wenzelm@43375
|
1 |
(* Title: Pure/Thy/rail.ML
|
wenzelm@43375
|
2 |
Author: Michael Kerscher, TU München
|
wenzelm@43375
|
3 |
Author: Makarius
|
wenzelm@43375
|
4 |
|
wenzelm@43375
|
5 |
Railroad diagrams in LaTeX.
|
wenzelm@43375
|
6 |
*)
|
wenzelm@43375
|
7 |
|
wenzelm@43375
|
8 |
structure Rail: sig end =
|
wenzelm@43375
|
9 |
struct
|
wenzelm@43375
|
10 |
|
wenzelm@43375
|
11 |
(** lexical syntax **)
|
wenzelm@43375
|
12 |
|
wenzelm@43375
|
13 |
(* datatype token *)
|
wenzelm@43375
|
14 |
|
wenzelm@43379
|
15 |
datatype kind =
|
wenzelm@43387
|
16 |
Keyword | Ident | String | Antiq of Symbol_Pos.T list * Position.range | EOF;
|
wenzelm@43375
|
17 |
|
wenzelm@43375
|
18 |
datatype token = Token of Position.range * (kind * string);
|
wenzelm@43375
|
19 |
|
wenzelm@43375
|
20 |
fun pos_of (Token ((pos, _), _)) = pos;
|
wenzelm@43375
|
21 |
fun end_pos_of (Token ((_, pos), _)) = pos;
|
wenzelm@43375
|
22 |
|
wenzelm@43375
|
23 |
fun kind_of (Token (_, (k, _))) = k;
|
wenzelm@43375
|
24 |
fun content_of (Token (_, (_, x))) = x;
|
wenzelm@43375
|
25 |
|
wenzelm@43375
|
26 |
|
wenzelm@43375
|
27 |
(* diagnostics *)
|
wenzelm@43375
|
28 |
|
wenzelm@43375
|
29 |
val print_kind =
|
wenzelm@43375
|
30 |
fn Keyword => "rail keyword"
|
wenzelm@43375
|
31 |
| Ident => "identifier"
|
wenzelm@43375
|
32 |
| String => "single-quoted string"
|
wenzelm@43379
|
33 |
| Antiq _ => "antiquotation"
|
wenzelm@43375
|
34 |
| EOF => "end-of-file";
|
wenzelm@43375
|
35 |
|
wenzelm@43375
|
36 |
fun print (Token ((pos, _), (k, x))) =
|
wenzelm@43375
|
37 |
(if k = EOF then print_kind k else print_kind k ^ " " ^ quote x) ^
|
wenzelm@43375
|
38 |
Position.str_of pos;
|
wenzelm@43375
|
39 |
|
wenzelm@43375
|
40 |
fun print_keyword x = print_kind Keyword ^ " " ^ quote x;
|
wenzelm@43375
|
41 |
|
wenzelm@43375
|
42 |
|
wenzelm@43375
|
43 |
(* stopper *)
|
wenzelm@43375
|
44 |
|
wenzelm@43375
|
45 |
fun mk_eof pos = Token ((pos, Position.none), (EOF, ""));
|
wenzelm@43375
|
46 |
val eof = mk_eof Position.none;
|
wenzelm@43375
|
47 |
|
wenzelm@43375
|
48 |
fun is_eof (Token (_, (EOF, _))) = true
|
wenzelm@43375
|
49 |
| is_eof _ = false;
|
wenzelm@43375
|
50 |
|
wenzelm@43375
|
51 |
val stopper =
|
wenzelm@43375
|
52 |
Scan.stopper (fn [] => eof | toks => mk_eof (end_pos_of (List.last toks))) is_eof;
|
wenzelm@43375
|
53 |
|
wenzelm@43375
|
54 |
|
wenzelm@43375
|
55 |
(* tokenize *)
|
wenzelm@43375
|
56 |
|
wenzelm@43375
|
57 |
local
|
wenzelm@43375
|
58 |
|
wenzelm@43375
|
59 |
fun token k ss = [Token (Symbol_Pos.range ss, (k, Symbol_Pos.content ss))];
|
wenzelm@43375
|
60 |
|
wenzelm@43375
|
61 |
val scan_space = Scan.many1 (Symbol.is_blank o Symbol_Pos.symbol);
|
wenzelm@43375
|
62 |
|
wenzelm@43375
|
63 |
val scan_keyword =
|
wenzelm@43387
|
64 |
Scan.one (member (op =) ["|", "*", "+", "?", "(", ")", "\\", ";", ":", "@"] o Symbol_Pos.symbol);
|
wenzelm@43375
|
65 |
|
wenzelm@43375
|
66 |
val scan_token =
|
wenzelm@43375
|
67 |
scan_space >> K [] ||
|
wenzelm@43387
|
68 |
Antiquote.scan_antiq >> (fn antiq as (ss, _) => token (Antiq antiq) ss) ||
|
wenzelm@43375
|
69 |
scan_keyword >> (token Keyword o single) ||
|
wenzelm@43375
|
70 |
Lexicon.scan_id >> token Ident ||
|
wenzelm@43387
|
71 |
Symbol_Pos.scan_string_q >> (token String o #1 o #2);
|
wenzelm@43375
|
72 |
|
wenzelm@43377
|
73 |
val scan =
|
wenzelm@43377
|
74 |
(Scan.repeat scan_token >> flat) --|
|
wenzelm@44818
|
75 |
Symbol_Pos.!!! (fn () => "Rail lexical error: bad input")
|
wenzelm@43377
|
76 |
(Scan.ahead (Scan.one Symbol_Pos.is_eof));
|
wenzelm@43377
|
77 |
|
wenzelm@43375
|
78 |
in
|
wenzelm@43375
|
79 |
|
wenzelm@43378
|
80 |
val tokenize = #1 o Scan.error (Scan.finite Symbol_Pos.stopper scan) o Symbol_Pos.explode;
|
wenzelm@43375
|
81 |
|
wenzelm@43375
|
82 |
end;
|
wenzelm@43375
|
83 |
|
wenzelm@43375
|
84 |
|
wenzelm@43375
|
85 |
|
wenzelm@43375
|
86 |
(** parsing **)
|
wenzelm@43375
|
87 |
|
wenzelm@43375
|
88 |
fun !!! scan =
|
wenzelm@43375
|
89 |
let
|
wenzelm@43375
|
90 |
val prefix = "Rail syntax error";
|
wenzelm@43375
|
91 |
|
wenzelm@43375
|
92 |
fun get_pos [] = " (past end-of-file!)"
|
wenzelm@43375
|
93 |
| get_pos (tok :: _) = Position.str_of (pos_of tok);
|
wenzelm@43375
|
94 |
|
wenzelm@44818
|
95 |
fun err (toks, NONE) = (fn () => prefix ^ get_pos toks)
|
wenzelm@43375
|
96 |
| err (toks, SOME msg) =
|
wenzelm@44818
|
97 |
(fn () =>
|
wenzelm@44818
|
98 |
let val s = msg () in
|
wenzelm@44818
|
99 |
if String.isPrefix prefix s then s
|
wenzelm@44818
|
100 |
else prefix ^ get_pos toks ^ ": " ^ s
|
wenzelm@44818
|
101 |
end);
|
wenzelm@43375
|
102 |
in Scan.!! err scan end;
|
wenzelm@43375
|
103 |
|
wenzelm@43375
|
104 |
fun $$$ x =
|
wenzelm@43375
|
105 |
Scan.one (fn tok => kind_of tok = Keyword andalso content_of tok = x) ||
|
wenzelm@43375
|
106 |
Scan.fail_with
|
wenzelm@44818
|
107 |
(fn [] => (fn () => print_keyword x ^ " expected (past end-of-file!)")
|
wenzelm@44818
|
108 |
| tok :: _ => (fn () => print_keyword x ^ " expected,\nbut " ^ print tok ^ " was found"));
|
wenzelm@43375
|
109 |
|
wenzelm@43375
|
110 |
fun enum1 sep scan = scan ::: Scan.repeat ($$$ sep |-- !!! scan);
|
wenzelm@43375
|
111 |
fun enum sep scan = enum1 sep scan || Scan.succeed [];
|
wenzelm@43375
|
112 |
|
wenzelm@43379
|
113 |
val ident = Scan.some (fn tok => if kind_of tok = Ident then SOME (content_of tok) else NONE);
|
wenzelm@43379
|
114 |
val string = Scan.some (fn tok => if kind_of tok = String then SOME (content_of tok) else NONE);
|
wenzelm@43375
|
115 |
|
wenzelm@43379
|
116 |
val antiq = Scan.some (fn tok => (case kind_of tok of Antiq a => SOME a | _ => NONE));
|
wenzelm@43375
|
117 |
|
wenzelm@43375
|
118 |
|
wenzelm@43375
|
119 |
|
wenzelm@43375
|
120 |
(** rail expressions **)
|
wenzelm@43375
|
121 |
|
wenzelm@43375
|
122 |
(* datatype *)
|
wenzelm@43375
|
123 |
|
wenzelm@43375
|
124 |
datatype rails =
|
wenzelm@43375
|
125 |
Cat of int * rail list
|
wenzelm@43375
|
126 |
and rail =
|
wenzelm@43375
|
127 |
Bar of rails list |
|
wenzelm@43375
|
128 |
Plus of rails * rails |
|
wenzelm@43375
|
129 |
Newline of int |
|
wenzelm@43375
|
130 |
Nonterminal of string |
|
wenzelm@43387
|
131 |
Terminal of bool * string |
|
wenzelm@43379
|
132 |
Antiquote of bool * (Symbol_Pos.T list * Position.range);
|
wenzelm@43375
|
133 |
|
wenzelm@43375
|
134 |
fun reverse_cat (Cat (y, rails)) = Cat (y, rev (map reverse rails))
|
wenzelm@43375
|
135 |
and reverse (Bar cats) = Bar (map reverse_cat cats)
|
wenzelm@43375
|
136 |
| reverse (Plus (cat1, cat2)) = Plus (reverse_cat cat1, reverse_cat cat2)
|
wenzelm@43375
|
137 |
| reverse x = x;
|
wenzelm@43375
|
138 |
|
wenzelm@43375
|
139 |
fun cat rails = Cat (0, rails);
|
wenzelm@43375
|
140 |
|
wenzelm@43375
|
141 |
val empty = cat [];
|
wenzelm@43375
|
142 |
fun is_empty (Cat (_, [])) = true | is_empty _ = false;
|
wenzelm@43375
|
143 |
|
wenzelm@43375
|
144 |
fun is_newline (Newline _) = true | is_newline _ = false;
|
wenzelm@43375
|
145 |
|
wenzelm@43375
|
146 |
fun bar [Cat (_, [rail])] = rail
|
wenzelm@43375
|
147 |
| bar cats = Bar cats;
|
wenzelm@43375
|
148 |
|
wenzelm@43375
|
149 |
fun plus cat1 cat2 = Plus (cat1, reverse_cat cat2);
|
wenzelm@43375
|
150 |
|
wenzelm@43375
|
151 |
fun star cat1 cat2 =
|
wenzelm@43375
|
152 |
if is_empty cat2 then plus empty cat1
|
wenzelm@43375
|
153 |
else bar [empty, cat [plus cat1 cat2]];
|
wenzelm@43375
|
154 |
|
wenzelm@43375
|
155 |
fun maybe rail = bar [empty, cat [rail]];
|
wenzelm@43375
|
156 |
|
wenzelm@43375
|
157 |
|
wenzelm@43375
|
158 |
(* read *)
|
wenzelm@43375
|
159 |
|
wenzelm@43375
|
160 |
local
|
wenzelm@43375
|
161 |
|
wenzelm@43387
|
162 |
val at_mode = Scan.option ($$$ "@") >> (fn NONE => false | _ => true);
|
wenzelm@43387
|
163 |
|
wenzelm@43375
|
164 |
fun body x = (enum1 "|" body1 >> bar) x
|
wenzelm@43375
|
165 |
and body0 x = (enum "|" body1 >> bar) x
|
wenzelm@43375
|
166 |
and body1 x =
|
wenzelm@43375
|
167 |
(body2 :|-- (fn a =>
|
wenzelm@43375
|
168 |
$$$ "*" |-- !!! body4e >> (cat o single o star a) ||
|
wenzelm@43375
|
169 |
$$$ "+" |-- !!! body4e >> (cat o single o plus a) ||
|
wenzelm@43375
|
170 |
Scan.succeed a)) x
|
wenzelm@43375
|
171 |
and body2 x = (Scan.repeat1 body3 >> cat) x
|
wenzelm@43375
|
172 |
and body3 x = (body4 :|-- (fn a => $$$ "?" >> K (maybe a) || Scan.succeed a)) x
|
wenzelm@43375
|
173 |
and body4 x =
|
wenzelm@43375
|
174 |
($$$ "(" |-- !!! (body0 --| $$$ ")") ||
|
wenzelm@43375
|
175 |
$$$ "\\" >> K (Newline 0) ||
|
wenzelm@43375
|
176 |
ident >> Nonterminal ||
|
wenzelm@43387
|
177 |
at_mode -- string >> Terminal ||
|
wenzelm@43387
|
178 |
at_mode -- antiq >> Antiquote) x
|
wenzelm@43375
|
179 |
and body4e x = (Scan.option body4 >> (cat o the_list)) x;
|
wenzelm@43375
|
180 |
|
wenzelm@43387
|
181 |
val rule_name = ident >> Antiquote.Text || antiq >> Antiquote.Antiq;
|
wenzelm@43379
|
182 |
val rule = rule_name -- ($$$ ":" |-- !!! body) || body >> pair (Antiquote.Text "");
|
wenzelm@43375
|
183 |
val rules = enum1 ";" (Scan.option rule) >> map_filter I;
|
wenzelm@43375
|
184 |
|
wenzelm@43375
|
185 |
in
|
wenzelm@43375
|
186 |
|
wenzelm@43378
|
187 |
val read =
|
wenzelm@43378
|
188 |
#1 o Scan.error (Scan.finite stopper (rules --| !!! (Scan.ahead (Scan.one is_eof)))) o tokenize;
|
wenzelm@43375
|
189 |
|
wenzelm@43375
|
190 |
end;
|
wenzelm@43375
|
191 |
|
wenzelm@43375
|
192 |
|
wenzelm@43375
|
193 |
(* latex output *)
|
wenzelm@43375
|
194 |
|
wenzelm@43375
|
195 |
local
|
wenzelm@43375
|
196 |
|
wenzelm@43375
|
197 |
fun vertical_range_cat (Cat (_, rails)) y =
|
wenzelm@43375
|
198 |
let val (rails', (_, y')) =
|
wenzelm@43375
|
199 |
fold_map (fn rail => fn (y0, y') =>
|
wenzelm@43375
|
200 |
if is_newline rail then (Newline (y' + 1), (y' + 1, y' + 2))
|
wenzelm@43375
|
201 |
else
|
wenzelm@43375
|
202 |
let val (rail', y0') = vertical_range rail y0;
|
wenzelm@43375
|
203 |
in (rail', (y0, Int.max (y0', y'))) end) rails (y, y + 1)
|
wenzelm@43375
|
204 |
in (Cat (y, rails'), y') end
|
wenzelm@43375
|
205 |
|
wenzelm@43375
|
206 |
and vertical_range (Bar cats) y =
|
wenzelm@43375
|
207 |
let val (cats', y') = fold_map vertical_range_cat cats y
|
wenzelm@43375
|
208 |
in (Bar cats', Int.max (y + 1, y')) end
|
wenzelm@43375
|
209 |
| vertical_range (Plus (cat1, cat2)) y =
|
wenzelm@43375
|
210 |
let val ([cat1', cat2'], y') = fold_map vertical_range_cat [cat1, cat2] y;
|
wenzelm@43375
|
211 |
in (Plus (cat1', cat2'), Int.max (y + 1, y')) end
|
wenzelm@43375
|
212 |
| vertical_range (Newline _) y = (Newline (y + 2), y + 3)
|
wenzelm@43375
|
213 |
| vertical_range atom y = (atom, y + 1);
|
wenzelm@43375
|
214 |
|
wenzelm@43379
|
215 |
fun output_rules state rules =
|
wenzelm@43379
|
216 |
let
|
wenzelm@43379
|
217 |
val output_antiq = Thy_Output.eval_antiq (#1 (Keyword.get_lexicons ())) state;
|
wenzelm@43387
|
218 |
fun output_text b s =
|
wenzelm@43387
|
219 |
Output.output s
|
wenzelm@43387
|
220 |
|> b ? enclose "\\isakeyword{" "}"
|
wenzelm@43387
|
221 |
|> enclose "\\isa{" "}";
|
wenzelm@43375
|
222 |
|
wenzelm@43379
|
223 |
fun output_cat c (Cat (_, rails)) = outputs c rails
|
wenzelm@43379
|
224 |
and outputs c [rail] = output c rail
|
wenzelm@43379
|
225 |
| outputs _ rails = implode (map (output "") rails)
|
wenzelm@43379
|
226 |
and output _ (Bar []) = ""
|
wenzelm@43379
|
227 |
| output c (Bar [cat]) = output_cat c cat
|
wenzelm@43379
|
228 |
| output _ (Bar (cat :: cats)) =
|
wenzelm@43379
|
229 |
"\\rail@bar\n" ^ output_cat "" cat ^
|
wenzelm@43379
|
230 |
implode (map (fn Cat (y, rails) =>
|
wenzelm@43379
|
231 |
"\\rail@nextbar{" ^ string_of_int y ^ "}\n" ^ outputs "" rails) cats) ^
|
wenzelm@43379
|
232 |
"\\rail@endbar\n"
|
wenzelm@43379
|
233 |
| output c (Plus (cat, Cat (y, rails))) =
|
wenzelm@43379
|
234 |
"\\rail@plus\n" ^ output_cat c cat ^
|
wenzelm@43379
|
235 |
"\\rail@nextplus{" ^ string_of_int y ^ "}\n" ^ outputs "c" rails ^
|
wenzelm@43379
|
236 |
"\\rail@endplus\n"
|
wenzelm@43379
|
237 |
| output _ (Newline y) = "\\rail@cr{" ^ string_of_int y ^ "}\n"
|
wenzelm@43387
|
238 |
| output c (Nonterminal s) = "\\rail@" ^ c ^ "nont{" ^ output_text false s ^ "}[]\n"
|
wenzelm@43387
|
239 |
| output c (Terminal (b, s)) = "\\rail@" ^ c ^ "term{" ^ output_text b s ^ "}[]\n"
|
wenzelm@43379
|
240 |
| output c (Antiquote (b, a)) =
|
wenzelm@43379
|
241 |
"\\rail@" ^ c ^ (if b then "term{" else "nont{") ^ output_antiq a ^ "}[]\n";
|
wenzelm@43375
|
242 |
|
wenzelm@43379
|
243 |
fun output_rule (name, rail) =
|
wenzelm@43379
|
244 |
let
|
wenzelm@43379
|
245 |
val (rail', y') = vertical_range rail 0;
|
wenzelm@43379
|
246 |
val out_name =
|
wenzelm@43379
|
247 |
(case name of
|
wenzelm@43534
|
248 |
Antiquote.Text "" => ""
|
wenzelm@43534
|
249 |
| Antiquote.Text s => output_text false s
|
wenzelm@43379
|
250 |
| Antiquote.Antiq a => output_antiq a);
|
wenzelm@43379
|
251 |
in
|
wenzelm@43379
|
252 |
"\\rail@begin{" ^ string_of_int y' ^ "}{" ^ out_name ^ "}\n" ^
|
wenzelm@43379
|
253 |
output "" rail' ^
|
wenzelm@43379
|
254 |
"\\rail@end\n"
|
wenzelm@43379
|
255 |
end;
|
wenzelm@43379
|
256 |
in
|
wenzelm@43379
|
257 |
"\\begin{railoutput}\n" ^
|
wenzelm@43379
|
258 |
implode (map output_rule rules) ^
|
wenzelm@43379
|
259 |
"\\end{railoutput}\n"
|
wenzelm@43375
|
260 |
end;
|
wenzelm@43375
|
261 |
|
wenzelm@43375
|
262 |
in
|
wenzelm@43375
|
263 |
|
wenzelm@43375
|
264 |
val _ =
|
wenzelm@44450
|
265 |
Context.>> (Context.map_theory
|
wenzelm@44450
|
266 |
(Thy_Output.antiquotation (Binding.name "rail")
|
wenzelm@44450
|
267 |
(Scan.lift (Parse.source_position Parse.string))
|
wenzelm@44450
|
268 |
(fn {state, ...} => output_rules state o read)));
|
wenzelm@43375
|
269 |
|
wenzelm@43375
|
270 |
end;
|
wenzelm@43375
|
271 |
|
wenzelm@43375
|
272 |
end;
|
wenzelm@43375
|
273 |
|