1 (* Title: Pure/General/yxml.ML |
|
2 Author: Makarius |
|
3 |
|
4 Efficient text representation of XML trees using extra characters X |
|
5 and Y -- no escaping, may nest marked text verbatim. |
|
6 |
|
7 Markup <elem att="val" ...>...body...</elem> is encoded as: |
|
8 |
|
9 X Y name Y att=val ... X |
|
10 ... |
|
11 body |
|
12 ... |
|
13 X Y X |
|
14 *) |
|
15 |
|
16 signature YXML = |
|
17 sig |
|
18 val X: Symbol.symbol |
|
19 val Y: Symbol.symbol |
|
20 val embed_controls: string -> string |
|
21 val detect: string -> bool |
|
22 val output_markup: Markup.T -> string * string |
|
23 val element: string -> XML.attributes -> string list -> string |
|
24 val string_of_body: XML.body -> string |
|
25 val string_of: XML.tree -> string |
|
26 val parse_body: string -> XML.body |
|
27 val parse: string -> XML.tree |
|
28 end; |
|
29 |
|
30 structure YXML: YXML = |
|
31 struct |
|
32 |
|
33 (** string representation **) |
|
34 |
|
35 (* idempotent recoding of certain low ASCII control characters *) |
|
36 |
|
37 fun pseudo_utf8 c = |
|
38 if Symbol.is_ascii_control c |
|
39 then chr 192 ^ chr (128 + ord c) |
|
40 else c; |
|
41 |
|
42 fun embed_controls str = |
|
43 if exists_string Symbol.is_ascii_control str |
|
44 then translate_string pseudo_utf8 str |
|
45 else str; |
|
46 |
|
47 |
|
48 (* markers *) |
|
49 |
|
50 val X = chr 5; |
|
51 val Y = chr 6; |
|
52 val XY = X ^ Y; |
|
53 val XYX = XY ^ X; |
|
54 |
|
55 val detect = exists_string (fn s => s = X orelse s = Y); |
|
56 |
|
57 |
|
58 (* output *) |
|
59 |
|
60 fun output_markup (markup as (name, atts)) = |
|
61 if Markup.is_empty markup then Markup.no_output |
|
62 else (XY ^ name ^ implode (map (fn (a, x) => Y ^ a ^ "=" ^ x) atts) ^ X, XYX); |
|
63 |
|
64 fun element name atts body = |
|
65 let val (pre, post) = output_markup (name, atts) |
|
66 in pre ^ implode body ^ post end; |
|
67 |
|
68 fun string_of_body body = |
|
69 let |
|
70 fun attrib (a, x) = |
|
71 Buffer.add Y #> |
|
72 Buffer.add a #> Buffer.add "=" #> Buffer.add x; |
|
73 fun tree (XML.Elem ((name, atts), ts)) = |
|
74 Buffer.add XY #> Buffer.add name #> fold attrib atts #> Buffer.add X #> |
|
75 trees ts #> |
|
76 Buffer.add XYX |
|
77 | tree (XML.Text s) = Buffer.add s |
|
78 and trees ts = fold tree ts; |
|
79 in Buffer.empty |> trees body |> Buffer.content end; |
|
80 |
|
81 val string_of = string_of_body o single; |
|
82 |
|
83 |
|
84 |
|
85 (** efficient YXML parsing **) |
|
86 |
|
87 local |
|
88 |
|
89 (* splitting *) |
|
90 |
|
91 fun is_char s c = ord s = Char.ord c; |
|
92 |
|
93 val split_string = |
|
94 Substring.full #> |
|
95 Substring.tokens (is_char X) #> |
|
96 map (Substring.fields (is_char Y) #> map Substring.string); |
|
97 |
|
98 |
|
99 (* structural errors *) |
|
100 |
|
101 fun err msg = raise Fail ("Malformed YXML encoding: " ^ msg); |
|
102 fun err_attribute () = err "bad attribute"; |
|
103 fun err_element () = err "bad element"; |
|
104 fun err_unbalanced "" = err "unbalanced element" |
|
105 | err_unbalanced name = err ("unbalanced element " ^ quote name); |
|
106 |
|
107 |
|
108 (* stack operations *) |
|
109 |
|
110 fun add x ((elem, body) :: pending) = (elem, x :: body) :: pending; |
|
111 |
|
112 fun push "" _ _ = err_element () |
|
113 | push name atts pending = ((name, atts), []) :: pending; |
|
114 |
|
115 fun pop ((("", _), _) :: _) = err_unbalanced "" |
|
116 | pop ((markup, body) :: pending) = add (XML.Elem (markup, rev body)) pending; |
|
117 |
|
118 |
|
119 (* parsing *) |
|
120 |
|
121 fun parse_attrib s = |
|
122 (case first_field "=" s of |
|
123 NONE => err_attribute () |
|
124 | SOME ("", _) => err_attribute () |
|
125 | SOME att => att); |
|
126 |
|
127 fun parse_chunk ["", ""] = pop |
|
128 | parse_chunk ("" :: name :: atts) = push name (map parse_attrib atts) |
|
129 | parse_chunk txts = fold (add o XML.Text) txts; |
|
130 |
|
131 in |
|
132 |
|
133 fun parse_body source = |
|
134 (case fold parse_chunk (split_string source) [(("", []), [])] of |
|
135 [(("", _), result)] => rev result |
|
136 | ((name, _), _) :: _ => err_unbalanced name); |
|
137 |
|
138 fun parse source = |
|
139 (case parse_body source of |
|
140 [result] => result |
|
141 | [] => XML.Text "" |
|
142 | _ => err "multiple results"); |
|
143 |
|
144 end; |
|
145 |
|
146 end; |
|
147 |
|