wenzelm@12416
|
1 |
(* Title: Pure/General/xml.ML
|
wenzelm@12416
|
2 |
ID: $Id$
|
wenzelm@14728
|
3 |
Author: David Aspinall, Stefan Berghofer and Markus Wenzel
|
wenzelm@12416
|
4 |
License: GPL (GNU GENERAL PUBLIC LICENSE)
|
wenzelm@12416
|
5 |
|
wenzelm@14969
|
6 |
Basic support for XML.
|
wenzelm@12416
|
7 |
*)
|
wenzelm@12416
|
8 |
|
wenzelm@12416
|
9 |
signature XML =
|
wenzelm@12416
|
10 |
sig
|
wenzelm@14969
|
11 |
val header: string
|
wenzelm@14969
|
12 |
val text: string -> string
|
wenzelm@14969
|
13 |
val cdata: string -> string
|
wenzelm@14969
|
14 |
val element: string -> (string * string) list -> string list -> string
|
berghofe@13729
|
15 |
datatype tree =
|
berghofe@13729
|
16 |
Elem of string * (string * string) list * tree list
|
berghofe@13729
|
17 |
| Text of string
|
berghofe@13729
|
18 |
val string_of_tree: tree -> string
|
berghofe@14185
|
19 |
val parse_content: string list -> tree list * string list
|
berghofe@14185
|
20 |
val parse_elem: string list -> tree * string list
|
berghofe@14185
|
21 |
val parse_document: string list -> (string option * tree) * string list
|
wenzelm@14969
|
22 |
val tree_of_string: string -> tree
|
wenzelm@12416
|
23 |
end;
|
wenzelm@12416
|
24 |
|
wenzelm@12416
|
25 |
structure XML: XML =
|
wenzelm@12416
|
26 |
struct
|
wenzelm@12416
|
27 |
|
wenzelm@14969
|
28 |
(** string based representation (small scale) **)
|
wenzelm@14969
|
29 |
|
wenzelm@14969
|
30 |
val header = "<?xml version=\"1.0\"?>\n";
|
wenzelm@14969
|
31 |
|
wenzelm@14969
|
32 |
|
wenzelm@14969
|
33 |
(* text and character data *)
|
wenzelm@14969
|
34 |
|
wenzelm@14969
|
35 |
fun decode "<" = "<"
|
wenzelm@14969
|
36 |
| decode ">" = ">"
|
wenzelm@14969
|
37 |
| decode "&" = "&"
|
wenzelm@14969
|
38 |
| decode "'" = "'"
|
wenzelm@14969
|
39 |
| decode """ = "\""
|
wenzelm@14969
|
40 |
| decode c = c;
|
wenzelm@12416
|
41 |
|
wenzelm@12416
|
42 |
fun encode "<" = "<"
|
wenzelm@12416
|
43 |
| encode ">" = ">"
|
wenzelm@12416
|
44 |
| encode "&" = "&"
|
wenzelm@12416
|
45 |
| encode "'" = "'"
|
wenzelm@12416
|
46 |
| encode "\"" = """
|
wenzelm@12416
|
47 |
| encode c = c;
|
wenzelm@12416
|
48 |
|
wenzelm@14928
|
49 |
val text = Library.translate_string encode;
|
wenzelm@12416
|
50 |
|
wenzelm@14969
|
51 |
val cdata = enclose "<![CDATA[" "]]>";
|
wenzelm@12416
|
52 |
|
wenzelm@14728
|
53 |
|
wenzelm@12416
|
54 |
(* elements *)
|
wenzelm@12416
|
55 |
|
wenzelm@14728
|
56 |
fun attribute (a, x) = a ^ " = \"" ^ text x ^ "\"";
|
wenzelm@12416
|
57 |
|
wenzelm@12416
|
58 |
fun element name atts cs =
|
wenzelm@12416
|
59 |
let val elem = space_implode " " (name :: map attribute atts) in
|
wenzelm@12416
|
60 |
if null cs then enclose "<" "/>" elem
|
wenzelm@12416
|
61 |
else enclose "<" ">" elem ^ implode cs ^ enclose "</" ">" name
|
wenzelm@12416
|
62 |
end;
|
wenzelm@12416
|
63 |
|
wenzelm@14728
|
64 |
|
wenzelm@12416
|
65 |
|
wenzelm@14969
|
66 |
(** explicit XML trees **)
|
berghofe@13729
|
67 |
|
wenzelm@14969
|
68 |
datatype tree =
|
wenzelm@14969
|
69 |
Elem of string * (string * string) list * tree list
|
wenzelm@14969
|
70 |
| Text of string;
|
wenzelm@14969
|
71 |
|
wenzelm@14969
|
72 |
fun string_of_tree tree =
|
wenzelm@14969
|
73 |
let
|
wenzelm@14969
|
74 |
fun string_of (Elem (name, atts, ts)) buf =
|
wenzelm@14969
|
75 |
let val buf' =
|
wenzelm@14969
|
76 |
buf |> Buffer.add "<"
|
wenzelm@14969
|
77 |
|> fold Buffer.add (separate " " (name :: map attribute atts))
|
wenzelm@14969
|
78 |
in
|
wenzelm@14969
|
79 |
if null ts then
|
wenzelm@14969
|
80 |
buf' |> Buffer.add "/>"
|
wenzelm@14969
|
81 |
else
|
wenzelm@14969
|
82 |
buf' |> Buffer.add ">"
|
wenzelm@14969
|
83 |
|> fold string_of ts
|
wenzelm@14969
|
84 |
|> Buffer.add "</" |> Buffer.add name |> Buffer.add ">"
|
wenzelm@14969
|
85 |
end
|
wenzelm@14969
|
86 |
| string_of (Text s) buf = Buffer.add (text s) buf;
|
wenzelm@14969
|
87 |
in Buffer.content (string_of tree Buffer.empty) end;
|
wenzelm@14969
|
88 |
|
wenzelm@14969
|
89 |
|
wenzelm@14969
|
90 |
|
wenzelm@14969
|
91 |
(** XML parsing **)
|
berghofe@13729
|
92 |
|
wenzelm@14728
|
93 |
fun err s (xs, _) =
|
wenzelm@14728
|
94 |
"XML parsing error: " ^ s ^ "\nfound: " ^ quote (Symbol.beginning 100 xs);
|
berghofe@13729
|
95 |
|
wenzelm@14728
|
96 |
val scan_whspc = Scan.any Symbol.is_blank;
|
berghofe@13729
|
97 |
|
berghofe@13729
|
98 |
val scan_special = $$ "&" ^^ Symbol.scan_id ^^ $$ ";" >> decode;
|
berghofe@13729
|
99 |
|
berghofe@14185
|
100 |
val parse_chars = Scan.repeat1 (Scan.unless (scan_whspc -- $$ "<")
|
berghofe@14185
|
101 |
(scan_special || Scan.one Symbol.not_eof)) >> implode;
|
berghofe@13729
|
102 |
|
wenzelm@14910
|
103 |
val parse_cdata = Scan.this_string "<![CDATA[" |--
|
wenzelm@14910
|
104 |
(Scan.repeat (Scan.unless (Scan.this_string "]]>") (Scan.one Symbol.not_eof)) >>
|
wenzelm@14910
|
105 |
implode) --| Scan.this_string "]]>";
|
berghofe@13729
|
106 |
|
berghofe@14185
|
107 |
val parse_att =
|
berghofe@14865
|
108 |
Symbol.scan_id --| scan_whspc --| $$ "=" --| scan_whspc --
|
berghofe@14865
|
109 |
(($$ "\"" || $$ "'") :-- (fn s => (Scan.repeat (Scan.unless ($$ s)
|
berghofe@14865
|
110 |
(scan_special || Scan.one Symbol.not_eof)) >> implode) --| $$ s) >> snd);
|
aspinall@14863
|
111 |
|
wenzelm@14910
|
112 |
val parse_comment = Scan.this_string "<!--" --
|
wenzelm@14910
|
113 |
Scan.repeat (Scan.unless (Scan.this_string "-->") (Scan.one Symbol.not_eof)) --
|
wenzelm@14910
|
114 |
Scan.this_string "-->";
|
berghofe@14185
|
115 |
|
wenzelm@14910
|
116 |
val parse_pi = Scan.this_string "<?" |--
|
wenzelm@14910
|
117 |
Scan.repeat (Scan.unless (Scan.this_string "?>") (Scan.one Symbol.not_eof)) --|
|
wenzelm@14910
|
118 |
Scan.this_string "?>";
|
berghofe@14185
|
119 |
|
berghofe@14185
|
120 |
fun parse_content xs =
|
berghofe@14185
|
121 |
((Scan.optional (scan_whspc |-- parse_chars >> (single o Text)) [] --
|
berghofe@14185
|
122 |
(Scan.repeat (scan_whspc |--
|
berghofe@14185
|
123 |
( parse_elem >> single
|
berghofe@14185
|
124 |
|| parse_cdata >> (single o Text)
|
berghofe@14185
|
125 |
|| parse_pi >> K []
|
berghofe@14185
|
126 |
|| parse_comment >> K []) --
|
berghofe@14185
|
127 |
Scan.optional (scan_whspc |-- parse_chars >> (single o Text)) []
|
berghofe@14185
|
128 |
>> op @) >> flat) >> op @) --| scan_whspc) xs
|
berghofe@14185
|
129 |
|
berghofe@14185
|
130 |
and parse_elem xs =
|
berghofe@14185
|
131 |
($$ "<" |-- Symbol.scan_id --
|
berghofe@14865
|
132 |
Scan.repeat (scan_whspc |-- parse_att) --| scan_whspc :-- (fn (s, _) =>
|
berghofe@14185
|
133 |
!! (err "Expected > or />")
|
wenzelm@14910
|
134 |
(Scan.this_string "/>" >> K []
|
berghofe@14185
|
135 |
|| $$ ">" |-- parse_content --|
|
berghofe@14185
|
136 |
!! (err ("Expected </" ^ s ^ ">"))
|
wenzelm@14910
|
137 |
(Scan.this_string ("</" ^ s) --| scan_whspc --| $$ ">"))) >>
|
berghofe@13729
|
138 |
(fn ((s, atts), ts) => Elem (s, atts, ts))) xs;
|
berghofe@13729
|
139 |
|
berghofe@14185
|
140 |
val parse_document =
|
wenzelm@14910
|
141 |
Scan.option (Scan.this_string "<!DOCTYPE" -- scan_whspc |--
|
berghofe@14185
|
142 |
(Scan.repeat (Scan.unless ($$ ">")
|
berghofe@14185
|
143 |
(Scan.one Symbol.not_eof)) >> implode) --| $$ ">" --| scan_whspc) --
|
berghofe@14185
|
144 |
parse_elem;
|
berghofe@14185
|
145 |
|
berghofe@13729
|
146 |
fun tree_of_string s =
|
berghofe@14185
|
147 |
(case Scan.finite Symbol.stopper (Scan.error (!! (err "Malformed element")
|
berghofe@14185
|
148 |
(scan_whspc |-- parse_elem --| scan_whspc))) (Symbol.explode s) of
|
wenzelm@14728
|
149 |
(x, []) => x
|
wenzelm@14728
|
150 |
| (_, ys) => error ("XML parsing error: Unprocessed input\n" ^ Symbol.beginning 100 ys));
|
berghofe@13729
|
151 |
|
wenzelm@12416
|
152 |
end;
|