Commit | Line | Data |
---|---|---|
0708f913 | 1 | (* Yoann Padioleau |
ae4735db C |
2 | * |
3 | * Copyright (C) 2010, University of Copenhagen DIKU and INRIA. | |
0708f913 | 4 | * Copyright (C) 2006, 2007, 2008 Ecole des Mines de Nantes |
34e49164 C |
5 | * |
6 | * This program is free software; you can redistribute it and/or | |
7 | * modify it under the terms of the GNU General Public License (GPL) | |
8 | * version 2 as published by the Free Software Foundation. | |
ae4735db | 9 | * |
34e49164 C |
10 | * This program is distributed in the hope that it will be useful, |
11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
13 | * file license.txt for more details. | |
14 | *) | |
15 | ||
16 | open Common | |
17 | ||
ae4735db | 18 | module TH = Token_helpers |
34e49164 C |
19 | module LP = Lexer_parser |
20 | ||
485bce71 C |
21 | module Stat = Parsing_stat |
22 | ||
34e49164 C |
23 | (*****************************************************************************) |
24 | (* Wrappers *) | |
25 | (*****************************************************************************) | |
ae4735db C |
26 | let pr2_err, pr2_once = Common.mk_pr2_wrappers Flag_parsing_c.verbose_parsing |
27 | ||
34e49164 C |
28 | (*****************************************************************************) |
29 | (* Helpers *) | |
30 | (*****************************************************************************) | |
31 | ||
ae4735db C |
32 | let lexbuf_to_strpos lexbuf = |
33 | (Lexing.lexeme lexbuf, Lexing.lexeme_start lexbuf) | |
34e49164 | 34 | |
ae4735db | 35 | let token_to_strpos tok = |
34e49164 C |
36 | (TH.str_of_tok tok, TH.pos_of_tok tok) |
37 | ||
38 | ||
ae4735db | 39 | let mk_info_item2 filename toks = |
34e49164 | 40 | let buf = Buffer.create 100 in |
ae4735db | 41 | let s = |
34e49164 C |
42 | (* old: get_slice_file filename (line1, line2) *) |
43 | begin | |
ae4735db | 44 | toks +> List.iter (fun tok -> |
34e49164 | 45 | match TH.pinfo_of_tok tok with |
ae4735db | 46 | | Ast_c.OriginTok _ -> |
91eba41f | 47 | Buffer.add_string buf (TH.str_of_tok tok) |
ae4735db | 48 | | Ast_c.AbstractLineTok _ -> |
91eba41f | 49 | raise Impossible |
34e49164 C |
50 | | _ -> () |
51 | ); | |
52 | Buffer.contents buf | |
53 | end | |
54 | in | |
ae4735db | 55 | (s, toks) |
34e49164 | 56 | |
ae4735db C |
57 | let mk_info_item a b = |
58 | Common.profile_code "C parsing.mk_info_item" | |
34e49164 C |
59 | (fun () -> mk_info_item2 a b) |
60 | ||
61 | ||
ae4735db | 62 | let info_same_line line xs = |
b1b2de81 | 63 | xs +> List.filter (fun info -> Ast_c.line_of_info info =|= line) |
34e49164 | 64 | |
34e49164 | 65 | |
978fd7e5 C |
66 | (* move in cpp_token_c ? *) |
67 | let is_define_passed passed = | |
68 | let xs = passed +> List.rev +> List.filter TH.is_not_comment in | |
ae4735db C |
69 | if List.length xs >= 2 |
70 | then | |
978fd7e5 | 71 | (match Common.head_middle_tail xs with |
ae4735db | 72 | | Parser_c.TDefine _, _, Parser_c.TDefEOL _ -> |
978fd7e5 C |
73 | true |
74 | | _ -> false | |
75 | ) | |
76 | else begin | |
77 | pr2_err "WEIRD: length list of error recovery tokens < 2 "; | |
ae4735db | 78 | false |
978fd7e5 C |
79 | end |
80 | ||
81 | ||
82 | (*****************************************************************************) | |
83 | (* Error diagnostic *) | |
84 | (*****************************************************************************) | |
85 | ||
ae4735db | 86 | let error_msg_tok tok = |
978fd7e5 C |
87 | let file = TH.file_of_tok tok in |
88 | if !Flag_parsing_c.verbose_parsing | |
ae4735db | 89 | then Common.error_message file (token_to_strpos tok) |
978fd7e5 C |
90 | else ("error in " ^ file ^ "; set verbose_parsing for more info") |
91 | ||
92 | ||
ae4735db | 93 | let print_bad line_error (start_line, end_line) filelines = |
978fd7e5 C |
94 | begin |
95 | pr2 ("badcount: " ^ i_to_s (end_line - start_line)); | |
96 | ||
ae4735db C |
97 | for i = start_line to end_line do |
98 | let line = filelines.(i) in | |
978fd7e5 | 99 | |
ae4735db C |
100 | if i =|= line_error |
101 | then pr2 ("BAD:!!!!!" ^ " " ^ line) | |
102 | else pr2 ("bad:" ^ " " ^ line) | |
978fd7e5 C |
103 | done |
104 | end | |
105 | ||
106 | ||
34e49164 C |
107 | (*****************************************************************************) |
108 | (* Stats on what was passed/commentized *) | |
109 | (*****************************************************************************) | |
110 | ||
c491d8ee | 111 | let commentized xs = xs +> Common.tail_map_filter (function |
ae4735db | 112 | | Parser_c.TCommentCpp (cppkind, ii) -> |
485bce71 | 113 | let s = Ast_c.str_of_info ii in |
ae4735db C |
114 | let legal_passing = |
115 | match !Flag_parsing_c.filter_passed_level with | |
485bce71 | 116 | | 0 -> false |
ae4735db | 117 | | 1 -> |
0708f913 | 118 | List.mem cppkind [Token_c.CppAttr] |
ae4735db | 119 | || |
485bce71 | 120 | (s =~ "__.*") |
ae4735db | 121 | | 2 -> |
0708f913 | 122 | List.mem cppkind [Token_c.CppAttr;Token_c.CppPassingNormal] |
ae4735db | 123 | || |
485bce71 | 124 | (s =~ "__.*") |
ae4735db | 125 | | 3 -> |
0708f913 | 126 | List.mem cppkind [Token_c.CppAttr;Token_c.CppPassingNormal;Token_c.CppDirective] |
ae4735db | 127 | || |
485bce71 | 128 | (s =~ "__.*") |
ae4735db | 129 | | 4 -> |
0708f913 | 130 | List.mem cppkind [Token_c.CppAttr;Token_c.CppPassingNormal;Token_c.CppMacro] |
ae4735db | 131 | || |
485bce71 C |
132 | (s =~ "__.*") |
133 | ||
134 | ||
ae4735db | 135 | | 5 -> |
0708f913 | 136 | List.mem cppkind [Token_c.CppAttr;Token_c.CppPassingNormal;Token_c.CppDirective;Token_c.CppMacro] |
ae4735db | 137 | || |
485bce71 C |
138 | (s =~ "__.*") |
139 | ||
140 | ||
141 | ||
142 | ||
143 | | _ -> failwith "not valid level passing number" | |
144 | in | |
145 | if legal_passing then None else Some (ii.Ast_c.pinfo) | |
146 | ||
147 | (* | |
ae4735db | 148 | | Ast_c.CppOther -> |
34e49164 C |
149 | (match s with |
150 | | s when s =~ "KERN_.*" -> None | |
151 | | s when s =~ "__.*" -> None | |
ae4735db | 152 | | _ -> |
485bce71 | 153 | Some (ii.Ast_c.pinfo) |
34e49164 | 154 | ) |
485bce71 C |
155 | *) |
156 | ||
ae4735db | 157 | |
34e49164 | 158 | | Parser_c.TCommentMisc ii |
ae4735db | 159 | | Parser_c.TAction ii |
34e49164 C |
160 | -> |
161 | Some (ii.Ast_c.pinfo) | |
ae4735db | 162 | | _ -> |
34e49164 C |
163 | None |
164 | ) | |
ae4735db C |
165 | |
166 | let count_lines_commentized xs = | |
34e49164 C |
167 | let line = ref (-1) in |
168 | let count = ref 0 in | |
169 | begin | |
170 | commentized xs +> | |
171 | List.iter | |
172 | (function | |
ae4735db | 173 | Ast_c.OriginTok pinfo | Ast_c.ExpandedTok (_,(pinfo,_)) -> |
34e49164 C |
174 | let newline = pinfo.Common.line in |
175 | if newline <> !line | |
176 | then begin | |
177 | line := newline; | |
178 | incr count | |
179 | end | |
180 | | _ -> ()); | |
181 | !count | |
182 | end | |
183 | ||
184 | ||
185 | ||
ae4735db | 186 | let print_commentized xs = |
34e49164 C |
187 | let line = ref (-1) in |
188 | begin | |
189 | let ys = commentized xs in | |
190 | ys +> | |
191 | List.iter | |
192 | (function | |
ae4735db | 193 | Ast_c.OriginTok pinfo | Ast_c.ExpandedTok (_,(pinfo,_)) -> |
34e49164 C |
194 | let newline = pinfo.Common.line in |
195 | let s = pinfo.Common.str in | |
ae4735db C |
196 | let s = Str.global_substitute |
197 | (Str.regexp "\n") (fun s -> "") s | |
34e49164 | 198 | in |
b1b2de81 | 199 | if newline =|= !line |
34e49164 C |
200 | then prerr_string (s ^ " ") |
201 | else begin | |
ae4735db C |
202 | if !line =|= -1 |
203 | then pr2_no_nl "passed:" | |
34e49164 C |
204 | else pr2_no_nl "\npassed:"; |
205 | line := newline; | |
206 | pr2_no_nl (s ^ " "); | |
207 | end | |
208 | | _ -> ()); | |
209 | if not (null ys) then pr2 ""; | |
210 | end | |
ae4735db | 211 | |
34e49164 C |
212 | |
213 | ||
214 | ||
215 | (*****************************************************************************) | |
216 | (* Lexing only *) | |
217 | (*****************************************************************************) | |
218 | ||
219 | (* called by parse_print_error_heuristic *) | |
ae4735db | 220 | let tokens2 file = |
708f4980 | 221 | let table = Common.full_charpos_to_pos_large file in |
34e49164 | 222 | |
ae4735db | 223 | Common.with_open_infile file (fun chan -> |
34e49164 | 224 | let lexbuf = Lexing.from_channel chan in |
ae4735db C |
225 | try |
226 | let rec tokens_aux acc = | |
34e49164 C |
227 | let tok = Lexer_c.token lexbuf in |
228 | (* fill in the line and col information *) | |
ae4735db | 229 | let tok = tok +> TH.visitor_info_of_tok (fun ii -> |
34e49164 C |
230 | { ii with Ast_c.pinfo= |
231 | (* could assert pinfo.filename = file ? *) | |
232 | match Ast_c.pinfo_of_info ii with | |
233 | Ast_c.OriginTok pi -> | |
708f4980 | 234 | Ast_c.OriginTok (Common.complete_parse_info_large file table pi) |
34e49164 | 235 | | Ast_c.ExpandedTok (pi,vpi) -> |
708f4980 | 236 | Ast_c.ExpandedTok((Common.complete_parse_info_large file table pi),vpi) |
34e49164 C |
237 | | Ast_c.FakeTok (s,vpi) -> Ast_c.FakeTok (s,vpi) |
238 | | Ast_c.AbstractLineTok pi -> failwith "should not occur" | |
239 | }) | |
240 | in | |
241 | ||
242 | if TH.is_eof tok | |
243 | then List.rev (tok::acc) | |
244 | else tokens_aux (tok::acc) | |
245 | in | |
246 | tokens_aux [] | |
247 | with | |
ae4735db C |
248 | | Lexer_c.Lexical s -> |
249 | failwith ("lexical error " ^ s ^ "\n =" ^ | |
34e49164 C |
250 | (Common.error_message file (lexbuf_to_strpos lexbuf))) |
251 | | e -> raise e | |
252 | ) | |
253 | ||
ae4735db C |
254 | let time_lexing ?(profile=true) a = |
255 | if profile | |
485bce71 | 256 | then Common.profile_code_exclusif "LEXING" (fun () -> tokens2 a) |
951c7801 | 257 | else tokens2 a |
ae4735db | 258 | let tokens ?profile a = |
485bce71 | 259 | Common.profile_code "C parsing.tokens" (fun () -> time_lexing ?profile a) |
34e49164 C |
260 | |
261 | ||
ae4735db | 262 | let tokens_of_string string = |
34e49164 | 263 | let lexbuf = Lexing.from_string string in |
ae4735db C |
264 | try |
265 | let rec tokens_s_aux () = | |
34e49164 C |
266 | let tok = Lexer_c.token lexbuf in |
267 | if TH.is_eof tok | |
268 | then [tok] | |
269 | else tok::(tokens_s_aux ()) | |
270 | in | |
271 | tokens_s_aux () | |
272 | with | |
273 | | Lexer_c.Lexical s -> failwith ("lexical error " ^ s ^ "\n =" ) | |
274 | | e -> raise e | |
275 | ||
276 | ||
277 | (*****************************************************************************) | |
278 | (* Parsing, but very basic, no more used *) | |
279 | (*****************************************************************************) | |
280 | ||
281 | (* | |
282 | * !!!Those function use refs, and are not reentrant !!! so take care. | |
283 | * It use globals defined in Lexer_parser. | |
ae4735db | 284 | * |
34e49164 C |
285 | * update: because now lexer return comments tokens, those functions |
286 | * may not work anymore. | |
287 | *) | |
288 | ||
ae4735db | 289 | let parse file = |
34e49164 C |
290 | let lexbuf = Lexing.from_channel (open_in file) in |
291 | let result = Parser_c.main Lexer_c.token lexbuf in | |
292 | result | |
293 | ||
294 | ||
ae4735db | 295 | let parse_print_error file = |
34e49164 C |
296 | let chan = (open_in file) in |
297 | let lexbuf = Lexing.from_channel chan in | |
298 | ||
299 | let error_msg () = Common.error_message file (lexbuf_to_strpos lexbuf) in | |
ae4735db | 300 | try |
34e49164 | 301 | lexbuf +> Parser_c.main Lexer_c.token |
ae4735db C |
302 | with |
303 | | Lexer_c.Lexical s -> | |
34e49164 | 304 | failwith ("lexical error " ^s^ "\n =" ^ error_msg ()) |
ae4735db | 305 | | Parsing.Parse_error -> |
34e49164 | 306 | failwith ("parse error \n = " ^ error_msg ()) |
ae4735db | 307 | | Semantic_c.Semantic (s, i) -> |
34e49164 C |
308 | failwith ("semantic error " ^ s ^ "\n =" ^ error_msg ()) |
309 | | e -> raise e | |
310 | ||
311 | ||
312 | ||
313 | ||
314 | (*****************************************************************************) | |
315 | (* Parsing subelements, useful to debug parser *) | |
316 | (*****************************************************************************) | |
317 | ||
318 | (* | |
319 | * !!!Those function use refs, and are not reentrant !!! so take care. | |
320 | * It use globals defined in Lexer_parser. | |
321 | *) | |
322 | ||
323 | ||
ae4735db C |
324 | (* old: |
325 | * let parse_gen parsefunc s = | |
34e49164 C |
326 | * let lexbuf = Lexing.from_string s in |
327 | * let result = parsefunc Lexer_c.token lexbuf in | |
328 | * result | |
329 | *) | |
330 | ||
ae4735db | 331 | let parse_gen parsefunc s = |
485bce71 | 332 | let toks = tokens_of_string s +> List.filter TH.is_not_comment in |
34e49164 C |
333 | |
334 | ||
335 | (* Why use this lexing scheme ? Why not classically give lexer func | |
ae4735db | 336 | * to parser ? Because I now keep comments in lexer. Could |
34e49164 C |
337 | * just do a simple wrapper that when comment ask again for a token, |
338 | * but maybe simpler to use cur_tok technique. | |
339 | *) | |
340 | let all_tokens = ref toks in | |
341 | let cur_tok = ref (List.hd !all_tokens) in | |
342 | ||
ae4735db C |
343 | let lexer_function = |
344 | (fun _ -> | |
34e49164 | 345 | if TH.is_eof !cur_tok |
708f4980 | 346 | then (pr2_err "LEXER: ALREADY AT END"; !cur_tok) |
34e49164 C |
347 | else |
348 | let v = Common.pop2 all_tokens in | |
349 | cur_tok := v; | |
350 | !cur_tok | |
ae4735db | 351 | ) |
34e49164 C |
352 | in |
353 | let lexbuf_fake = Lexing.from_function (fun buf n -> raise Impossible) in | |
354 | let result = parsefunc lexer_function lexbuf_fake in | |
355 | result | |
356 | ||
357 | ||
358 | let type_of_string = parse_gen Parser_c.type_name | |
359 | let statement_of_string = parse_gen Parser_c.statement | |
360 | let expression_of_string = parse_gen Parser_c.expr | |
361 | ||
362 | (* ex: statement_of_string "(struct us_data* )psh->hostdata = NULL;" *) | |
363 | ||
364 | ||
365 | ||
366 | ||
367 | ||
34e49164 | 368 | (*****************************************************************************) |
485bce71 | 369 | (* Parsing default define macros, usually in a standard.h file *) |
34e49164 C |
370 | (*****************************************************************************) |
371 | ||
ae4735db C |
372 | let extract_macros2 file = |
373 | Common.save_excursion Flag_parsing_c.verbose_lexing (fun () -> | |
708f4980 C |
374 | Flag_parsing_c.verbose_lexing := false; |
375 | let toks = tokens ~profile:false file in | |
978fd7e5 C |
376 | let toks = Parsing_hacks.fix_tokens_define toks in |
377 | Cpp_token_c.extract_macros toks | |
708f4980 | 378 | ) |
34e49164 | 379 | |
ae4735db | 380 | let extract_macros a = |
978fd7e5 | 381 | Common.profile_code_exclusif "HACK" (fun () -> extract_macros2 a) |
485bce71 | 382 | |
34e49164 C |
383 | |
384 | (*****************************************************************************) | |
978fd7e5 | 385 | (* Helper for main entry point *) |
34e49164 C |
386 | (*****************************************************************************) |
387 | ||
34e49164 C |
388 | |
389 | (* The use of local refs (remaining_tokens, passed_tokens, ...) makes | |
390 | * possible error recovery. Indeed, they allow to skip some tokens and | |
391 | * still be able to call again the ocamlyacc parser. It is ugly code | |
392 | * because we cant modify ocamllex and ocamlyacc. As we want some | |
393 | * extended lexing tricks, we have to use such refs. | |
ae4735db | 394 | * |
34e49164 C |
395 | * Those refs are now also used for my lalr(k) technique. Indeed They |
396 | * store the futur and previous tokens that were parsed, and so | |
397 | * provide enough context information for powerful lex trick. | |
ae4735db | 398 | * |
34e49164 | 399 | * - passed_tokens_last_ckp stores the passed tokens since last |
485bce71 | 400 | * checkpoint. Used for NotParsedCorrectly and also to build the |
34e49164 C |
401 | * info_item attached to each program_element. |
402 | * - passed_tokens_clean is used for lookahead, in fact for lookback. | |
403 | * - remaining_tokens_clean is used for lookahead. Now remaining_tokens | |
404 | * contain some comments and so would make pattern matching difficult | |
ae4735db | 405 | * in lookahead. Hence this variable. We would like also to get rid |
34e49164 C |
406 | * of cpp instruction because sometimes a cpp instruction is between |
407 | * two tokens and makes a pattern matching fail. But lookahead also | |
408 | * transform some cpp instruction (in comment) so can't remove them. | |
ae4735db | 409 | * |
34e49164 C |
410 | * So remaining_tokens, passed_tokens_last_ckp contain comment-tokens, |
411 | * whereas passed_tokens_clean and remaining_tokens_clean does not contain | |
412 | * comment-tokens. | |
ae4735db | 413 | * |
34e49164 | 414 | * Normally we have: |
ae4735db | 415 | * toks = (reverse passed_tok) ++ cur_tok ++ remaining_tokens |
34e49164 | 416 | * after the call to pop2. |
ae4735db | 417 | * toks = (reverse passed_tok) ++ remaining_tokens |
34e49164 C |
418 | * at the and of the lexer_function call. |
419 | * At the very beginning, cur_tok and remaining_tokens overlap, but not after. | |
420 | * At the end of lexer_function call, cur_tok overlap with passed_tok. | |
ae4735db | 421 | * |
34e49164 | 422 | * convention: I use "tr" for "tokens refs" |
ae4735db | 423 | * |
485bce71 C |
424 | * I now also need this lexing trick because the lexer return comment |
425 | * tokens. | |
34e49164 C |
426 | *) |
427 | ||
428 | type tokens_state = { | |
429 | mutable rest : Parser_c.token list; | |
430 | mutable rest_clean : Parser_c.token list; | |
431 | mutable current : Parser_c.token; | |
432 | (* it's passed since last "checkpoint", not passed from the beginning *) | |
433 | mutable passed : Parser_c.token list; | |
434 | mutable passed_clean : Parser_c.token list; | |
435 | } | |
708f4980 | 436 | |
ae4735db C |
437 | let mk_tokens_state toks = |
438 | { | |
708f4980 C |
439 | rest = toks; |
440 | rest_clean = (toks +> List.filter TH.is_not_comment); | |
441 | current = (List.hd toks); | |
ae4735db | 442 | passed = []; |
708f4980 C |
443 | passed_clean = []; |
444 | } | |
445 | ||
446 | ||
447 | ||
ae4735db | 448 | let clone_tokens_state tr = |
485bce71 C |
449 | { rest = tr.rest; |
450 | rest_clean = tr.rest_clean; | |
451 | current = tr.current; | |
452 | passed = tr.passed; | |
453 | passed_clean = tr.passed_clean; | |
454 | } | |
ae4735db | 455 | let copy_tokens_state ~src ~dst = |
485bce71 C |
456 | dst.rest <- src.rest; |
457 | dst.rest_clean <- src.rest_clean; | |
458 | dst.current <- src.current; | |
459 | dst.passed <- src.passed; | |
460 | dst.passed_clean <- src.passed_clean; | |
461 | () | |
462 | ||
708f4980 | 463 | (* todo? agglomerate the x##b ? *) |
485bce71 C |
464 | let rec filter_noise n xs = |
465 | match n, xs with | |
466 | | _, [] -> [] | |
467 | | 0, xs -> xs | |
ae4735db | 468 | | n, x::xs -> |
485bce71 | 469 | (match x with |
ae4735db | 470 | | Parser_c.TMacroAttr _ -> |
485bce71 | 471 | filter_noise (n-1) xs |
ae4735db | 472 | | _ -> |
485bce71 C |
473 | x::filter_noise (n-1) xs |
474 | ) | |
475 | ||
ae4735db | 476 | let clean_for_lookahead xs = |
485bce71 C |
477 | match xs with |
478 | | [] -> [] | |
479 | | [x] -> [x] | |
ae4735db | 480 | | x::xs -> |
485bce71 C |
481 | x::filter_noise 10 xs |
482 | ||
34e49164 | 483 | |
485bce71 | 484 | |
ae4735db | 485 | (* Hacked lex. This function use refs passed by parse_print_error_heuristic |
485bce71 C |
486 | * tr means token refs. |
487 | *) | |
ae4735db | 488 | let rec lexer_function ~pass tr = fun lexbuf -> |
34e49164 | 489 | match tr.rest with |
708f4980 | 490 | | [] -> pr2_err "ALREADY AT END"; tr.current |
ae4735db | 491 | | v::xs -> |
34e49164 C |
492 | tr.rest <- xs; |
493 | tr.current <- v; | |
494 | ||
495 | if !Flag_parsing_c.debug_lexer then Common.pr2_gen v; | |
496 | ||
497 | if TH.is_comment v | |
498 | then begin | |
499 | tr.passed <- v::tr.passed; | |
485bce71 | 500 | lexer_function ~pass tr lexbuf |
34e49164 C |
501 | end |
502 | else begin | |
503 | let x = List.hd tr.rest_clean in | |
504 | tr.rest_clean <- List.tl tr.rest_clean; | |
b1b2de81 | 505 | assert (x =*= v); |
ae4735db | 506 | |
34e49164 | 507 | (match v with |
113803cf | 508 | |
ae4735db | 509 | (* fix_define1. |
113803cf C |
510 | * |
511 | * Why not in parsing_hacks lookahead and do passing like | |
ae4735db C |
512 | * I do for some ifdef directives ? Because here I also need to |
513 | * generate some tokens sometimes and so I need access to the | |
113803cf | 514 | * tr.passed, tr.rest, etc. |
485bce71 | 515 | *) |
9f8e26f4 | 516 | | Parser_c.TDefine (tok) -> |
ae4735db | 517 | if not (LP.current_context () =*= LP.InTopLevel) && |
708f4980 | 518 | (!Flag_parsing_c.cpp_directive_passing || (pass >= 2)) |
34e49164 | 519 | then begin |
485bce71 | 520 | incr Stat.nDefinePassing; |
34e49164 | 521 | pr2_once ("CPP-DEFINE: inside function, I treat it as comment"); |
0708f913 | 522 | let v' = Parser_c.TCommentCpp (Token_c.CppDirective,TH.info_of_tok v) |
34e49164 C |
523 | in |
524 | tr.passed <- v'::tr.passed; | |
978fd7e5 C |
525 | tr.rest <- Parsing_hacks.comment_until_defeol tr.rest; |
526 | tr.rest_clean <- Parsing_hacks.drop_until_defeol tr.rest_clean; | |
485bce71 | 527 | lexer_function ~pass tr lexbuf |
34e49164 C |
528 | end |
529 | else begin | |
530 | tr.passed <- v::tr.passed; | |
531 | tr.passed_clean <- v::tr.passed_clean; | |
532 | v | |
533 | end | |
ae4735db C |
534 | |
535 | | Parser_c.TInclude (includes, filename, inifdef, info) -> | |
b1b2de81 | 536 | if not (LP.current_context () =*= LP.InTopLevel) && |
708f4980 | 537 | (!Flag_parsing_c.cpp_directive_passing || (pass >= 2)) |
34e49164 | 538 | then begin |
485bce71 | 539 | incr Stat.nIncludePassing; |
34e49164 | 540 | pr2_once ("CPP-INCLUDE: inside function, I treat it as comment"); |
0708f913 | 541 | let v = Parser_c.TCommentCpp(Token_c.CppDirective, info) in |
34e49164 | 542 | tr.passed <- v::tr.passed; |
485bce71 | 543 | lexer_function ~pass tr lexbuf |
34e49164 C |
544 | end |
545 | else begin | |
ae4735db | 546 | let (v,new_tokens) = |
978fd7e5 | 547 | Parsing_hacks.tokens_include (info, includes, filename, inifdef) in |
ae4735db | 548 | let new_tokens_clean = |
34e49164 C |
549 | new_tokens +> List.filter TH.is_not_comment in |
550 | ||
551 | tr.passed <- v::tr.passed; | |
552 | tr.passed_clean <- v::tr.passed_clean; | |
553 | tr.rest <- new_tokens ++ tr.rest; | |
554 | tr.rest_clean <- new_tokens_clean ++ tr.rest_clean; | |
555 | v | |
556 | end | |
ae4735db C |
557 | |
558 | | _ -> | |
559 | ||
34e49164 C |
560 | (* typedef_fix1 *) |
561 | let v = match v with | |
ae4735db C |
562 | | Parser_c.TIdent (s, ii) -> |
563 | if | |
564 | LP.is_typedef s && | |
485bce71 | 565 | not (!Flag_parsing_c.disable_add_typedef) && |
b1b2de81 | 566 | pass =|= 1 |
34e49164 C |
567 | then Parser_c.TypedefIdent (s, ii) |
568 | else Parser_c.TIdent (s, ii) | |
569 | | x -> x | |
570 | in | |
ae4735db | 571 | |
485bce71 C |
572 | let v = Parsing_hacks.lookahead ~pass |
573 | (clean_for_lookahead (v::tr.rest_clean)) | |
574 | tr.passed_clean in | |
34e49164 C |
575 | |
576 | tr.passed <- v::tr.passed; | |
ae4735db | 577 | |
485bce71 | 578 | (* the lookahead may have changed the status of the token and |
34e49164 | 579 | * consider it as a comment, for instance some #include are |
485bce71 | 580 | * turned into comments, hence this code. *) |
34e49164 | 581 | match v with |
485bce71 | 582 | | Parser_c.TCommentCpp _ -> lexer_function ~pass tr lexbuf |
ae4735db | 583 | | v -> |
34e49164 C |
584 | tr.passed_clean <- v::tr.passed_clean; |
585 | v | |
586 | ) | |
587 | end | |
588 | ||
589 | ||
708f4980 C |
590 | let max_pass = 4 |
591 | ||
34e49164 | 592 | |
ae4735db | 593 | let get_one_elem ~pass tr (file, filelines) = |
485bce71 C |
594 | |
595 | if not (LP.is_enabled_typedef()) && !Flag_parsing_c.debug_typedef | |
708f4980 | 596 | then pr2_err "TYPEDEF:_handle_typedef=false. Not normal if dont come from exn"; |
485bce71 C |
597 | |
598 | (* normally have to do that only when come from an exception in which | |
ae4735db | 599 | * case the dt() may not have been done |
485bce71 C |
600 | * TODO but if was in scoped scope ? have to let only the last scope |
601 | * so need do a LP.lexer_reset_typedef (); | |
602 | *) | |
ae4735db | 603 | LP.enable_typedef(); |
485bce71 C |
604 | LP._lexer_hint := (LP.default_hint ()); |
605 | LP.save_typedef_state(); | |
606 | ||
607 | tr.passed <- []; | |
608 | ||
609 | let lexbuf_fake = Lexing.from_function (fun buf n -> raise Impossible) in | |
ae4735db C |
610 | |
611 | (try | |
485bce71 C |
612 | (* -------------------------------------------------- *) |
613 | (* Call parser *) | |
614 | (* -------------------------------------------------- *) | |
ae4735db | 615 | Common.profile_code_exclusif "YACC" (fun () -> |
c491d8ee | 616 | Left (Parser_c.celem (lexer_function ~pass tr) lexbuf_fake) |
485bce71 | 617 | ) |
ae4735db | 618 | with e -> |
485bce71 C |
619 | LP.restore_typedef_state(); |
620 | ||
621 | (* must keep here, before the code that adjusts the tr fields *) | |
622 | let line_error = TH.line_of_tok tr.current in | |
708f4980 C |
623 | |
624 | let passed_before_error = tr.passed in | |
625 | let current = tr.current in | |
485bce71 | 626 | (* error recovery, go to next synchro point *) |
ae4735db | 627 | let (passed', rest') = |
978fd7e5 | 628 | Parsing_recovery_c.find_next_synchro tr.rest tr.passed in |
485bce71 C |
629 | tr.rest <- rest'; |
630 | tr.passed <- passed'; | |
ae4735db | 631 | |
485bce71 C |
632 | tr.current <- List.hd passed'; |
633 | tr.passed_clean <- []; (* enough ? *) | |
634 | (* with error recovery, rest and rest_clean may not be in sync *) | |
635 | tr.rest_clean <- (tr.rest +> List.filter TH.is_not_comment); | |
ae4735db C |
636 | |
637 | ||
638 | let info_of_bads = Common.map_eff_rev TH.info_of_tok tr.passed in | |
639 | Right (info_of_bads, line_error, | |
640 | tr.passed, passed_before_error, | |
708f4980 | 641 | current, e) |
485bce71 C |
642 | ) |
643 | ||
644 | ||
645 | ||
978fd7e5 C |
646 | (* Macro problem recovery *) |
647 | (* used by the multi-pass error recovery expand-on-demand *) | |
648 | (* | |
ae4735db C |
649 | val candidate_macros_in_passed: |
650 | defs: (string, define_def) Hashtbl.t -> | |
978fd7e5 C |
651 | Parser_c.token list -> (string * define_def) list |
652 | *) | |
653 | ||
ae4735db | 654 | let candidate_macros_in_passed2 ~defs passed = |
978fd7e5 C |
655 | let res = ref [] in |
656 | let res2 = ref [] in | |
657 | ||
658 | passed +> List.iter (function | |
659 | | Parser_c.TIdent (s,_) | |
660 | (* bugfix: may have to undo some infered things *) | |
661 | | Parser_c.TMacroIterator (s,_) | |
662 | | Parser_c.TypedefIdent (s,_) | |
ae4735db | 663 | -> |
978fd7e5 | 664 | (match Common.hfind_option s defs with |
ae4735db C |
665 | | Some def -> |
666 | if s ==~ Parsing_hacks.regexp_macro | |
978fd7e5 C |
667 | then |
668 | (* pr2 (spf "candidate: %s" s); *) | |
ae4735db C |
669 | Common.push2 (s, def) res |
670 | else | |
978fd7e5 C |
671 | Common.push2 (s, def) res2 |
672 | | None -> () | |
673 | ) | |
674 | ||
675 | | _ -> () | |
676 | ); | |
ae4735db C |
677 | if null !res |
678 | then !res2 | |
978fd7e5 C |
679 | else !res |
680 | ||
ae4735db C |
681 | let candidate_macros_in_passed ~defs b = |
682 | Common.profile_code "MACRO managment" (fun () -> | |
978fd7e5 | 683 | candidate_macros_in_passed2 ~defs b) |
ae4735db | 684 | |
978fd7e5 C |
685 | |
686 | ||
687 | ||
688 | ||
689 | let find_optional_macro_to_expand2 ~defs toks = | |
690 | ||
691 | let defs = Common.hash_of_list defs in | |
692 | ||
c491d8ee | 693 | let toks = toks +> Common.tail_map (function |
978fd7e5 C |
694 | |
695 | (* special cases to undo *) | |
ae4735db | 696 | | Parser_c.TMacroIterator (s, ii) -> |
978fd7e5 C |
697 | if Hashtbl.mem defs s |
698 | then Parser_c.TIdent (s, ii) | |
699 | else Parser_c.TMacroIterator (s, ii) | |
700 | ||
ae4735db | 701 | | Parser_c.TypedefIdent (s, ii) -> |
978fd7e5 C |
702 | if Hashtbl.mem defs s |
703 | then Parser_c.TIdent (s, ii) | |
704 | else Parser_c.TypedefIdent (s, ii) | |
705 | ||
706 | | x -> x | |
707 | ) in | |
708 | ||
709 | let tokens = toks in | |
710 | Parsing_hacks.fix_tokens_cpp ~macro_defs:defs tokens | |
711 | ||
712 | (* just calling apply_macro_defs and having a specialized version | |
ae4735db | 713 | * of the code in fix_tokens_cpp is not enough as some work such |
978fd7e5 C |
714 | * as the passing of the body of attribute in Parsing_hacks.find_macro_paren |
715 | * will not get the chance to be run on the new expanded tokens. | |
ae4735db | 716 | * Hence even if it's expensive, it's currently better to |
978fd7e5 C |
717 | * just call directly fix_tokens_cpp again here. |
718 | ||
719 | let tokens2 = ref (tokens +> Common.acc_map TV.mk_token_extended) in | |
720 | let cleaner = !tokens2 +> Parsing_hacks.filter_cpp_stuff in | |
721 | let paren_grouped = TV.mk_parenthised cleaner in | |
722 | Cpp_token_c.apply_macro_defs | |
723 | ~msg_apply_known_macro:(fun s -> pr2 (spf "APPLYING: %s" s)) | |
724 | ~msg_apply_known_macro_hint:(fun s -> pr2 "hint") | |
725 | defs paren_grouped; | |
726 | (* because the before field is used by apply_macro_defs *) | |
ae4735db C |
727 | tokens2 := TV.rebuild_tokens_extented !tokens2; |
728 | Parsing_hacks.insert_virtual_positions | |
978fd7e5 C |
729 | (!tokens2 +> Common.acc_map (fun x -> x.TV.tok)) |
730 | *) | |
ae4735db C |
731 | let find_optional_macro_to_expand ~defs a = |
732 | Common.profile_code "MACRO managment" (fun () -> | |
978fd7e5 | 733 | find_optional_macro_to_expand2 ~defs a) |
ae4735db | 734 | |
978fd7e5 C |
735 | |
736 | ||
737 | ||
738 | ||
739 | (*****************************************************************************) | |
740 | (* Main entry points *) | |
741 | (*****************************************************************************) | |
742 | ||
ae4735db | 743 | let (_defs : (string, Cpp_token_c.define_def) Hashtbl.t ref) = |
978fd7e5 C |
744 | ref (Hashtbl.create 101) |
745 | ||
ae4735db | 746 | let (_defs_builtins : (string, Cpp_token_c.define_def) Hashtbl.t ref) = |
978fd7e5 C |
747 | ref (Hashtbl.create 101) |
748 | ||
749 | ||
750 | (* can not be put in parsing_hack, cos then mutually recursive problem as | |
751 | * we also want to parse the standard.h file. | |
752 | *) | |
ae4735db | 753 | let init_defs_macros std_h = |
978fd7e5 C |
754 | if not (Common.lfile_exists std_h) |
755 | then pr2 ("warning: Can't find default macro file: " ^ std_h) | |
756 | else begin | |
757 | pr2 ("init_defs: " ^ std_h); | |
758 | _defs := Common.hash_of_list (extract_macros std_h); | |
759 | end | |
760 | ||
ae4735db | 761 | let init_defs_builtins file_h = |
978fd7e5 C |
762 | if not (Common.lfile_exists file_h) |
763 | then pr2 ("warning: Can't find macro file: " ^ file_h) | |
764 | else begin | |
765 | pr2 ("init_defs_builtins: " ^ file_h); | |
ae4735db | 766 | _defs_builtins := |
978fd7e5 C |
767 | Common.hash_of_list (extract_macros file_h); |
768 | end | |
769 | ||
770 | ||
771 | ||
772 | type info_item = string * Parser_c.token list | |
773 | ||
774 | type program2 = toplevel2 list | |
775 | and toplevel2 = Ast_c.toplevel * info_item | |
776 | ||
ae4735db | 777 | let program_of_program2 xs = |
978fd7e5 C |
778 | xs +> List.map fst |
779 | ||
ae4735db C |
780 | let with_program2 f program2 = |
781 | program2 | |
782 | +> Common.unzip | |
783 | +> (fun (program, infos) -> | |
978fd7e5 C |
784 | f program, infos |
785 | ) | |
786 | +> Common.uncurry Common.zip | |
787 | ||
788 | ||
789 | ||
790 | ||
791 | ||
485bce71 | 792 | |
34e49164 | 793 | (* note: as now we go in 2 passes, there is first all the error message of |
485bce71 | 794 | * the lexer, and then the error of the parser. It is not anymore |
34e49164 | 795 | * interwinded. |
ae4735db | 796 | * |
34e49164 C |
797 | * !!!This function use refs, and is not reentrant !!! so take care. |
798 | * It use globals defined in Lexer_parser and also the _defs global | |
ae4735db C |
799 | * in parsing_hack.ml. |
800 | * | |
485bce71 C |
801 | * This function uses internally some semi globals in the |
802 | * tokens_stat record and parsing_stat record. | |
34e49164 C |
803 | *) |
804 | ||
ae4735db | 805 | let parse_print_error_heuristic2 file = |
34e49164 | 806 | |
91eba41f | 807 | let filelines = Common.cat_array file in |
485bce71 C |
808 | let stat = Parsing_stat.default_stat file in |
809 | ||
34e49164 C |
810 | (* -------------------------------------------------- *) |
811 | (* call lexer and get all the tokens *) | |
812 | (* -------------------------------------------------- *) | |
ae4735db | 813 | LP.lexer_reset_typedef(); |
485bce71 | 814 | Parsing_hacks.ifdef_paren_cnt := 0; |
708f4980 | 815 | |
485bce71 | 816 | let toks_orig = tokens file in |
978fd7e5 | 817 | let toks = Parsing_hacks.fix_tokens_define toks_orig in |
708f4980 | 818 | let toks = Parsing_hacks.fix_tokens_cpp ~macro_defs:!_defs_builtins toks in |
34e49164 | 819 | |
708f4980 | 820 | (* expand macros on demand trick, preparation phase *) |
ae4735db C |
821 | let macros = |
822 | Common.profile_code "MACRO mgmt prep 1" (fun () -> | |
708f4980 C |
823 | let macros = Hashtbl.copy !_defs in |
824 | (* include also builtins as some macros may generate some builtins too | |
825 | * like __decl_spec or __stdcall | |
826 | *) | |
ae4735db | 827 | !_defs_builtins +> Hashtbl.iter (fun s def -> |
708f4980 C |
828 | Hashtbl.replace macros s def; |
829 | ); | |
830 | macros | |
831 | ) | |
832 | in | |
ae4735db | 833 | Common.profile_code "MACRO mgmt prep 2" (fun () -> |
978fd7e5 | 834 | let local_macros = extract_macros file in |
ae4735db | 835 | local_macros +> List.iter (fun (s, def) -> |
708f4980 C |
836 | Hashtbl.replace macros s def; |
837 | ); | |
838 | ); | |
34e49164 | 839 | |
708f4980 | 840 | let tr = mk_tokens_state toks in |
485bce71 C |
841 | |
842 | let rec loop tr = | |
34e49164 C |
843 | |
844 | (* todo?: I am not sure that it represents current_line, cos maybe | |
845 | * tr.current partipated in the previous parsing phase, so maybe tr.current | |
846 | * is not the first token of the next parsing phase. Same with checkpoint2. | |
847 | * It would be better to record when we have a } or ; in parser.mly, | |
848 | * cos we know that they are the last symbols of external_declaration2. | |
485bce71 C |
849 | * |
850 | * bugfix: may not be equal to 'file' as after macro expansions we can | |
851 | * start to parse a new entity from the body of a macro, for instance | |
852 | * when parsing a define_machine() body, cf standard.h | |
34e49164 C |
853 | *) |
854 | let checkpoint = TH.line_of_tok tr.current in | |
485bce71 | 855 | let checkpoint_file = TH.file_of_tok tr.current in |
34e49164 | 856 | |
485bce71 | 857 | (* call the parser *) |
ae4735db C |
858 | let elem = |
859 | let pass1 = | |
860 | Common.profile_code "Parsing: 1st pass" (fun () -> | |
708f4980 C |
861 | get_one_elem ~pass:1 tr (file, filelines) |
862 | ) in | |
485bce71 C |
863 | match pass1 with |
864 | | Left e -> Left e | |
ae4735db | 865 | | Right (info,line_err, passed, passed_before_error, cur, exn) -> |
708f4980 C |
866 | if !Flag_parsing_c.disable_multi_pass |
867 | then pass1 | |
485bce71 | 868 | else begin |
ae4735db | 869 | Common.profile_code "Parsing: multi pass" (fun () -> |
708f4980 C |
870 | |
871 | pr2_err "parsing pass2: try again"; | |
872 | let toks = List.rev passed ++ tr.rest in | |
873 | let new_tr = mk_tokens_state toks in | |
874 | copy_tokens_state ~src:new_tr ~dst:tr; | |
875 | let passx = get_one_elem ~pass:2 tr (file, filelines) in | |
876 | ||
877 | (match passx with | |
878 | | Left e -> passx | |
ae4735db C |
879 | | Right (info,line_err,passed,passed_before_error,cur,exn) -> |
880 | let candidates = | |
881 | candidate_macros_in_passed ~defs:macros passed | |
708f4980 | 882 | in |
ae4735db | 883 | |
978fd7e5 | 884 | |
708f4980 C |
885 | if is_define_passed passed || null candidates |
886 | then passx | |
887 | else begin | |
888 | (* todo factorize code *) | |
889 | ||
890 | pr2_err "parsing pass3: try again"; | |
891 | let toks = List.rev passed ++ tr.rest in | |
ae4735db | 892 | let toks' = |
708f4980 C |
893 | find_optional_macro_to_expand ~defs:candidates toks in |
894 | let new_tr = mk_tokens_state toks' in | |
895 | copy_tokens_state ~src:new_tr ~dst:tr; | |
896 | let passx = get_one_elem ~pass:3 tr (file, filelines) in | |
897 | ||
898 | (match passx with | |
899 | | Left e -> passx | |
ae4735db | 900 | | Right (info,line_err,passed,passed_before_error,cur,exn) -> |
708f4980 C |
901 | pr2_err "parsing pass4: try again"; |
902 | ||
ae4735db C |
903 | let candidates = |
904 | candidate_macros_in_passed | |
905 | ~defs:macros passed | |
978fd7e5 | 906 | in |
708f4980 C |
907 | |
908 | let toks = List.rev passed ++ tr.rest in | |
ae4735db | 909 | let toks' = |
708f4980 C |
910 | find_optional_macro_to_expand ~defs:candidates toks in |
911 | let new_tr = mk_tokens_state toks' in | |
912 | copy_tokens_state ~src:new_tr ~dst:tr; | |
913 | let passx = get_one_elem ~pass:4 tr (file, filelines) in | |
914 | passx | |
915 | ) | |
916 | end | |
917 | ) | |
918 | ) | |
485bce71 C |
919 | end |
920 | in | |
921 | ||
922 | ||
923 | (* again not sure if checkpoint2 corresponds to end of bad region *) | |
924 | let checkpoint2 = TH.line_of_tok tr.current in (* <> line_error *) | |
925 | let checkpoint2_file = TH.file_of_tok tr.current in | |
926 | ||
ae4735db | 927 | let diffline = |
b1b2de81 | 928 | if (checkpoint_file =$= checkpoint2_file) && (checkpoint_file =$= file) |
ae4735db | 929 | then (checkpoint2 - checkpoint) |
485bce71 C |
930 | else 0 |
931 | (* TODO? so if error come in middle of something ? where the | |
932 | * start token was from original file but synchro found in body | |
933 | * of macro ? then can have wrong number of lines stat. | |
934 | * Maybe simpler just to look at tr.passed and count | |
935 | * the lines in the token from the correct file ? | |
936 | *) | |
34e49164 | 937 | in |
ae4735db | 938 | let info = mk_info_item file (List.rev tr.passed) in |
34e49164 | 939 | |
485bce71 | 940 | (* some stat updates *) |
ae4735db | 941 | stat.Stat.commentized <- |
485bce71 C |
942 | stat.Stat.commentized + count_lines_commentized (snd info); |
943 | ||
ae4735db | 944 | let elem = |
485bce71 | 945 | match elem with |
ae4735db | 946 | | Left e -> |
91eba41f C |
947 | stat.Stat.correct <- stat.Stat.correct + diffline; |
948 | e | |
ae4735db C |
949 | | Right (info_of_bads, line_error, toks_of_bads, |
950 | _passed_before_error, cur, exn) -> | |
708f4980 C |
951 | |
952 | let was_define = is_define_passed tr.passed in | |
ae4735db | 953 | |
708f4980 C |
954 | if was_define && !Flag_parsing_c.filter_msg_define_error |
955 | then () | |
956 | else begin | |
957 | ||
958 | (match exn with | |
ae4735db | 959 | | Lexer_c.Lexical _ |
708f4980 C |
960 | | Parsing.Parse_error |
961 | | Semantic_c.Semantic _ -> () | |
962 | | e -> raise e | |
963 | ); | |
964 | ||
965 | if !Flag_parsing_c.show_parsing_error | |
ae4735db | 966 | then begin |
708f4980 C |
967 | (match exn with |
968 | (* Lexical is not anymore launched I think *) | |
ae4735db | 969 | | Lexer_c.Lexical s -> |
708f4980 | 970 | pr2 ("lexical error " ^s^ "\n =" ^ error_msg_tok cur) |
ae4735db | 971 | | Parsing.Parse_error -> |
708f4980 | 972 | pr2 ("parse error \n = " ^ error_msg_tok cur) |
ae4735db | 973 | | Semantic_c.Semantic (s, i) -> |
708f4980 C |
974 | pr2 ("semantic error " ^s^ "\n ="^ error_msg_tok cur) |
975 | | e -> raise Impossible | |
976 | ); | |
977 | (* bugfix: *) | |
ae4735db | 978 | if (checkpoint_file =$= checkpoint2_file) && |
708f4980 C |
979 | checkpoint_file =$= file |
980 | then print_bad line_error (checkpoint, checkpoint2) filelines | |
981 | else pr2 "PB: bad: but on tokens not from original file" | |
982 | end; | |
983 | ||
ae4735db C |
984 | |
985 | let pbline = | |
986 | toks_of_bads | |
708f4980 | 987 | +> Common.filter (TH.is_same_line_or_close line_error) |
ae4735db | 988 | +> Common.filter TH.is_ident_like |
708f4980 | 989 | in |
ae4735db | 990 | let error_info = |
708f4980 C |
991 | (pbline +> List.map TH.str_of_tok), line_error |
992 | in | |
ae4735db | 993 | stat.Stat.problematic_lines <- |
708f4980 C |
994 | error_info::stat.Stat.problematic_lines; |
995 | ||
996 | end; | |
997 | ||
91eba41f C |
998 | if was_define && !Flag_parsing_c.filter_define_error |
999 | then stat.Stat.correct <- stat.Stat.correct + diffline | |
1000 | else stat.Stat.bad <- stat.Stat.bad + diffline; | |
1001 | ||
485bce71 C |
1002 | Ast_c.NotParsedCorrectly info_of_bads |
1003 | in | |
34e49164 C |
1004 | |
1005 | (match elem with | |
1006 | | Ast_c.FinalDef x -> [(Ast_c.FinalDef x, info)] | |
485bce71 | 1007 | | xs -> (xs, info):: loop tr (* recurse *) |
34e49164 C |
1008 | ) |
1009 | in | |
485bce71 | 1010 | let v = loop tr in |
978fd7e5 | 1011 | let v = with_program2 Parsing_consistency_c.consistency_checking v in |
34e49164 C |
1012 | (v, stat) |
1013 | ||
1014 | ||
ae4735db | 1015 | let time_total_parsing a = |
485bce71 C |
1016 | Common.profile_code "TOTAL" (fun () -> parse_print_error_heuristic2 a) |
1017 | ||
ae4735db | 1018 | let parse_print_error_heuristic a = |
485bce71 C |
1019 | Common.profile_code "C parsing" (fun () -> time_total_parsing a) |
1020 | ||
34e49164 C |
1021 | |
1022 | (* alias *) | |
1023 | let parse_c_and_cpp a = parse_print_error_heuristic a | |
1024 | ||
1025 | (*****************************************************************************) | |
1026 | (* Same but faster cos memoize stuff *) | |
1027 | (*****************************************************************************) | |
ae4735db C |
1028 | let parse_cache file = |
1029 | if not !Flag_parsing_c.use_cache then parse_print_error_heuristic file | |
1030 | else | |
485bce71 | 1031 | let _ = pr2 "TOFIX" in |
ae4735db | 1032 | let need_no_changed_files = |
34e49164 | 1033 | (* should use Sys.argv.(0), would be safer. *) |
485bce71 C |
1034 | |
1035 | [ | |
1036 | (* TOFIX | |
1037 | Config.path ^ "/parsing_c/c_parser.cma"; | |
ae4735db | 1038 | (* we may also depend now on the semantic patch because |
485bce71 | 1039 | the SP may use macro and so we will disable some of the |
ae4735db | 1040 | macro expansions from standard.h. |
485bce71 C |
1041 | *) |
1042 | !Config.std_h; | |
1043 | *) | |
ae4735db | 1044 | ] |
34e49164 | 1045 | in |
ae4735db | 1046 | let need_no_changed_variables = |
34e49164 | 1047 | (* could add some of the flags of flag_parsing_c.ml *) |
ae4735db | 1048 | [] |
34e49164 | 1049 | in |
ae4735db C |
1050 | Common.cache_computation_robust |
1051 | file ".ast_raw" | |
1052 | (need_no_changed_files, need_no_changed_variables) ".depend_raw" | |
34e49164 C |
1053 | (fun () -> parse_print_error_heuristic file) |
1054 | ||
1055 | ||
1056 | ||
1057 | (*****************************************************************************) | |
485bce71 | 1058 | (* Some special cases *) |
34e49164 C |
1059 | (*****************************************************************************) |
1060 | ||
485bce71 | 1061 | let (cstatement_of_string: string -> Ast_c.statement) = fun s -> |
708f4980 C |
1062 | let tmpfile = Common.new_temp_file "cocci_stmt_of_s" "c" in |
1063 | Common.write_file tmpfile ("void main() { \n" ^ s ^ "\n}"); | |
1064 | let program = parse_c_and_cpp tmpfile +> fst in | |
ae4735db | 1065 | program +> Common.find_some (fun (e,_) -> |
485bce71 C |
1066 | match e with |
1067 | | Ast_c.Definition ({Ast_c.f_body = [Ast_c.StmtElem st]},_) -> Some st | |
1068 | | _ -> None | |
1069 | ) | |
1070 | ||
1071 | let (cexpression_of_string: string -> Ast_c.expression) = fun s -> | |
708f4980 C |
1072 | let tmpfile = Common.new_temp_file "cocci_expr_of_s" "c" in |
1073 | Common.write_file tmpfile ("void main() { \n" ^ s ^ ";\n}"); | |
1074 | let program = parse_c_and_cpp tmpfile +> fst in | |
ae4735db | 1075 | program +> Common.find_some (fun (e,_) -> |
485bce71 | 1076 | match e with |
ae4735db | 1077 | | Ast_c.Definition ({Ast_c.f_body = compound},_) -> |
485bce71 | 1078 | (match compound with |
ae4735db | 1079 | | [Ast_c.StmtElem st] -> |
708f4980 C |
1080 | (match Ast_c.unwrap_st st with |
1081 | | Ast_c.ExprStatement (Some e) -> Some e | |
1082 | | _ -> None | |
1083 | ) | |
485bce71 C |
1084 | | _ -> None |
1085 | ) | |
1086 | | _ -> None | |
1087 | ) |