18
|
1 |
(* Title: Pure/Syntax/syntax.ML
|
0
|
2 |
ID: $Id$
|
|
3 |
Author: Tobias Nipkow and Markus Wenzel, TU Muenchen
|
18
|
4 |
|
|
5 |
Root of Isabelle's syntax module.
|
|
6 |
|
|
7 |
TODO:
|
|
8 |
extend_tables (requires extend_gram) (roots!)
|
|
9 |
replace add_synrules by extend_tables
|
|
10 |
extend, merge: make roots handling more robust
|
|
11 |
extend: read_typ (incl check) as arg, remove def_sort
|
|
12 |
extend: use extend_tables
|
0
|
13 |
*)
|
|
14 |
|
|
15 |
signature SYNTAX =
|
|
16 |
sig
|
18
|
17 |
include AST0
|
0
|
18 |
include LEXICON0
|
|
19 |
include EXTENSION0
|
|
20 |
include TYPE_EXT0
|
|
21 |
include SEXTENSION1
|
|
22 |
include PRINTER0
|
18
|
23 |
type syntax
|
|
24 |
val type_syn: syntax
|
|
25 |
val extend: syntax * (indexname -> sort) -> string list * string list * sext
|
|
26 |
-> syntax
|
|
27 |
val merge: syntax * syntax -> syntax
|
|
28 |
val print_gram: syntax -> unit
|
|
29 |
val print_trans: syntax -> unit
|
|
30 |
val print_syntax: syntax -> unit
|
|
31 |
val test_read: syntax -> string -> string -> unit
|
|
32 |
val read: syntax -> typ -> string -> term
|
|
33 |
val pretty_term: syntax -> term -> Pretty.T
|
|
34 |
val pretty_typ: syntax -> typ -> Pretty.T
|
|
35 |
val string_of_term: syntax -> term -> string
|
|
36 |
val string_of_typ: syntax -> typ -> string
|
0
|
37 |
end;
|
|
38 |
|
|
39 |
functor SyntaxFun(structure Symtab: SYMTAB and TypeExt: TYPE_EXT
|
|
40 |
and Parser: PARSER and SExtension: SEXTENSION and Printer: PRINTER
|
18
|
41 |
sharing TypeExt.Extension = SExtension.Extension
|
|
42 |
and Parser.XGram = TypeExt.Extension.XGram = Printer.XGram
|
|
43 |
and Parser.XGram.Ast = Parser.ParseTree.Ast)(*: SYNTAX *) = (* FIXME *)
|
0
|
44 |
struct
|
|
45 |
|
|
46 |
structure Extension = TypeExt.Extension;
|
|
47 |
structure XGram = Extension.XGram;
|
|
48 |
structure Lexicon = Parser.ParseTree.Lexicon;
|
18
|
49 |
open Lexicon Parser Parser.ParseTree Extension TypeExt SExtension Printer
|
|
50 |
XGram XGram.Ast;
|
0
|
51 |
|
|
52 |
|
|
53 |
fun lookup tab a = Symtab.lookup (tab, a);
|
|
54 |
|
|
55 |
|
|
56 |
|
|
57 |
(** datatype syntax **)
|
|
58 |
|
|
59 |
datatype tables =
|
18
|
60 |
Tabs of {
|
|
61 |
lexicon: lexicon,
|
|
62 |
roots: string list,
|
|
63 |
gram: gram,
|
|
64 |
consts: string list,
|
0
|
65 |
parse_ast_trtab: (ast list -> ast) Symtab.table,
|
|
66 |
parse_ruletab: (ast * ast) list Symtab.table,
|
|
67 |
parse_trtab: (term list -> term) Symtab.table,
|
|
68 |
print_trtab: (term list -> term) Symtab.table,
|
|
69 |
print_ruletab: (ast * ast) list Symtab.table,
|
18
|
70 |
prtab: prtab};
|
0
|
71 |
|
|
72 |
datatype gramgraph =
|
|
73 |
EmptyGG |
|
|
74 |
ExtGG of gramgraph ref * (ext * synrules) |
|
|
75 |
MergeGG of gramgraph ref * gramgraph ref;
|
|
76 |
|
|
77 |
datatype syntax = Syntax of gramgraph ref * tables;
|
|
78 |
|
|
79 |
|
|
80 |
|
|
81 |
(*** compile syntax ***)
|
|
82 |
|
18
|
83 |
(* translation funs *)
|
|
84 |
|
|
85 |
fun extend_trtab tab trfuns name =
|
|
86 |
Symtab.balance (Symtab.st_of_alist (trfuns, tab)) handle Symtab.DUPLICATE s
|
|
87 |
=> error ("More than one " ^ name ^ " for " ^ quote s);
|
|
88 |
|
|
89 |
val mk_trtab = extend_trtab Symtab.null;
|
|
90 |
|
|
91 |
|
|
92 |
(* translation rules *)
|
|
93 |
|
|
94 |
fun mk_ruletab rules =
|
|
95 |
let
|
|
96 |
fun add_rule (r, tab) =
|
|
97 |
let val a = head_of_rule r;
|
|
98 |
in
|
|
99 |
(case lookup tab a of
|
|
100 |
None => Symtab.update ((a, [r]), tab)
|
|
101 |
| Some rs => Symtab.update ((a, r :: rs), tab))
|
|
102 |
end;
|
|
103 |
in
|
|
104 |
Symtab.balance (foldr add_rule (rules, Symtab.null))
|
|
105 |
end;
|
|
106 |
|
|
107 |
fun extend_ruletab tab rules =
|
|
108 |
mk_ruletab (flat (map #2 (Symtab.alist_of tab)) @ rules);
|
|
109 |
|
|
110 |
|
|
111 |
(* mk_tables *)
|
|
112 |
|
|
113 |
fun mk_tables (XGram xgram) =
|
|
114 |
let
|
|
115 |
val {roots, prods, consts, parse_ast_translation, parse_rules,
|
|
116 |
parse_translation, print_translation, print_rules,
|
|
117 |
print_ast_translation} = xgram;
|
|
118 |
in
|
|
119 |
Tabs {
|
|
120 |
lexicon = mk_lexicon (literals_of prods),
|
|
121 |
roots = roots,
|
|
122 |
gram = mk_gram roots prods,
|
|
123 |
consts = consts,
|
|
124 |
parse_ast_trtab = mk_trtab parse_ast_translation "parse ast translation",
|
|
125 |
parse_ruletab = mk_ruletab parse_rules,
|
|
126 |
parse_trtab = mk_trtab parse_translation "parse translation",
|
|
127 |
print_trtab = mk_trtab print_translation "print translation",
|
|
128 |
print_ruletab = mk_ruletab print_rules,
|
|
129 |
prtab = mk_prtab prods print_ast_translation}
|
|
130 |
end;
|
|
131 |
|
|
132 |
|
|
133 |
(* add_synrules *)
|
|
134 |
|
|
135 |
fun add_synrules (Tabs tabs) (SynRules rules) =
|
|
136 |
let
|
|
137 |
val {lexicon, roots, gram, consts, parse_ast_trtab, parse_ruletab,
|
|
138 |
parse_trtab, print_trtab, print_ruletab, prtab} = tabs;
|
|
139 |
val {parse_rules, print_rules} = rules;
|
|
140 |
in
|
|
141 |
Tabs {
|
|
142 |
lexicon = lexicon, roots = roots, gram = gram, consts = consts,
|
|
143 |
parse_ast_trtab = parse_ast_trtab,
|
|
144 |
parse_ruletab = extend_ruletab parse_ruletab parse_rules,
|
|
145 |
parse_trtab = parse_trtab,
|
|
146 |
print_trtab = print_trtab,
|
|
147 |
print_ruletab = extend_ruletab print_ruletab print_rules,
|
|
148 |
prtab = prtab}
|
|
149 |
end;
|
|
150 |
|
|
151 |
|
0
|
152 |
(* ggr_to_xgram *)
|
|
153 |
|
|
154 |
fun ggr_to_xgram ggr =
|
|
155 |
let
|
|
156 |
fun flatGG ggr (xg, v) =
|
|
157 |
if ggr mem v then (xg, v) else flatGG' ggr (xg, ggr :: v)
|
|
158 |
and flatGG' (ref EmptyGG) xgv = xgv
|
|
159 |
| flatGG' (ref (ExtGG (ggr, ext))) xgv =
|
|
160 |
let
|
18
|
161 |
val (xg', v') = flatGG ggr xgv;
|
0
|
162 |
in
|
18
|
163 |
(extend_xgram xg' ext, v')
|
0
|
164 |
end
|
|
165 |
| flatGG' (ref (MergeGG (ggr1, ggr2))) xgv =
|
|
166 |
flatGG ggr1 (flatGG ggr2 xgv);
|
|
167 |
in
|
18
|
168 |
#1 (flatGG ggr (empty_xgram, []))
|
0
|
169 |
end;
|
|
170 |
|
|
171 |
|
|
172 |
(* make_syntax *)
|
|
173 |
|
18
|
174 |
fun make_syntax ggr = Syntax (ggr, mk_tables (ggr_to_xgram ggr));
|
0
|
175 |
|
|
176 |
|
|
177 |
|
|
178 |
(*** inspect syntax ***)
|
|
179 |
|
|
180 |
fun xgram_of (Syntax (ggr, _)) = ggr_to_xgram ggr;
|
|
181 |
|
|
182 |
fun string_of_big_list name prts =
|
|
183 |
Pretty.string_of (Pretty.blk (2,
|
|
184 |
separate Pretty.fbrk (Pretty.str name :: prts)));
|
|
185 |
|
18
|
186 |
fun string_of_strings name strs =
|
|
187 |
Pretty.string_of (Pretty.blk (2,
|
|
188 |
separate (Pretty.brk 1)
|
|
189 |
(map Pretty.str (name :: map quote (sort_strings strs)))));
|
0
|
190 |
|
18
|
191 |
|
|
192 |
(* print_gram *)
|
0
|
193 |
|
|
194 |
fun prt_gram (XGram {roots, prods, ...}) =
|
|
195 |
let
|
|
196 |
fun pretty_name name = [Pretty.str (name ^ " ="), Pretty.brk 1];
|
|
197 |
|
|
198 |
fun pretty_xsymbs (Terminal s :: xs) =
|
|
199 |
Pretty.str (quote s) :: Pretty.brk 1 :: pretty_xsymbs xs
|
|
200 |
| pretty_xsymbs (Nonterminal (s, p) :: xs) =
|
|
201 |
(if is_terminal s then Pretty.str s
|
|
202 |
else Pretty.str (s ^ "[" ^ string_of_int p ^ "]"))
|
|
203 |
:: Pretty.brk 1 :: pretty_xsymbs xs
|
|
204 |
| pretty_xsymbs (_ :: xs) = pretty_xsymbs xs
|
|
205 |
| pretty_xsymbs [] = [];
|
|
206 |
|
|
207 |
fun pretty_const "" = [Pretty.brk 1]
|
|
208 |
| pretty_const c = [Pretty.str (" => " ^ quote c), Pretty.brk 1];
|
|
209 |
|
|
210 |
fun pretty_pri p = [Pretty.str ("(" ^ string_of_int p ^ ")")];
|
|
211 |
|
|
212 |
fun pretty_prod (Prod (name, xsymbs, const, pri)) =
|
|
213 |
Pretty.blk (2, pretty_name name @ pretty_xsymbs xsymbs @
|
|
214 |
pretty_const const @ pretty_pri pri);
|
|
215 |
in
|
18
|
216 |
writeln (string_of_strings "lexicon:" (literals_of prods));
|
0
|
217 |
writeln (Pretty.string_of (Pretty.blk (2,
|
|
218 |
separate (Pretty.brk 1) (map Pretty.str ("roots:" :: roots)))));
|
|
219 |
writeln (string_of_big_list "prods:" (map pretty_prod prods))
|
|
220 |
end;
|
|
221 |
|
|
222 |
val print_gram = prt_gram o xgram_of;
|
|
223 |
|
|
224 |
|
18
|
225 |
(* print_trans *)
|
0
|
226 |
|
|
227 |
fun prt_trans (XGram xgram) =
|
|
228 |
let
|
|
229 |
fun string_of_trs name trs = string_of_strings name (map fst trs);
|
|
230 |
|
|
231 |
fun string_of_rules name rules =
|
|
232 |
string_of_big_list name (map pretty_rule rules);
|
|
233 |
|
18
|
234 |
val {consts, parse_ast_translation, parse_rules, parse_translation,
|
|
235 |
print_translation, print_rules, print_ast_translation, ...} = xgram;
|
0
|
236 |
in
|
|
237 |
writeln (string_of_strings "consts:" consts);
|
|
238 |
writeln (string_of_trs "parse_ast_translation:" parse_ast_translation);
|
|
239 |
writeln (string_of_rules "parse_rules:" parse_rules);
|
|
240 |
writeln (string_of_trs "parse_translation:" parse_translation);
|
|
241 |
writeln (string_of_trs "print_translation:" print_translation);
|
|
242 |
writeln (string_of_rules "print_rules:" print_rules);
|
|
243 |
writeln (string_of_trs "print_ast_translation:" print_ast_translation)
|
|
244 |
end;
|
|
245 |
|
|
246 |
val print_trans = prt_trans o xgram_of;
|
|
247 |
|
|
248 |
|
|
249 |
(* print_syntax *)
|
|
250 |
|
|
251 |
fun print_syntax syn =
|
|
252 |
let
|
|
253 |
val xgram = xgram_of syn;
|
|
254 |
in
|
|
255 |
prt_gram xgram; prt_trans xgram
|
|
256 |
end;
|
|
257 |
|
|
258 |
|
|
259 |
|
|
260 |
(*** parsing and printing ***)
|
|
261 |
|
18
|
262 |
(* mk_get_rules *)
|
0
|
263 |
|
18
|
264 |
fun mk_get_rules ruletab =
|
0
|
265 |
let
|
|
266 |
fun get_rules a =
|
|
267 |
(case lookup ruletab a of
|
|
268 |
Some rules => rules
|
|
269 |
| None => []);
|
|
270 |
in
|
18
|
271 |
if Symtab.is_null ruletab then None else Some get_rules
|
|
272 |
end;
|
|
273 |
|
|
274 |
|
|
275 |
(* read_ast *)
|
|
276 |
|
|
277 |
fun read_ast (Syntax (_, tabs)) xids root str =
|
|
278 |
let
|
|
279 |
val Tabs {lexicon, gram, parse_ast_trtab, ...} = tabs;
|
|
280 |
in
|
|
281 |
pt_to_ast (lookup parse_ast_trtab)
|
|
282 |
(parse gram root (tokenize lexicon xids str))
|
0
|
283 |
end;
|
|
284 |
|
|
285 |
|
|
286 |
|
18
|
287 |
(** test_read **)
|
|
288 |
|
|
289 |
fun test_read (Syntax (_, tabs)) root str =
|
|
290 |
let
|
|
291 |
val Tabs {lexicon, gram, parse_ast_trtab, parse_ruletab, ...} = tabs;
|
|
292 |
|
|
293 |
val toks = tokenize lexicon false str;
|
|
294 |
val _ = writeln ("tokens: " ^ space_implode " " (map display_token toks));
|
|
295 |
|
|
296 |
val pt = parse gram root toks;
|
|
297 |
val raw_ast = pt_to_ast (K None) pt;
|
|
298 |
val _ = writeln ("raw: " ^ str_of_ast raw_ast);
|
|
299 |
|
|
300 |
val pre_ast = pt_to_ast (lookup parse_ast_trtab) pt;
|
|
301 |
val _ = normalize true true (mk_get_rules parse_ruletab) pre_ast;
|
|
302 |
in () end;
|
|
303 |
|
|
304 |
|
|
305 |
|
0
|
306 |
(** read **)
|
|
307 |
|
18
|
308 |
fun read (syn as Syntax (_, tabs)) ty str =
|
0
|
309 |
let
|
18
|
310 |
val Tabs {parse_ruletab, parse_trtab, ...} = tabs;
|
|
311 |
val ast = read_ast syn false (typ_to_nonterm ty) str;
|
0
|
312 |
in
|
|
313 |
ast_to_term (lookup parse_trtab)
|
18
|
314 |
(normalize_ast (mk_get_rules parse_ruletab) ast)
|
0
|
315 |
end;
|
|
316 |
|
|
317 |
|
|
318 |
|
18
|
319 |
(** read_rule **)
|
|
320 |
|
|
321 |
fun read_rule syn (xrule as ((_, lhs_src), (_, rhs_src))) =
|
|
322 |
let
|
|
323 |
val Syntax (_, Tabs {consts, ...}) = syn;
|
|
324 |
|
|
325 |
fun constantify (ast as Constant _) = ast
|
|
326 |
| constantify (ast as Variable x) =
|
|
327 |
if x mem consts then Constant x else ast
|
|
328 |
| constantify (Appl asts) = Appl (map constantify asts);
|
0
|
329 |
|
18
|
330 |
fun read_pat (root, str) =
|
|
331 |
constantify (read_ast syn true root str)
|
|
332 |
handle ERROR => error ("The error above occurred in " ^ quote str);
|
|
333 |
|
|
334 |
val rule as (lhs, rhs) = (pairself read_pat) xrule;
|
|
335 |
in
|
|
336 |
(case rule_error rule of
|
|
337 |
Some msg =>
|
|
338 |
error ("Error in syntax translation rule: " ^ msg ^
|
|
339 |
"\nexternal: " ^ quote lhs_src ^ " -> " ^ quote rhs_src ^
|
|
340 |
"\ninternal: " ^ str_of_ast lhs ^ " -> " ^ str_of_ast rhs)
|
|
341 |
| None => rule)
|
|
342 |
end;
|
0
|
343 |
|
|
344 |
|
|
345 |
|
18
|
346 |
(** read_xrules **)
|
|
347 |
|
|
348 |
fun read_xrules syn xrules =
|
|
349 |
let
|
|
350 |
fun right_rule (xpat1 |-> xpat2) = Some (xpat1, xpat2)
|
|
351 |
| right_rule (xpat1 <-| xpat2) = None
|
|
352 |
| right_rule (xpat1 <-> xpat2) = Some (xpat1, xpat2);
|
0
|
353 |
|
18
|
354 |
fun left_rule (xpat1 |-> xpat2) = None
|
|
355 |
| left_rule (xpat1 <-| xpat2) = Some (xpat2, xpat1)
|
|
356 |
| left_rule (xpat1 <-> xpat2) = Some (xpat2, xpat1);
|
|
357 |
in
|
|
358 |
(map (read_rule syn) (mapfilter right_rule xrules),
|
|
359 |
map (read_rule syn) (mapfilter left_rule xrules))
|
|
360 |
end;
|
|
361 |
|
|
362 |
|
|
363 |
|
|
364 |
(** pretty terms or typs **)
|
|
365 |
|
|
366 |
fun pretty_t t_to_ast pretty_t (syn as Syntax (_, tabs)) t =
|
0
|
367 |
let
|
18
|
368 |
val Tabs {print_trtab, print_ruletab, prtab, ...} = tabs;
|
0
|
369 |
val ast = t_to_ast (lookup print_trtab) t;
|
|
370 |
in
|
18
|
371 |
pretty_t prtab (normalize_ast (mk_get_rules print_ruletab) ast)
|
0
|
372 |
end;
|
|
373 |
|
|
374 |
val pretty_term = pretty_t term_to_ast pretty_term_ast;
|
|
375 |
|
|
376 |
val pretty_typ = pretty_t typ_to_ast pretty_typ_ast;
|
|
377 |
|
|
378 |
fun string_of_term syn t = Pretty.string_of (pretty_term syn t);
|
|
379 |
|
|
380 |
fun string_of_typ syn ty = Pretty.string_of (pretty_typ syn ty);
|
|
381 |
|
|
382 |
|
|
383 |
|
|
384 |
(*** build syntax ***)
|
|
385 |
|
|
386 |
(* type_syn *)
|
|
387 |
|
|
388 |
val type_syn = make_syntax (ref (ExtGG (ref EmptyGG, (type_ext, empty_synrules))));
|
|
389 |
|
|
390 |
|
18
|
391 |
(** extend **)
|
0
|
392 |
|
|
393 |
fun extend (old_syn as Syntax (ggr, _), def_sort) (roots, xconsts, sext) =
|
|
394 |
let
|
|
395 |
fun read_typ s = typ_of_term def_sort (read old_syn typeT s);
|
|
396 |
val ext = ext_of_sext roots xconsts read_typ sext;
|
|
397 |
|
|
398 |
val (tmp_syn as Syntax (_, tmp_tabs)) =
|
|
399 |
make_syntax (ref (ExtGG (ggr, (ext, empty_synrules))));
|
|
400 |
|
18
|
401 |
val (parse_rules, print_rules) = read_xrules tmp_syn (xrules_of sext);
|
|
402 |
val rules =
|
0
|
403 |
SynRules {
|
18
|
404 |
parse_rules = parse_rules,
|
|
405 |
print_rules = print_rules};
|
0
|
406 |
in
|
18
|
407 |
Syntax (ref (ExtGG (ggr, (ext, rules))), add_synrules tmp_tabs rules)
|
0
|
408 |
end;
|
|
409 |
|
|
410 |
|
|
411 |
(* merge *)
|
|
412 |
|
|
413 |
fun merge (Syntax (ggr1, _), Syntax (ggr2, _)) =
|
|
414 |
make_syntax (ref (MergeGG (ggr1, ggr2)));
|
|
415 |
|
|
416 |
|
|
417 |
end;
|
|
418 |
|