389
|
1 |
(* Title: Pure/Thy/thy_parse.ML
|
|
2 |
ID: $Id$
|
|
3 |
Author: Markus Wenzel, TU Muenchen
|
|
4 |
|
|
5 |
The parser for theory files.
|
|
6 |
|
|
7 |
TODO:
|
|
8 |
remove quote in syn_err (?)
|
|
9 |
check: names vs names1
|
|
10 |
*)
|
|
11 |
|
|
12 |
infix 5 -- --$$ $$-- ^^;
|
|
13 |
infix 3 >>;
|
|
14 |
infix 0 ||;
|
|
15 |
|
|
16 |
signature THY_PARSE =
|
|
17 |
sig
|
|
18 |
type token
|
|
19 |
val !! : ('a -> 'b * 'c) -> 'a -> 'b * 'c
|
|
20 |
val >> : ('a -> 'b * 'c) * ('b -> 'd) -> 'a -> 'd * 'c
|
|
21 |
val || : ('a -> 'b) * ('a -> 'b) -> 'a -> 'b
|
|
22 |
val -- : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> ('b * 'd) * 'e
|
|
23 |
val ^^ : ('a -> string * 'b) * ('b -> string * 'c) -> 'a -> string * 'c
|
|
24 |
val $$ : string -> token list -> string * token list
|
|
25 |
val $$-- : string * (token list -> 'b * 'c) -> token list -> 'b * 'c
|
|
26 |
val --$$ : ('a -> 'b * token list) * string -> 'a -> 'b * token list
|
|
27 |
val ident: token list -> string * token list
|
|
28 |
val long_ident: token list -> string * token list
|
|
29 |
val long_id: token list -> string * token list
|
|
30 |
val type_var: token list -> string * token list
|
|
31 |
val nat: token list -> string * token list
|
|
32 |
val string: token list -> string * token list
|
|
33 |
val verbatim: token list -> string * token list
|
|
34 |
val empty: 'a -> 'b list * 'a
|
|
35 |
val optional: ('a -> 'b * 'a) -> 'b -> 'a -> 'b * 'a
|
|
36 |
val repeat: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
|
37 |
val repeat1: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
|
38 |
val enum: string -> (token list -> 'a * token list)
|
|
39 |
-> token list -> 'a list * token list
|
|
40 |
val enum1: string -> (token list -> 'a * token list)
|
|
41 |
-> token list -> 'a list * token list
|
|
42 |
val list: (token list -> 'a * token list)
|
|
43 |
-> token list -> 'a list * token list
|
|
44 |
val list1: (token list -> 'a * token list)
|
|
45 |
-> token list -> 'a list * token list
|
|
46 |
val name: token list -> string * token list
|
|
47 |
val sort: token list -> string * token list
|
|
48 |
type syntax
|
|
49 |
val make_syntax: string list ->
|
|
50 |
(string * (token list -> (string * string) * token list)) list -> syntax
|
|
51 |
val parse_thy: syntax -> string -> string
|
|
52 |
val section: string -> string -> (token list -> string * token list)
|
|
53 |
-> (string * (token list -> (string * string) * token list))
|
|
54 |
val axm_section: string -> string
|
|
55 |
-> (token list -> (string * string list) * token list)
|
|
56 |
-> (string * (token list -> (string * string) * token list))
|
|
57 |
val pure_keywords: string list
|
|
58 |
val pure_sections:
|
|
59 |
(string * (token list -> (string * string) * token list)) list
|
|
60 |
val pure_syntax: syntax
|
|
61 |
end;
|
|
62 |
|
|
63 |
functor ThyParseFun(structure Symtab: SYMTAB and ThyScan: THY_SCAN)(*: THY_PARSE *) = (* FIXME *)
|
|
64 |
struct
|
|
65 |
|
|
66 |
open ThyScan;
|
|
67 |
|
|
68 |
|
|
69 |
(** parser toolbox **)
|
|
70 |
|
|
71 |
type token = token_kind * string * int;
|
|
72 |
|
|
73 |
|
|
74 |
(* errors *)
|
|
75 |
|
|
76 |
exception SYNTAX_ERROR of string * string * int;
|
|
77 |
|
|
78 |
fun syn_err s1 s2 n = raise SYNTAX_ERROR (s1, s2, n);
|
|
79 |
|
|
80 |
fun eof_err () = error "Unexpected end-of-file";
|
|
81 |
|
|
82 |
fun !! parse toks = parse toks
|
|
83 |
handle SYNTAX_ERROR (s1, s2, n) => error ("Syntax error on line " ^
|
|
84 |
string_of_int n ^ ": " ^ s1 ^ " expected and " ^ s2 ^ " was found");
|
|
85 |
|
|
86 |
|
|
87 |
(* parser combinators *)
|
|
88 |
|
|
89 |
fun (parse >> f) toks = apfst f (parse toks);
|
|
90 |
|
|
91 |
fun (parse1 || parse2) toks =
|
|
92 |
parse1 toks handle SYNTAX_ERROR _ => parse2 toks;
|
|
93 |
|
|
94 |
fun (parse1 -- parse2) toks =
|
|
95 |
let
|
|
96 |
val (x, toks') = parse1 toks;
|
|
97 |
val (y, toks'') = parse2 toks';
|
|
98 |
in
|
|
99 |
((x, y), toks'')
|
|
100 |
end;
|
|
101 |
|
|
102 |
fun (parse1 ^^ parse2) = parse1 -- parse2 >> op ^;
|
|
103 |
|
|
104 |
|
|
105 |
(* generic parsers *)
|
|
106 |
|
|
107 |
fun $$ a ((k, b, n) :: toks) =
|
|
108 |
if k = Keyword andalso a = b then (a, toks)
|
|
109 |
else syn_err (quote a) (quote b) n
|
|
110 |
| $$ _ [] = eof_err ();
|
|
111 |
|
|
112 |
fun (a $$-- parse) = $$ a -- parse >> #2;
|
|
113 |
|
|
114 |
fun (parse --$$ a) = parse -- $$ a >> #1;
|
|
115 |
|
|
116 |
|
|
117 |
fun kind k1 ((k2, s, n) :: toks) =
|
|
118 |
if k1 = k2 then (s, toks)
|
|
119 |
else syn_err (name_of_kind k1) (quote s) n
|
|
120 |
| kind _ [] = eof_err ();
|
|
121 |
|
|
122 |
val ident = kind Ident;
|
|
123 |
val long_ident = kind LongIdent;
|
|
124 |
val long_id = ident || long_ident;
|
|
125 |
val type_var = kind TypeVar >> quote;
|
|
126 |
val nat = kind Nat;
|
|
127 |
val string = kind String;
|
|
128 |
val verbatim = kind Verbatim;
|
|
129 |
val eof = kind EOF;
|
|
130 |
|
|
131 |
fun empty toks = ([], toks);
|
|
132 |
|
|
133 |
fun optional parse def = parse || empty >> K def;
|
|
134 |
|
|
135 |
fun repeat parse toks = (parse -- repeat parse >> op :: || empty) toks;
|
|
136 |
fun repeat1 parse = parse -- repeat parse >> op ::;
|
|
137 |
|
|
138 |
fun enum1 sep parse = parse -- repeat (sep $$-- parse) >> op ::;
|
|
139 |
fun enum sep parse = enum1 sep parse || empty;
|
|
140 |
|
|
141 |
val list = enum ",";
|
|
142 |
val list1 = enum1 ",";
|
|
143 |
|
|
144 |
|
|
145 |
|
|
146 |
(** theory parsers **)
|
|
147 |
|
|
148 |
(* misc utilities *)
|
|
149 |
|
|
150 |
fun cat s1 s2 = s1 ^ " " ^ s2;
|
|
151 |
|
|
152 |
val pars = parents "(" ")";
|
|
153 |
val brackets = parents "[" "]";
|
|
154 |
|
|
155 |
val mk_list = brackets o commas;
|
|
156 |
val mk_big_list = brackets o space_implode ",\n ";
|
|
157 |
|
|
158 |
fun mk_pair (x, y) = pars (commas [x, y]);
|
|
159 |
fun mk_triple (x, y, z) = pars (commas [x, y, z]);
|
|
160 |
fun mk_triple1 ((x, y), z) = mk_triple (x, y, z);
|
|
161 |
fun mk_triple2 (x, (y, z)) = mk_triple (x, y, z);
|
|
162 |
|
|
163 |
val split_decls = flat o map (fn (xs, y) => map (rpair y) xs);
|
|
164 |
|
|
165 |
fun strip_quotes str =
|
|
166 |
implode (tl (take (size str - 1, explode str)));
|
|
167 |
|
|
168 |
|
|
169 |
(* names *)
|
|
170 |
|
|
171 |
val name = ident >> quote || string;
|
|
172 |
val names = list name;
|
|
173 |
val names1 = list1 name;
|
|
174 |
val name_list = names >> mk_list;
|
|
175 |
val name_list1 = names1 >> mk_list;
|
|
176 |
|
|
177 |
|
|
178 |
(* classes *)
|
|
179 |
|
|
180 |
fun mk_subclass (c, cs) = mk_triple ("[]", c, cs);
|
|
181 |
|
|
182 |
val subclass = name -- optional ("<" $$-- !! name_list1) "[]";
|
|
183 |
|
|
184 |
val class_decls = repeat1 (subclass >> mk_subclass) >> mk_big_list;
|
|
185 |
|
|
186 |
|
|
187 |
(* arities *)
|
|
188 |
|
|
189 |
val sort =
|
|
190 |
name >> brackets ||
|
|
191 |
"{" $$-- name_list --$$ "}";
|
|
192 |
|
|
193 |
val sort_list1 = list1 sort >> mk_list;
|
|
194 |
|
|
195 |
|
|
196 |
val arity = optional ("(" $$-- !! (sort_list1 --$$")")) "[]" -- sort;
|
|
197 |
|
|
198 |
val arity_decls = repeat1 (names1 --$$ "::" -- !! arity)
|
|
199 |
>> (mk_big_list o map mk_triple2 o split_decls);
|
|
200 |
|
|
201 |
|
|
202 |
(* mixfix annotations *)
|
|
203 |
|
|
204 |
val infxl = "infixl" $$-- !! nat >> cat "Infixl";
|
|
205 |
val infxr = "infixr" $$-- !! nat >> cat "Infixr";
|
|
206 |
|
|
207 |
val binder = "binder" $$-- !! (string -- nat) >> (cat "Binder" o mk_pair);
|
|
208 |
|
|
209 |
val opt_pris = optional ("[" $$-- !! (list nat --$$ "]")) [] >> mk_list;
|
|
210 |
|
|
211 |
val mixfix = string -- !! (opt_pris -- optional nat "max_pri")
|
|
212 |
>> (cat "Mixfix" o mk_triple2);
|
|
213 |
|
|
214 |
fun opt_syn fx =
|
|
215 |
"(" $$-- fx --$$ ")" ||
|
|
216 |
empty >> K "NoSyn";
|
|
217 |
|
|
218 |
val opt_infix = opt_syn (infxl || infxr);
|
|
219 |
val opt_mixfix = opt_syn (mixfix || infxl || infxr || binder);
|
|
220 |
|
|
221 |
|
|
222 |
(* types *)
|
|
223 |
|
|
224 |
fun mk_old_type_decl ((ts, n), syn) =
|
|
225 |
map (fn t => (mk_triple (t, n, syn), false)) ts;
|
|
226 |
|
|
227 |
fun mk_type_decl (((xs, t), None), syn) =
|
|
228 |
[(mk_triple (t, string_of_int (length xs), syn), false)]
|
|
229 |
| mk_type_decl (((xs, t), Some rhs), syn) =
|
|
230 |
[(pars (commas [t, mk_list xs, rhs, syn]), true)];
|
|
231 |
|
|
232 |
fun mk_type_decls tys =
|
|
233 |
"also add_types\n" ^ mk_big_list (keyfilter tys false) ^ "\n\n\
|
|
234 |
\also add_tyabbrs\n" ^ mk_big_list (keyfilter tys true);
|
|
235 |
|
|
236 |
|
|
237 |
val old_type_decl = names -- nat -- opt_infix >> mk_old_type_decl;
|
|
238 |
|
|
239 |
val type_args =
|
|
240 |
type_var >> (fn x => [x]) ||
|
|
241 |
"(" $$-- !! (list1 type_var --$$ ")") ||
|
|
242 |
empty >> K [];
|
|
243 |
|
|
244 |
val type_decl = type_args -- name -- optional ("=" $$-- !! string >> Some) None
|
|
245 |
-- opt_infix >> mk_type_decl;
|
|
246 |
|
|
247 |
val type_decls = repeat1 (old_type_decl || type_decl)
|
|
248 |
>> (rpair "" o mk_type_decls o flat);
|
|
249 |
|
|
250 |
|
|
251 |
(* consts *)
|
|
252 |
|
|
253 |
val const_decls = repeat1 (names1 --$$ "::" -- !! (string -- opt_mixfix))
|
|
254 |
>> (mk_big_list o map mk_triple2 o split_decls);
|
|
255 |
|
|
256 |
|
|
257 |
(* translations *)
|
|
258 |
|
|
259 |
val trans_pat =
|
|
260 |
optional ("(" $$-- !! (name --$$ ")")) "\"logic\"" -- string >> mk_pair;
|
|
261 |
|
|
262 |
val trans_arrow =
|
|
263 |
$$ "=>" >> K " |-> " ||
|
|
264 |
$$ "<=" >> K " <-| " ||
|
|
265 |
$$ "==" >> K " <-> ";
|
|
266 |
|
|
267 |
val trans_decls = repeat1 (trans_pat ^^ !! (trans_arrow ^^ trans_pat))
|
|
268 |
>> mk_big_list;
|
|
269 |
|
|
270 |
|
|
271 |
(* ML translations *)
|
|
272 |
|
|
273 |
val trfun_defs =
|
|
274 |
" val parse_ast_translation = [];\n\
|
|
275 |
\ val parse_translation = [];\n\
|
|
276 |
\ val print_translation = [];\n\
|
|
277 |
\ val print_ast_translation = [];";
|
|
278 |
|
|
279 |
val trfun_args =
|
|
280 |
"(parse_ast_translation, parse_translation, \
|
|
281 |
\print_translation, print_ast_translation)";
|
|
282 |
|
|
283 |
fun mk_mltrans txt =
|
|
284 |
"let\n"
|
|
285 |
^ trfun_defs ^ "\n"
|
|
286 |
^ txt ^ "\n\
|
|
287 |
\in\n\
|
|
288 |
\ " ^ trfun_args ^ "\n\
|
|
289 |
\end";
|
|
290 |
|
|
291 |
val mltrans = verbatim >> mk_mltrans;
|
|
292 |
|
|
293 |
|
|
294 |
(* axioms *)
|
|
295 |
|
|
296 |
val mk_axms = mk_big_list o map (mk_pair o apfst quote);
|
|
297 |
|
|
298 |
fun mk_axiom_decls axms = (mk_axms axms, map fst axms);
|
|
299 |
|
|
300 |
val axiom_decls = repeat1 (ident -- !! string) >> mk_axiom_decls;
|
|
301 |
|
|
302 |
|
|
303 |
(* axclass *)
|
|
304 |
|
|
305 |
fun mk_axclass_decl ((c, cs), axms) =
|
|
306 |
(mk_pair (c, cs) ^ "\n" ^ mk_axms axms,
|
|
307 |
(strip_quotes c ^ "I") :: map fst axms);
|
|
308 |
|
|
309 |
val axclass_decl = subclass -- repeat (ident -- !! string) >> mk_axclass_decl;
|
|
310 |
|
|
311 |
|
|
312 |
(* instance *)
|
|
313 |
|
|
314 |
fun mk_instance_decl ((((t, ss), c), axths), opt_tac) =
|
|
315 |
mk_triple (t, ss, c) ^ "\n" ^
|
|
316 |
mk_list (keyfilter axths false) ^ "\n" ^
|
|
317 |
mk_list (keyfilter axths true) ^ "\n" ^
|
|
318 |
opt_tac;
|
|
319 |
|
|
320 |
val axm_or_thm =
|
|
321 |
string >> rpair false ||
|
|
322 |
long_id >> rpair true;
|
|
323 |
|
|
324 |
val instance_decl =
|
|
325 |
name --$$ "::" -- optional ("(" $$-- sort_list1 --$$")") "[]" -- name --
|
|
326 |
optional ("(" $$-- list1 axm_or_thm --$$ ")") [] --
|
|
327 |
optional (verbatim >> (pars o cat "Some" o pars)) "None"
|
|
328 |
>> mk_instance_decl;
|
|
329 |
|
|
330 |
|
|
331 |
|
|
332 |
(** theory syntax **)
|
|
333 |
|
|
334 |
type syntax =
|
|
335 |
lexicon * (token list -> (string * string) * token list) Symtab.table;
|
|
336 |
|
|
337 |
fun make_syntax keywords sects =
|
|
338 |
(make_lexicon keywords, Symtab.make sects handle Symtab.DUPS names
|
|
339 |
=> error ("Duplicate sections in thy syntax: " ^ commas_quote names));
|
|
340 |
|
|
341 |
|
|
342 |
(* header *)
|
|
343 |
|
|
344 |
fun mk_header (thy_name, bases) =
|
|
345 |
(thy_name, "(base_on " ^ mk_list bases ^ " " ^ quote thy_name ^ ")");
|
|
346 |
|
|
347 |
val base =
|
|
348 |
ident >> (cat "Thy" o quote) ||
|
|
349 |
string >> cat "File";
|
|
350 |
|
|
351 |
val header = ident --$$ "=" -- enum1 "+" base >> mk_header;
|
|
352 |
|
|
353 |
|
|
354 |
(* extension *)
|
|
355 |
|
|
356 |
fun mk_extension (txts, mltxt) =
|
|
357 |
let
|
|
358 |
val cat_sects = space_implode "\n\n" o filter_out (equal "");
|
|
359 |
val (extxts, postxts) = split_list txts;
|
|
360 |
in
|
|
361 |
(cat_sects extxts, cat_sects postxts, mltxt)
|
|
362 |
end;
|
|
363 |
|
|
364 |
fun sect tab ((Keyword, s, n) :: toks) =
|
|
365 |
(case Symtab.lookup (tab, s) of
|
|
366 |
Some parse => !! parse toks
|
|
367 |
| None => syn_err "section" s n)
|
|
368 |
| sect _ ((_, s, n) :: _) = syn_err "section" s n
|
|
369 |
| sect _ [] = eof_err ();
|
|
370 |
|
|
371 |
fun extension sectab = "+" $$-- !! (repeat (sect sectab) --$$ "end") --
|
|
372 |
optional ("ML" $$-- verbatim) "" >> mk_extension;
|
|
373 |
|
|
374 |
|
|
375 |
(* theory definition *)
|
|
376 |
|
|
377 |
fun mk_structure ((thy_name, old_thys), Some (extxt, postxt, mltxt)) =
|
|
378 |
"structure " ^ thy_name ^ " =\n\
|
|
379 |
\struct\n\
|
|
380 |
\\n\
|
|
381 |
\local\n" ^ " open Mixfix;\n" (* FIXME tmp *)
|
|
382 |
^ trfun_defs ^ "\n\
|
|
383 |
\in\n\
|
|
384 |
\\n"
|
|
385 |
^ mltxt ^ "\n\
|
|
386 |
\\n\
|
|
387 |
\val thy = " ^ old_thys ^ "\n\n\
|
|
388 |
\also add_trfuns\n"
|
|
389 |
^ trfun_args ^ "\n\
|
|
390 |
\\n"
|
|
391 |
^ extxt ^ "\n\
|
|
392 |
\\n\
|
|
393 |
\also add_thyname " ^ quote thy_name ^ ";\n\
|
|
394 |
\\n\
|
|
395 |
\\n"
|
|
396 |
^ postxt ^ "\n\
|
|
397 |
\\n\
|
|
398 |
\end;\n\
|
|
399 |
\end;\n"
|
|
400 |
| mk_structure ((thy_name, old_thys), None) =
|
|
401 |
"structure " ^ thy_name ^ " =\n\
|
|
402 |
\struct\n\
|
|
403 |
\\n\
|
|
404 |
\val thy = " ^ old_thys ^ ";\n\
|
|
405 |
\\n\
|
|
406 |
\end;\n";
|
|
407 |
|
|
408 |
fun theory_defn sectab =
|
|
409 |
header -- optional (extension sectab >> Some) None -- eof
|
|
410 |
>> (mk_structure o #1);
|
|
411 |
|
|
412 |
fun parse_thy (lex, sectab) str =
|
|
413 |
#1 (!! (theory_defn sectab) (tokenize lex str));
|
|
414 |
|
|
415 |
|
|
416 |
(* standard sections *)
|
|
417 |
|
|
418 |
fun mk_val ax = "val " ^ ax ^ " = get_axiom thy " ^ quote ax ^ ";";
|
|
419 |
|
|
420 |
fun mk_axm_sect pretxt (txt, axs) =
|
|
421 |
(pretxt ^ "\n" ^ txt, cat_lines (map mk_val axs));
|
|
422 |
|
|
423 |
fun axm_section name pretxt parse =
|
|
424 |
(name, parse >> mk_axm_sect pretxt);
|
|
425 |
|
|
426 |
fun section name pretxt parse =
|
|
427 |
axm_section name pretxt (parse >> rpair []);
|
|
428 |
|
|
429 |
|
|
430 |
val pure_keywords =
|
|
431 |
["classes", "default", "types", "arities", "consts", "syntax",
|
|
432 |
"translations", "MLtrans", "MLtext", "rules", "defns", "axclass",
|
|
433 |
"instance", "end", "ML", "mixfix", "infixr", "infixl", "binder", "=", "+",
|
|
434 |
",", "<", "{", "}", "(", ")", "[", "]", "::", "==", "=>", "<="];
|
|
435 |
|
|
436 |
val pure_sections =
|
|
437 |
[section "classes" "also add_classes" class_decls,
|
|
438 |
section "default" "also add_defsort" sort,
|
|
439 |
("types", type_decls),
|
|
440 |
section "arities" "also add_arities" arity_decls,
|
|
441 |
section "consts" "also add_consts" const_decls,
|
|
442 |
section "syntax" "also add_syntax" const_decls,
|
|
443 |
section "translations" "also add_trrules" trans_decls,
|
|
444 |
section "MLtrans" "also add_trfuns" mltrans,
|
|
445 |
("MLtext", verbatim >> rpair ""),
|
|
446 |
axm_section "rules" "also add_axioms" axiom_decls,
|
|
447 |
axm_section "defns" "also add_defns" axiom_decls,
|
|
448 |
axm_section "axclass" "also add_axclass" axclass_decl,
|
|
449 |
section "instance" "also add_instance" instance_decl];
|
|
450 |
|
|
451 |
|
|
452 |
(* FIXME -> thy_read.ML *)
|
|
453 |
val pure_syntax = make_syntax pure_keywords pure_sections;
|
|
454 |
|
|
455 |
end;
|
|
456 |
|