wenzelm@240
|
1 |
(* Title: Pure/Syntax/syn_ext.ML
|
wenzelm@240
|
2 |
ID: $Id$
|
wenzelm@911
|
3 |
Author: Markus Wenzel and Carsten Clasohm, TU Muenchen
|
wenzelm@240
|
4 |
|
wenzelm@240
|
5 |
Syntax extension (internal interface).
|
wenzelm@240
|
6 |
*)
|
wenzelm@240
|
7 |
|
wenzelm@240
|
8 |
signature SYN_EXT0 =
|
paulson@1510
|
9 |
sig
|
wenzelm@240
|
10 |
val typeT: typ
|
wenzelm@240
|
11 |
val constrainC: string
|
paulson@1510
|
12 |
end;
|
wenzelm@240
|
13 |
|
wenzelm@240
|
14 |
signature SYN_EXT =
|
paulson@1510
|
15 |
sig
|
wenzelm@240
|
16 |
include SYN_EXT0
|
paulson@1510
|
17 |
val logic: string
|
paulson@1510
|
18 |
val args: string
|
paulson@1510
|
19 |
val cargs: string
|
paulson@1510
|
20 |
val any: string
|
paulson@1510
|
21 |
val sprop: string
|
paulson@1510
|
22 |
val typ_to_nonterm: typ -> string
|
paulson@1510
|
23 |
datatype xsymb =
|
paulson@1510
|
24 |
Delim of string |
|
paulson@1510
|
25 |
Argument of string * int |
|
paulson@1510
|
26 |
Space of string |
|
paulson@1510
|
27 |
Bg of int | Brk of int | En
|
paulson@1510
|
28 |
datatype xprod = XProd of string * xsymb list * string * int
|
paulson@1510
|
29 |
val max_pri: int
|
paulson@1510
|
30 |
val chain_pri: int
|
paulson@1510
|
31 |
val delims_of: xprod list -> string list
|
paulson@1510
|
32 |
datatype mfix = Mfix of string * typ * string * int list * int
|
paulson@1510
|
33 |
datatype syn_ext =
|
paulson@1510
|
34 |
SynExt of {
|
paulson@1510
|
35 |
logtypes: string list,
|
paulson@1510
|
36 |
xprods: xprod list,
|
paulson@1510
|
37 |
consts: string list,
|
paulson@1510
|
38 |
parse_ast_translation: (string * (Ast.ast list -> Ast.ast)) list,
|
paulson@1510
|
39 |
parse_rules: (Ast.ast * Ast.ast) list,
|
paulson@1510
|
40 |
parse_translation: (string * (term list -> term)) list,
|
paulson@1510
|
41 |
print_translation: (string * (term list -> term)) list,
|
paulson@1510
|
42 |
print_rules: (Ast.ast * Ast.ast) list,
|
paulson@1510
|
43 |
print_ast_translation: (string * (Ast.ast list -> Ast.ast)) list}
|
paulson@1510
|
44 |
val mk_syn_ext: bool -> string list -> mfix list ->
|
paulson@1510
|
45 |
string list -> (string * (Ast.ast list -> Ast.ast)) list *
|
paulson@1510
|
46 |
(string * (term list -> term)) list *
|
paulson@1510
|
47 |
(string * (term list -> term)) list * (string * (Ast.ast list -> Ast.ast)) list
|
paulson@1510
|
48 |
-> (Ast.ast * Ast.ast) list * (Ast.ast * Ast.ast) list -> syn_ext
|
paulson@1510
|
49 |
val syn_ext: string list -> mfix list -> string list ->
|
paulson@1510
|
50 |
(string * (Ast.ast list -> Ast.ast)) list * (string * (term list -> term)) list *
|
paulson@1510
|
51 |
(string * (term list -> term)) list * (string * (Ast.ast list -> Ast.ast)) list
|
paulson@1510
|
52 |
-> (Ast.ast * Ast.ast) list * (Ast.ast * Ast.ast) list -> syn_ext
|
paulson@1510
|
53 |
val syn_ext_logtypes: string list -> syn_ext
|
paulson@1510
|
54 |
val syn_ext_const_names: string list -> string list -> syn_ext
|
paulson@1510
|
55 |
val syn_ext_rules: string list -> (Ast.ast * Ast.ast) list * (Ast.ast * Ast.ast) list -> syn_ext
|
paulson@1510
|
56 |
val syn_ext_trfuns: string list ->
|
paulson@1510
|
57 |
(string * (Ast.ast list -> Ast.ast)) list * (string * (term list -> term)) list *
|
paulson@1510
|
58 |
(string * (term list -> term)) list * (string * (Ast.ast list -> Ast.ast)) list
|
paulson@1510
|
59 |
-> syn_ext
|
paulson@1510
|
60 |
val pure_ext: syn_ext
|
paulson@1510
|
61 |
end;
|
wenzelm@240
|
62 |
|
paulson@1510
|
63 |
structure SynExt : SYN_EXT =
|
wenzelm@240
|
64 |
struct
|
wenzelm@240
|
65 |
|
wenzelm@240
|
66 |
open Lexicon Ast;
|
wenzelm@240
|
67 |
|
wenzelm@240
|
68 |
(** misc definitions **)
|
wenzelm@240
|
69 |
|
wenzelm@240
|
70 |
(* syntactic categories *)
|
wenzelm@240
|
71 |
|
wenzelm@240
|
72 |
val logic = "logic";
|
wenzelm@240
|
73 |
val logicT = Type (logic, []);
|
wenzelm@240
|
74 |
|
wenzelm@240
|
75 |
val args = "args";
|
clasohm@1178
|
76 |
val cargs = "cargs";
|
wenzelm@240
|
77 |
|
clasohm@330
|
78 |
val typeT = Type ("type", []);
|
clasohm@330
|
79 |
|
clasohm@764
|
80 |
val sprop = "#prop";
|
clasohm@764
|
81 |
val spropT = Type (sprop, []);
|
wenzelm@240
|
82 |
|
clasohm@764
|
83 |
val any = "any";
|
clasohm@624
|
84 |
val anyT = Type (any, []);
|
clasohm@624
|
85 |
|
wenzelm@780
|
86 |
|
wenzelm@240
|
87 |
(* constants *)
|
wenzelm@240
|
88 |
|
wenzelm@240
|
89 |
val constrainC = "_constrain";
|
wenzelm@240
|
90 |
|
wenzelm@240
|
91 |
|
wenzelm@240
|
92 |
|
wenzelm@240
|
93 |
(** datatype xprod **)
|
wenzelm@240
|
94 |
|
wenzelm@240
|
95 |
(*Delim s: delimiter s
|
wenzelm@240
|
96 |
Argument (s, p): nonterminal s requiring priority >= p, or valued token
|
wenzelm@240
|
97 |
Space s: some white space for printing
|
wenzelm@240
|
98 |
Bg, Brk, En: blocks and breaks for pretty printing*)
|
wenzelm@240
|
99 |
|
wenzelm@240
|
100 |
datatype xsymb =
|
wenzelm@240
|
101 |
Delim of string |
|
wenzelm@240
|
102 |
Argument of string * int |
|
wenzelm@240
|
103 |
Space of string |
|
wenzelm@240
|
104 |
Bg of int | Brk of int | En;
|
wenzelm@240
|
105 |
|
wenzelm@240
|
106 |
|
wenzelm@240
|
107 |
(*XProd (lhs, syms, c, p):
|
wenzelm@240
|
108 |
lhs: name of nonterminal on the lhs of the production
|
wenzelm@240
|
109 |
syms: list of symbols on the rhs of the production
|
wenzelm@240
|
110 |
c: head of parse tree
|
wenzelm@240
|
111 |
p: priority of this production*)
|
wenzelm@240
|
112 |
|
wenzelm@240
|
113 |
datatype xprod = XProd of string * xsymb list * string * int;
|
wenzelm@240
|
114 |
|
wenzelm@240
|
115 |
val max_pri = 1000; (*maximum legal priority*)
|
wenzelm@240
|
116 |
val chain_pri = ~1; (*dummy for chain productions*)
|
wenzelm@240
|
117 |
|
wenzelm@240
|
118 |
|
wenzelm@240
|
119 |
(* delims_of *)
|
wenzelm@240
|
120 |
|
wenzelm@240
|
121 |
fun delims_of xprods =
|
wenzelm@240
|
122 |
let
|
wenzelm@240
|
123 |
fun del_of (Delim s) = Some s
|
wenzelm@240
|
124 |
| del_of _ = None;
|
wenzelm@240
|
125 |
|
wenzelm@240
|
126 |
fun dels_of (XProd (_, xsymbs, _, _)) =
|
wenzelm@240
|
127 |
mapfilter del_of xsymbs;
|
wenzelm@240
|
128 |
in
|
wenzelm@240
|
129 |
distinct (flat (map dels_of xprods))
|
wenzelm@240
|
130 |
end;
|
wenzelm@240
|
131 |
|
wenzelm@240
|
132 |
|
wenzelm@240
|
133 |
|
wenzelm@240
|
134 |
(** datatype mfix **)
|
wenzelm@240
|
135 |
|
wenzelm@240
|
136 |
(*Mfix (sy, ty, c, ps, p):
|
wenzelm@240
|
137 |
sy: rhs of production as symbolic string
|
wenzelm@240
|
138 |
ty: type description of production
|
wenzelm@240
|
139 |
c: head of parse tree
|
wenzelm@240
|
140 |
ps: priorities of arguments in sy
|
wenzelm@240
|
141 |
p: priority of production*)
|
wenzelm@240
|
142 |
|
wenzelm@240
|
143 |
datatype mfix = Mfix of string * typ * string * int list * int;
|
wenzelm@240
|
144 |
|
wenzelm@240
|
145 |
|
wenzelm@240
|
146 |
(* typ_to_nonterm *)
|
wenzelm@240
|
147 |
|
clasohm@865
|
148 |
fun typ_to_nt _ (Type (c, _)) = c
|
clasohm@865
|
149 |
| typ_to_nt default _ = default;
|
clasohm@865
|
150 |
|
clasohm@764
|
151 |
(*get nonterminal for rhs*)
|
clasohm@865
|
152 |
val typ_to_nonterm = typ_to_nt any;
|
wenzelm@240
|
153 |
|
clasohm@764
|
154 |
(*get nonterminal for lhs*)
|
clasohm@865
|
155 |
val typ_to_nonterm1 = typ_to_nt logic;
|
wenzelm@240
|
156 |
|
wenzelm@240
|
157 |
|
wenzelm@240
|
158 |
(* mfix_to_xprod *)
|
wenzelm@240
|
159 |
|
clasohm@764
|
160 |
fun mfix_to_xprod convert logtypes (Mfix (sy, typ, const, pris, pri)) =
|
wenzelm@240
|
161 |
let
|
wenzelm@240
|
162 |
fun err msg =
|
wenzelm@780
|
163 |
(writeln ("Error in mixfix annotation " ^ quote sy ^ " for "
|
clasohm@764
|
164 |
^ quote const);
|
wenzelm@240
|
165 |
error msg);
|
wenzelm@240
|
166 |
|
wenzelm@240
|
167 |
fun check_pri p =
|
wenzelm@240
|
168 |
if p >= 0 andalso p <= max_pri then ()
|
wenzelm@240
|
169 |
else err ("precedence out of range: " ^ string_of_int p);
|
wenzelm@240
|
170 |
|
wenzelm@240
|
171 |
fun blocks_ok [] 0 = true
|
wenzelm@240
|
172 |
| blocks_ok [] _ = false
|
wenzelm@240
|
173 |
| blocks_ok (Bg _ :: syms) n = blocks_ok syms (n + 1)
|
wenzelm@240
|
174 |
| blocks_ok (En :: _) 0 = false
|
wenzelm@240
|
175 |
| blocks_ok (En :: syms) n = blocks_ok syms (n - 1)
|
wenzelm@240
|
176 |
| blocks_ok (_ :: syms) n = blocks_ok syms n;
|
wenzelm@240
|
177 |
|
wenzelm@240
|
178 |
fun check_blocks syms =
|
wenzelm@240
|
179 |
if blocks_ok syms 0 then ()
|
wenzelm@240
|
180 |
else err "unbalanced block parentheses";
|
wenzelm@240
|
181 |
|
wenzelm@240
|
182 |
|
wenzelm@911
|
183 |
local
|
wenzelm@911
|
184 |
fun is_meta c = c mem ["(", ")", "/", "_"];
|
wenzelm@240
|
185 |
|
wenzelm@911
|
186 |
fun scan_delim_char ("'" :: c :: cs) =
|
wenzelm@911
|
187 |
if is_blank c then raise LEXICAL_ERROR else (c, cs)
|
wenzelm@911
|
188 |
| scan_delim_char ["'"] = err "trailing escape character"
|
wenzelm@911
|
189 |
| scan_delim_char (chs as c :: cs) =
|
wenzelm@911
|
190 |
if is_blank c orelse is_meta c then raise LEXICAL_ERROR else (c, cs)
|
wenzelm@911
|
191 |
| scan_delim_char [] = raise LEXICAL_ERROR;
|
wenzelm@240
|
192 |
|
wenzelm@911
|
193 |
val scan_sym =
|
wenzelm@911
|
194 |
$$ "_" >> K (Argument ("", 0)) ||
|
wenzelm@911
|
195 |
$$ "(" -- scan_int >> (Bg o #2) ||
|
wenzelm@911
|
196 |
$$ ")" >> K En ||
|
wenzelm@911
|
197 |
$$ "/" -- $$ "/" >> K (Brk ~1) ||
|
wenzelm@911
|
198 |
$$ "/" -- scan_any is_blank >> (Brk o length o #2) ||
|
wenzelm@911
|
199 |
scan_any1 is_blank >> (Space o implode) ||
|
wenzelm@911
|
200 |
repeat1 scan_delim_char >> (Delim o implode);
|
wenzelm@911
|
201 |
|
wenzelm@911
|
202 |
val scan_symb =
|
wenzelm@911
|
203 |
scan_sym >> Some ||
|
wenzelm@911
|
204 |
$$ "'" -- scan_one is_blank >> K None;
|
wenzelm@911
|
205 |
in
|
wenzelm@911
|
206 |
val scan_symbs = mapfilter I o #1 o repeat scan_symb;
|
wenzelm@911
|
207 |
end;
|
wenzelm@240
|
208 |
|
wenzelm@240
|
209 |
|
wenzelm@240
|
210 |
val cons_fst = apfst o cons;
|
wenzelm@240
|
211 |
|
wenzelm@240
|
212 |
fun add_args [] ty [] = ([], typ_to_nonterm1 ty)
|
wenzelm@240
|
213 |
| add_args [] _ _ = err "too many precedences"
|
wenzelm@240
|
214 |
| add_args (Argument _ :: syms) (Type ("fun", [ty, tys])) [] =
|
wenzelm@240
|
215 |
cons_fst (Argument (typ_to_nonterm ty, 0)) (add_args syms tys [])
|
wenzelm@240
|
216 |
| add_args (Argument _ :: syms) (Type ("fun", [ty, tys])) (p :: ps) =
|
wenzelm@240
|
217 |
cons_fst (Argument (typ_to_nonterm ty, p)) (add_args syms tys ps)
|
wenzelm@240
|
218 |
| add_args (Argument _ :: _) _ _ =
|
wenzelm@240
|
219 |
err "more arguments than in corresponding type"
|
wenzelm@240
|
220 |
| add_args (sym :: syms) ty ps = cons_fst sym (add_args syms ty ps);
|
wenzelm@240
|
221 |
|
wenzelm@240
|
222 |
|
wenzelm@240
|
223 |
fun is_arg (Argument _) = true
|
wenzelm@240
|
224 |
| is_arg _ = false;
|
wenzelm@240
|
225 |
|
wenzelm@240
|
226 |
fun is_term (Delim _) = true
|
wenzelm@240
|
227 |
| is_term (Argument (s, _)) = is_terminal s
|
wenzelm@240
|
228 |
| is_term _ = false;
|
wenzelm@240
|
229 |
|
wenzelm@240
|
230 |
fun rem_pri (Argument (s, _)) = Argument (s, chain_pri)
|
wenzelm@240
|
231 |
| rem_pri sym = sym;
|
wenzelm@240
|
232 |
|
clasohm@764
|
233 |
fun is_delim (Delim _) = true
|
clasohm@764
|
234 |
| is_delim _ = false;
|
clasohm@764
|
235 |
|
clasohm@764
|
236 |
(*replace logical types on rhs by "logic"*)
|
clasohm@764
|
237 |
fun unify_logtypes copy_prod (a as (Argument (s, p))) =
|
clasohm@764
|
238 |
if s mem logtypes then Argument (logic, p)
|
clasohm@764
|
239 |
else a
|
clasohm@764
|
240 |
| unify_logtypes _ a = a;
|
wenzelm@240
|
241 |
|
wenzelm@911
|
242 |
val raw_symbs = scan_symbs (explode sy);
|
wenzelm@240
|
243 |
val (symbs, lhs) = add_args raw_symbs typ pris;
|
clasohm@764
|
244 |
val copy_prod = lhs mem ["prop", "logic"]
|
clasohm@764
|
245 |
andalso const <> ""
|
clasohm@1145
|
246 |
andalso not (null symbs)
|
clasohm@764
|
247 |
andalso not (exists is_delim symbs);
|
clasohm@764
|
248 |
val lhs' = if convert andalso not copy_prod then
|
clasohm@764
|
249 |
(if lhs mem logtypes then logic
|
clasohm@764
|
250 |
else if lhs = "prop" then sprop else lhs)
|
clasohm@764
|
251 |
else lhs;
|
clasohm@764
|
252 |
val symbs' = map (unify_logtypes copy_prod) symbs;
|
clasohm@764
|
253 |
val xprod = XProd (lhs', symbs', const, pri);
|
wenzelm@240
|
254 |
in
|
wenzelm@240
|
255 |
seq check_pri pris;
|
wenzelm@240
|
256 |
check_pri pri;
|
clasohm@764
|
257 |
check_blocks symbs';
|
wenzelm@240
|
258 |
|
clasohm@764
|
259 |
if is_terminal lhs' then err ("illegal lhs: " ^ lhs')
|
wenzelm@240
|
260 |
else if const <> "" then xprod
|
clasohm@764
|
261 |
else if length (filter is_arg symbs') <> 1 then
|
wenzelm@240
|
262 |
err "copy production must have exactly one argument"
|
clasohm@764
|
263 |
else if exists is_term symbs' then xprod
|
clasohm@764
|
264 |
else XProd (lhs', map rem_pri symbs', "", chain_pri)
|
wenzelm@240
|
265 |
end;
|
wenzelm@240
|
266 |
|
wenzelm@240
|
267 |
|
wenzelm@240
|
268 |
(** datatype syn_ext **)
|
wenzelm@240
|
269 |
|
wenzelm@240
|
270 |
datatype syn_ext =
|
wenzelm@240
|
271 |
SynExt of {
|
clasohm@764
|
272 |
logtypes: string list,
|
wenzelm@240
|
273 |
xprods: xprod list,
|
wenzelm@240
|
274 |
consts: string list,
|
paulson@1510
|
275 |
parse_ast_translation: (string * (Ast.ast list -> Ast.ast)) list,
|
paulson@1510
|
276 |
parse_rules: (Ast.ast * Ast.ast) list,
|
wenzelm@240
|
277 |
parse_translation: (string * (term list -> term)) list,
|
wenzelm@240
|
278 |
print_translation: (string * (term list -> term)) list,
|
paulson@1510
|
279 |
print_rules: (Ast.ast * Ast.ast) list,
|
paulson@1510
|
280 |
print_ast_translation: (string * (Ast.ast list -> Ast.ast)) list};
|
wenzelm@240
|
281 |
|
wenzelm@240
|
282 |
|
wenzelm@240
|
283 |
(* syn_ext *)
|
wenzelm@240
|
284 |
|
clasohm@764
|
285 |
fun mk_syn_ext convert logtypes mfixes consts trfuns rules =
|
wenzelm@240
|
286 |
let
|
wenzelm@240
|
287 |
val (parse_ast_translation, parse_translation, print_translation,
|
wenzelm@240
|
288 |
print_ast_translation) = trfuns;
|
wenzelm@240
|
289 |
val (parse_rules, print_rules) = rules;
|
clasohm@764
|
290 |
val logtypes' = logtypes \ "prop";
|
wenzelm@240
|
291 |
|
clasohm@624
|
292 |
val mfix_consts = distinct (map (fn (Mfix (_, _, c, _, _)) => c) mfixes);
|
clasohm@764
|
293 |
val xprods = map (mfix_to_xprod convert logtypes') mfixes;
|
wenzelm@240
|
294 |
in
|
wenzelm@240
|
295 |
SynExt {
|
clasohm@764
|
296 |
logtypes = logtypes',
|
clasohm@624
|
297 |
xprods = xprods,
|
clasohm@368
|
298 |
consts = filter is_xid (consts union mfix_consts),
|
wenzelm@240
|
299 |
parse_ast_translation = parse_ast_translation,
|
wenzelm@240
|
300 |
parse_rules = parse_rules,
|
wenzelm@240
|
301 |
parse_translation = parse_translation,
|
wenzelm@240
|
302 |
print_translation = print_translation,
|
wenzelm@240
|
303 |
print_rules = print_rules,
|
wenzelm@240
|
304 |
print_ast_translation = print_ast_translation}
|
wenzelm@240
|
305 |
end;
|
wenzelm@240
|
306 |
|
clasohm@764
|
307 |
val syn_ext = mk_syn_ext true;
|
wenzelm@240
|
308 |
|
clasohm@764
|
309 |
fun syn_ext_logtypes logtypes =
|
clasohm@764
|
310 |
syn_ext logtypes [] [] ([], [], [], []) ([], []);
|
wenzelm@555
|
311 |
|
clasohm@764
|
312 |
fun syn_ext_const_names logtypes cs =
|
clasohm@764
|
313 |
syn_ext logtypes [] cs ([], [], [], []) ([], []);
|
wenzelm@240
|
314 |
|
clasohm@764
|
315 |
fun syn_ext_rules logtypes rules =
|
clasohm@764
|
316 |
syn_ext logtypes [] [] ([], [], [], []) rules;
|
wenzelm@240
|
317 |
|
clasohm@764
|
318 |
fun syn_ext_trfuns logtypes trfuns =
|
clasohm@764
|
319 |
syn_ext logtypes [] [] trfuns ([], []);
|
wenzelm@240
|
320 |
|
clasohm@764
|
321 |
(* pure_ext *)
|
clasohm@764
|
322 |
|
clasohm@764
|
323 |
val pure_ext = mk_syn_ext false []
|
clasohm@764
|
324 |
[Mfix ("_", spropT --> propT, "", [0], 0),
|
clasohm@764
|
325 |
Mfix ("_", logicT --> anyT, "", [0], 0),
|
clasohm@764
|
326 |
Mfix ("_", spropT --> anyT, "", [0], 0),
|
clasohm@764
|
327 |
Mfix ("'(_')", logicT --> logicT, "", [0], max_pri),
|
clasohm@764
|
328 |
Mfix ("'(_')", spropT --> spropT, "", [0], max_pri),
|
clasohm@764
|
329 |
Mfix ("_::_", [logicT, typeT] ---> logicT, "_constrain", [4, 0], 3),
|
clasohm@764
|
330 |
Mfix ("_::_", [spropT, typeT] ---> spropT, "_constrain", [4, 0], 3)]
|
clasohm@764
|
331 |
[]
|
clasohm@764
|
332 |
([], [], [], [])
|
clasohm@764
|
333 |
([], []);
|
wenzelm@240
|
334 |
|
wenzelm@240
|
335 |
end;
|
paulson@1510
|
336 |
|