wenzelm@41
|
1 |
(* Title: Pure/library.ML
|
clasohm@0
|
2 |
ID: $Id$
|
wenzelm@233
|
3 |
Author: Lawrence C Paulson, Cambridge University Computer Laboratory
|
clasohm@0
|
4 |
Copyright 1992 University of Cambridge
|
clasohm@0
|
5 |
|
wenzelm@233
|
6 |
Basic library: functions, options, pairs, booleans, lists, integers,
|
wenzelm@4212
|
7 |
strings, lists as sets, association lists, generic tables, balanced
|
wenzelm@4212
|
8 |
trees, orders, diagnostics, timing, misc functions.
|
clasohm@0
|
9 |
*)
|
clasohm@0
|
10 |
|
wenzelm@4212
|
11 |
infix |> ~~ \ \\ ins ins_string ins_int orf andf prefix upto downto
|
wenzelm@4212
|
12 |
mem mem_int mem_string union union_int union_string inter inter_int
|
wenzelm@4212
|
13 |
inter_string subset subset_int subset_string;
|
clasohm@1364
|
14 |
|
clasohm@1364
|
15 |
|
clasohm@1364
|
16 |
structure Library =
|
clasohm@1364
|
17 |
struct
|
clasohm@0
|
18 |
|
wenzelm@233
|
19 |
(** functions **)
|
clasohm@0
|
20 |
|
wenzelm@233
|
21 |
(*handy combinators*)
|
wenzelm@233
|
22 |
fun curry f x y = f (x, y);
|
wenzelm@233
|
23 |
fun uncurry f (x, y) = f x y;
|
wenzelm@233
|
24 |
fun I x = x;
|
wenzelm@233
|
25 |
fun K x y = x;
|
clasohm@0
|
26 |
|
wenzelm@380
|
27 |
(*reverse apply*)
|
wenzelm@410
|
28 |
fun (x |> f) = f x;
|
wenzelm@380
|
29 |
|
wenzelm@233
|
30 |
(*application of (infix) operator to its left or right argument*)
|
wenzelm@233
|
31 |
fun apl (x, f) y = f (x, y);
|
wenzelm@233
|
32 |
fun apr (f, y) x = f (x, y);
|
clasohm@0
|
33 |
|
wenzelm@233
|
34 |
(*function exponentiation: f(...(f x)...) with n applications of f*)
|
wenzelm@233
|
35 |
fun funpow n f x =
|
wenzelm@233
|
36 |
let fun rep (0, x) = x
|
wenzelm@233
|
37 |
| rep (n, x) = rep (n - 1, f x)
|
wenzelm@233
|
38 |
in rep (n, x) end;
|
wenzelm@160
|
39 |
|
wenzelm@160
|
40 |
|
wenzelm@160
|
41 |
|
wenzelm@2471
|
42 |
(** stamps **)
|
wenzelm@2471
|
43 |
|
wenzelm@2471
|
44 |
type stamp = unit ref;
|
wenzelm@2471
|
45 |
val stamp: unit -> stamp = ref;
|
wenzelm@2471
|
46 |
|
wenzelm@2471
|
47 |
|
wenzelm@2471
|
48 |
|
wenzelm@233
|
49 |
(** options **)
|
clasohm@0
|
50 |
|
clasohm@0
|
51 |
datatype 'a option = None | Some of 'a;
|
clasohm@0
|
52 |
|
wenzelm@4139
|
53 |
exception OPTION;
|
clasohm@0
|
54 |
|
clasohm@0
|
55 |
fun the (Some x) = x
|
wenzelm@4139
|
56 |
| the None = raise OPTION;
|
clasohm@0
|
57 |
|
wenzelm@4212
|
58 |
(*strict!*)
|
wenzelm@255
|
59 |
fun if_none None y = y
|
wenzelm@255
|
60 |
| if_none (Some x) _ = x;
|
wenzelm@255
|
61 |
|
clasohm@0
|
62 |
fun is_some (Some _) = true
|
clasohm@0
|
63 |
| is_some None = false;
|
clasohm@0
|
64 |
|
clasohm@0
|
65 |
fun is_none (Some _) = false
|
clasohm@0
|
66 |
| is_none None = true;
|
clasohm@0
|
67 |
|
wenzelm@233
|
68 |
fun apsome f (Some x) = Some (f x)
|
wenzelm@233
|
69 |
| apsome _ None = None;
|
clasohm@0
|
70 |
|
wenzelm@4139
|
71 |
(*handle partial functions*)
|
wenzelm@4181
|
72 |
fun can f x = (f x; true) handle _ => false;
|
wenzelm@4139
|
73 |
fun try f x = Some (f x) handle _ => None;
|
wenzelm@4139
|
74 |
|
wenzelm@4139
|
75 |
|
wenzelm@4139
|
76 |
|
wenzelm@233
|
77 |
(** pairs **)
|
wenzelm@233
|
78 |
|
wenzelm@233
|
79 |
fun pair x y = (x, y);
|
wenzelm@233
|
80 |
fun rpair x y = (y, x);
|
wenzelm@233
|
81 |
|
wenzelm@233
|
82 |
fun fst (x, y) = x;
|
wenzelm@233
|
83 |
fun snd (x, y) = y;
|
wenzelm@233
|
84 |
|
wenzelm@233
|
85 |
fun eq_fst ((x1, _), (x2, _)) = x1 = x2;
|
wenzelm@233
|
86 |
fun eq_snd ((_, y1), (_, y2)) = y1 = y2;
|
wenzelm@233
|
87 |
|
wenzelm@233
|
88 |
fun swap (x, y) = (y, x);
|
wenzelm@233
|
89 |
|
wenzelm@4212
|
90 |
(*apply function to components*)
|
wenzelm@233
|
91 |
fun apfst f (x, y) = (f x, y);
|
wenzelm@233
|
92 |
fun apsnd f (x, y) = (x, f y);
|
wenzelm@4212
|
93 |
fun pairself f (x, y) = (f x, f y);
|
wenzelm@233
|
94 |
|
wenzelm@233
|
95 |
|
wenzelm@233
|
96 |
|
wenzelm@233
|
97 |
(** booleans **)
|
wenzelm@233
|
98 |
|
wenzelm@233
|
99 |
(* equality *)
|
wenzelm@233
|
100 |
|
wenzelm@233
|
101 |
fun equal x y = x = y;
|
wenzelm@233
|
102 |
fun not_equal x y = x <> y;
|
wenzelm@233
|
103 |
|
wenzelm@233
|
104 |
|
wenzelm@233
|
105 |
(* operators for combining predicates *)
|
wenzelm@233
|
106 |
|
paulson@2175
|
107 |
fun (p orf q) = fn x => p x orelse q x;
|
paulson@2175
|
108 |
fun (p andf q) = fn x => p x andalso q x;
|
wenzelm@233
|
109 |
|
wenzelm@233
|
110 |
|
wenzelm@233
|
111 |
(* predicates on lists *)
|
wenzelm@233
|
112 |
|
wenzelm@233
|
113 |
(*exists pred [x1, ..., xn] ===> pred x1 orelse ... orelse pred xn*)
|
wenzelm@233
|
114 |
fun exists (pred: 'a -> bool) : 'a list -> bool =
|
wenzelm@233
|
115 |
let fun boolf [] = false
|
wenzelm@233
|
116 |
| boolf (x :: xs) = pred x orelse boolf xs
|
wenzelm@233
|
117 |
in boolf end;
|
wenzelm@233
|
118 |
|
wenzelm@233
|
119 |
(*forall pred [x1, ..., xn] ===> pred x1 andalso ... andalso pred xn*)
|
wenzelm@233
|
120 |
fun forall (pred: 'a -> bool) : 'a list -> bool =
|
wenzelm@233
|
121 |
let fun boolf [] = true
|
wenzelm@233
|
122 |
| boolf (x :: xs) = pred x andalso boolf xs
|
wenzelm@233
|
123 |
in boolf end;
|
clasohm@0
|
124 |
|
wenzelm@233
|
125 |
|
wenzelm@380
|
126 |
(* flags *)
|
wenzelm@380
|
127 |
|
wenzelm@380
|
128 |
fun set flag = (flag := true; true);
|
wenzelm@380
|
129 |
fun reset flag = (flag := false; false);
|
wenzelm@380
|
130 |
fun toggle flag = (flag := not (! flag); ! flag);
|
wenzelm@380
|
131 |
|
wenzelm@4212
|
132 |
(*temporarily set flag, handling errors*)
|
wenzelm@2978
|
133 |
fun setmp flag value f x =
|
wenzelm@2958
|
134 |
let
|
wenzelm@2958
|
135 |
val orig_value = ! flag;
|
wenzelm@2958
|
136 |
fun return y = (flag := orig_value; y);
|
wenzelm@2958
|
137 |
in
|
wenzelm@2958
|
138 |
flag := value;
|
wenzelm@2958
|
139 |
return (f x handle exn => (return (); raise exn))
|
wenzelm@2958
|
140 |
end;
|
wenzelm@2958
|
141 |
|
wenzelm@380
|
142 |
|
wenzelm@233
|
143 |
|
wenzelm@233
|
144 |
(** lists **)
|
wenzelm@233
|
145 |
|
wenzelm@233
|
146 |
exception LIST of string;
|
wenzelm@233
|
147 |
|
wenzelm@233
|
148 |
fun null [] = true
|
wenzelm@233
|
149 |
| null (_ :: _) = false;
|
wenzelm@233
|
150 |
|
wenzelm@233
|
151 |
fun hd [] = raise LIST "hd"
|
wenzelm@233
|
152 |
| hd (x :: _) = x;
|
wenzelm@233
|
153 |
|
wenzelm@233
|
154 |
fun tl [] = raise LIST "tl"
|
wenzelm@233
|
155 |
| tl (_ :: xs) = xs;
|
wenzelm@233
|
156 |
|
wenzelm@233
|
157 |
fun cons x xs = x :: xs;
|
wenzelm@233
|
158 |
|
wenzelm@233
|
159 |
|
wenzelm@233
|
160 |
(* fold *)
|
wenzelm@233
|
161 |
|
wenzelm@233
|
162 |
(*the following versions of fold are designed to fit nicely with infixes*)
|
clasohm@0
|
163 |
|
wenzelm@233
|
164 |
(* (op @) (e, [x1, ..., xn]) ===> ((e @ x1) @ x2) ... @ xn
|
wenzelm@233
|
165 |
for operators that associate to the left (TAIL RECURSIVE)*)
|
wenzelm@233
|
166 |
fun foldl (f: 'a * 'b -> 'a) : 'a * 'b list -> 'a =
|
wenzelm@233
|
167 |
let fun itl (e, []) = e
|
wenzelm@233
|
168 |
| itl (e, a::l) = itl (f(e, a), l)
|
wenzelm@233
|
169 |
in itl end;
|
wenzelm@233
|
170 |
|
wenzelm@233
|
171 |
(* (op @) ([x1, ..., xn], e) ===> x1 @ (x2 ... @ (xn @ e))
|
wenzelm@233
|
172 |
for operators that associate to the right (not tail recursive)*)
|
wenzelm@233
|
173 |
fun foldr f (l, e) =
|
wenzelm@233
|
174 |
let fun itr [] = e
|
wenzelm@233
|
175 |
| itr (a::l) = f(a, itr l)
|
wenzelm@233
|
176 |
in itr l end;
|
wenzelm@233
|
177 |
|
wenzelm@233
|
178 |
(* (op @) [x1, ..., xn] ===> x1 @ (x2 ... @ (x[n-1] @ xn))
|
wenzelm@233
|
179 |
for n > 0, operators that associate to the right (not tail recursive)*)
|
wenzelm@233
|
180 |
fun foldr1 f l =
|
wenzelm@4181
|
181 |
let fun itr [x] = x
|
wenzelm@233
|
182 |
| itr (x::l) = f(x, itr l)
|
wenzelm@233
|
183 |
in itr l end;
|
wenzelm@233
|
184 |
|
wenzelm@233
|
185 |
|
wenzelm@233
|
186 |
(* basic list functions *)
|
wenzelm@233
|
187 |
|
wenzelm@233
|
188 |
(*length of a list, should unquestionably be a standard function*)
|
wenzelm@233
|
189 |
local fun length1 (n, []) = n (*TAIL RECURSIVE*)
|
wenzelm@233
|
190 |
| length1 (n, x :: xs) = length1 (n + 1, xs)
|
wenzelm@233
|
191 |
in fun length l = length1 (0, l) end;
|
wenzelm@233
|
192 |
|
wenzelm@233
|
193 |
(*take the first n elements from a list*)
|
wenzelm@233
|
194 |
fun take (n, []) = []
|
wenzelm@233
|
195 |
| take (n, x :: xs) =
|
wenzelm@233
|
196 |
if n > 0 then x :: take (n - 1, xs) else [];
|
wenzelm@233
|
197 |
|
wenzelm@233
|
198 |
(*drop the first n elements from a list*)
|
wenzelm@233
|
199 |
fun drop (n, []) = []
|
wenzelm@233
|
200 |
| drop (n, x :: xs) =
|
wenzelm@233
|
201 |
if n > 0 then drop (n - 1, xs) else x :: xs;
|
clasohm@0
|
202 |
|
wenzelm@233
|
203 |
(*return nth element of a list, where 0 designates the first element;
|
wenzelm@233
|
204 |
raise EXCEPTION if list too short*)
|
wenzelm@233
|
205 |
fun nth_elem NL =
|
wenzelm@233
|
206 |
(case drop NL of
|
wenzelm@233
|
207 |
[] => raise LIST "nth_elem"
|
wenzelm@233
|
208 |
| x :: _ => x);
|
wenzelm@233
|
209 |
|
wenzelm@233
|
210 |
(*last element of a list*)
|
wenzelm@233
|
211 |
fun last_elem [] = raise LIST "last_elem"
|
wenzelm@233
|
212 |
| last_elem [x] = x
|
wenzelm@233
|
213 |
| last_elem (_ :: xs) = last_elem xs;
|
wenzelm@233
|
214 |
|
wenzelm@3762
|
215 |
(*rear decomposition*)
|
wenzelm@3762
|
216 |
fun split_last [] = raise LIST "split_last"
|
wenzelm@3762
|
217 |
| split_last [x] = ([], x)
|
wenzelm@3762
|
218 |
| split_last (x :: xs) = apfst (cons x) (split_last xs);
|
wenzelm@3762
|
219 |
|
wenzelm@4212
|
220 |
(*find the position of an element in a list*)
|
wenzelm@4212
|
221 |
fun find_index pred =
|
wenzelm@4212
|
222 |
let fun find _ [] = ~1
|
wenzelm@4212
|
223 |
| find n (x :: xs) = if pred x then n else find (n + 1) xs;
|
wenzelm@4212
|
224 |
in find 0 end;
|
wenzelm@3762
|
225 |
|
wenzelm@4224
|
226 |
fun find_index_eq x = find_index (equal x);
|
wenzelm@4212
|
227 |
|
wenzelm@4212
|
228 |
(*find first element satisfying predicate*)
|
wenzelm@4212
|
229 |
fun find_first _ [] = None
|
wenzelm@4212
|
230 |
| find_first pred (x :: xs) =
|
wenzelm@4212
|
231 |
if pred x then Some x else find_first pred xs;
|
wenzelm@233
|
232 |
|
wenzelm@233
|
233 |
(*flatten a list of lists to a list*)
|
wenzelm@233
|
234 |
fun flat (ls: 'c list list) : 'c list = foldr (op @) (ls, []);
|
wenzelm@233
|
235 |
|
wenzelm@233
|
236 |
(*like Lisp's MAPC -- seq proc [x1, ..., xn] evaluates
|
wenzelm@233
|
237 |
(proc x1; ...; proc xn) for side effects*)
|
wenzelm@233
|
238 |
fun seq (proc: 'a -> unit) : 'a list -> unit =
|
wenzelm@233
|
239 |
let fun seqf [] = ()
|
wenzelm@233
|
240 |
| seqf (x :: xs) = (proc x; seqf xs)
|
wenzelm@233
|
241 |
in seqf end;
|
wenzelm@233
|
242 |
|
wenzelm@233
|
243 |
(*separate s [x1, x2, ..., xn] ===> [x1, s, x2, s, ..., s, xn]*)
|
wenzelm@233
|
244 |
fun separate s (x :: (xs as _ :: _)) = x :: s :: separate s xs
|
wenzelm@233
|
245 |
| separate _ xs = xs;
|
wenzelm@233
|
246 |
|
wenzelm@233
|
247 |
(*make the list [x, x, ..., x] of length n*)
|
wenzelm@233
|
248 |
fun replicate n (x: 'a) : 'a list =
|
wenzelm@233
|
249 |
let fun rep (0, xs) = xs
|
wenzelm@233
|
250 |
| rep (n, xs) = rep (n - 1, x :: xs)
|
wenzelm@233
|
251 |
in
|
wenzelm@233
|
252 |
if n < 0 then raise LIST "replicate"
|
wenzelm@233
|
253 |
else rep (n, [])
|
wenzelm@233
|
254 |
end;
|
wenzelm@233
|
255 |
|
wenzelm@4248
|
256 |
(*multiply [a, b, c, ...] * [xs, ys, zs, ...]*)
|
wenzelm@4248
|
257 |
fun multiply ([], _) = []
|
wenzelm@4248
|
258 |
| multiply (x :: xs, yss) = map (cons x) yss @ multiply (xs, yss);
|
wenzelm@4248
|
259 |
|
wenzelm@233
|
260 |
|
wenzelm@233
|
261 |
(* filter *)
|
wenzelm@233
|
262 |
|
wenzelm@233
|
263 |
(*copy the list preserving elements that satisfy the predicate*)
|
wenzelm@233
|
264 |
fun filter (pred: 'a->bool) : 'a list -> 'a list =
|
clasohm@0
|
265 |
let fun filt [] = []
|
wenzelm@233
|
266 |
| filt (x :: xs) = if pred x then x :: filt xs else filt xs
|
wenzelm@233
|
267 |
in filt end;
|
clasohm@0
|
268 |
|
clasohm@0
|
269 |
fun filter_out f = filter (not o f);
|
clasohm@0
|
270 |
|
wenzelm@233
|
271 |
fun mapfilter (f: 'a -> 'b option) ([]: 'a list) = [] : 'b list
|
wenzelm@233
|
272 |
| mapfilter f (x :: xs) =
|
wenzelm@233
|
273 |
(case f x of
|
wenzelm@233
|
274 |
None => mapfilter f xs
|
wenzelm@233
|
275 |
| Some y => y :: mapfilter f xs);
|
wenzelm@233
|
276 |
|
wenzelm@233
|
277 |
|
wenzelm@233
|
278 |
(* lists of pairs *)
|
wenzelm@233
|
279 |
|
wenzelm@380
|
280 |
fun map2 _ ([], []) = []
|
wenzelm@380
|
281 |
| map2 f (x :: xs, y :: ys) = (f (x, y) :: map2 f (xs, ys))
|
wenzelm@380
|
282 |
| map2 _ _ = raise LIST "map2";
|
wenzelm@380
|
283 |
|
wenzelm@380
|
284 |
fun exists2 _ ([], []) = false
|
wenzelm@380
|
285 |
| exists2 pred (x :: xs, y :: ys) = pred (x, y) orelse exists2 pred (xs, ys)
|
wenzelm@380
|
286 |
| exists2 _ _ = raise LIST "exists2";
|
wenzelm@380
|
287 |
|
wenzelm@380
|
288 |
fun forall2 _ ([], []) = true
|
wenzelm@380
|
289 |
| forall2 pred (x :: xs, y :: ys) = pred (x, y) andalso forall2 pred (xs, ys)
|
wenzelm@380
|
290 |
| forall2 _ _ = raise LIST "forall2";
|
wenzelm@380
|
291 |
|
wenzelm@233
|
292 |
(*combine two lists forming a list of pairs:
|
wenzelm@233
|
293 |
[x1, ..., xn] ~~ [y1, ..., yn] ===> [(x1, y1), ..., (xn, yn)]*)
|
wenzelm@233
|
294 |
fun [] ~~ [] = []
|
wenzelm@233
|
295 |
| (x :: xs) ~~ (y :: ys) = (x, y) :: (xs ~~ ys)
|
wenzelm@233
|
296 |
| _ ~~ _ = raise LIST "~~";
|
wenzelm@233
|
297 |
|
wenzelm@233
|
298 |
(*inverse of ~~; the old 'split':
|
wenzelm@233
|
299 |
[(x1, y1), ..., (xn, yn)] ===> ([x1, ..., xn], [y1, ..., yn])*)
|
wenzelm@233
|
300 |
fun split_list (l: ('a * 'b) list) = (map #1 l, map #2 l);
|
wenzelm@233
|
301 |
|
wenzelm@233
|
302 |
|
wenzelm@233
|
303 |
(* prefixes, suffixes *)
|
wenzelm@233
|
304 |
|
wenzelm@233
|
305 |
fun [] prefix _ = true
|
wenzelm@233
|
306 |
| (x :: xs) prefix (y :: ys) = x = y andalso (xs prefix ys)
|
wenzelm@233
|
307 |
| _ prefix _ = false;
|
wenzelm@233
|
308 |
|
wenzelm@233
|
309 |
(* [x1, ..., xi, ..., xn] ---> ([x1, ..., x(i-1)], [xi, ..., xn])
|
wenzelm@233
|
310 |
where xi is the first element that does not satisfy the predicate*)
|
wenzelm@233
|
311 |
fun take_prefix (pred : 'a -> bool) (xs: 'a list) : 'a list * 'a list =
|
wenzelm@233
|
312 |
let fun take (rxs, []) = (rev rxs, [])
|
wenzelm@255
|
313 |
| take (rxs, x :: xs) =
|
wenzelm@255
|
314 |
if pred x then take(x :: rxs, xs) else (rev rxs, x :: xs)
|
wenzelm@233
|
315 |
in take([], xs) end;
|
wenzelm@233
|
316 |
|
wenzelm@233
|
317 |
(* [x1, ..., xi, ..., xn] ---> ([x1, ..., xi], [x(i+1), ..., xn])
|
wenzelm@233
|
318 |
where xi is the last element that does not satisfy the predicate*)
|
wenzelm@233
|
319 |
fun take_suffix _ [] = ([], [])
|
wenzelm@233
|
320 |
| take_suffix pred (x :: xs) =
|
wenzelm@233
|
321 |
(case take_suffix pred xs of
|
wenzelm@233
|
322 |
([], sffx) => if pred x then ([], x :: sffx) else ([x], sffx)
|
wenzelm@233
|
323 |
| (prfx, sffx) => (x :: prfx, sffx));
|
wenzelm@233
|
324 |
|
wenzelm@233
|
325 |
|
wenzelm@233
|
326 |
|
wenzelm@233
|
327 |
(** integers **)
|
wenzelm@233
|
328 |
|
wenzelm@2958
|
329 |
fun inc i = (i := ! i + 1; ! i);
|
wenzelm@2958
|
330 |
fun dec i = (i := ! i - 1; ! i);
|
wenzelm@233
|
331 |
|
wenzelm@233
|
332 |
|
wenzelm@233
|
333 |
(* lists of integers *)
|
wenzelm@233
|
334 |
|
wenzelm@233
|
335 |
(*make the list [from, from + 1, ..., to]*)
|
paulson@2175
|
336 |
fun (from upto to) =
|
wenzelm@233
|
337 |
if from > to then [] else from :: ((from + 1) upto to);
|
wenzelm@233
|
338 |
|
wenzelm@233
|
339 |
(*make the list [from, from - 1, ..., to]*)
|
paulson@2175
|
340 |
fun (from downto to) =
|
wenzelm@233
|
341 |
if from < to then [] else from :: ((from - 1) downto to);
|
wenzelm@233
|
342 |
|
wenzelm@233
|
343 |
(*predicate: downto0 (is, n) <=> is = [n, n - 1, ..., 0]*)
|
wenzelm@233
|
344 |
fun downto0 (i :: is, n) = i = n andalso downto0 (is, n - 1)
|
wenzelm@233
|
345 |
| downto0 ([], n) = n = ~1;
|
wenzelm@233
|
346 |
|
wenzelm@233
|
347 |
|
wenzelm@233
|
348 |
(* convert integers to strings *)
|
wenzelm@233
|
349 |
|
wenzelm@233
|
350 |
(*expand the number in the given base;
|
wenzelm@233
|
351 |
example: radixpand (2, 8) gives [1, 0, 0, 0]*)
|
wenzelm@233
|
352 |
fun radixpand (base, num) : int list =
|
wenzelm@233
|
353 |
let
|
wenzelm@233
|
354 |
fun radix (n, tail) =
|
wenzelm@233
|
355 |
if n < base then n :: tail
|
wenzelm@233
|
356 |
else radix (n div base, (n mod base) :: tail)
|
wenzelm@233
|
357 |
in radix (num, []) end;
|
wenzelm@233
|
358 |
|
wenzelm@233
|
359 |
(*expands a number into a string of characters starting from "zerochar";
|
wenzelm@233
|
360 |
example: radixstring (2, "0", 8) gives "1000"*)
|
wenzelm@233
|
361 |
fun radixstring (base, zerochar, num) =
|
wenzelm@233
|
362 |
let val offset = ord zerochar;
|
wenzelm@233
|
363 |
fun chrof n = chr (offset + n)
|
wenzelm@233
|
364 |
in implode (map chrof (radixpand (base, num))) end;
|
wenzelm@233
|
365 |
|
wenzelm@233
|
366 |
|
paulson@3407
|
367 |
val string_of_int = Int.toString;
|
wenzelm@233
|
368 |
|
paulson@3407
|
369 |
fun string_of_indexname (a,0) = a
|
paulson@3407
|
370 |
| string_of_indexname (a,i) = a ^ "_" ^ Int.toString i;
|
wenzelm@233
|
371 |
|
wenzelm@233
|
372 |
|
wenzelm@4212
|
373 |
|
wenzelm@233
|
374 |
(** strings **)
|
wenzelm@233
|
375 |
|
wenzelm@233
|
376 |
fun is_letter ch =
|
wenzelm@233
|
377 |
ord "A" <= ord ch andalso ord ch <= ord "Z" orelse
|
wenzelm@233
|
378 |
ord "a" <= ord ch andalso ord ch <= ord "z";
|
wenzelm@233
|
379 |
|
wenzelm@233
|
380 |
fun is_digit ch =
|
wenzelm@233
|
381 |
ord "0" <= ord ch andalso ord ch <= ord "9";
|
wenzelm@233
|
382 |
|
wenzelm@233
|
383 |
(*letter or _ or prime (')*)
|
wenzelm@233
|
384 |
fun is_quasi_letter "_" = true
|
wenzelm@233
|
385 |
| is_quasi_letter "'" = true
|
wenzelm@233
|
386 |
| is_quasi_letter ch = is_letter ch;
|
wenzelm@233
|
387 |
|
lcp@512
|
388 |
(*white space: blanks, tabs, newlines, formfeeds*)
|
wenzelm@233
|
389 |
val is_blank : string -> bool =
|
wenzelm@3393
|
390 |
fn " " => true | "\t" => true | "\n" => true | "\^L" => true | "\160" => true
|
wenzelm@3063
|
391 |
| _ => false;
|
wenzelm@233
|
392 |
|
wenzelm@233
|
393 |
val is_letdig = is_quasi_letter orf is_digit;
|
wenzelm@233
|
394 |
|
wenzelm@2196
|
395 |
(*printable chars*)
|
wenzelm@2196
|
396 |
fun is_printable c = ord c > ord " " andalso ord c <= ord "~";
|
wenzelm@2196
|
397 |
|
wenzelm@233
|
398 |
(*lower all chars of string*)
|
wenzelm@233
|
399 |
val to_lower =
|
wenzelm@233
|
400 |
let
|
wenzelm@233
|
401 |
fun lower ch =
|
wenzelm@233
|
402 |
if ch >= "A" andalso ch <= "Z" then
|
wenzelm@233
|
403 |
chr (ord ch - ord "A" + ord "a")
|
wenzelm@233
|
404 |
else ch;
|
wenzelm@233
|
405 |
in implode o (map lower) o explode end;
|
wenzelm@233
|
406 |
|
lcp@512
|
407 |
(*enclose in brackets*)
|
lcp@512
|
408 |
fun enclose lpar rpar str = lpar ^ str ^ rpar;
|
wenzelm@255
|
409 |
|
wenzelm@233
|
410 |
(*simple quoting (does not escape special chars)*)
|
lcp@512
|
411 |
val quote = enclose "\"" "\"";
|
wenzelm@233
|
412 |
|
wenzelm@4212
|
413 |
(*space_implode "..." (explode "hello") = "h...e...l...l...o"*)
|
wenzelm@233
|
414 |
fun space_implode a bs = implode (separate a bs);
|
wenzelm@233
|
415 |
|
wenzelm@255
|
416 |
val commas = space_implode ", ";
|
wenzelm@380
|
417 |
val commas_quote = commas o map quote;
|
wenzelm@255
|
418 |
|
wenzelm@233
|
419 |
(*concatenate messages, one per line, into a string*)
|
wenzelm@255
|
420 |
val cat_lines = space_implode "\n";
|
wenzelm@233
|
421 |
|
wenzelm@4212
|
422 |
(*space_explode "." "h.e..l.lo" = ["h", "e", "", "l", "lo"]*)
|
wenzelm@3832
|
423 |
fun space_explode _ "" = []
|
wenzelm@3832
|
424 |
| space_explode sep str =
|
wenzelm@3832
|
425 |
let
|
wenzelm@3832
|
426 |
fun expl chs =
|
wenzelm@3832
|
427 |
(case take_prefix (not_equal sep) chs of
|
wenzelm@3832
|
428 |
(cs, []) => [implode cs]
|
wenzelm@3832
|
429 |
| (cs, _ :: cs') => implode cs :: expl cs');
|
wenzelm@3832
|
430 |
in expl (explode str) end;
|
wenzelm@3832
|
431 |
|
wenzelm@3832
|
432 |
val split_lines = space_explode "\n";
|
wenzelm@3832
|
433 |
|
wenzelm@3832
|
434 |
|
wenzelm@233
|
435 |
|
wenzelm@233
|
436 |
(** lists as sets **)
|
wenzelm@233
|
437 |
|
wenzelm@233
|
438 |
(*membership in a list*)
|
wenzelm@233
|
439 |
fun x mem [] = false
|
wenzelm@233
|
440 |
| x mem (y :: ys) = x = y orelse x mem ys;
|
clasohm@0
|
441 |
|
paulson@2175
|
442 |
(*membership in a list, optimized version for ints*)
|
berghofe@1576
|
443 |
fun (x:int) mem_int [] = false
|
berghofe@1576
|
444 |
| x mem_int (y :: ys) = x = y orelse x mem_int ys;
|
berghofe@1576
|
445 |
|
paulson@2175
|
446 |
(*membership in a list, optimized version for strings*)
|
berghofe@1576
|
447 |
fun (x:string) mem_string [] = false
|
berghofe@1576
|
448 |
| x mem_string (y :: ys) = x = y orelse x mem_string ys;
|
berghofe@1576
|
449 |
|
clasohm@0
|
450 |
(*generalized membership test*)
|
wenzelm@233
|
451 |
fun gen_mem eq (x, []) = false
|
wenzelm@233
|
452 |
| gen_mem eq (x, y :: ys) = eq (x, y) orelse gen_mem eq (x, ys);
|
wenzelm@233
|
453 |
|
wenzelm@233
|
454 |
|
wenzelm@233
|
455 |
(*insertion into list if not already there*)
|
paulson@2175
|
456 |
fun (x ins xs) = if x mem xs then xs else x :: xs;
|
clasohm@0
|
457 |
|
paulson@2175
|
458 |
(*insertion into list, optimized version for ints*)
|
paulson@2175
|
459 |
fun (x ins_int xs) = if x mem_int xs then xs else x :: xs;
|
berghofe@1576
|
460 |
|
paulson@2175
|
461 |
(*insertion into list, optimized version for strings*)
|
paulson@2175
|
462 |
fun (x ins_string xs) = if x mem_string xs then xs else x :: xs;
|
berghofe@1576
|
463 |
|
clasohm@0
|
464 |
(*generalized insertion*)
|
wenzelm@233
|
465 |
fun gen_ins eq (x, xs) = if gen_mem eq (x, xs) then xs else x :: xs;
|
wenzelm@233
|
466 |
|
wenzelm@233
|
467 |
|
wenzelm@233
|
468 |
(*union of sets represented as lists: no repetitions*)
|
wenzelm@233
|
469 |
fun xs union [] = xs
|
wenzelm@233
|
470 |
| [] union ys = ys
|
wenzelm@233
|
471 |
| (x :: xs) union ys = xs union (x ins ys);
|
clasohm@0
|
472 |
|
paulson@2175
|
473 |
(*union of sets, optimized version for ints*)
|
berghofe@1576
|
474 |
fun (xs:int list) union_int [] = xs
|
berghofe@1576
|
475 |
| [] union_int ys = ys
|
berghofe@1576
|
476 |
| (x :: xs) union_int ys = xs union_int (x ins_int ys);
|
berghofe@1576
|
477 |
|
paulson@2175
|
478 |
(*union of sets, optimized version for strings*)
|
berghofe@1576
|
479 |
fun (xs:string list) union_string [] = xs
|
berghofe@1576
|
480 |
| [] union_string ys = ys
|
berghofe@1576
|
481 |
| (x :: xs) union_string ys = xs union_string (x ins_string ys);
|
berghofe@1576
|
482 |
|
clasohm@0
|
483 |
(*generalized union*)
|
wenzelm@233
|
484 |
fun gen_union eq (xs, []) = xs
|
wenzelm@233
|
485 |
| gen_union eq ([], ys) = ys
|
wenzelm@233
|
486 |
| gen_union eq (x :: xs, ys) = gen_union eq (xs, gen_ins eq (x, ys));
|
wenzelm@233
|
487 |
|
wenzelm@233
|
488 |
|
wenzelm@233
|
489 |
(*intersection*)
|
wenzelm@233
|
490 |
fun [] inter ys = []
|
wenzelm@233
|
491 |
| (x :: xs) inter ys =
|
wenzelm@233
|
492 |
if x mem ys then x :: (xs inter ys) else xs inter ys;
|
wenzelm@233
|
493 |
|
paulson@2175
|
494 |
(*intersection, optimized version for ints*)
|
berghofe@1576
|
495 |
fun ([]:int list) inter_int ys = []
|
berghofe@1576
|
496 |
| (x :: xs) inter_int ys =
|
berghofe@1576
|
497 |
if x mem_int ys then x :: (xs inter_int ys) else xs inter_int ys;
|
berghofe@1576
|
498 |
|
paulson@2175
|
499 |
(*intersection, optimized version for strings *)
|
berghofe@1576
|
500 |
fun ([]:string list) inter_string ys = []
|
berghofe@1576
|
501 |
| (x :: xs) inter_string ys =
|
berghofe@1576
|
502 |
if x mem_string ys then x :: (xs inter_string ys) else xs inter_string ys;
|
berghofe@1576
|
503 |
|
wenzelm@233
|
504 |
|
wenzelm@233
|
505 |
(*subset*)
|
wenzelm@233
|
506 |
fun [] subset ys = true
|
wenzelm@233
|
507 |
| (x :: xs) subset ys = x mem ys andalso xs subset ys;
|
wenzelm@233
|
508 |
|
paulson@2175
|
509 |
(*subset, optimized version for ints*)
|
berghofe@1576
|
510 |
fun ([]:int list) subset_int ys = true
|
berghofe@1576
|
511 |
| (x :: xs) subset_int ys = x mem_int ys andalso xs subset_int ys;
|
berghofe@1576
|
512 |
|
paulson@2175
|
513 |
(*subset, optimized version for strings*)
|
berghofe@1576
|
514 |
fun ([]:string list) subset_string ys = true
|
berghofe@1576
|
515 |
| (x :: xs) subset_string ys = x mem_string ys andalso xs subset_string ys;
|
berghofe@1576
|
516 |
|
wenzelm@4363
|
517 |
(*set equality*)
|
wenzelm@4363
|
518 |
fun eq_set (xs, ys) =
|
wenzelm@4363
|
519 |
xs = ys orelse (xs subset ys andalso ys subset xs);
|
wenzelm@4363
|
520 |
|
paulson@2182
|
521 |
(*set equality for strings*)
|
berghofe@1576
|
522 |
fun eq_set_string ((xs:string list), ys) =
|
berghofe@1576
|
523 |
xs = ys orelse (xs subset_string ys andalso ys subset_string xs);
|
berghofe@1576
|
524 |
|
paulson@2182
|
525 |
fun gen_subset eq (xs, ys) = forall (fn x => gen_mem eq (x, ys)) xs;
|
paulson@2182
|
526 |
|
wenzelm@265
|
527 |
|
wenzelm@233
|
528 |
(*removing an element from a list WITHOUT duplicates*)
|
wenzelm@233
|
529 |
fun (y :: ys) \ x = if x = y then ys else y :: (ys \ x)
|
wenzelm@233
|
530 |
| [] \ x = [];
|
wenzelm@233
|
531 |
|
paulson@2243
|
532 |
fun ys \\ xs = foldl (op \) (ys,xs);
|
clasohm@0
|
533 |
|
wenzelm@233
|
534 |
(*removing an element from a list -- possibly WITH duplicates*)
|
wenzelm@233
|
535 |
fun gen_rem eq (xs, y) = filter_out (fn x => eq (x, y)) xs;
|
wenzelm@233
|
536 |
|
paulson@2243
|
537 |
fun gen_rems eq = foldl (gen_rem eq);
|
wenzelm@233
|
538 |
|
wenzelm@233
|
539 |
|
wenzelm@233
|
540 |
(*makes a list of the distinct members of the input; preserves order, takes
|
wenzelm@233
|
541 |
first of equal elements*)
|
wenzelm@233
|
542 |
fun gen_distinct eq lst =
|
wenzelm@233
|
543 |
let
|
wenzelm@233
|
544 |
val memb = gen_mem eq;
|
clasohm@0
|
545 |
|
wenzelm@233
|
546 |
fun dist (rev_seen, []) = rev rev_seen
|
wenzelm@233
|
547 |
| dist (rev_seen, x :: xs) =
|
wenzelm@233
|
548 |
if memb (x, rev_seen) then dist (rev_seen, xs)
|
wenzelm@233
|
549 |
else dist (x :: rev_seen, xs);
|
wenzelm@233
|
550 |
in
|
wenzelm@233
|
551 |
dist ([], lst)
|
wenzelm@233
|
552 |
end;
|
wenzelm@233
|
553 |
|
paulson@2243
|
554 |
fun distinct l = gen_distinct (op =) l;
|
wenzelm@233
|
555 |
|
wenzelm@233
|
556 |
|
wenzelm@233
|
557 |
(*returns the tail beginning with the first repeated element, or []*)
|
wenzelm@233
|
558 |
fun findrep [] = []
|
wenzelm@233
|
559 |
| findrep (x :: xs) = if x mem xs then x :: xs else findrep xs;
|
wenzelm@233
|
560 |
|
wenzelm@233
|
561 |
|
wenzelm@255
|
562 |
(*returns a list containing all repeated elements exactly once; preserves
|
wenzelm@255
|
563 |
order, takes first of equal elements*)
|
wenzelm@255
|
564 |
fun gen_duplicates eq lst =
|
wenzelm@255
|
565 |
let
|
wenzelm@255
|
566 |
val memb = gen_mem eq;
|
wenzelm@255
|
567 |
|
wenzelm@255
|
568 |
fun dups (rev_dups, []) = rev rev_dups
|
wenzelm@255
|
569 |
| dups (rev_dups, x :: xs) =
|
wenzelm@255
|
570 |
if memb (x, rev_dups) orelse not (memb (x, xs)) then
|
wenzelm@255
|
571 |
dups (rev_dups, xs)
|
wenzelm@255
|
572 |
else dups (x :: rev_dups, xs);
|
wenzelm@255
|
573 |
in
|
wenzelm@255
|
574 |
dups ([], lst)
|
wenzelm@255
|
575 |
end;
|
wenzelm@255
|
576 |
|
paulson@2243
|
577 |
fun duplicates l = gen_duplicates (op =) l;
|
wenzelm@255
|
578 |
|
wenzelm@255
|
579 |
|
wenzelm@233
|
580 |
|
wenzelm@233
|
581 |
(** association lists **)
|
clasohm@0
|
582 |
|
wenzelm@233
|
583 |
(*association list lookup*)
|
wenzelm@233
|
584 |
fun assoc ([], key) = None
|
wenzelm@233
|
585 |
| assoc ((keyi, xi) :: pairs, key) =
|
wenzelm@233
|
586 |
if key = keyi then Some xi else assoc (pairs, key);
|
wenzelm@233
|
587 |
|
paulson@2175
|
588 |
(*association list lookup, optimized version for ints*)
|
berghofe@1576
|
589 |
fun assoc_int ([], (key:int)) = None
|
berghofe@1576
|
590 |
| assoc_int ((keyi, xi) :: pairs, key) =
|
berghofe@1576
|
591 |
if key = keyi then Some xi else assoc_int (pairs, key);
|
berghofe@1576
|
592 |
|
paulson@2175
|
593 |
(*association list lookup, optimized version for strings*)
|
berghofe@1576
|
594 |
fun assoc_string ([], (key:string)) = None
|
berghofe@1576
|
595 |
| assoc_string ((keyi, xi) :: pairs, key) =
|
berghofe@1576
|
596 |
if key = keyi then Some xi else assoc_string (pairs, key);
|
berghofe@1576
|
597 |
|
paulson@2175
|
598 |
(*association list lookup, optimized version for string*ints*)
|
berghofe@1576
|
599 |
fun assoc_string_int ([], (key:string*int)) = None
|
berghofe@1576
|
600 |
| assoc_string_int ((keyi, xi) :: pairs, key) =
|
berghofe@1576
|
601 |
if key = keyi then Some xi else assoc_string_int (pairs, key);
|
berghofe@1576
|
602 |
|
wenzelm@233
|
603 |
fun assocs ps x =
|
wenzelm@233
|
604 |
(case assoc (ps, x) of
|
wenzelm@233
|
605 |
None => []
|
wenzelm@233
|
606 |
| Some ys => ys);
|
wenzelm@233
|
607 |
|
wenzelm@255
|
608 |
(*two-fold association list lookup*)
|
wenzelm@255
|
609 |
fun assoc2 (aal, (key1, key2)) =
|
wenzelm@255
|
610 |
(case assoc (aal, key1) of
|
wenzelm@255
|
611 |
Some al => assoc (al, key2)
|
wenzelm@255
|
612 |
| None => None);
|
wenzelm@255
|
613 |
|
wenzelm@233
|
614 |
(*generalized association list lookup*)
|
wenzelm@233
|
615 |
fun gen_assoc eq ([], key) = None
|
wenzelm@233
|
616 |
| gen_assoc eq ((keyi, xi) :: pairs, key) =
|
wenzelm@233
|
617 |
if eq (key, keyi) then Some xi else gen_assoc eq (pairs, key);
|
wenzelm@233
|
618 |
|
wenzelm@233
|
619 |
(*association list update*)
|
wenzelm@233
|
620 |
fun overwrite (al, p as (key, _)) =
|
wenzelm@233
|
621 |
let fun over ((q as (keyi, _)) :: pairs) =
|
wenzelm@233
|
622 |
if keyi = key then p :: pairs else q :: (over pairs)
|
wenzelm@233
|
623 |
| over [] = [p]
|
wenzelm@233
|
624 |
in over al end;
|
wenzelm@233
|
625 |
|
wenzelm@2522
|
626 |
fun gen_overwrite eq (al, p as (key, _)) =
|
wenzelm@2522
|
627 |
let fun over ((q as (keyi, _)) :: pairs) =
|
wenzelm@2522
|
628 |
if eq (keyi, key) then p :: pairs else q :: (over pairs)
|
wenzelm@2522
|
629 |
| over [] = [p]
|
wenzelm@2522
|
630 |
in over al end;
|
wenzelm@2522
|
631 |
|
wenzelm@233
|
632 |
|
wenzelm@233
|
633 |
|
wenzelm@233
|
634 |
(** generic tables **)
|
clasohm@0
|
635 |
|
wenzelm@233
|
636 |
(*Tables are supposed to be 'efficient' encodings of lists of elements distinct
|
wenzelm@233
|
637 |
wrt. an equality "eq". The extend and merge operations below are optimized
|
wenzelm@233
|
638 |
for long-term space efficiency.*)
|
wenzelm@233
|
639 |
|
wenzelm@233
|
640 |
(*append (new) elements to a table*)
|
wenzelm@233
|
641 |
fun generic_extend _ _ _ tab [] = tab
|
wenzelm@233
|
642 |
| generic_extend eq dest_tab mk_tab tab1 lst2 =
|
wenzelm@233
|
643 |
let
|
wenzelm@233
|
644 |
val lst1 = dest_tab tab1;
|
wenzelm@233
|
645 |
val new_lst2 = gen_rems eq (lst2, lst1);
|
wenzelm@233
|
646 |
in
|
wenzelm@233
|
647 |
if null new_lst2 then tab1
|
wenzelm@233
|
648 |
else mk_tab (lst1 @ new_lst2)
|
wenzelm@233
|
649 |
end;
|
clasohm@0
|
650 |
|
wenzelm@233
|
651 |
(*append (new) elements of 2nd table to 1st table*)
|
wenzelm@233
|
652 |
fun generic_merge eq dest_tab mk_tab tab1 tab2 =
|
wenzelm@233
|
653 |
let
|
wenzelm@233
|
654 |
val lst1 = dest_tab tab1;
|
wenzelm@233
|
655 |
val lst2 = dest_tab tab2;
|
wenzelm@233
|
656 |
val new_lst2 = gen_rems eq (lst2, lst1);
|
wenzelm@233
|
657 |
in
|
wenzelm@233
|
658 |
if null new_lst2 then tab1
|
wenzelm@233
|
659 |
else if gen_subset eq (lst1, lst2) then tab2
|
wenzelm@233
|
660 |
else mk_tab (lst1 @ new_lst2)
|
wenzelm@233
|
661 |
end;
|
clasohm@0
|
662 |
|
wenzelm@233
|
663 |
|
wenzelm@233
|
664 |
(*lists as tables*)
|
paulson@2243
|
665 |
fun extend_list tab = generic_extend (op =) I I tab;
|
paulson@2243
|
666 |
fun merge_lists tab = generic_merge (op =) I I tab;
|
wenzelm@233
|
667 |
|
wenzelm@380
|
668 |
fun merge_rev_lists xs [] = xs
|
wenzelm@380
|
669 |
| merge_rev_lists [] ys = ys
|
wenzelm@380
|
670 |
| merge_rev_lists xs (y :: ys) =
|
wenzelm@380
|
671 |
(if y mem xs then I else cons y) (merge_rev_lists xs ys);
|
wenzelm@380
|
672 |
|
clasohm@0
|
673 |
|
clasohm@0
|
674 |
|
wenzelm@233
|
675 |
(** balanced trees **)
|
wenzelm@233
|
676 |
|
wenzelm@233
|
677 |
exception Balance; (*indicates non-positive argument to balancing fun*)
|
wenzelm@233
|
678 |
|
wenzelm@233
|
679 |
(*balanced folding; avoids deep nesting*)
|
wenzelm@233
|
680 |
fun fold_bal f [x] = x
|
wenzelm@233
|
681 |
| fold_bal f [] = raise Balance
|
wenzelm@233
|
682 |
| fold_bal f xs =
|
wenzelm@233
|
683 |
let val k = length xs div 2
|
wenzelm@233
|
684 |
in f (fold_bal f (take(k, xs)),
|
wenzelm@233
|
685 |
fold_bal f (drop(k, xs)))
|
wenzelm@233
|
686 |
end;
|
wenzelm@233
|
687 |
|
wenzelm@233
|
688 |
(*construct something of the form f(...g(...(x)...)) for balanced access*)
|
wenzelm@233
|
689 |
fun access_bal (f, g, x) n i =
|
wenzelm@233
|
690 |
let fun acc n i = (*1<=i<=n*)
|
wenzelm@233
|
691 |
if n=1 then x else
|
wenzelm@233
|
692 |
let val n2 = n div 2
|
wenzelm@233
|
693 |
in if i<=n2 then f (acc n2 i)
|
wenzelm@233
|
694 |
else g (acc (n-n2) (i-n2))
|
wenzelm@233
|
695 |
end
|
wenzelm@233
|
696 |
in if 1<=i andalso i<=n then acc n i else raise Balance end;
|
wenzelm@233
|
697 |
|
wenzelm@233
|
698 |
(*construct ALL such accesses; could try harder to share recursive calls!*)
|
wenzelm@233
|
699 |
fun accesses_bal (f, g, x) n =
|
wenzelm@233
|
700 |
let fun acc n =
|
wenzelm@233
|
701 |
if n=1 then [x] else
|
wenzelm@233
|
702 |
let val n2 = n div 2
|
wenzelm@233
|
703 |
val acc2 = acc n2
|
wenzelm@233
|
704 |
in if n-n2=n2 then map f acc2 @ map g acc2
|
wenzelm@233
|
705 |
else map f acc2 @ map g (acc (n-n2)) end
|
wenzelm@233
|
706 |
in if 1<=n then acc n else raise Balance end;
|
wenzelm@233
|
707 |
|
wenzelm@233
|
708 |
|
wenzelm@233
|
709 |
|
wenzelm@2506
|
710 |
(** orders **)
|
wenzelm@2506
|
711 |
|
wenzelm@2506
|
712 |
datatype order = LESS | EQUAL | GREATER;
|
wenzelm@2506
|
713 |
|
wenzelm@4445
|
714 |
fun rev_order LESS = GREATER
|
wenzelm@4445
|
715 |
| rev_order EQUAL = EQUAL
|
wenzelm@4445
|
716 |
| rev_order GREATER = LESS;
|
wenzelm@4445
|
717 |
|
wenzelm@4479
|
718 |
(*assume rel is a linear strict order*)
|
wenzelm@4445
|
719 |
fun make_ord rel (x, y) =
|
wenzelm@4445
|
720 |
if rel (x, y) then LESS
|
wenzelm@4445
|
721 |
else if rel (y, x) then GREATER
|
wenzelm@4445
|
722 |
else EQUAL;
|
wenzelm@4445
|
723 |
|
wenzelm@4343
|
724 |
fun int_ord (i, j: int) =
|
wenzelm@2506
|
725 |
if i < j then LESS
|
wenzelm@2506
|
726 |
else if i = j then EQUAL
|
wenzelm@2506
|
727 |
else GREATER;
|
wenzelm@2506
|
728 |
|
wenzelm@4343
|
729 |
fun string_ord (a, b: string) =
|
wenzelm@2506
|
730 |
if a < b then LESS
|
wenzelm@2506
|
731 |
else if a = b then EQUAL
|
wenzelm@2506
|
732 |
else GREATER;
|
wenzelm@2506
|
733 |
|
wenzelm@4343
|
734 |
(*lexicographic product*)
|
wenzelm@4343
|
735 |
fun prod_ord a_ord b_ord ((x, y), (x', y')) =
|
wenzelm@4343
|
736 |
(case a_ord (x, x') of EQUAL => b_ord (y, y') | ord => ord);
|
wenzelm@4343
|
737 |
|
wenzelm@4343
|
738 |
(*dictionary order -- in general NOT well-founded!*)
|
wenzelm@4343
|
739 |
fun dict_ord _ ([], []) = EQUAL
|
wenzelm@4343
|
740 |
| dict_ord _ ([], _ :: _) = LESS
|
wenzelm@4343
|
741 |
| dict_ord _ (_ :: _, []) = GREATER
|
wenzelm@4343
|
742 |
| dict_ord elem_ord (x :: xs, y :: ys) =
|
wenzelm@4343
|
743 |
(case elem_ord (x, y) of EQUAL => dict_ord elem_ord (xs, ys) | ord => ord);
|
wenzelm@4343
|
744 |
|
wenzelm@4343
|
745 |
(*lexicographic product of lists*)
|
wenzelm@4343
|
746 |
fun list_ord elem_ord (xs, ys) =
|
wenzelm@4343
|
747 |
prod_ord int_ord (dict_ord elem_ord) ((length xs, xs), (length ys, ys));
|
wenzelm@4343
|
748 |
|
wenzelm@2506
|
749 |
|
wenzelm@2506
|
750 |
|
wenzelm@3525
|
751 |
(** input / output and diagnostics **)
|
wenzelm@233
|
752 |
|
paulson@2243
|
753 |
val cd = OS.FileSys.chDir;
|
wenzelm@2317
|
754 |
val pwd = OS.FileSys.getDir;
|
paulson@2243
|
755 |
|
wenzelm@3525
|
756 |
|
wenzelm@3525
|
757 |
local
|
wenzelm@3525
|
758 |
fun out s =
|
wenzelm@3525
|
759 |
(TextIO.output (TextIO.stdOut, s); TextIO.flushOut TextIO.stdOut);
|
wenzelm@3525
|
760 |
|
wenzelm@3525
|
761 |
fun prefix_lines prfx txt =
|
wenzelm@3832
|
762 |
txt |> split_lines |> map (fn s => prfx ^ s ^ "\n") |> implode;
|
wenzelm@3525
|
763 |
in
|
wenzelm@3525
|
764 |
|
wenzelm@3525
|
765 |
(*hooks for output channels: normal, warning, error*)
|
wenzelm@3525
|
766 |
val prs_fn = ref (fn s => out s);
|
wenzelm@3525
|
767 |
val warning_fn = ref (fn s => out (prefix_lines "### " s));
|
wenzelm@3525
|
768 |
val error_fn = ref (fn s => out (prefix_lines "*** " s));
|
wenzelm@3525
|
769 |
|
wenzelm@3525
|
770 |
end;
|
berghofe@1580
|
771 |
|
berghofe@1580
|
772 |
fun prs s = !prs_fn s;
|
wenzelm@233
|
773 |
fun writeln s = prs (s ^ "\n");
|
wenzelm@233
|
774 |
|
wenzelm@3525
|
775 |
fun warning s = !warning_fn s;
|
wenzelm@233
|
776 |
|
wenzelm@233
|
777 |
(*print error message and abort to top level*)
|
wenzelm@233
|
778 |
exception ERROR;
|
wenzelm@4212
|
779 |
fun error_msg s = !error_fn s; (*promise to raise ERROR later!*)
|
wenzelm@3553
|
780 |
fun error s = (error_msg s; raise ERROR);
|
wenzelm@3553
|
781 |
fun sys_error msg = (error_msg " !! SYSTEM ERROR !!\n"; error msg);
|
wenzelm@233
|
782 |
|
wenzelm@233
|
783 |
fun assert p msg = if p then () else error msg;
|
wenzelm@233
|
784 |
fun deny p msg = if p then error msg else ();
|
wenzelm@233
|
785 |
|
lcp@544
|
786 |
(*Assert pred for every member of l, generating a message if pred fails*)
|
wenzelm@4212
|
787 |
fun assert_all pred l msg_fn =
|
lcp@544
|
788 |
let fun asl [] = ()
|
wenzelm@4212
|
789 |
| asl (x::xs) = if pred x then asl xs else error (msg_fn x)
|
wenzelm@4212
|
790 |
in asl l end;
|
wenzelm@233
|
791 |
|
wenzelm@3624
|
792 |
|
wenzelm@4212
|
793 |
(* handle errors capturing messages *)
|
wenzelm@3699
|
794 |
|
wenzelm@3699
|
795 |
datatype 'a error =
|
wenzelm@3699
|
796 |
Error of string |
|
wenzelm@3699
|
797 |
OK of 'a;
|
wenzelm@3699
|
798 |
|
wenzelm@4248
|
799 |
fun get_error (Error msg) = Some msg
|
wenzelm@4248
|
800 |
| get_error _ = None;
|
wenzelm@4248
|
801 |
|
wenzelm@4248
|
802 |
fun get_ok (OK x) = Some x
|
wenzelm@4248
|
803 |
| get_ok _ = None;
|
wenzelm@4248
|
804 |
|
wenzelm@3699
|
805 |
fun handle_error f x =
|
wenzelm@3699
|
806 |
let
|
wenzelm@3699
|
807 |
val buffer = ref "";
|
wenzelm@3699
|
808 |
fun capture s = buffer := ! buffer ^ s ^ "\n";
|
wenzelm@3699
|
809 |
val result = Some (setmp error_fn capture f x) handle ERROR => None;
|
wenzelm@3699
|
810 |
in
|
wenzelm@4212
|
811 |
(case result of
|
wenzelm@3699
|
812 |
None => Error (! buffer)
|
wenzelm@4212
|
813 |
| Some y => OK y)
|
wenzelm@3624
|
814 |
end;
|
wenzelm@3624
|
815 |
|
wenzelm@3624
|
816 |
|
wenzelm@233
|
817 |
|
wenzelm@233
|
818 |
(** timing **)
|
wenzelm@233
|
819 |
|
paulson@4326
|
820 |
(*a conditional timing function: applies f to () and, if the flag is true,
|
paulson@4326
|
821 |
prints its runtime*)
|
paulson@4326
|
822 |
fun cond_timeit flag f =
|
paulson@4326
|
823 |
if flag then
|
paulson@4326
|
824 |
let val start = startTiming()
|
paulson@4326
|
825 |
val result = f ()
|
paulson@4326
|
826 |
in
|
paulson@4326
|
827 |
writeln (endTiming start); result
|
paulson@4326
|
828 |
end
|
paulson@4326
|
829 |
else f ();
|
paulson@4326
|
830 |
|
wenzelm@233
|
831 |
(*unconditional timing function*)
|
paulson@2243
|
832 |
fun timeit x = cond_timeit true x;
|
wenzelm@233
|
833 |
|
wenzelm@233
|
834 |
(*timed application function*)
|
wenzelm@233
|
835 |
fun timeap f x = timeit (fn () => f x);
|
wenzelm@233
|
836 |
|
berghofe@3606
|
837 |
|
wenzelm@233
|
838 |
|
wenzelm@233
|
839 |
(** misc functions **)
|
wenzelm@233
|
840 |
|
wenzelm@233
|
841 |
(*use the keyfun to make a list of (x, key) pairs*)
|
clasohm@0
|
842 |
fun make_keylist (keyfun: 'a->'b) : 'a list -> ('a * 'b) list =
|
wenzelm@233
|
843 |
let fun keypair x = (x, keyfun x)
|
wenzelm@233
|
844 |
in map keypair end;
|
clasohm@0
|
845 |
|
wenzelm@233
|
846 |
(*given a list of (x, key) pairs and a searchkey
|
clasohm@0
|
847 |
return the list of xs from each pair whose key equals searchkey*)
|
clasohm@0
|
848 |
fun keyfilter [] searchkey = []
|
wenzelm@233
|
849 |
| keyfilter ((x, key) :: pairs) searchkey =
|
wenzelm@233
|
850 |
if key = searchkey then x :: keyfilter pairs searchkey
|
wenzelm@233
|
851 |
else keyfilter pairs searchkey;
|
clasohm@0
|
852 |
|
clasohm@0
|
853 |
|
clasohm@0
|
854 |
(*Partition list into elements that satisfy predicate and those that don't.
|
wenzelm@233
|
855 |
Preserves order of elements in both lists.*)
|
clasohm@0
|
856 |
fun partition (pred: 'a->bool) (ys: 'a list) : ('a list * 'a list) =
|
clasohm@0
|
857 |
let fun part ([], answer) = answer
|
wenzelm@233
|
858 |
| part (x::xs, (ys, ns)) = if pred(x)
|
wenzelm@233
|
859 |
then part (xs, (x::ys, ns))
|
wenzelm@233
|
860 |
else part (xs, (ys, x::ns))
|
wenzelm@233
|
861 |
in part (rev ys, ([], [])) end;
|
clasohm@0
|
862 |
|
clasohm@0
|
863 |
|
clasohm@0
|
864 |
fun partition_eq (eq:'a * 'a -> bool) =
|
clasohm@0
|
865 |
let fun part [] = []
|
wenzelm@233
|
866 |
| part (x::ys) = let val (xs, xs') = partition (apl(x, eq)) ys
|
wenzelm@233
|
867 |
in (x::xs)::(part xs') end
|
clasohm@0
|
868 |
in part end;
|
clasohm@0
|
869 |
|
clasohm@0
|
870 |
|
wenzelm@233
|
871 |
(*Partition a list into buckets [ bi, b(i+1), ..., bj ]
|
clasohm@0
|
872 |
putting x in bk if p(k)(x) holds. Preserve order of elements if possible.*)
|
clasohm@0
|
873 |
fun partition_list p i j =
|
wenzelm@233
|
874 |
let fun part k xs =
|
wenzelm@233
|
875 |
if k>j then
|
clasohm@0
|
876 |
(case xs of [] => []
|
clasohm@0
|
877 |
| _ => raise LIST "partition_list")
|
clasohm@0
|
878 |
else
|
wenzelm@233
|
879 |
let val (ns, rest) = partition (p k) xs;
|
wenzelm@233
|
880 |
in ns :: part(k+1)rest end
|
clasohm@0
|
881 |
in part i end;
|
clasohm@0
|
882 |
|
clasohm@0
|
883 |
|
wenzelm@233
|
884 |
(* sorting *)
|
wenzelm@233
|
885 |
|
wenzelm@4445
|
886 |
(*quicksort (stable, i.e. does not reorder equal elements)*)
|
wenzelm@4445
|
887 |
fun sort ord =
|
wenzelm@4445
|
888 |
let
|
wenzelm@4445
|
889 |
fun qsort xs =
|
wenzelm@4445
|
890 |
let val len = length xs in
|
wenzelm@4445
|
891 |
if len <= 1 then xs
|
wenzelm@4445
|
892 |
else
|
wenzelm@4445
|
893 |
let val (lts, eqs, gts) = part (nth_elem (len div 2, xs)) xs in
|
wenzelm@4445
|
894 |
qsort lts @ eqs @ qsort gts
|
wenzelm@4445
|
895 |
end
|
wenzelm@4445
|
896 |
end
|
wenzelm@4445
|
897 |
and part _ [] = ([], [], [])
|
wenzelm@4445
|
898 |
| part pivot (x :: xs) = add (ord (x, pivot)) x (part pivot xs)
|
wenzelm@4445
|
899 |
and add LESS x (lts, eqs, gts) = (x :: lts, eqs, gts)
|
wenzelm@4445
|
900 |
| add EQUAL x (lts, eqs, gts) = (lts, x :: eqs, gts)
|
wenzelm@4445
|
901 |
| add GREATER x (lts, eqs, gts) = (lts, eqs, x :: gts);
|
wenzelm@4445
|
902 |
in qsort end;
|
clasohm@0
|
903 |
|
wenzelm@41
|
904 |
(*sort strings*)
|
wenzelm@4445
|
905 |
val sort_strings = sort string_ord;
|
wenzelm@4445
|
906 |
fun sort_wrt sel xs = sort (string_ord o pairself sel) xs;
|
wenzelm@41
|
907 |
|
wenzelm@41
|
908 |
|
wenzelm@233
|
909 |
(* transitive closure (not Warshall's algorithm) *)
|
clasohm@0
|
910 |
|
wenzelm@233
|
911 |
fun transitive_closure [] = []
|
wenzelm@233
|
912 |
| transitive_closure ((x, ys)::ps) =
|
wenzelm@233
|
913 |
let val qs = transitive_closure ps
|
paulson@2182
|
914 |
val zs = foldl (fn (zs, y) => assocs qs y union_string zs) (ys, ys)
|
paulson@2182
|
915 |
fun step(u, us) = (u, if x mem_string us then zs union_string us
|
paulson@2243
|
916 |
else us)
|
wenzelm@233
|
917 |
in (x, zs) :: map step qs end;
|
clasohm@0
|
918 |
|
clasohm@0
|
919 |
|
wenzelm@233
|
920 |
(* generating identifiers *)
|
clasohm@0
|
921 |
|
paulson@4063
|
922 |
(** Freshly generated identifiers; supplied prefix MUST start with a letter **)
|
clasohm@0
|
923 |
local
|
paulson@4063
|
924 |
(*Maps 0-63 to A-Z, a-z, 0-9 or _ or ' for generating random identifiers*)
|
paulson@4063
|
925 |
fun char i = if i<26 then chr (ord "A" + i)
|
paulson@4063
|
926 |
else if i<52 then chr (ord "a" + i - 26)
|
paulson@4063
|
927 |
else if i<62 then chr (ord"0" + i - 52)
|
paulson@4063
|
928 |
else if i=62 then "_"
|
paulson@4063
|
929 |
else (*i=63*) "'";
|
paulson@4063
|
930 |
|
paulson@4063
|
931 |
val charVec = Vector.tabulate (64, char);
|
paulson@4063
|
932 |
|
paulson@4063
|
933 |
fun newid n =
|
paulson@4063
|
934 |
let
|
wenzelm@4284
|
935 |
in implode (map (fn i => Vector.sub(charVec,i)) (radixpand (64,n))) end;
|
paulson@2003
|
936 |
|
wenzelm@4284
|
937 |
val seedr = ref 0;
|
clasohm@0
|
938 |
|
paulson@4063
|
939 |
in
|
wenzelm@4284
|
940 |
|
paulson@4063
|
941 |
fun init_gensym() = (seedr := 0);
|
paulson@2003
|
942 |
|
wenzelm@4284
|
943 |
fun gensym pre = pre ^ (#1(newid (!seedr), inc seedr));
|
paulson@4063
|
944 |
end;
|
paulson@4063
|
945 |
|
paulson@4063
|
946 |
|
paulson@4063
|
947 |
local
|
paulson@4063
|
948 |
(*Identifies those character codes legal in identifiers.
|
paulson@4063
|
949 |
chould use Basis Library character functions if Poly/ML provided characters*)
|
paulson@4063
|
950 |
fun idCode k = (ord "a" <= k andalso k < ord "z") orelse
|
paulson@4063
|
951 |
(ord "A" <= k andalso k < ord "Z") orelse
|
paulson@4063
|
952 |
(ord "0" <= k andalso k < ord "9");
|
paulson@4063
|
953 |
|
paulson@4063
|
954 |
val idCodeVec = Vector.tabulate (256, idCode);
|
paulson@4063
|
955 |
|
paulson@4063
|
956 |
in
|
paulson@2003
|
957 |
|
clasohm@0
|
958 |
(*Increment a list of letters like a reversed base 26 number.
|
wenzelm@233
|
959 |
If head is "z", bumps chars in tail.
|
clasohm@0
|
960 |
Digits are incremented as if they were integers.
|
clasohm@0
|
961 |
"_" and "'" are not changed.
|
wenzelm@233
|
962 |
For making variants of identifiers.*)
|
clasohm@0
|
963 |
|
paulson@4063
|
964 |
fun bump_int_list(c::cs) =
|
paulson@4063
|
965 |
if c="9" then "0" :: bump_int_list cs
|
paulson@4063
|
966 |
else
|
paulson@4063
|
967 |
if "0" <= c andalso c < "9" then chr(ord(c)+1) :: cs
|
wenzelm@233
|
968 |
else "1" :: c :: cs
|
clasohm@0
|
969 |
| bump_int_list([]) = error("bump_int_list: not an identifier");
|
clasohm@0
|
970 |
|
wenzelm@233
|
971 |
fun bump_list([], d) = [d]
|
wenzelm@233
|
972 |
| bump_list(["'"], d) = [d, "'"]
|
wenzelm@233
|
973 |
| bump_list("z"::cs, _) = "a" :: bump_list(cs, "a")
|
wenzelm@233
|
974 |
| bump_list("Z"::cs, _) = "A" :: bump_list(cs, "A")
|
wenzelm@233
|
975 |
| bump_list("9"::cs, _) = "0" :: bump_int_list cs
|
paulson@4063
|
976 |
| bump_list(c::cs, _) =
|
paulson@4063
|
977 |
let val k = ord(c)
|
paulson@4063
|
978 |
in if Vector.sub(idCodeVec,k) then chr(k+1) :: cs
|
paulson@4063
|
979 |
else
|
paulson@4063
|
980 |
if c="'" orelse c="_" then c :: bump_list(cs, "")
|
paulson@4063
|
981 |
else error("bump_list: not legal in identifier: " ^
|
paulson@4063
|
982 |
implode(rev(c::cs)))
|
wenzelm@233
|
983 |
end;
|
clasohm@0
|
984 |
|
clasohm@0
|
985 |
end;
|
clasohm@0
|
986 |
|
wenzelm@233
|
987 |
fun bump_string s : string = implode (rev (bump_list(rev(explode s), "")));
|
wenzelm@41
|
988 |
|
wenzelm@41
|
989 |
|
wenzelm@233
|
990 |
(* lexical scanning *)
|
clasohm@0
|
991 |
|
wenzelm@233
|
992 |
(*scan a list of characters into "words" composed of "letters" (recognized by
|
wenzelm@233
|
993 |
is_let) and separated by any number of non-"letters"*)
|
wenzelm@233
|
994 |
fun scanwords is_let cs =
|
clasohm@0
|
995 |
let fun scan1 [] = []
|
wenzelm@233
|
996 |
| scan1 cs =
|
wenzelm@233
|
997 |
let val (lets, rest) = take_prefix is_let cs
|
wenzelm@233
|
998 |
in implode lets :: scanwords is_let rest end;
|
wenzelm@233
|
999 |
in scan1 (#2 (take_prefix (not o is_let) cs)) end;
|
clasohm@24
|
1000 |
|
wenzelm@4212
|
1001 |
|
wenzelm@4212
|
1002 |
|
wenzelm@4212
|
1003 |
(* Variable-branching trees: for proof terms etc. *)
|
wenzelm@4212
|
1004 |
datatype 'a mtree = Join of 'a * 'a mtree list;
|
wenzelm@4212
|
1005 |
|
wenzelm@4212
|
1006 |
|
wenzelm@4255
|
1007 |
(* generic objects -- fool the ML type system via exception constructors *)
|
wenzelm@4255
|
1008 |
type object = exn;
|
wenzelm@4255
|
1009 |
|
wenzelm@4255
|
1010 |
|
clasohm@1364
|
1011 |
end;
|
clasohm@1364
|
1012 |
|
clasohm@1364
|
1013 |
open Library;
|