defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
1 (* Title: Pure/Thy/rail.ML
2 Author: Michael Kerscher, TU München
5 Railroad diagrams in LaTeX.
8 structure Rail: sig end =
11 (** lexical syntax **)
16 Keyword | Ident | String | Antiq of Symbol_Pos.T list * Position.range | EOF;
18 datatype token = Token of Position.range * (kind * string);
20 fun pos_of (Token ((pos, _), _)) = pos;
21 fun end_pos_of (Token ((_, pos), _)) = pos;
23 fun kind_of (Token (_, (k, _))) = k;
24 fun content_of (Token (_, (_, x))) = x;
30 fn Keyword => "rail keyword"
31 | Ident => "identifier"
32 | String => "single-quoted string"
33 | Antiq _ => "antiquotation"
34 | EOF => "end-of-file";
36 fun print (Token ((pos, _), (k, x))) =
37 (if k = EOF then print_kind k else print_kind k ^ " " ^ quote x) ^
40 fun print_keyword x = print_kind Keyword ^ " " ^ quote x;
45 fun mk_eof pos = Token ((pos, Position.none), (EOF, ""));
46 val eof = mk_eof Position.none;
48 fun is_eof (Token (_, (EOF, _))) = true
52 Scan.stopper (fn [] => eof | toks => mk_eof (end_pos_of (List.last toks))) is_eof;
59 fun token k ss = [Token (Symbol_Pos.range ss, (k, Symbol_Pos.content ss))];
61 val scan_space = Scan.many1 (Symbol.is_blank o Symbol_Pos.symbol);
64 Scan.one (member (op =) ["|", "*", "+", "?", "(", ")", "\\", ";", ":", "@"] o Symbol_Pos.symbol);
68 Antiquote.scan_antiq >> (fn antiq as (ss, _) => token (Antiq antiq) ss) ||
69 scan_keyword >> (token Keyword o single) ||
70 Lexicon.scan_id >> token Ident ||
71 Symbol_Pos.scan_string_q >> (token String o #1 o #2);
74 (Scan.repeat scan_token >> flat) --|
75 Symbol_Pos.!!! (fn () => "Rail lexical error: bad input")
76 (Scan.ahead (Scan.one Symbol_Pos.is_eof));
80 val tokenize = #1 o Scan.error (Scan.finite Symbol_Pos.stopper scan) o Symbol_Pos.explode;
90 val prefix = "Rail syntax error";
92 fun get_pos [] = " (past end-of-file!)"
93 | get_pos (tok :: _) = Position.str_of (pos_of tok);
95 fun err (toks, NONE) = (fn () => prefix ^ get_pos toks)
96 | err (toks, SOME msg) =
99 if String.isPrefix prefix s then s
100 else prefix ^ get_pos toks ^ ": " ^ s
102 in Scan.!! err scan end;
105 Scan.one (fn tok => kind_of tok = Keyword andalso content_of tok = x) ||
107 (fn [] => (fn () => print_keyword x ^ " expected (past end-of-file!)")
108 | tok :: _ => (fn () => print_keyword x ^ " expected,\nbut " ^ print tok ^ " was found"));
110 fun enum1 sep scan = scan ::: Scan.repeat ($$$ sep |-- !!! scan);
111 fun enum sep scan = enum1 sep scan || Scan.succeed [];
113 val ident = Scan.some (fn tok => if kind_of tok = Ident then SOME (content_of tok) else NONE);
114 val string = Scan.some (fn tok => if kind_of tok = String then SOME (content_of tok) else NONE);
116 val antiq = Scan.some (fn tok => (case kind_of tok of Antiq a => SOME a | _ => NONE));
120 (** rail expressions **)
125 Cat of int * rail list
128 Plus of rails * rails |
130 Nonterminal of string |
131 Terminal of bool * string |
132 Antiquote of bool * (Symbol_Pos.T list * Position.range);
134 fun reverse_cat (Cat (y, rails)) = Cat (y, rev (map reverse rails))
135 and reverse (Bar cats) = Bar (map reverse_cat cats)
136 | reverse (Plus (cat1, cat2)) = Plus (reverse_cat cat1, reverse_cat cat2)
139 fun cat rails = Cat (0, rails);
142 fun is_empty (Cat (_, [])) = true | is_empty _ = false;
144 fun is_newline (Newline _) = true | is_newline _ = false;
146 fun bar [Cat (_, [rail])] = rail
147 | bar cats = Bar cats;
149 fun plus cat1 cat2 = Plus (cat1, reverse_cat cat2);
152 if is_empty cat2 then plus empty cat1
153 else bar [empty, cat [plus cat1 cat2]];
155 fun maybe rail = bar [empty, cat [rail]];
162 val at_mode = Scan.option ($$$ "@") >> (fn NONE => false | _ => true);
164 fun body x = (enum1 "|" body1 >> bar) x
165 and body0 x = (enum "|" body1 >> bar) x
168 $$$ "*" |-- !!! body4e >> (cat o single o star a) ||
169 $$$ "+" |-- !!! body4e >> (cat o single o plus a) ||
171 and body2 x = (Scan.repeat1 body3 >> cat) x
172 and body3 x = (body4 :|-- (fn a => $$$ "?" >> K (maybe a) || Scan.succeed a)) x
174 ($$$ "(" |-- !!! (body0 --| $$$ ")") ||
175 $$$ "\\" >> K (Newline 0) ||
176 ident >> Nonterminal ||
177 at_mode -- string >> Terminal ||
178 at_mode -- antiq >> Antiquote) x
179 and body4e x = (Scan.option body4 >> (cat o the_list)) x;
181 val rule_name = ident >> Antiquote.Text || antiq >> Antiquote.Antiq;
182 val rule = rule_name -- ($$$ ":" |-- !!! body) || body >> pair (Antiquote.Text "");
183 val rules = enum1 ";" (Scan.option rule) >> map_filter I;
188 #1 o Scan.error (Scan.finite stopper (rules --| !!! (Scan.ahead (Scan.one is_eof)))) o tokenize;
197 fun vertical_range_cat (Cat (_, rails)) y =
198 let val (rails', (_, y')) =
199 fold_map (fn rail => fn (y0, y') =>
200 if is_newline rail then (Newline (y' + 1), (y' + 1, y' + 2))
202 let val (rail', y0') = vertical_range rail y0;
203 in (rail', (y0, Int.max (y0', y'))) end) rails (y, y + 1)
204 in (Cat (y, rails'), y') end
206 and vertical_range (Bar cats) y =
207 let val (cats', y') = fold_map vertical_range_cat cats y
208 in (Bar cats', Int.max (y + 1, y')) end
209 | vertical_range (Plus (cat1, cat2)) y =
210 let val ([cat1', cat2'], y') = fold_map vertical_range_cat [cat1, cat2] y;
211 in (Plus (cat1', cat2'), Int.max (y + 1, y')) end
212 | vertical_range (Newline _) y = (Newline (y + 2), y + 3)
213 | vertical_range atom y = (atom, y + 1);
215 fun output_rules state rules =
217 val output_antiq = Thy_Output.eval_antiq (#1 (Keyword.get_lexicons ())) state;
218 fun output_text b s =
220 |> b ? enclose "\\isakeyword{" "}"
221 |> enclose "\\isa{" "}";
223 fun output_cat c (Cat (_, rails)) = outputs c rails
224 and outputs c [rail] = output c rail
225 | outputs _ rails = implode (map (output "") rails)
226 and output _ (Bar []) = ""
227 | output c (Bar [cat]) = output_cat c cat
228 | output _ (Bar (cat :: cats)) =
229 "\\rail@bar\n" ^ output_cat "" cat ^
230 implode (map (fn Cat (y, rails) =>
231 "\\rail@nextbar{" ^ string_of_int y ^ "}\n" ^ outputs "" rails) cats) ^
233 | output c (Plus (cat, Cat (y, rails))) =
234 "\\rail@plus\n" ^ output_cat c cat ^
235 "\\rail@nextplus{" ^ string_of_int y ^ "}\n" ^ outputs "c" rails ^
237 | output _ (Newline y) = "\\rail@cr{" ^ string_of_int y ^ "}\n"
238 | output c (Nonterminal s) = "\\rail@" ^ c ^ "nont{" ^ output_text false s ^ "}[]\n"
239 | output c (Terminal (b, s)) = "\\rail@" ^ c ^ "term{" ^ output_text b s ^ "}[]\n"
240 | output c (Antiquote (b, a)) =
241 "\\rail@" ^ c ^ (if b then "term{" else "nont{") ^ output_antiq a ^ "}[]\n";
243 fun output_rule (name, rail) =
245 val (rail', y') = vertical_range rail 0;
248 Antiquote.Text "" => ""
249 | Antiquote.Text s => output_text false s
250 | Antiquote.Antiq a => output_antiq a);
252 "\\rail@begin{" ^ string_of_int y' ^ "}{" ^ out_name ^ "}\n" ^
257 "\\begin{railoutput}\n" ^
258 implode (map output_rule rules) ^
259 "\\end{railoutput}\n"
265 Context.>> (Context.map_theory
266 (Thy_Output.antiquotation (Binding.name "rail")
267 (Scan.lift (Parse.source_position Parse.string))
268 (fn {state, ...} => output_rules state o read)));