return input->at(i); |
---|
144 | } |
---|
145 | TokenList *tokenRange(Int start, Int end) { |
---|
146 | return input->subarr(start, end - start); |
---|
147 | } |
---|
148 | Token ¤t() { |
---|
149 | return input->at(pos); |
---|
150 | } |
---|
151 | Token &next() { |
---|
152 | return input->at(pos+1); |
---|
153 | } |
---|
154 | Symbol current_sym() { |
---|
155 | return input->at(pos).sym; |
---|
156 | } |
---|
157 | void emit(Token token) { |
---|
158 | output->add(token); |
---|
159 | } |
---|
160 | void emit_gen(Str *s) { |
---|
161 | output->add(Token(SymGen, s)); |
---|
162 | } |
---|
163 | void mark() { |
---|
164 | marker = pos; |
---|
165 | } |
---|
166 | Int current_pos() { |
---|
167 | return pos; |
---|
168 | } |
---|
169 | Int markerPos() { |
---|
170 | return marker; |
---|
171 | } |
---|
172 | noinline void push_marked_until(Int p) { |
---|
173 | while (marker < p) { |
---|
174 | output->add(input->at(marker)); |
---|
175 | marker++; |
---|
176 | } |
---|
177 | } |
---|
178 | noinline void emit_range(Int start, Int end) { |
---|
179 | while (start < end) { |
---|
180 | output->add(input->at(start)); |
---|
181 | start++; |
---|
182 | } |
---|
183 | } |
---|
184 | void emit_tokens(TokenList *tokens) { |
---|
185 | output->add(tokens); |
---|
186 | } |
---|
187 | void push_marked() { |
---|
188 | push_marked_until(pos); |
---|
189 | } |
---|
190 | State save() { |
---|
191 | State result; |
---|
192 | result.pos = pos; |
---|
193 | result.marker = marker; |
---|
194 | return result; |
---|
195 | } |
---|
196 | void restore(State state) { |
---|
197 | pos = state.pos; |
---|
198 | marker = state.marker; |
---|
199 | } |
---|
200 | }; |
---|
201 | |
---|
202 | bool IsLiteral(Parser *parser, Int start, Int end) { |
---|
203 | Int op = 0; |
---|
204 | Int lit = 0; |
---|
205 | for (Int i = start; i < end; i++) { |
---|
206 | Token &token = parser->token(i); |
---|
207 | switch (token.sym) { |
---|
208 | case SymWS: |
---|
209 | case SymEOL: |
---|
210 | case SymComment: |
---|
211 | break; |
---|
212 | case SymOp: |
---|
213 | if (token.str->eq("-") || token.str->eq("+")) |
---|
214 | op++; |
---|
215 | else |
---|
216 | return false; |
---|
217 | break; |
---|
218 | case SymLiteral: |
---|
219 | lit++; |
---|
220 | break; |
---|
221 | default: |
---|
222 | return false; |
---|
223 | } |
---|
224 | } |
---|
225 | return lit >= 1 && op <= 1; |
---|
226 | } |
---|
227 | |
---|
228 | void EmitDecl(Parser *parser, Str *storage_class, |
---|
229 | Int type_start, Int type_end, |
---|
230 | Int var_start, Int var_end, |
---|
231 | Int init_start, Int init_end, Int var_pos, |
---|
232 | bool is_class, bool is_toplevel, DeclType decl_type) { |
---|
233 | bool is_static = (decl_type == StaticDecl); |
---|
234 | bool is_extern = (decl_type == ExternDecl); |
---|
235 | parser->emit(Token(SymGen, storage_class)); |
---|
236 | parser->emit(space); |
---|
237 | parser->emit_range(type_start, type_end); |
---|
238 | parser->emit(space); |
---|
239 | if (is_class) |
---|
240 | parser->emit(asterisk); |
---|
241 | Str *var_name = parser->token(var_pos).str; |
---|
242 | parser->emit_range(var_start, var_pos); |
---|
243 | parser->emit(Token(SymGen, var_name)); |
---|
244 | parser->emit_range(var_pos+1, var_end); |
---|
245 | if (is_static && !is_class && IsLiteral(parser, init_start+1, init_end)) { |
---|
246 | parser->emit(space); |
---|
247 | parser->emit_range(init_start, init_end); |
---|
248 | parser->emit(semicolon); |
---|
249 | return; |
---|
250 | } |
---|
251 | parser->emit(semicolon); |
---|
252 | if (is_class) { |
---|
253 | class_vars->add(var_name, S("(*")->add(var_name)->add(")")); |
---|
254 | class_types->add(var_name, parser->token(type_start).str); |
---|
255 | } |
---|
256 | if ((is_class && !is_extern) || init_start >= 0) { |
---|
257 | if (is_toplevel) { |
---|
258 | init_list->add(var_name); |
---|
259 | } |
---|
260 | Token var_init = Token(SymGen, var_name->clone()->add("__INIT__")); |
---|
261 | parser->emit(static_token); |
---|
262 | parser->emit(space); |
---|
263 | parser->emit_range(type_start, type_end); |
---|
264 | parser->emit(space); |
---|
265 | parser->emit_range(var_start, var_pos); |
---|
266 | parser->emit(var_init); |
---|
267 | parser->emit_range(var_pos+1, var_end); |
---|
268 | if (init_start >= 0) { |
---|
269 | parser->emit(space); |
---|
270 | parser->emit_range(init_start, init_end); |
---|
271 | } |
---|
272 | parser->emit(semicolon); |
---|
273 | if (!is_toplevel && !parser->c_source()) { |
---|
274 | parser->init_count++; |
---|
275 | parser->emit(Token(SymGen, S( |
---|
276 | "void pSingular_register_init_var(void *, void *, long);" |
---|
277 | "class %s__CONSTR__ {\n" |
---|
278 | " public: %s__CONSTR__() {\n" |
---|
279 | " pSingular_register_init_var(&%s, &%s__INIT__, sizeof(%s));\n" |
---|
280 | " }\n" |
---|
281 | "} %s__AUX__;\n" |
---|
282 | )->replace_all(S("%s"), var_name))); |
---|
283 | } |
---|
284 | } |
---|
285 | } |
---|
286 | |
---|
287 | void EmitEpilogue(Parser *parser) { |
---|
288 | if (init_list->len() == 0 && class_vars->count() == 0) |
---|
289 | return; |
---|
290 | Str *modulename = parser->source->modulename; |
---|
291 | TokenList *output = parser->output; |
---|
292 | for (Int i = 0; i < output->len(); i++) { |
---|
293 | Token &token = output->at(i); |
---|
294 | if (token.sym == SymIdent && class_vars->contains(token.str)) { |
---|
295 | Int j = i-1; |
---|
296 | while (j >= 0 && TEST(SymsWS | BIT(SymAst), output->at(j).sym)) |
---|
297 | j--; |
---|
298 | if (j < 0 || !type_prefix_set->contains(output->at(j).str)) |
---|
299 | token.str = class_vars->at(token.str); |
---|
300 | } |
---|
301 | } |
---|
302 | if (init_list->len() == 0) |
---|
303 | return; |
---|
304 | Str *init_part; |
---|
305 | if (parser->c_source()) { |
---|
306 | init_part = S("\n" |
---|
307 | "void pSingular_init_var(const void *s, const void *t, long n);\n" |
---|
308 | "void *pSingular_alloc_var(long n);\n" |
---|
309 | "void pSingular_register_init(void (*f)());\n" |
---|
310 | "static void pSingular_mod_init() {\n" |
---|
311 | ); |
---|
312 | |
---|
313 | } else { |
---|
314 | init_part = S("\n" |
---|
315 | "extern \"C\" {\n" |
---|
316 | "void pSingular_init_var(const void *s, const void *t, long n);\n" |
---|
317 | "void *pSingular_alloc_var(long n);\n" |
---|
318 | "void pSingular_register_init(void (*f)());\n" |
---|
319 | "}\n" |
---|
320 | "typedef struct {\n" |
---|
321 | " void *target; void *source; long size;\n" |
---|
322 | "} pSingular_var_desc;\n" |
---|
323 | "static pSingular_var_desc pSingular_var_descs[%n];\n" |
---|
324 | "static void pSingular_register_init_var(void *t, void *s, long n) {\n" |
---|
325 | " pSingular_var_desc * p = pSingular_var_descs;\n" |
---|
326 | " while (p->target) p++;\n" |
---|
327 | " p->target = t; p->source = s; p->size = n;\n" |
---|
328 | "}\n" |
---|
329 | "static void pSingular_mod_init() {\n" |
---|
330 | ); |
---|
331 | } |
---|
332 | for (Int i = 0; i < init_list->len(); i++) { |
---|
333 | Str *var_name = init_list->at(i); |
---|
334 | if (class_vars->contains(var_name)) { |
---|
335 | Str *type = class_types->at(var_name); |
---|
336 | if (namespaced->contains(var_name)) { |
---|
337 | type = namespaced->at(var_name)->clone()->add("::")->add(type); |
---|
338 | var_name = namespaced->at(var_name)->clone()->add("::")->add(var_name); |
---|
339 | } |
---|
340 | init_part->add(S( |
---|
341 | " %s = (%c *)pSingular_alloc_var((long)sizeof(%c));\n" |
---|
342 | " pSingular_init_var(%s, &%s__INIT__, (long) sizeof(%s));\n" |
---|
343 | )->replace_all(S("%c"), type)->replace_all(S("%s"), var_name)); |
---|
344 | } else { |
---|
345 | init_part->add( |
---|
346 | S(" pSingular_init_var((void *)&%s, (void *)&%s__INIT__, (long) sizeof(%s));\n") |
---|
347 | ->replace_all(S("%s"), var_name) |
---|
348 | ); |
---|
349 | } |
---|
350 | } |
---|
351 | init_part->add("}\n"); |
---|
352 | init_part = init_part->replace_all(S("%n"), S(parser->init_count+1)); |
---|
353 | parser->emit(Token(SymGen, init_part)); |
---|
354 | Str *init_rest; |
---|
355 | if (parser->c_source()) { |
---|
356 | init_rest = S( |
---|
357 | "__attribute__((constructor))" |
---|
358 | "static void pSingular_init_%s(void) {\n" |
---|
359 | " pSingular_register_init(pSingular_mod_init);\n" |
---|
360 | "}\n" |
---|
361 | ); |
---|
362 | } else { |
---|
363 | init_rest = S( |
---|
364 | "static struct pSingular_Init_%s {\n" |
---|
365 | " pSingular_Init_%s() {\n" |
---|
366 | " pSingular_register_init(pSingular_mod_init);\n" |
---|
367 | " }\n" |
---|
368 | "} pSingular_init_%s;\n" |
---|
369 | ); |
---|
370 | } |
---|
371 | init_rest = init_rest->replace_all(S("%s"), modulename); |
---|
372 | parser->emit(Token(SymGen, init_rest)); |
---|
373 | } |
---|
374 | |
---|
375 | void TransformVarDecl(Parser *parser, Str *storage_class, |
---|
376 | bool is_class, bool is_toplevel, DeclType decl_type) { |
---|
377 | // We rewrite: VAR type a, b = init, c; |
---|
378 | // as: |
---|
379 | // storage_class type a; |
---|
380 | // storage_class type b; |
---|
381 | // static type b__INIT__; |
---|
382 | // storage_class type c; |
---|
383 | // Init(b, init); |
---|
384 | State saved = parser->save(); |
---|
385 | Int special_pos = parser->current_pos(); |
---|
386 | parser->current().str = storage_class; // rewrite contents |
---|
387 | parser->advance(); // skip past special token |
---|
388 | parser->skip_while(SymsWS); |
---|
389 | Int type_start = parser->current_pos(); |
---|
390 | parser->advance(); |
---|
391 | parser->skip_while(SymIdent | SymColonColon | SymsWS); |
---|
392 | Int type_end = parser->current_pos(); |
---|
393 | for(;;) { |
---|
394 | Int var_pos = -1, var_start = -1, var_end = -1; |
---|
395 | var_start = parser->current_pos(); |
---|
396 | parser->skip_while(SymsTypePrefix | BIT(SymWS) | BIT(SymClass)); |
---|
397 | if (TEST(SymsEndDecl | BIT(SymLBrkt), parser->current_sym())) { |
---|
398 | // We are one symbol past the initial variable identifier. |
---|
399 | var_pos = parser->find_back_until(BIT(SymIdent)); |
---|
400 | var_end = var_pos + 1; |
---|
401 | } else { |
---|
402 | // We have a function pointer declaration |
---|
403 | State tmp = parser->save(); |
---|
404 | var_start = parser->current_pos(); |
---|
405 | parser->skip_until(BIT(SymIdent)); |
---|
406 | if (parser->current_sym() == SymIdent) |
---|
407 | var_pos = parser->current_pos(); |
---|
408 | parser->restore(tmp); |
---|
409 | parser->safe_skip_until(SymsEndDecl); |
---|
410 | var_end = parser->current_pos(); |
---|
411 | } |
---|
412 | Int init_start = -1, init_end = -1; |
---|
413 | if (parser->current_sym() == SymEqual) { |
---|
414 | init_start = parser->current_pos(); |
---|
415 | parser->advance(); |
---|
416 | parser->safe_skip_until(BIT(SymComma) | BIT(SymSemicolon)); |
---|
417 | init_end = parser->current_pos(); |
---|
418 | } |
---|
419 | switch (parser->current_sym()) { |
---|
420 | case SymComma: |
---|
421 | EmitDecl(parser, storage_class, |
---|
422 | type_start, type_end, var_start, var_end, |
---|
423 | init_start, init_end, var_pos, is_class, is_toplevel, decl_type); |
---|
424 | parser->advance(); |
---|
425 | break; |
---|
426 | case SymSemicolon: |
---|
427 | EmitDecl(parser, storage_class, |
---|
428 | type_start, type_end, var_start, var_end, |
---|
429 | init_start, init_end, var_pos, is_class, is_toplevel, decl_type); |
---|
430 | parser->advance(); |
---|
431 | parser->mark(); |
---|
432 | return; |
---|
433 | case SymEOF: |
---|
434 | return; // error |
---|
435 | default: |
---|
436 | assert(0, "exhaustive switch hits default case"); |
---|
437 | return; |
---|
438 | } |
---|
439 | } |
---|
440 | } |
---|
441 | |
---|
442 | bool IsToplevel(Arr<int> *stack) { |
---|
443 | if (stack->len() == 0) return true; |
---|
444 | for (Int i = 0; i < stack->len(); i++) { |
---|
445 | if (!stack->at(i)) return false; |
---|
446 | } |
---|
447 | return true; |
---|
448 | } |
---|
449 | |
---|
450 | TokenList *Transform(SourceFile *source) { |
---|
451 | Parser *parser = new Parser(source); |
---|
452 | Arr<int> *toplevel = new Arr<int>(); |
---|
453 | int tl = 0; |
---|
454 | while (parser->current().sym != SymEOF) { |
---|
455 | parser->skip_until(SymsSpecial); |
---|
456 | parser->push_marked(); |
---|
457 | bool is_toplevel = IsToplevel(toplevel); |
---|
458 | switch (parser->current().sym) { |
---|
459 | case SymVAR: |
---|
460 | TransformVarDecl(parser, decl_var, |
---|
461 | false, is_toplevel, NormalDecl); |
---|
462 | break; |
---|
463 | case SymEXTERN_VAR: |
---|
464 | parser->current().str = decl_extern_var; |
---|
465 | parser->advance(); |
---|
466 | parser->push_marked(); |
---|
467 | break; |
---|
468 | case SymSTATIC_VAR: |
---|
469 | TransformVarDecl(parser, decl_static_var, |
---|
470 | false, is_toplevel, StaticDecl); |
---|
471 | break; |
---|
472 | case SymINST_VAR: |
---|
473 | TransformVarDecl(parser, decl_var, |
---|
474 | true, is_toplevel, NormalDecl); |
---|
475 | break; |
---|
476 | case SymEXTERN_INST_VAR: |
---|
477 | TransformVarDecl(parser, decl_extern_var, |
---|
478 | true, is_toplevel, ExternDecl); |
---|
479 | break; |
---|
480 | case SymSTATIC_INST_VAR: |
---|
481 | TransformVarDecl(parser, decl_static_var, |
---|
482 | true, is_toplevel, StaticDecl); |
---|
483 | break; |
---|
484 | case SymExtern: |
---|
485 | parser->advance(); |
---|
486 | if (parser->current().sym != SymLiteral) |
---|
487 | break; |
---|
488 | if (!parser->current().str->eq("\"C\"")) |
---|
489 | break; |
---|
490 | tl = 1; |
---|
491 | break; |
---|
492 | case SymNamespace: |
---|
493 | tl = 1; |
---|
494 | parser->advance(); |
---|
495 | break; |
---|
496 | case SymLBrace: |
---|
497 | toplevel->add(tl); |
---|
498 | parser->advance(); |
---|
499 | break; |
---|
500 | case SymRBrace: |
---|
501 | if (toplevel->len() > 0) |
---|
502 | toplevel->pop(); |
---|
503 | tl = 0; |
---|
504 | parser->advance(); |
---|
505 | break; |
---|
506 | case SymEOF: |
---|
507 | break; |
---|
508 | default: |
---|
509 | assert(0, "exhaustive switch hits default case"); |
---|
510 | return NULL; |
---|
511 | } |
---|
512 | } |
---|
513 | EmitEpilogue(parser); |
---|
514 | return parser->output; |
---|
515 | } |
---|
516 | |
---|
517 | Str *TestPreProcessor(Str *filename) { |
---|
518 | SourceFile *source = ReadSource(filename); |
---|
519 | if (!source) { |
---|
520 | return S("ERROR: File not found: ")->add(filename)->add("\n"); |
---|
521 | } |
---|
522 | TokenList *tokens = Transform(source); |
---|
523 | Str *result = new Str(); |
---|
524 | for (Int i = 0; i < tokens->len(); i++) { |
---|
525 | result->add(tokens->at(i).str); |
---|
526 | } |
---|
527 | return result; |
---|
528 | } |
---|
529 | |
---|
530 | Str *RunPreProcessor(Str *filename, Str *filedata) { |
---|
531 | SourceFile *source = ReadSource(filename, filedata); |
---|
532 | TokenList *tokens = Transform(source); |
---|
533 | Str *result = new Str(); for (Int i = 0; i < tokens->len(); i++) { |
---|
534 | result->add(tokens->at(i).str); |
---|
535 | } |
---|
536 | return result; |
---|
537 | } |
---|