Commit | Line | Data |
---|---|---|
9f8e26f4 | 1 | (* |
ae4735db | 2 | * Copyright 2005-2010, Ecole des Mines de Nantes, University of Copenhagen |
9f8e26f4 C |
3 | * Yoann Padioleau, Julia Lawall, Rene Rydhof Hansen, Henrik Stuart, Gilles Muller, Nicolas Palix |
4 | * This file is part of Coccinelle. | |
5 | * | |
6 | * Coccinelle is free software: you can redistribute it and/or modify | |
7 | * it under the terms of the GNU General Public License as published by | |
8 | * the Free Software Foundation, according to version 2 of the License. | |
9 | * | |
10 | * Coccinelle 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 | * GNU General Public License for more details. | |
14 | * | |
15 | * You should have received a copy of the GNU General Public License | |
16 | * along with Coccinelle. If not, see <http://www.gnu.org/licenses/>. | |
17 | * | |
18 | * The authors reserve the right to distribute this or future versions of | |
19 | * Coccinelle under other licenses. | |
20 | *) | |
21 | ||
22 | ||
34e49164 C |
23 | { |
24 | open Parser_cocci_menhir | |
25 | module D = Data | |
26 | module Ast = Ast_cocci | |
27 | module Ast0 = Ast0_cocci | |
28 | module P = Parse_aux | |
29 | exception Lexical of string | |
30 | let tok = Lexing.lexeme | |
31 | ||
32 | let line = ref 1 | |
33 | let logical_line = ref 0 | |
34 | ||
35 | (* ---------------------------------------------------------------------- *) | |
36 | (* control codes *) | |
37 | ||
38 | (* Defined in data.ml | |
39 | type line_type = MINUS | OPTMINUS | UNIQUEMINUS | PLUS | CONTEXT | UNIQUE | OPT | |
40 | *) | |
41 | ||
42 | let current_line_type = ref (D.CONTEXT,!line,!logical_line) | |
43 | ||
44 | let prev_plus = ref false | |
45 | let line_start = ref 0 (* offset of the beginning of the line *) | |
46 | let get_current_line_type lexbuf = | |
47 | let (c,l,ll) = !current_line_type in | |
48 | let lex_start = Lexing.lexeme_start lexbuf in | |
49 | let preceeding_spaces = | |
50 | if !line_start < 0 then 0 else lex_start - !line_start in | |
708f4980 | 51 | (*line_start := -1;*) |
951c7801 | 52 | prev_plus := (c = D.PLUS) or (c = D.PLUSPLUS); |
34e49164 C |
53 | (c,l,ll,lex_start,preceeding_spaces,[],[],Ast0.NoMetaPos) |
54 | let current_line_started = ref false | |
55 | let col_zero = ref true | |
56 | ||
57 | let reset_line lexbuf = | |
58 | line := !line + 1; | |
59 | current_line_type := (D.CONTEXT,!line,!logical_line); | |
60 | current_line_started := false; | |
61 | col_zero := true; | |
62 | line_start := Lexing.lexeme_start lexbuf + 1 | |
63 | ||
64 | let started_line = ref (-1) | |
65 | ||
66 | let start_line seen_char = | |
67 | current_line_started := true; | |
68 | col_zero := false; | |
69 | (if seen_char && not(!line = !started_line) | |
70 | then | |
71 | begin | |
72 | started_line := !line; | |
73 | logical_line := !logical_line + 1 | |
74 | end) | |
75 | ||
76 | let pass_zero _ = col_zero := false | |
77 | ||
78 | let lexerr s1 s2 = raise (Lexical (Printf.sprintf "%s%s" s1 s2)) | |
79 | ||
80 | let add_current_line_type x = | |
81 | match (x,!current_line_type) with | |
82 | (D.MINUS,(D.CONTEXT,ln,lln)) -> | |
83 | current_line_type := (D.MINUS,ln,lln) | |
84 | | (D.MINUS,(D.UNIQUE,ln,lln)) -> | |
85 | current_line_type := (D.UNIQUEMINUS,ln,lln) | |
86 | | (D.MINUS,(D.OPT,ln,lln)) -> | |
87 | current_line_type := (D.OPTMINUS,ln,lln) | |
88 | | (D.PLUS,(D.CONTEXT,ln,lln)) -> | |
89 | current_line_type := (D.PLUS,ln,lln) | |
951c7801 C |
90 | | (D.PLUSPLUS,(D.CONTEXT,ln,lln)) -> |
91 | current_line_type := (D.PLUSPLUS,ln,lln) | |
34e49164 C |
92 | | (D.UNIQUE,(D.CONTEXT,ln,lln)) -> |
93 | current_line_type := (D.UNIQUE,ln,lln) | |
94 | | (D.OPT,(D.CONTEXT,ln,lln)) -> | |
95 | current_line_type := (D.OPT,ln,lln) | |
96 | | _ -> lexerr "invalid control character combination" "" | |
97 | ||
98 | let check_minus_context_linetype s = | |
99 | match !current_line_type with | |
951c7801 | 100 | (D.PLUS,_,_) | (D.PLUSPLUS,_,_) -> lexerr "invalid in a + context: " s |
34e49164 C |
101 | | _ -> () |
102 | ||
103 | let check_context_linetype s = | |
104 | match !current_line_type with | |
105 | (D.CONTEXT,_,_) -> () | |
106 | | _ -> lexerr "invalid in a nonempty context: " s | |
107 | ||
108 | let check_plus_linetype s = | |
109 | match !current_line_type with | |
951c7801 | 110 | (D.PLUS,_,_) | (D.PLUSPLUS,_,_) -> () |
34e49164 C |
111 | | _ -> lexerr "invalid in a non + context: " s |
112 | ||
113 | let check_arity_context_linetype s = | |
114 | match !current_line_type with | |
951c7801 C |
115 | (D.CONTEXT,_,_) | (D.PLUS,_,_) | (D.PLUSPLUS,_,_) |
116 | | (D.UNIQUE,_,_) | (D.OPT,_,_) -> () | |
34e49164 C |
117 | | _ -> lexerr "invalid in a nonempty context: " s |
118 | ||
119 | let process_include start finish str = | |
120 | (match !current_line_type with | |
951c7801 | 121 | (D.PLUS,_,_) | (D.PLUSPLUS,_,_) -> |
34e49164 C |
122 | (try |
123 | let _ = Str.search_forward (Str.regexp "\\.\\.\\.") str start in | |
124 | lexerr "... not allowed in + include" "" | |
125 | with Not_found -> ()) | |
126 | | _ -> ()); | |
127 | String.sub str (start + 1) (finish - start - 1) | |
128 | ||
129 | (* ---------------------------------------------------------------------- *) | |
130 | type pm = PATCH | MATCH | UNKNOWN | |
131 | ||
132 | let pm = ref UNKNOWN | |
133 | ||
134 | let patch_or_match = function | |
135 | PATCH -> | |
7f004419 C |
136 | if not !D.ignore_patch_or_match |
137 | then | |
138 | (match !pm with | |
139 | MATCH -> | |
140 | lexerr "- or + not allowed in the first column for a match" "" | |
141 | | PATCH -> () | |
142 | | UNKNOWN -> Flag.sgrep_mode2 := false; pm := PATCH) | |
34e49164 | 143 | | MATCH -> |
7f004419 C |
144 | if not !D.ignore_patch_or_match |
145 | then | |
146 | (match !pm with | |
147 | PATCH -> lexerr "* not allowed in the first column for a patch" "" | |
148 | | MATCH -> () | |
149 | | UNKNOWN -> Flag.sgrep_mode2 := true; pm := MATCH) | |
34e49164 C |
150 | | _ -> failwith "unexpected argument" |
151 | ||
152 | (* ---------------------------------------------------------------------- *) | |
153 | (* identifiers, including metavariables *) | |
154 | ||
155 | let metavariables = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t) | |
156 | ||
157 | let all_metavariables = | |
158 | (Hashtbl.create(100) : (string,(string * (D.clt -> token)) list) Hashtbl.t) | |
159 | ||
160 | let type_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t) | |
161 | ||
162 | let declarer_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t) | |
163 | ||
164 | let iterator_names = (Hashtbl.create(100) : (string, D.clt -> token) Hashtbl.t) | |
165 | ||
166 | let rule_names = (Hashtbl.create(100) : (string, unit) Hashtbl.t) | |
167 | ||
168 | let check_var s linetype = | |
169 | let fail _ = | |
170 | if (!Data.in_prolog || !Data.in_rule_name) && | |
171 | Str.string_match (Str.regexp "<.*>") s 0 | |
172 | then TPathIsoFile s | |
173 | else | |
174 | try (Hashtbl.find metavariables s) linetype | |
175 | with Not_found -> | |
176 | (try (Hashtbl.find type_names s) linetype | |
177 | with Not_found -> | |
178 | (try (Hashtbl.find declarer_names s) linetype | |
faf9a90c | 179 | with Not_found -> |
34e49164 C |
180 | (try (Hashtbl.find iterator_names s) linetype |
181 | with Not_found -> TIdent (s,linetype)))) in | |
182 | if !Data.in_meta or !Data.in_rule_name | |
183 | then (try Hashtbl.find rule_names s; TRuleName s with Not_found -> fail()) | |
184 | else fail() | |
185 | ||
186 | let id_tokens lexbuf = | |
187 | let s = tok lexbuf in | |
188 | let linetype = get_current_line_type lexbuf in | |
189 | let in_rule_name = !Data.in_rule_name in | |
978fd7e5 | 190 | let in_meta = !Data.in_meta && not !Data.saw_struct in |
34e49164 C |
191 | let in_iso = !Data.in_iso in |
192 | let in_prolog = !Data.in_prolog in | |
193 | match s with | |
194 | "identifier" when in_meta -> check_arity_context_linetype s; TIdentifier | |
195 | | "type" when in_meta -> check_arity_context_linetype s; TType | |
196 | | "parameter" when in_meta -> check_arity_context_linetype s; TParameter | |
197 | | "constant" when in_meta -> check_arity_context_linetype s; TConstant | |
faf9a90c C |
198 | | "generated" when in_rule_name && not (!Flag.make_hrule = None) -> |
199 | check_arity_context_linetype s; TGenerated | |
34e49164 C |
200 | | "expression" when in_meta || in_rule_name -> |
201 | check_arity_context_linetype s; TExpression | |
113803cf C |
202 | | "initialiser" when in_meta || in_rule_name -> |
203 | check_arity_context_linetype s; TInitialiser | |
204 | | "initializer" when in_meta || in_rule_name -> | |
205 | check_arity_context_linetype s; TInitialiser | |
34e49164 C |
206 | | "idexpression" when in_meta -> |
207 | check_arity_context_linetype s; TIdExpression | |
208 | | "statement" when in_meta -> check_arity_context_linetype s; TStatement | |
209 | | "function" when in_meta -> check_arity_context_linetype s; TFunction | |
210 | | "local" when in_meta -> check_arity_context_linetype s; TLocal | |
211 | | "list" when in_meta -> check_arity_context_linetype s; Tlist | |
212 | | "fresh" when in_meta -> check_arity_context_linetype s; TFresh | |
213 | | "typedef" when in_meta -> check_arity_context_linetype s; TTypedef | |
214 | | "declarer" when in_meta -> check_arity_context_linetype s; TDeclarer | |
215 | | "iterator" when in_meta -> check_arity_context_linetype s; TIterator | |
216 | | "name" when in_meta -> check_arity_context_linetype s; TName | |
217 | | "position" when in_meta -> check_arity_context_linetype s; TPosition | |
218 | | "any" when in_meta -> check_arity_context_linetype s; TPosAny | |
219 | | "pure" when in_meta && in_iso -> | |
220 | check_arity_context_linetype s; TPure | |
221 | | "context" when in_meta && in_iso -> | |
222 | check_arity_context_linetype s; TContext | |
223 | | "error" when in_meta -> check_arity_context_linetype s; TError | |
224 | | "words" when in_meta -> check_context_linetype s; TWords | |
225 | ||
226 | | "using" when in_rule_name || in_prolog -> check_context_linetype s; TUsing | |
ae4735db C |
227 | | "virtual" when in_prolog or in_rule_name or in_meta -> |
228 | (* don't want to allow virtual as a rule name *) | |
229 | check_context_linetype s; TVirtual | |
34e49164 C |
230 | | "disable" when in_rule_name -> check_context_linetype s; TDisable |
231 | | "extends" when in_rule_name -> check_context_linetype s; TExtends | |
232 | | "depends" when in_rule_name -> check_context_linetype s; TDepends | |
233 | | "on" when in_rule_name -> check_context_linetype s; TOn | |
234 | | "ever" when in_rule_name -> check_context_linetype s; TEver | |
235 | | "never" when in_rule_name -> check_context_linetype s; TNever | |
978fd7e5 | 236 | (* exists and forall for when are reparsed in parse_cocci.ml *) |
34e49164 C |
237 | | "exists" when in_rule_name -> check_context_linetype s; TExists |
238 | | "forall" when in_rule_name -> check_context_linetype s; TForall | |
b1b2de81 C |
239 | | "script" when in_rule_name -> check_context_linetype s; TScript |
240 | | "initialize" when in_rule_name -> check_context_linetype s; TInitialize | |
241 | | "finalize" when in_rule_name -> check_context_linetype s; TFinalize | |
34e49164 C |
242 | |
243 | | "char" -> Tchar linetype | |
244 | | "short" -> Tshort linetype | |
245 | | "int" -> Tint linetype | |
246 | | "double" -> Tdouble linetype | |
247 | | "float" -> Tfloat linetype | |
248 | | "long" -> Tlong linetype | |
249 | | "void" -> Tvoid linetype | |
978fd7e5 C |
250 | (* in_meta is only for the first keyword; drop it now to allow any type |
251 | name *) | |
252 | | "struct" -> Data.saw_struct := true; Tstruct linetype | |
253 | | "union" -> Data.saw_struct := true; Tunion linetype | |
254 | | "enum" -> Data.saw_struct := true; Tenum linetype | |
34e49164 C |
255 | | "unsigned" -> Tunsigned linetype |
256 | | "signed" -> Tsigned linetype | |
faf9a90c | 257 | |
34e49164 C |
258 | | "auto" -> Tauto linetype |
259 | | "register" -> Tregister linetype | |
260 | | "extern" -> Textern linetype | |
261 | | "static" -> Tstatic linetype | |
262 | | "inline" -> Tinline linetype | |
263 | | "typedef" -> Ttypedef linetype | |
264 | ||
265 | | "const" -> Tconst linetype | |
266 | | "volatile" -> Tvolatile linetype | |
267 | ||
268 | | "if" -> TIf linetype | |
269 | | "else" -> TElse linetype | |
270 | | "while" -> TWhile linetype | |
271 | | "do" -> TDo linetype | |
272 | | "for" -> TFor linetype | |
273 | | "switch" -> TSwitch linetype | |
274 | | "case" -> TCase linetype | |
275 | | "default" -> TDefault linetype | |
276 | | "return" -> TReturn linetype | |
277 | | "break" -> TBreak linetype | |
278 | | "continue" -> TContinue linetype | |
279 | | "goto" -> TGoto linetype | |
280 | ||
281 | | "sizeof" -> TSizeof linetype | |
282 | ||
283 | | "Expression" -> TIsoExpression | |
284 | | "ArgExpression" -> TIsoArgExpression | |
285 | | "TestExpression" -> TIsoTestExpression | |
286 | | "Statement" -> TIsoStatement | |
287 | | "Declaration" -> TIsoDeclaration | |
288 | | "Type" -> TIsoType | |
289 | | "TopLevel" -> TIsoTopLevel | |
290 | ||
291 | | s -> check_var s linetype | |
292 | ||
293 | let mkassign op lexbuf = | |
294 | TAssign (Ast.OpAssign op, (get_current_line_type lexbuf)) | |
295 | ||
296 | let init _ = | |
297 | line := 1; | |
298 | logical_line := 0; | |
299 | prev_plus := false; | |
300 | line_start := 0; | |
301 | current_line_started := false; | |
302 | col_zero := true; | |
303 | pm := UNKNOWN; | |
304 | Data.in_rule_name := false; | |
305 | Data.in_meta := false; | |
306 | Data.in_prolog := false; | |
978fd7e5 | 307 | Data.saw_struct := false; |
34e49164 C |
308 | Data.inheritable_positions := []; |
309 | Hashtbl.clear all_metavariables; | |
310 | Hashtbl.clear Data.all_metadecls; | |
311 | Hashtbl.clear metavariables; | |
312 | Hashtbl.clear type_names; | |
313 | Hashtbl.clear rule_names; | |
708f4980 C |
314 | Hashtbl.clear iterator_names; |
315 | Hashtbl.clear declarer_names; | |
34e49164 C |
316 | let get_name (_,x) = x in |
317 | Data.add_id_meta := | |
318 | (fun name constraints pure -> | |
319 | let fn clt = TMetaId(name,constraints,pure,clt) in | |
320 | Hashtbl.replace metavariables (get_name name) fn); | |
ae4735db C |
321 | Data.add_virt_id_meta_found := |
322 | (fun name vl -> | |
323 | let fn clt = TIdent(vl,clt) in | |
324 | Hashtbl.replace metavariables name fn); | |
325 | Data.add_virt_id_meta_not_found := | |
326 | (fun name pure -> | |
327 | let fn clt = TMetaId(name,Ast.IdNoConstraint,pure,clt) in | |
328 | Hashtbl.replace metavariables (get_name name) fn); | |
b1b2de81 C |
329 | Data.add_fresh_id_meta := |
330 | (fun name -> | |
951c7801 | 331 | let fn clt = TMetaId(name,Ast.IdNoConstraint,Ast0.Impure,clt) in |
b1b2de81 | 332 | Hashtbl.replace metavariables (get_name name) fn); |
34e49164 C |
333 | Data.add_type_meta := |
334 | (fun name pure -> | |
335 | let fn clt = TMetaType(name,pure,clt) in | |
336 | Hashtbl.replace metavariables (get_name name) fn); | |
113803cf C |
337 | Data.add_init_meta := |
338 | (fun name pure -> | |
339 | let fn clt = TMetaInit(name,pure,clt) in | |
340 | Hashtbl.replace metavariables (get_name name) fn); | |
34e49164 C |
341 | Data.add_param_meta := |
342 | (function name -> function pure -> | |
343 | let fn clt = TMetaParam(name,pure,clt) in | |
344 | Hashtbl.replace metavariables (get_name name) fn); | |
345 | Data.add_paramlist_meta := | |
346 | (function name -> function lenname -> function pure -> | |
347 | let fn clt = TMetaParamList(name,lenname,pure,clt) in | |
348 | Hashtbl.replace metavariables (get_name name) fn); | |
349 | Data.add_const_meta := | |
350 | (fun tyopt name constraints pure -> | |
351 | let fn clt = TMetaConst(name,constraints,pure,tyopt,clt) in | |
352 | Hashtbl.replace metavariables (get_name name) fn); | |
353 | Data.add_err_meta := | |
354 | (fun name constraints pure -> | |
355 | let fn clt = TMetaErr(name,constraints,pure,clt) in | |
356 | Hashtbl.replace metavariables (get_name name) fn); | |
357 | Data.add_exp_meta := | |
358 | (fun tyopt name constraints pure -> | |
359 | let fn clt = TMetaExp(name,constraints,pure,tyopt,clt) in | |
360 | Hashtbl.replace metavariables (get_name name) fn); | |
361 | Data.add_idexp_meta := | |
362 | (fun tyopt name constraints pure -> | |
363 | let fn clt = TMetaIdExp(name,constraints,pure,tyopt,clt) in | |
364 | Hashtbl.replace metavariables (get_name name) fn); | |
365 | Data.add_local_idexp_meta := | |
366 | (fun tyopt name constraints pure -> | |
367 | let fn clt = TMetaLocalIdExp(name,constraints,pure,tyopt,clt) in | |
368 | Hashtbl.replace metavariables (get_name name) fn); | |
369 | Data.add_explist_meta := | |
370 | (function name -> function lenname -> function pure -> | |
371 | let fn clt = TMetaExpList(name,lenname,pure,clt) in | |
372 | Hashtbl.replace metavariables (get_name name) fn); | |
373 | Data.add_stm_meta := | |
374 | (function name -> function pure -> | |
375 | let fn clt = TMetaStm(name,pure,clt) in | |
376 | Hashtbl.replace metavariables (get_name name) fn); | |
377 | Data.add_stmlist_meta := | |
378 | (function name -> function pure -> | |
379 | let fn clt = TMetaStmList(name,pure,clt) in | |
380 | Hashtbl.replace metavariables (get_name name) fn); | |
381 | Data.add_func_meta := | |
382 | (fun name constraints pure -> | |
383 | let fn clt = TMetaFunc(name,constraints,pure,clt) in | |
384 | Hashtbl.replace metavariables (get_name name) fn); | |
385 | Data.add_local_func_meta := | |
386 | (fun name constraints pure -> | |
387 | let fn clt = TMetaLocalFunc(name,constraints,pure,clt) in | |
388 | Hashtbl.replace metavariables (get_name name) fn); | |
389 | Data.add_iterator_meta := | |
390 | (fun name constraints pure -> | |
391 | let fn clt = TMetaIterator(name,constraints,pure,clt) in | |
392 | Hashtbl.replace metavariables (get_name name) fn); | |
393 | Data.add_declarer_meta := | |
394 | (fun name constraints pure -> | |
395 | let fn clt = TMetaDeclarer(name,constraints,pure,clt) in | |
396 | Hashtbl.replace metavariables (get_name name) fn); | |
397 | Data.add_pos_meta := | |
398 | (fun name constraints any -> | |
399 | let fn ((d,ln,_,_,_,_,_,_) as clt) = | |
400 | (if d = Data.PLUS | |
401 | then | |
402 | failwith | |
403 | (Printf.sprintf "%d: positions only allowed in minus code" ln)); | |
404 | TMetaPos(name,constraints,any,clt) in | |
405 | Hashtbl.replace metavariables (get_name name) fn); | |
406 | Data.add_type_name := | |
407 | (function name -> | |
408 | let fn clt = TTypeId(name,clt) in | |
409 | Hashtbl.replace type_names name fn); | |
410 | Data.add_declarer_name := | |
411 | (function name -> | |
412 | let fn clt = TDeclarerId(name,clt) in | |
413 | Hashtbl.replace declarer_names name fn); | |
414 | Data.add_iterator_name := | |
415 | (function name -> | |
416 | let fn clt = TIteratorId(name,clt) in | |
417 | Hashtbl.replace iterator_names name fn); | |
418 | Data.init_rule := (function _ -> Hashtbl.clear metavariables); | |
419 | Data.install_bindings := | |
420 | (function parent -> | |
421 | List.iter (function (name,fn) -> Hashtbl.add metavariables name fn) | |
422 | (Hashtbl.find all_metavariables parent)) | |
423 | ||
424 | let drop_spaces s = | |
425 | let len = String.length s in | |
426 | let rec loop n = | |
427 | if n = len | |
428 | then n | |
429 | else | |
430 | if List.mem (String.get s n) [' ';'\t'] | |
431 | then loop (n+1) | |
432 | else n in | |
433 | let start = loop 0 in | |
434 | String.sub s start (len - start) | |
435 | } | |
436 | ||
437 | (* ---------------------------------------------------------------------- *) | |
438 | (* tokens *) | |
439 | ||
440 | let letter = ['A'-'Z' 'a'-'z' '_'] | |
441 | let digit = ['0'-'9'] | |
442 | ||
443 | let dec = ['0'-'9'] | |
444 | let oct = ['0'-'7'] | |
445 | let hex = ['0'-'9' 'a'-'f' 'A'-'F'] | |
446 | ||
447 | let decimal = ('0' | (['1'-'9'] dec*)) | |
448 | let octal = ['0'] oct+ | |
faf9a90c | 449 | let hexa = ("0x" |"0X") hex+ |
34e49164 C |
450 | |
451 | let pent = dec+ | |
452 | let pfract = dec+ | |
453 | let sign = ['-' '+'] | |
454 | let exp = ['e''E'] sign? dec+ | |
455 | let real = pent exp | ((pent? '.' pfract | pent '.' pfract? ) exp?) | |
456 | ||
457 | ||
458 | rule token = parse | |
459 | | [' ' '\t' ]+ { start_line false; token lexbuf } | |
460 | | ['\n' '\r' '\011' '\012'] { reset_line lexbuf; token lexbuf } | |
461 | ||
462 | | "//" [^ '\n']* { start_line false; token lexbuf } | |
463 | ||
464 | | "@@" { start_line true; TArobArob } | |
465 | | "@" { pass_zero(); | |
466 | if !Data.in_rule_name or not !current_line_started | |
467 | then (start_line true; TArob) | |
468 | else (check_minus_context_linetype "@"; TPArob) } | |
469 | ||
951c7801 C |
470 | | "~=" { start_line true; TTildeEq (get_current_line_type lexbuf) } |
471 | | "!~=" { start_line true; TTildeExclEq (get_current_line_type lexbuf) } | |
34e49164 C |
472 | | "WHEN" | "when" |
473 | { start_line true; check_minus_context_linetype (tok lexbuf); | |
474 | TWhen (get_current_line_type lexbuf) } | |
475 | ||
476 | | "..." | |
477 | { start_line true; check_minus_context_linetype (tok lexbuf); | |
478 | TEllipsis (get_current_line_type lexbuf) } | |
479 | (* | |
480 | | "ooo" | |
481 | { start_line true; check_minus_context_linetype (tok lexbuf); | |
482 | TCircles (get_current_line_type lexbuf) } | |
483 | ||
484 | | "***" | |
485 | { start_line true; check_minus_context_linetype (tok lexbuf); | |
486 | TStars (get_current_line_type lexbuf) } | |
487 | *) | |
488 | | "<..." { start_line true; check_context_linetype (tok lexbuf); | |
489 | TOEllipsis (get_current_line_type lexbuf) } | |
490 | | "...>" { start_line true; check_context_linetype (tok lexbuf); | |
491 | TCEllipsis (get_current_line_type lexbuf) } | |
492 | | "<+..." { start_line true; check_context_linetype (tok lexbuf); | |
493 | TPOEllipsis (get_current_line_type lexbuf) } | |
494 | | "...+>" { start_line true; check_context_linetype (tok lexbuf); | |
495 | TPCEllipsis (get_current_line_type lexbuf) } | |
496 | (* | |
497 | | "<ooo" { start_line true; check_context_linetype (tok lexbuf); | |
498 | TOCircles (get_current_line_type lexbuf) } | |
499 | | "ooo>" { start_line true; check_context_linetype (tok lexbuf); | |
500 | TCCircles (get_current_line_type lexbuf) } | |
501 | ||
502 | | "<***" { start_line true; check_context_linetype (tok lexbuf); | |
503 | TOStars (get_current_line_type lexbuf) } | |
504 | | "***>" { start_line true; check_context_linetype (tok lexbuf); | |
505 | TCStars (get_current_line_type lexbuf) } | |
506 | *) | |
507 | | "-" { pass_zero(); | |
508 | if !current_line_started | |
509 | then (start_line true; TMinus (get_current_line_type lexbuf)) | |
510 | else (patch_or_match PATCH; | |
511 | add_current_line_type D.MINUS; token lexbuf) } | |
512 | | "+" { pass_zero(); | |
513 | if !current_line_started | |
514 | then (start_line true; TPlus (get_current_line_type lexbuf)) | |
515 | else if !Data.in_meta | |
516 | then TPlus0 | |
517 | else (patch_or_match PATCH; | |
518 | add_current_line_type D.PLUS; token lexbuf) } | |
519 | | "?" { pass_zero(); | |
520 | if !current_line_started | |
521 | then (start_line true; TWhy (get_current_line_type lexbuf)) | |
522 | else if !Data.in_meta | |
523 | then TWhy0 | |
524 | else (add_current_line_type D.OPT; token lexbuf) } | |
525 | | "!" { pass_zero(); | |
526 | if !current_line_started | |
527 | then (start_line true; TBang (get_current_line_type lexbuf)) | |
528 | else if !Data.in_meta | |
529 | then TBang0 | |
530 | else (add_current_line_type D.UNIQUE; token lexbuf) } | |
531 | | "(" { if not !col_zero | |
532 | then (start_line true; TOPar (get_current_line_type lexbuf)) | |
533 | else | |
534 | (start_line true; check_context_linetype (tok lexbuf); | |
535 | TOPar0 (get_current_line_type lexbuf))} | |
536 | | "\\(" { start_line true; TOPar0 (get_current_line_type lexbuf) } | |
537 | | "|" { if not (!col_zero) | |
538 | then (start_line true; TOr(get_current_line_type lexbuf)) | |
539 | else (start_line true; | |
540 | check_context_linetype (tok lexbuf); | |
541 | TMid0 (get_current_line_type lexbuf))} | |
542 | | "\\|" { start_line true; TMid0 (get_current_line_type lexbuf) } | |
543 | | ")" { if not !col_zero | |
544 | then (start_line true; TCPar (get_current_line_type lexbuf)) | |
545 | else | |
546 | (start_line true; check_context_linetype (tok lexbuf); | |
547 | TCPar0 (get_current_line_type lexbuf))} | |
548 | | "\\)" { start_line true; TCPar0 (get_current_line_type lexbuf) } | |
549 | ||
550 | | '[' { start_line true; TOCro (get_current_line_type lexbuf) } | |
551 | | ']' { start_line true; TCCro (get_current_line_type lexbuf) } | |
552 | | '{' { start_line true; TOBrace (get_current_line_type lexbuf) } | |
553 | | '}' { start_line true; TCBrace (get_current_line_type lexbuf) } | |
554 | ||
555 | | "->" { start_line true; TPtrOp (get_current_line_type lexbuf) } | |
556 | | '.' { start_line true; TDot (get_current_line_type lexbuf) } | |
557 | | ',' { start_line true; TComma (get_current_line_type lexbuf) } | |
558 | | ";" { start_line true; | |
559 | if !Data.in_meta | |
560 | then TMPtVirg (* works better with tokens_all *) | |
561 | else TPtVirg (get_current_line_type lexbuf) } | |
562 | ||
faf9a90c | 563 | |
34e49164 C |
564 | | '*' { pass_zero(); |
565 | if !current_line_started | |
566 | then | |
567 | (start_line true; TMul (get_current_line_type lexbuf)) | |
568 | else | |
569 | (patch_or_match MATCH; | |
570 | add_current_line_type D.MINUS; token lexbuf) } | |
571 | | '/' { start_line true; | |
faf9a90c | 572 | TDmOp (Ast.Div,get_current_line_type lexbuf) } |
34e49164 | 573 | | '%' { start_line true; |
faf9a90c C |
574 | TDmOp (Ast.Mod,get_current_line_type lexbuf) } |
575 | | '~' { start_line true; TTilde (get_current_line_type lexbuf) } | |
576 | ||
951c7801 C |
577 | | "++" { pass_zero(); |
578 | if !current_line_started | |
579 | then | |
580 | (start_line true; TInc (get_current_line_type lexbuf)) | |
581 | else (patch_or_match PATCH; | |
582 | add_current_line_type D.PLUSPLUS; token lexbuf) } | |
34e49164 | 583 | | "--" { start_line true; TDec (get_current_line_type lexbuf) } |
faf9a90c C |
584 | |
585 | | "=" { start_line true; TEq (get_current_line_type lexbuf) } | |
586 | ||
34e49164 C |
587 | | "-=" { start_line true; mkassign Ast.Minus lexbuf } |
588 | | "+=" { start_line true; mkassign Ast.Plus lexbuf } | |
faf9a90c | 589 | |
34e49164 C |
590 | | "*=" { start_line true; mkassign Ast.Mul lexbuf } |
591 | | "/=" { start_line true; mkassign Ast.Div lexbuf } | |
592 | | "%=" { start_line true; mkassign Ast.Mod lexbuf } | |
faf9a90c | 593 | |
34e49164 C |
594 | | "&=" { start_line true; mkassign Ast.And lexbuf } |
595 | | "|=" { start_line true; mkassign Ast.Or lexbuf } | |
596 | | "^=" { start_line true; mkassign Ast.Xor lexbuf } | |
faf9a90c | 597 | |
34e49164 C |
598 | | "<<=" { start_line true; mkassign Ast.DecLeft lexbuf } |
599 | | ">>=" { start_line true; mkassign Ast.DecRight lexbuf } | |
600 | ||
601 | | ":" { start_line true; TDotDot (get_current_line_type lexbuf) } | |
faf9a90c | 602 | |
951c7801 C |
603 | | "==" { start_line true; TEqEq (get_current_line_type lexbuf) } |
604 | | "!=" { start_line true; TNotEq (get_current_line_type lexbuf) } | |
34e49164 C |
605 | | ">=" { start_line true; |
606 | TLogOp(Ast.SupEq,get_current_line_type lexbuf) } | |
607 | | "<=" { start_line true; | |
608 | TLogOp(Ast.InfEq,get_current_line_type lexbuf) } | |
609 | | "<" { start_line true; | |
faf9a90c | 610 | TLogOp(Ast.Inf,get_current_line_type lexbuf) } |
34e49164 C |
611 | | ">" { start_line true; |
612 | TLogOp(Ast.Sup,get_current_line_type lexbuf) } | |
faf9a90c C |
613 | |
614 | | "&&" { start_line true; TAndLog (get_current_line_type lexbuf) } | |
34e49164 | 615 | | "||" { start_line true; TOrLog (get_current_line_type lexbuf) } |
faf9a90c | 616 | |
34e49164 C |
617 | | ">>" { start_line true; |
618 | TShOp(Ast.DecRight,get_current_line_type lexbuf) } | |
619 | | "<<" { start_line true; | |
620 | TShOp(Ast.DecLeft,get_current_line_type lexbuf) } | |
faf9a90c | 621 | |
34e49164 C |
622 | | "&" { start_line true; TAnd (get_current_line_type lexbuf) } |
623 | | "^" { start_line true; TXor(get_current_line_type lexbuf) } | |
624 | ||
978fd7e5 | 625 | | "##" { start_line true; TCppConcatOp } |
708f4980 | 626 | | (( ("#" [' ' '\t']* "define" [' ' '\t']+)) as def) |
faf9a90c | 627 | ( (letter (letter |digit)*) as ident) |
34e49164 C |
628 | { start_line true; |
629 | let (arity,line,lline,offset,col,strbef,straft,pos) as lt = | |
630 | get_current_line_type lexbuf in | |
708f4980 | 631 | let off = String.length def in |
34e49164 C |
632 | (* -1 in the code below because the ident is not at the line start *) |
633 | TDefine | |
634 | (lt, | |
635 | check_var ident | |
708f4980 C |
636 | (arity,line,lline,offset+off,col+off,[],[],Ast0.NoMetaPos)) } |
637 | | (( ("#" [' ' '\t']* "define" [' ' '\t']+)) as def) | |
faf9a90c | 638 | ( (letter (letter | digit)*) as ident) |
34e49164 C |
639 | '(' |
640 | { start_line true; | |
641 | let (arity,line,lline,offset,col,strbef,straft,pos) as lt = | |
642 | get_current_line_type lexbuf in | |
708f4980 | 643 | let off = String.length def in |
34e49164 C |
644 | TDefineParam |
645 | (lt, | |
646 | check_var ident | |
647 | (* why pos here but not above? *) | |
708f4980 C |
648 | (arity,line,lline,offset+off,col+off,strbef,straft,pos), |
649 | offset + off + (String.length ident), | |
650 | col + off + (String.length ident)) } | |
34e49164 C |
651 | | "#" [' ' '\t']* "include" [' ' '\t']* '"' [^ '"']+ '"' |
652 | { TIncludeL | |
653 | (let str = tok lexbuf in | |
654 | let start = String.index str '"' in | |
655 | let finish = String.rindex str '"' in | |
656 | start_line true; | |
657 | (process_include start finish str,get_current_line_type lexbuf)) } | |
658 | | "#" [' ' '\t']* "include" [' ' '\t']* '<' [^ '>']+ '>' | |
659 | { TIncludeNL | |
660 | (let str = tok lexbuf in | |
661 | let start = String.index str '<' in | |
662 | let finish = String.rindex str '>' in | |
663 | start_line true; | |
664 | (process_include start finish str,get_current_line_type lexbuf)) } | |
665 | | "#" [' ' '\t']* "if" [^'\n']* | |
666 | | "#" [' ' '\t']* "ifdef" [^'\n']* | |
667 | | "#" [' ' '\t']* "ifndef" [^'\n']* | |
668 | | "#" [' ' '\t']* "else" [^'\n']* | |
669 | | "#" [' ' '\t']* "elif" [^'\n']* | |
670 | | "#" [' ' '\t']* "endif" [^'\n']* | |
671 | | "#" [' ' '\t']* "error" [^'\n']* | |
672 | { start_line true; check_plus_linetype (tok lexbuf); | |
0708f913 C |
673 | TPragma (tok lexbuf, get_current_line_type lexbuf) } |
674 | | "/*" | |
675 | { start_line true; check_plus_linetype (tok lexbuf); | |
676 | (* second argument to TPragma is not quite right, because | |
677 | it represents only the first token of the comemnt, but that | |
678 | should be good enough *) | |
679 | TPragma ("/*"^(comment lexbuf), get_current_line_type lexbuf) } | |
34e49164 C |
680 | | "---" [^'\n']* |
681 | { (if !current_line_started | |
682 | then lexerr "--- must be at the beginning of the line" ""); | |
683 | start_line true; | |
684 | TMinusFile | |
685 | (let str = tok lexbuf in | |
686 | (drop_spaces(String.sub str 3 (String.length str - 3)), | |
687 | (get_current_line_type lexbuf))) } | |
688 | | "+++" [^'\n']* | |
689 | { (if !current_line_started | |
690 | then lexerr "+++ must be at the beginning of the line" ""); | |
691 | start_line true; | |
692 | TPlusFile | |
693 | (let str = tok lexbuf in | |
694 | (drop_spaces(String.sub str 3 (String.length str - 3)), | |
695 | (get_current_line_type lexbuf))) } | |
696 | ||
697 | | letter (letter | digit)* | |
faf9a90c | 698 | { start_line true; id_tokens lexbuf } |
34e49164 C |
699 | |
700 | | "'" { start_line true; | |
701 | TChar(char lexbuf,get_current_line_type lexbuf) } | |
702 | | '"' { start_line true; | |
703 | TString(string lexbuf,(get_current_line_type lexbuf)) } | |
704 | | (real as x) { start_line true; | |
705 | TFloat(x,(get_current_line_type lexbuf)) } | |
faf9a90c C |
706 | | ((( decimal | hexa | octal) |
707 | ( ['u' 'U'] | |
708 | | ['l' 'L'] | |
34e49164 C |
709 | | (['l' 'L'] ['u' 'U']) |
710 | | (['u' 'U'] ['l' 'L']) | |
711 | | (['u' 'U'] ['l' 'L'] ['l' 'L']) | |
712 | | (['l' 'L'] ['l' 'L']) | |
713 | )? | |
714 | ) as x) { start_line true; TInt(x,(get_current_line_type lexbuf)) } | |
715 | ||
716 | | "<=>" { TIso } | |
717 | | "=>" { TRightIso } | |
718 | ||
719 | | eof { EOF } | |
720 | ||
721 | | _ { lexerr "unrecognised symbol, in token rule: " (tok lexbuf) } | |
722 | ||
723 | ||
724 | and char = parse | |
725 | | (_ as x) "'" { String.make 1 x } | |
726 | | (("\\" (oct | oct oct | oct oct oct)) as x "'") { x } | |
727 | | (("\\x" (hex | hex hex)) as x "'") { x } | |
728 | | (("\\" (_ as v)) as x "'") | |
729 | { (match v with | |
730 | | 'n' -> () | 't' -> () | 'v' -> () | 'b' -> () | |
731 | | 'r' -> () | 'f' -> () | 'a' -> () | |
732 | | '\\' -> () | '?' -> () | '\'' -> () | '"' -> () | |
733 | | 'e' -> () | |
734 | | _ -> lexerr "unrecognised symbol: " (tok lexbuf) | |
735 | ); | |
736 | x | |
faf9a90c | 737 | } |
34e49164 C |
738 | | _ { lexerr "unrecognised symbol: " (tok lexbuf) } |
739 | ||
740 | and string = parse | |
741 | | '"' { "" } | |
742 | | (_ as x) { Common.string_of_char x ^ string lexbuf } | |
743 | | ("\\" (oct | oct oct | oct oct oct)) as x { x ^ string lexbuf } | |
744 | | ("\\x" (hex | hex hex)) as x { x ^ string lexbuf } | |
faf9a90c C |
745 | | ("\\" (_ as v)) as x |
746 | { | |
34e49164 | 747 | (match v with |
9f8e26f4 C |
748 | | 'n' -> () | 't' -> () | 'v' -> () | 'b' -> () | 'r' -> () |
749 | | 'f' -> () | 'a' -> () | |
750 | | '\\' -> () | '?' -> () | '\'' -> () | '"' -> () | |
751 | | 'e' -> () | |
752 | | '\n' -> () | |
753 | | '(' -> () | '|' -> () | ')' -> () | |
754 | | _ -> lexerr "unrecognised symbol:" (tok lexbuf) | |
34e49164 C |
755 | ); |
756 | x ^ string lexbuf | |
757 | } | |
758 | | _ { lexerr "unrecognised symbol: " (tok lexbuf) } | |
0708f913 C |
759 | |
760 | and comment = parse | |
761 | | "*/" { start_line true; tok lexbuf } | |
762 | | ['\n' '\r' '\011' '\012'] | |
763 | { reset_line lexbuf; let s = tok lexbuf in s ^ comment lexbuf } | |
764 | | "+" { pass_zero(); | |
765 | if !current_line_started | |
766 | then (start_line true; let s = tok lexbuf in s^(comment lexbuf)) | |
767 | else comment lexbuf } | |
768 | (* noteopti: *) | |
769 | | [^ '*'] { start_line true; let s = tok lexbuf in s ^ comment lexbuf } | |
770 | | [ '*'] { start_line true; let s = tok lexbuf in s ^ comment lexbuf } | |
951c7801 | 771 | | _ |
0708f913 C |
772 | { start_line true; let s = tok lexbuf in |
773 | Common.pr2 ("LEXER: unrecognised symbol in comment:"^s); | |
774 | s ^ comment lexbuf | |
775 | } | |
776 |