berghofe@11523
|
1 |
(* Title: Pure/General/scan.ML
|
berghofe@11523
|
2 |
ID: $Id$
|
berghofe@11523
|
3 |
Author: Markus Wenzel and Tobias Nipkow, TU Muenchen
|
wenzelm@8806
|
4 |
License: GPL (GNU GENERAL PUBLIC LICENSE)
|
wenzelm@6116
|
5 |
|
wenzelm@6116
|
6 |
Generic scanners (for potentially infinite input).
|
wenzelm@6116
|
7 |
*)
|
wenzelm@6116
|
8 |
|
wenzelm@6116
|
9 |
infix 5 -- :-- |-- --| ^^;
|
wenzelm@6116
|
10 |
infix 3 >>;
|
wenzelm@6116
|
11 |
infix 0 ||;
|
wenzelm@6116
|
12 |
|
wenzelm@6116
|
13 |
signature BASIC_SCAN =
|
wenzelm@6116
|
14 |
sig
|
wenzelm@6116
|
15 |
val !! : ('a * string option -> string) -> ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
16 |
val >> : ('a -> 'b * 'c) * ('b -> 'd) -> 'a -> 'd * 'c
|
wenzelm@6116
|
17 |
val || : ('a -> 'b) * ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
18 |
val -- : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> ('b * 'd) * 'e
|
wenzelm@6116
|
19 |
val :-- : ('a -> 'b * 'c) * ('b -> 'c -> 'd * 'e) -> 'a -> ('b * 'd) * 'e
|
wenzelm@6116
|
20 |
val |-- : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> 'd * 'e
|
wenzelm@6116
|
21 |
val --| : ('a -> 'b * 'c) * ('c -> 'd * 'e) -> 'a -> 'b * 'e
|
wenzelm@6116
|
22 |
val ^^ : ('a -> string * 'b) * ('b -> string * 'c) -> 'a -> string * 'c
|
wenzelm@6116
|
23 |
val $$ : ''a -> ''a list -> ''a * ''a list
|
wenzelm@6116
|
24 |
end;
|
wenzelm@6116
|
25 |
|
wenzelm@6116
|
26 |
signature SCAN =
|
wenzelm@6116
|
27 |
sig
|
wenzelm@6116
|
28 |
include BASIC_SCAN
|
wenzelm@6116
|
29 |
val fail: 'a -> 'b
|
wenzelm@6116
|
30 |
val fail_with: ('a -> string) -> 'a -> 'b
|
wenzelm@6116
|
31 |
val succeed: 'a -> 'b -> 'a * 'b
|
wenzelm@6116
|
32 |
val one: ('a -> bool) -> 'a list -> 'a * 'a list
|
wenzelm@6116
|
33 |
val any: ('a -> bool) -> 'a list -> 'a list * 'a list
|
wenzelm@6116
|
34 |
val any1: ('a -> bool) -> 'a list -> 'a list * 'a list
|
wenzelm@6116
|
35 |
val optional: ('a -> 'b * 'a) -> 'b -> 'a -> 'b * 'a
|
wenzelm@6116
|
36 |
val option: ('a -> 'b * 'a) -> 'a -> 'b option * 'a
|
wenzelm@6116
|
37 |
val repeat: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
wenzelm@6116
|
38 |
val repeat1: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
wenzelm@6116
|
39 |
val max: ('a * 'a -> bool) -> ('b -> 'a * 'b) -> ('b -> 'a * 'b) -> 'b -> 'a * 'b
|
wenzelm@6116
|
40 |
val ahead: ('a -> 'b * 'c) -> 'a -> 'b * 'a
|
wenzelm@6116
|
41 |
val unless: ('a -> 'b * 'a) -> ('a -> 'c * 'd) -> 'a -> 'c * 'd
|
wenzelm@6116
|
42 |
val first: ('a -> 'b) list -> 'a -> 'b
|
wenzelm@9122
|
43 |
val state: 'a * 'b -> 'a * ('a * 'b)
|
wenzelm@6116
|
44 |
val depend: ('a -> 'b -> ('c * 'd) * 'e) -> 'a * 'b -> 'd * ('c * 'e)
|
wenzelm@6116
|
45 |
val lift: ('a -> 'b * 'c) -> 'd * 'a -> 'b * ('d * 'c)
|
wenzelm@6116
|
46 |
val pass: 'a -> ('a * 'b -> 'c * ('d * 'e)) -> 'b -> 'c * 'e
|
wenzelm@6116
|
47 |
val try: ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
48 |
val force: ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
49 |
val prompt: string -> ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
50 |
val finite': 'a * ('a -> bool) -> ('b * 'a list -> 'c * ('d * 'a list))
|
wenzelm@6116
|
51 |
-> 'b * 'a list -> 'c * ('d * 'a list)
|
wenzelm@6116
|
52 |
val finite: 'a * ('a -> bool) -> ('a list -> 'b * 'a list) -> 'a list -> 'b * 'a list
|
wenzelm@6116
|
53 |
val read: 'a * ('a -> bool) -> ('a list -> 'b * 'a list) -> 'a list -> 'b option
|
wenzelm@6116
|
54 |
val catch: ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
55 |
val error: ('a -> 'b) -> 'a -> 'b
|
wenzelm@6116
|
56 |
val source': string -> (string -> 'a -> 'b list * 'a) -> ('b list * 'a -> 'c) ->
|
wenzelm@6116
|
57 |
'b * ('b -> bool) -> ('d * 'b list -> 'e list * ('d * 'b list)) ->
|
wenzelm@10746
|
58 |
('d * 'b list -> 'e list * ('d * 'b list)) option -> 'd * 'a -> 'e list * ('d * 'c)
|
wenzelm@6116
|
59 |
val source: string -> (string -> 'a -> 'b list * 'a) -> ('b list * 'a -> 'c) ->
|
wenzelm@6116
|
60 |
'b * ('b -> bool) -> ('b list -> 'd list * 'b list) ->
|
wenzelm@10746
|
61 |
('b list -> 'd list * 'b list) option -> 'a -> 'd list * 'c
|
wenzelm@6116
|
62 |
val single: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
wenzelm@6116
|
63 |
val bulk: ('a -> 'b * 'a) -> 'a -> 'b list * 'a
|
wenzelm@6116
|
64 |
type lexicon
|
wenzelm@7025
|
65 |
val dest_lexicon: lexicon -> string list
|
wenzelm@6116
|
66 |
val make_lexicon: string list list -> lexicon
|
wenzelm@6116
|
67 |
val empty_lexicon: lexicon
|
wenzelm@6116
|
68 |
val extend_lexicon: lexicon -> string list list -> lexicon
|
wenzelm@6116
|
69 |
val merge_lexicons: lexicon -> lexicon -> lexicon
|
wenzelm@6116
|
70 |
val literal: lexicon -> string list -> string list * string list
|
wenzelm@6116
|
71 |
end;
|
wenzelm@6116
|
72 |
|
wenzelm@6116
|
73 |
structure Scan: SCAN =
|
wenzelm@6116
|
74 |
struct
|
wenzelm@6116
|
75 |
|
wenzelm@6116
|
76 |
|
wenzelm@6116
|
77 |
(** scanners **)
|
wenzelm@6116
|
78 |
|
berghofe@11523
|
79 |
exception MORE of string option; (*need more input (prompt)*)
|
berghofe@11523
|
80 |
exception FAIL of string option; (*try alternatives (reason of failure)*)
|
berghofe@11523
|
81 |
exception ABORT of string; (*dead end*)
|
wenzelm@6116
|
82 |
|
wenzelm@6116
|
83 |
|
wenzelm@6116
|
84 |
(* scanner combinators *)
|
wenzelm@6116
|
85 |
|
wenzelm@6116
|
86 |
fun (scan >> f) xs = apfst f (scan xs);
|
wenzelm@6116
|
87 |
|
wenzelm@6116
|
88 |
fun (scan1 || scan2) xs = scan1 xs handle FAIL _ => scan2 xs;
|
wenzelm@6116
|
89 |
|
wenzelm@6116
|
90 |
(*dependent pairing*)
|
wenzelm@6116
|
91 |
fun (scan1 :-- scan2) xs =
|
wenzelm@6116
|
92 |
let
|
wenzelm@6116
|
93 |
val (x, ys) = scan1 xs;
|
wenzelm@6116
|
94 |
val (y, zs) = scan2 x ys;
|
wenzelm@6116
|
95 |
in ((x, y), zs) end;
|
wenzelm@6116
|
96 |
|
wenzelm@6116
|
97 |
fun (scan1 -- scan2) = scan1 :-- (fn _ => scan2);
|
wenzelm@6116
|
98 |
fun (scan1 |-- scan2) = scan1 -- scan2 >> #2;
|
wenzelm@6116
|
99 |
fun (scan1 --| scan2) = scan1 -- scan2 >> #1;
|
wenzelm@6116
|
100 |
fun (scan1 ^^ scan2) = scan1 -- scan2 >> op ^;
|
wenzelm@6116
|
101 |
|
wenzelm@6116
|
102 |
|
wenzelm@6116
|
103 |
(* generic scanners *)
|
wenzelm@6116
|
104 |
|
wenzelm@6116
|
105 |
fun fail _ = raise FAIL None;
|
wenzelm@6116
|
106 |
fun fail_with msg_of xs = raise FAIL (Some (msg_of xs));
|
wenzelm@6116
|
107 |
fun succeed y xs = (y, xs);
|
wenzelm@6116
|
108 |
|
wenzelm@6116
|
109 |
fun one _ [] = raise MORE None
|
wenzelm@6116
|
110 |
| one pred (x :: xs) =
|
wenzelm@6116
|
111 |
if pred x then (x, xs) else raise FAIL None;
|
wenzelm@6116
|
112 |
|
wenzelm@6116
|
113 |
fun $$ _ [] = raise MORE None
|
wenzelm@6116
|
114 |
| $$ a (x :: xs) =
|
wenzelm@6116
|
115 |
if a = x then (x, xs) else raise FAIL None;
|
wenzelm@6116
|
116 |
|
wenzelm@6116
|
117 |
fun any _ [] = raise MORE None
|
wenzelm@6116
|
118 |
| any pred (lst as x :: xs) =
|
wenzelm@6116
|
119 |
if pred x then apfst (cons x) (any pred xs)
|
wenzelm@6116
|
120 |
else ([], lst);
|
wenzelm@6116
|
121 |
|
wenzelm@6116
|
122 |
fun any1 pred = one pred -- any pred >> op ::;
|
wenzelm@6116
|
123 |
|
wenzelm@6116
|
124 |
fun optional scan def = scan || succeed def;
|
wenzelm@6116
|
125 |
fun option scan = optional (scan >> Some) None;
|
wenzelm@6116
|
126 |
|
wenzelm@6116
|
127 |
fun repeat scan xs = (scan -- repeat scan >> op :: || succeed []) xs;
|
wenzelm@6116
|
128 |
fun repeat1 scan = scan -- repeat scan >> op ::;
|
wenzelm@6116
|
129 |
|
wenzelm@6116
|
130 |
fun max leq scan1 scan2 xs =
|
wenzelm@6116
|
131 |
(case (option scan1 xs, option scan2 xs) of
|
berghofe@11523
|
132 |
((None, _), (None, _)) => raise FAIL None (*looses FAIL msg!*)
|
wenzelm@6116
|
133 |
| ((Some tok1, xs'), (None, _)) => (tok1, xs')
|
wenzelm@6116
|
134 |
| ((None, _), (Some tok2, xs')) => (tok2, xs')
|
wenzelm@6116
|
135 |
| ((Some tok1, xs1'), (Some tok2, xs2')) =>
|
wenzelm@6116
|
136 |
if leq (tok2, tok1) then (tok1, xs1') else (tok2, xs2'));
|
wenzelm@6116
|
137 |
|
wenzelm@6116
|
138 |
fun ahead scan xs = (fst (scan xs), xs);
|
wenzelm@6116
|
139 |
|
wenzelm@6116
|
140 |
fun unless test scan =
|
wenzelm@6116
|
141 |
ahead (option test) :-- (fn None => scan | _ => fail) >> #2;
|
wenzelm@6116
|
142 |
|
wenzelm@6116
|
143 |
fun first [] = fail
|
wenzelm@6116
|
144 |
| first (scan :: scans) = scan || first scans;
|
wenzelm@6116
|
145 |
|
wenzelm@6116
|
146 |
|
wenzelm@6116
|
147 |
(* state based scanners *)
|
wenzelm@6116
|
148 |
|
wenzelm@9122
|
149 |
fun state (st, xs) = (st, (st, xs));
|
wenzelm@9122
|
150 |
|
wenzelm@6116
|
151 |
fun depend scan (st, xs) =
|
wenzelm@6116
|
152 |
let val ((st', y), xs') = scan st xs
|
wenzelm@6116
|
153 |
in (y, (st', xs')) end;
|
wenzelm@6116
|
154 |
|
wenzelm@6116
|
155 |
fun lift scan (st, xs) =
|
wenzelm@6116
|
156 |
let val (y, xs') = scan xs
|
wenzelm@6116
|
157 |
in (y, (st, xs')) end;
|
wenzelm@6116
|
158 |
|
wenzelm@6116
|
159 |
fun pass st scan xs =
|
wenzelm@6116
|
160 |
let val (y, (_, xs')) = scan (st, xs)
|
wenzelm@6116
|
161 |
in (y, xs') end;
|
wenzelm@6116
|
162 |
|
wenzelm@6116
|
163 |
|
wenzelm@6116
|
164 |
(* exception handling *)
|
wenzelm@6116
|
165 |
|
wenzelm@6116
|
166 |
fun !! err scan xs = scan xs handle FAIL msg => raise ABORT (err (xs, msg));
|
wenzelm@6116
|
167 |
fun try scan xs = scan xs handle MORE _ => raise FAIL None | ABORT _ => raise FAIL None;
|
wenzelm@6116
|
168 |
fun force scan xs = scan xs handle MORE _ => raise FAIL None;
|
wenzelm@6116
|
169 |
fun prompt str scan xs = scan xs handle MORE None => raise MORE (Some str);
|
wenzelm@6116
|
170 |
fun catch scan xs = scan xs handle ABORT msg => raise FAIL (Some msg);
|
wenzelm@6116
|
171 |
fun error scan xs = scan xs handle ABORT msg => Library.error msg;
|
wenzelm@6116
|
172 |
|
wenzelm@6116
|
173 |
|
wenzelm@6116
|
174 |
(* finite scans *)
|
wenzelm@6116
|
175 |
|
wenzelm@6116
|
176 |
fun finite' (stopper, is_stopper) scan (state, input) =
|
wenzelm@6116
|
177 |
let
|
wenzelm@6116
|
178 |
fun lost () = raise ABORT "Scanner bug: lost stopper of finite scan!";
|
wenzelm@6116
|
179 |
|
wenzelm@6116
|
180 |
fun stop [] = lost ()
|
wenzelm@6116
|
181 |
| stop lst =
|
wenzelm@6116
|
182 |
let val (xs, x) = split_last lst
|
wenzelm@6116
|
183 |
in if is_stopper x then ((), xs) else lost () end;
|
wenzelm@6116
|
184 |
in
|
wenzelm@6116
|
185 |
if exists is_stopper input then
|
wenzelm@6116
|
186 |
raise ABORT "Stopper may not occur in input of finite scan!"
|
wenzelm@6116
|
187 |
else (force scan --| lift stop) (state, input @ [stopper])
|
wenzelm@6116
|
188 |
end;
|
wenzelm@6116
|
189 |
|
wenzelm@6116
|
190 |
fun finite stopper scan xs =
|
wenzelm@6116
|
191 |
let val (y, ((), xs')) = finite' stopper (lift scan) ((), xs)
|
wenzelm@6116
|
192 |
in (y, xs') end;
|
wenzelm@6116
|
193 |
|
wenzelm@6116
|
194 |
fun read stopper scan xs =
|
wenzelm@6116
|
195 |
(case error (finite stopper (option scan)) xs of
|
wenzelm@6116
|
196 |
(y as Some _, []) => y
|
wenzelm@6116
|
197 |
| _ => None);
|
wenzelm@6116
|
198 |
|
wenzelm@6116
|
199 |
|
wenzelm@6116
|
200 |
(* infinite scans -- draining state-based source *)
|
wenzelm@6116
|
201 |
|
wenzelm@6116
|
202 |
fun drain def_prmpt get stopper scan ((state, xs), src) =
|
wenzelm@6116
|
203 |
(scan (state, xs), src) handle MORE prmpt =>
|
wenzelm@6116
|
204 |
(case get (if_none prmpt def_prmpt) src of
|
wenzelm@6116
|
205 |
([], _) => (finite' stopper scan (state, xs), src)
|
wenzelm@6116
|
206 |
| (xs', src') => drain def_prmpt get stopper scan ((state, xs @ xs'), src'));
|
wenzelm@6116
|
207 |
|
wenzelm@6116
|
208 |
fun source' def_prmpt get unget stopper scanner opt_recover (state, src) =
|
wenzelm@6116
|
209 |
let
|
wenzelm@10746
|
210 |
val drain_with = drain def_prmpt get stopper;
|
wenzelm@6116
|
211 |
|
wenzelm@6116
|
212 |
fun drain_loop recover inp =
|
wenzelm@6116
|
213 |
drain_with (catch scanner) inp handle FAIL msg =>
|
wenzelm@10746
|
214 |
(error_msg (if_none msg "Syntax error."); drain_with recover inp);
|
wenzelm@6116
|
215 |
|
wenzelm@6116
|
216 |
val ((ys, (state', xs')), src') =
|
wenzelm@6116
|
217 |
(case (get def_prmpt src, opt_recover) of
|
wenzelm@6116
|
218 |
(([], s), _) => (([], (state, [])), s)
|
wenzelm@6116
|
219 |
| ((xs, s), None) => drain_with (error scanner) ((state, xs), s)
|
wenzelm@8653
|
220 |
| ((xs, s), Some r) => drain_loop (unless (lift (one (#2 stopper))) r) ((state, xs), s));
|
wenzelm@8653
|
221 |
in (ys, (state', unget (xs', src'))) end;
|
wenzelm@6116
|
222 |
|
wenzelm@6116
|
223 |
fun source def_prmpt get unget stopper scan opt_recover src =
|
wenzelm@6116
|
224 |
let val (ys, ((), src')) =
|
wenzelm@6116
|
225 |
source' def_prmpt get unget stopper (lift scan) (apsome lift opt_recover) ((), src)
|
wenzelm@6116
|
226 |
in (ys, src') end;
|
wenzelm@6116
|
227 |
|
wenzelm@6116
|
228 |
fun single scan = scan >> (fn x => [x]);
|
wenzelm@6116
|
229 |
fun bulk scan = scan -- repeat (try scan) >> (op ::);
|
wenzelm@6116
|
230 |
|
wenzelm@6116
|
231 |
|
wenzelm@6116
|
232 |
|
wenzelm@6116
|
233 |
(** datatype lexicon **)
|
wenzelm@6116
|
234 |
|
wenzelm@6116
|
235 |
datatype lexicon =
|
wenzelm@6116
|
236 |
Empty |
|
wenzelm@6116
|
237 |
Branch of string * string list * lexicon * lexicon * lexicon;
|
wenzelm@6116
|
238 |
|
wenzelm@6116
|
239 |
val no_literal = [];
|
wenzelm@6116
|
240 |
|
wenzelm@6116
|
241 |
|
wenzelm@6116
|
242 |
(* dest_lexicon *)
|
wenzelm@6116
|
243 |
|
wenzelm@7025
|
244 |
fun dest_lex Empty = []
|
wenzelm@7025
|
245 |
| dest_lex (Branch (_, [], lt, eq, gt)) =
|
wenzelm@7025
|
246 |
dest_lex lt @ dest_lex eq @ dest_lex gt
|
wenzelm@7025
|
247 |
| dest_lex (Branch (_, cs, lt, eq, gt)) =
|
wenzelm@7025
|
248 |
dest_lex lt @ [cs] @ dest_lex eq @ dest_lex gt;
|
wenzelm@7025
|
249 |
|
wenzelm@7025
|
250 |
val dest_lexicon = map implode o dest_lex;
|
wenzelm@6116
|
251 |
|
wenzelm@6116
|
252 |
|
wenzelm@6116
|
253 |
(* empty, extend, make, merge lexicons *)
|
wenzelm@6116
|
254 |
|
wenzelm@6116
|
255 |
val empty_lexicon = Empty;
|
wenzelm@6116
|
256 |
|
berghofe@11523
|
257 |
fun extend_lexicon lexicon [] = lexicon
|
berghofe@11523
|
258 |
| extend_lexicon lexicon chrss =
|
wenzelm@6116
|
259 |
let
|
berghofe@11523
|
260 |
fun ext (lex, chrs) =
|
berghofe@11523
|
261 |
let
|
berghofe@11523
|
262 |
fun add (Branch (d, a, lt, eq, gt)) (chs as c :: cs) =
|
berghofe@11523
|
263 |
if c < d then Branch (d, a, add lt chs, eq, gt)
|
berghofe@11523
|
264 |
else if c > d then Branch (d, a, lt, eq, add gt chs)
|
berghofe@11523
|
265 |
else Branch (d, if null cs then chrs else a, lt, add eq cs, gt)
|
berghofe@11523
|
266 |
| add Empty [c] =
|
berghofe@11523
|
267 |
Branch (c, chrs, Empty, Empty, Empty)
|
berghofe@11523
|
268 |
| add Empty (c :: cs) =
|
berghofe@11523
|
269 |
Branch (c, no_literal, Empty, add Empty cs, Empty)
|
berghofe@11523
|
270 |
| add lex [] = lex;
|
berghofe@11523
|
271 |
in add lex chrs end;
|
berghofe@11523
|
272 |
in foldl ext (lexicon, chrss \\ dest_lex lexicon) end;
|
wenzelm@6116
|
273 |
|
wenzelm@6116
|
274 |
val make_lexicon = extend_lexicon empty_lexicon;
|
wenzelm@6116
|
275 |
|
wenzelm@6116
|
276 |
fun merge_lexicons lex1 lex2 =
|
wenzelm@6116
|
277 |
let
|
wenzelm@7025
|
278 |
val chss1 = dest_lex lex1;
|
wenzelm@7025
|
279 |
val chss2 = dest_lex lex2;
|
wenzelm@6116
|
280 |
in
|
wenzelm@6116
|
281 |
if chss2 subset chss1 then lex1
|
wenzelm@6116
|
282 |
else if chss1 subset chss2 then lex2
|
wenzelm@6116
|
283 |
else extend_lexicon lex1 chss2
|
wenzelm@6116
|
284 |
end;
|
wenzelm@6116
|
285 |
|
wenzelm@6116
|
286 |
|
wenzelm@6116
|
287 |
(* scan literal *)
|
wenzelm@6116
|
288 |
|
wenzelm@6116
|
289 |
fun literal lex chrs =
|
wenzelm@6116
|
290 |
let
|
wenzelm@6116
|
291 |
fun lit Empty res _ = res
|
wenzelm@6116
|
292 |
| lit (Branch _) _ [] = raise MORE None
|
wenzelm@6116
|
293 |
| lit (Branch (d, a, lt, eq, gt)) res (chs as c :: cs) =
|
berghofe@11523
|
294 |
if c < d then lit lt res chs
|
berghofe@11523
|
295 |
else if c > d then lit gt res chs
|
berghofe@11523
|
296 |
else lit eq (if a = no_literal then res else Some (a, cs)) cs;
|
wenzelm@6116
|
297 |
in
|
wenzelm@6116
|
298 |
(case lit lex None chrs of
|
wenzelm@6116
|
299 |
None => raise FAIL None
|
wenzelm@6116
|
300 |
| Some res => res)
|
wenzelm@6116
|
301 |
end;
|
wenzelm@6116
|
302 |
|
wenzelm@6116
|
303 |
|
wenzelm@6116
|
304 |
end;
|
wenzelm@6116
|
305 |
|
wenzelm@6116
|
306 |
|
wenzelm@6116
|
307 |
structure BasicScan: BASIC_SCAN = Scan;
|
wenzelm@6116
|
308 |
open BasicScan;
|