src/Pure/PIDE/xml.ML
author wenzelm
Mon, 28 Nov 2011 22:18:19 +0100
changeset 45667 546d78f0d81f
parent 45155 3216d65d8f34
child 46837 5bdd68f380b3
permissions -rw-r--r--
explicit indication of modules for independent Scala library;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
44698
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
     1
(*  Title:      Pure/PIDE/xml.ML
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
     2
    Author:     David Aspinall
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
     3
    Author:     Stefan Berghofer
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
     4
    Author:     Makarius
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
     5
44698
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
     6
Untyped XML trees and basic data representation.
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
     7
*)
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
     8
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
     9
signature XML_DATA_OPS =
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    10
sig
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    11
  type 'a A
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    12
  type 'a T
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    13
  type 'a V
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    14
  val int_atom: int A
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    15
  val bool_atom: bool A
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    16
  val unit_atom: unit A
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    17
  val properties: Properties.T T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    18
  val string: string T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    19
  val int: int T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    20
  val bool: bool T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    21
  val unit: unit T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    22
  val pair: 'a T -> 'b T -> ('a * 'b) T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    23
  val triple: 'a T -> 'b T -> 'c T -> ('a * 'b * 'c) T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    24
  val list: 'a T -> 'a list T
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    25
  val option: 'a T -> 'a option T
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    26
  val variant: 'a V list -> 'a T
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    27
end;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    28
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    29
signature XML =
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    30
sig
28017
4919bd124a58 type Properties.T;
wenzelm
parents: 27884
diff changeset
    31
  type attributes = Properties.T
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    32
  datatype tree =
38228
ada3ab6b9085 simplified type XML.tree: embed Markup.T directly, avoid slightly odd triple;
wenzelm
parents: 31469
diff changeset
    33
      Elem of Markup.T * tree list
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    34
    | Text of string
38266
492d377ecfe2 type XML.body as basic data representation language;
wenzelm
parents: 38228
diff changeset
    35
  type body = tree list
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    36
  val add_content: tree -> Buffer.T -> Buffer.T
39555
ccb223a4d49c added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents: 38474
diff changeset
    37
  val content_of: body -> string
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    38
  val header: string
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    39
  val text: string -> string
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    40
  val element: string -> attributes -> string list -> string
40131
7cbebd636e79 explicitly qualify type Output.output, which is a slightly odd internal feature;
wenzelm
parents: 39555
diff changeset
    41
  val output_markup: Markup.T -> Output.output * Output.output
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
    42
  val string_of: tree -> string
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
    43
  val pretty: int -> tree -> Pretty.T
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    44
  val output: tree -> TextIO.outstream -> unit
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
    45
  val parse_comments: string list -> unit * string list
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    46
  val parse_string : string -> string option
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    47
  val parse_element: string list -> tree * string list
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
    48
  val parse_document: string list -> tree * string list
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
    49
  val parse: string -> tree
44808
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
    50
  val cache: unit -> tree -> tree
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    51
  exception XML_ATOM of string
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
    52
  exception XML_BODY of body
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    53
  structure Encode: XML_DATA_OPS
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
    54
  structure Decode: XML_DATA_OPS
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    55
end;
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    56
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    57
structure XML: XML =
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    58
struct
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    59
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    60
(** XML trees **)
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    61
28017
4919bd124a58 type Properties.T;
wenzelm
parents: 27884
diff changeset
    62
type attributes = Properties.T;
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    63
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    64
datatype tree =
38228
ada3ab6b9085 simplified type XML.tree: embed Markup.T directly, avoid slightly odd triple;
wenzelm
parents: 31469
diff changeset
    65
    Elem of Markup.T * tree list
28033
f03b5856f286 removed obsolete XML.Output workaround;
wenzelm
parents: 28017
diff changeset
    66
  | Text of string;
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    67
38266
492d377ecfe2 type XML.body as basic data representation language;
wenzelm
parents: 38228
diff changeset
    68
type body = tree list;
492d377ecfe2 type XML.body as basic data representation language;
wenzelm
parents: 38228
diff changeset
    69
38228
ada3ab6b9085 simplified type XML.tree: embed Markup.T directly, avoid slightly odd triple;
wenzelm
parents: 31469
diff changeset
    70
fun add_content (Elem (_, ts)) = fold add_content ts
28033
f03b5856f286 removed obsolete XML.Output workaround;
wenzelm
parents: 28017
diff changeset
    71
  | add_content (Text s) = Buffer.add s;
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    72
39555
ccb223a4d49c added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents: 38474
diff changeset
    73
fun content_of body = Buffer.empty |> fold add_content body |> Buffer.content;
ccb223a4d49c added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents: 38474
diff changeset
    74
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    75
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    76
26525
14a56f013469 added detect;
wenzelm
parents: 25838
diff changeset
    77
(** string representation **)
14a56f013469 added detect;
wenzelm
parents: 25838
diff changeset
    78
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    79
val header = "<?xml version=\"1.0\"?>\n";
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    80
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    81
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
    82
(* escaped text *)
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    83
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    84
fun decode "&lt;" = "<"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    85
  | decode "&gt;" = ">"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    86
  | decode "&amp;" = "&"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    87
  | decode "&apos;" = "'"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    88
  | decode "&quot;" = "\""
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    89
  | decode c = c;
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    90
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    91
fun encode "<" = "&lt;"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    92
  | encode ">" = "&gt;"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    93
  | encode "&" = "&amp;"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    94
  | encode "'" = "&apos;"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    95
  | encode "\"" = "&quot;"
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    96
  | encode c = c;
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    97
25838
00b2a1b2c4e9 removed unused text_charref, cdata;
wenzelm
parents: 24584
diff changeset
    98
val text = translate_string encode;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
    99
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   100
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   101
(* elements *)
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   102
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   103
fun elem name atts =
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   104
  space_implode " " (name :: map (fn (a, x) => a ^ "=\"" ^ text x ^ "\"") atts);
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   105
26525
14a56f013469 added detect;
wenzelm
parents: 25838
diff changeset
   106
fun element name atts body =
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   107
  let val b = implode body in
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   108
    if b = "" then enclose "<" "/>" (elem name atts)
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   109
    else enclose "<" ">" (elem name atts) ^ b ^ enclose "</" ">" name
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   110
  end;
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   111
27884
10c927e4abf5 output_markup: check Markup.is_none;
wenzelm
parents: 26984
diff changeset
   112
fun output_markup (markup as (name, atts)) =
38474
e498dc2eb576 uniform Markup.empty/Markup.Empty in ML and Scala;
wenzelm
parents: 38266
diff changeset
   113
  if Markup.is_empty markup then Markup.no_output
27884
10c927e4abf5 output_markup: check Markup.is_none;
wenzelm
parents: 26984
diff changeset
   114
  else (enclose "<" ">" (elem name atts), enclose "</" ">" name);
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   115
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   116
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   117
(* output *)
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   118
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   119
fun buffer_of depth tree =
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   120
  let
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   121
    fun traverse _ (Elem ((name, atts), [])) =
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   122
          Buffer.add "<" #> Buffer.add (elem name atts) #> Buffer.add "/>"
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   123
      | traverse d (Elem ((name, atts), ts)) =
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   124
          Buffer.add "<" #> Buffer.add (elem name atts) #> Buffer.add ">" #>
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   125
          traverse_body d ts #>
26525
14a56f013469 added detect;
wenzelm
parents: 25838
diff changeset
   126
          Buffer.add "</" #> Buffer.add name #> Buffer.add ">"
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   127
      | traverse _ (Text s) = Buffer.add (text s)
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   128
    and traverse_body 0 _ = Buffer.add "..."
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   129
      | traverse_body d ts = fold (traverse (d - 1)) ts;
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   130
  in Buffer.empty |> traverse depth tree end;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   131
43791
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   132
val string_of = Buffer.content o buffer_of ~1;
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   133
val output = Buffer.output o buffer_of ~1;
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   134
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   135
fun pretty depth tree =
5e9a1d71f94d XML.pretty with depth limit;
wenzelm
parents: 43783
diff changeset
   136
  Pretty.str (Buffer.content (buffer_of (Int.max (0, depth)) tree));
25838
00b2a1b2c4e9 removed unused text_charref, cdata;
wenzelm
parents: 24584
diff changeset
   137
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   138
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   139
44698
0385292321a0 moved XML/YXML to src/Pure/PIDE;
wenzelm
parents: 43949
diff changeset
   140
(** XML parsing **)
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   141
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   142
local
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   143
43947
9b00f09f7721 defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents: 43844
diff changeset
   144
fun err msg (xs, _) =
9b00f09f7721 defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents: 43844
diff changeset
   145
  fn () => "XML parsing error: " ^ msg () ^ "\nfound: " ^ quote (Symbol.beginning 100 xs);
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   146
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   147
fun ignored _ = [];
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   148
45155
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   149
fun name_start_char c = Symbol.is_ascii_letter c orelse c = ":" orelse c = "_";
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   150
fun name_char c = name_start_char c orelse Symbol.is_ascii_digit c orelse c = "-" orelse c = ".";
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   151
val parse_name = Scan.one name_start_char ::: Scan.many name_char;
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   152
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   153
val blanks = Scan.many Symbol.is_blank;
45155
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   154
val special = $$ "&" ^^ (parse_name >> implode) ^^ $$ ";" >> decode;
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   155
val regular = Scan.one Symbol.is_regular;
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   156
fun regular_except x = Scan.one (fn c => Symbol.is_regular c andalso c <> x);
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   157
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   158
val parse_chars = Scan.repeat1 (special || regular_except "<") >> implode;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   159
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   160
val parse_cdata =
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   161
  Scan.this_string "<![CDATA[" |--
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   162
  (Scan.repeat (Scan.unless (Scan.this_string "]]>") regular) >> implode) --|
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   163
  Scan.this_string "]]>";
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   164
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   165
val parse_att =
45155
3216d65d8f34 slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents: 44809
diff changeset
   166
  ((parse_name >> implode) --| (blanks -- $$ "=" -- blanks)) --
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   167
  (($$ "\"" || $$ "'") :|-- (fn s =>
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   168
    (Scan.repeat (special || regular_except s) >> implode) --| $$ s));
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   169
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   170
val parse_comment =
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   171
  Scan.this_string "<!--" --
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   172
  Scan.repeat (Scan.unless (Scan.this_string "-->") regular) --
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   173
  Scan.this_string "-->" >> ignored;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   174
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   175
val parse_processing_instruction =
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   176
  Scan.this_string "<?" --
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   177
  Scan.repeat (Scan.unless (Scan.this_string "?>") regular) --
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   178
  Scan.this_string "?>" >> ignored;
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   179
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   180
val parse_doctype =
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   181
  Scan.this_string "<!DOCTYPE" --
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   182
  Scan.repeat (Scan.unless ($$ ">") regular) --
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   183
  $$ ">" >> ignored;
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   184
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   185
val parse_misc =
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   186
  Scan.one Symbol.is_blank >> ignored ||
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   187
  parse_processing_instruction ||
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   188
  parse_comment;
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   189
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   190
val parse_optional_text =
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   191
  Scan.optional (parse_chars >> (single o Text)) [];
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   192
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   193
in
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   194
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   195
val parse_comments =
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   196
  blanks -- Scan.repeat (parse_comment -- blanks >> K ()) >> K ();
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   197
40627
becf5d5187cc renamed raw "explode" function to "raw_explode" to emphasize its meaning;
wenzelm
parents: 40131
diff changeset
   198
val parse_string = Scan.read Symbol.stopper parse_chars o raw_explode;
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   199
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   200
fun parse_content xs =
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   201
  (parse_optional_text @@@
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   202
    (Scan.repeat
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   203
      ((parse_element >> single ||
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   204
        parse_cdata >> (single o Text) ||
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   205
        parse_processing_instruction ||
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   206
        parse_comment)
26551
da1cd11d8a25 renamed parse_comment_whspc to parse_comments;
wenzelm
parents: 26546
diff changeset
   207
      @@@ parse_optional_text) >> flat)) xs
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   208
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   209
and parse_element xs =
43949
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   210
  ($$ "<" |-- parse_name -- Scan.repeat (blanks |-- parse_att) --| blanks :--
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   211
    (fn (name, _) =>
43947
9b00f09f7721 defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents: 43844
diff changeset
   212
      !! (err (fn () => "Expected > or />"))
43949
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   213
       ($$ "/" -- $$ ">" >> ignored ||
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   214
        $$ ">" |-- parse_content --|
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   215
          !! (err (fn () => "Expected </" ^ implode name ^ ">"))
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   216
              ($$ "<" -- $$ "/" -- Scan.this name -- blanks -- $$ ">")))
94033767ef9b more precise parse_name according to XML standard;
wenzelm
parents: 43947
diff changeset
   217
    >> (fn ((name, atts), body) => Elem ((implode name, atts), body))) xs;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   218
26984
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   219
val parse_document =
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   220
  (Scan.repeat parse_misc -- Scan.option parse_doctype -- Scan.repeat parse_misc)
d0e098e206f3 added parse_document (optional unchecked header material);
wenzelm
parents: 26554
diff changeset
   221
  |-- parse_element;
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   222
26539
a0754be538ab added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents: 26525
diff changeset
   223
fun parse s =
43947
9b00f09f7721 defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents: 43844
diff changeset
   224
  (case Scan.finite Symbol.stopper (Scan.error (!! (err (fn () => "Malformed element"))
40627
becf5d5187cc renamed raw "explode" function to "raw_explode" to emphasize its meaning;
wenzelm
parents: 40131
diff changeset
   225
      (blanks |-- parse_document --| blanks))) (raw_explode s) of
24264
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   226
    (x, []) => x
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   227
  | (_, ys) => error ("XML parsing error: Unprocessed input\n" ^ Symbol.beginning 100 ys));
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   228
d6935e7dac8b moved Tools/xml.ML to General/xml.ML (again);
wenzelm
parents:
diff changeset
   229
end;
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   230
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   231
44809
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   232
(** cache for substructural sharing **)
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   233
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   234
fun tree_ord tu =
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   235
  if pointer_eq tu then EQUAL
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   236
  else
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   237
    (case tu of
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   238
      (Text _, Elem _) => LESS
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   239
    | (Elem _, Text _) => GREATER
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   240
    | (Text s, Text s') => fast_string_ord (s, s')
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   241
    | (Elem e, Elem e') =>
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   242
        prod_ord
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   243
          (prod_ord fast_string_ord (list_ord (prod_ord fast_string_ord fast_string_ord)))
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   244
          (list_ord tree_ord) (e, e'));
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   245
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   246
structure Treetab = Table(type key = tree val ord = tree_ord);
44808
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   247
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   248
fun cache () =
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   249
  let
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   250
    val strings = Unsynchronized.ref (Symtab.empty: unit Symtab.table);
44809
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   251
    val trees = Unsynchronized.ref (Treetab.empty: unit Treetab.table);
44808
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   252
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   253
    fun string s =
44809
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   254
      if size s <= 1 then s
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   255
      else
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   256
        (case Symtab.lookup_key (! strings) s of
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   257
          SOME (s', ()) => s'
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   258
        | NONE => (Unsynchronized.change strings (Symtab.update (s, ())); s));
44808
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   259
44809
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   260
    fun tree t =
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   261
      (case Treetab.lookup_key (! trees) t of
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   262
        SOME (t', ()) => t'
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   263
      | NONE =>
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   264
          let
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   265
            val t' =
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   266
              (case t of
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   267
                Elem ((a, ps), b) => Elem ((string a, map (pairself string) ps), map tree b)
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   268
              | Text s => Text (string s));
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   269
            val _ = Unsynchronized.change trees (Treetab.update (t', ()));
df3626d1066e more substructural sharing to gain significant compression;
wenzelm
parents: 44808
diff changeset
   270
          in t' end);
44808
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   271
  in tree end;
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   272
05b8997899a2 XML.cache for partial sharing (strings only);
wenzelm
parents: 44698
diff changeset
   273
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   274
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   275
(** XML as data representation language **)
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   276
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   277
exception XML_ATOM of string;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   278
exception XML_BODY of tree list;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   279
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   280
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   281
structure Encode =
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   282
struct
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   283
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   284
type 'a A = 'a -> string;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   285
type 'a T = 'a -> body;
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   286
type 'a V = 'a -> string list * body;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   287
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   288
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   289
(* atomic values *)
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   290
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   291
fun int_atom i = signed_string_of_int i;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   292
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   293
fun bool_atom false = "0"
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   294
  | bool_atom true = "1";
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   295
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   296
fun unit_atom () = "";
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   297
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   298
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   299
(* structural nodes *)
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   300
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   301
fun node ts = Elem ((":", []), ts);
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   302
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   303
fun vector xs = map_index (fn (i, x) => (int_atom i, x)) xs;
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   304
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   305
fun tagged (tag, (xs, ts)) = Elem ((int_atom tag, vector xs), ts);
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   306
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   307
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   308
(* representation of standard types *)
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   309
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   310
fun properties props = [Elem ((":", props), [])];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   311
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   312
fun string "" = []
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   313
  | string s = [Text s];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   314
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   315
val int = string o int_atom;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   316
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   317
val bool = string o bool_atom;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   318
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   319
val unit = string o unit_atom;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   320
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   321
fun pair f g (x, y) = [node (f x), node (g y)];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   322
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   323
fun triple f g h (x, y, z) = [node (f x), node (g y), node (h z)];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   324
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   325
fun list f xs = map (node o f) xs;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   326
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   327
fun option _ NONE = []
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   328
  | option f (SOME x) = [node (f x)];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   329
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   330
fun variant fs x = [tagged (the (get_index (fn f => try f x) fs))];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   331
26546
ba4cdf92c7c4 further cleanup of XML signature;
wenzelm
parents: 26539
diff changeset
   332
end;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   333
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   334
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   335
structure Decode =
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   336
struct
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   337
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   338
type 'a A = string -> 'a;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   339
type 'a T = body -> 'a;
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   340
type 'a V = string list * body -> 'a;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   341
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   342
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   343
(* atomic values *)
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   344
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   345
fun int_atom s =
43797
fad7758421bf more precise integer Markup.properties/XML.attributes: disallow ML-style ~ minus;
wenzelm
parents: 43791
diff changeset
   346
  Markup.parse_int s
fad7758421bf more precise integer Markup.properties/XML.attributes: disallow ML-style ~ minus;
wenzelm
parents: 43791
diff changeset
   347
    handle Fail _ => raise XML_ATOM s;
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   348
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   349
fun bool_atom "0" = false
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   350
  | bool_atom "1" = true
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   351
  | bool_atom s = raise XML_ATOM s;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   352
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   353
fun unit_atom "" = ()
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   354
  | unit_atom s = raise XML_ATOM s;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   355
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   356
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   357
(* structural nodes *)
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   358
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   359
fun node (Elem ((":", []), ts)) = ts
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   360
  | node t = raise XML_BODY [t];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   361
43783
ef45eaf2775f made SML/NJ happy;
wenzelm
parents: 43781
diff changeset
   362
fun vector atts =
43844
wenzelm
parents: 43797
diff changeset
   363
  #1 (fold_map (fn (a, x) =>
wenzelm
parents: 43797
diff changeset
   364
        fn i => if int_atom a = i then (x, i + 1) else raise XML_ATOM a) atts 0);
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   365
43844
wenzelm
parents: 43797
diff changeset
   366
fun tagged (Elem ((name, atts), ts)) = (int_atom name, (vector atts, ts))
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   367
  | tagged t = raise XML_BODY [t];
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   368
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   369
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   370
(* representation of standard types *)
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   371
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   372
fun properties [Elem ((":", props), [])] = props
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   373
  | properties ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   374
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   375
fun string [] = ""
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   376
  | string [Text s] = s
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   377
  | string ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   378
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   379
val int = int_atom o string;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   380
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   381
val bool = bool_atom o string;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   382
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   383
val unit = unit_atom o string;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   384
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   385
fun pair f g [t1, t2] = (f (node t1), g (node t2))
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   386
  | pair _ _ ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   387
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   388
fun triple f g h [t1, t2, t3] = (f (node t1), g (node t2), h (node t3))
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   389
  | triple _ _ _ ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   390
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   391
fun list f ts = map (f o node) ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   392
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   393
fun option _ [] = NONE
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   394
  | option f [t] = SOME (f (node t))
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   395
  | option _ ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   396
43768
d52ab827d62b more precise exceptions;
wenzelm
parents: 43767
diff changeset
   397
fun variant fs [t] =
d52ab827d62b more precise exceptions;
wenzelm
parents: 43767
diff changeset
   398
      let
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   399
        val (tag, (xs, ts)) = tagged t;
43768
d52ab827d62b more precise exceptions;
wenzelm
parents: 43767
diff changeset
   400
        val f = nth fs tag handle General.Subscript => raise XML_BODY [t];
43778
ce9189450447 more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents: 43768
diff changeset
   401
      in f (xs, ts) end
43767
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   402
  | variant _ ts = raise XML_BODY ts;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   403
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   404
end;
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   405
e0219ef7f84c tuned XML modules;
wenzelm
parents: 40627
diff changeset
   406
end;