src/Pure/General/yxml.ML
author wenzelm
Thu, 17 Dec 2009 13:58:15 +0100
changeset 34095 c2f176a38448
parent 31469 40f815edbde4
child 38228 ada3ab6b9085
permissions -rw-r--r--
robust representation of low ASCII control characters within XML/YXML text;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
     1
(*  Title:      Pure/General/yxml.ML
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
     2
    Author:     Makarius
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
     3
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
     4
Efficient text representation of XML trees using extra characters X
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
     5
and Y -- no escaping, may nest marked text verbatim.
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
     6
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
     7
Markup <elem att="val" ...>...body...</elem> is encoded as:
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
     8
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
     9
  X Y name Y att=val ... X
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    10
  ...
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    11
  body
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    12
  ...
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    13
  X Y X
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    14
*)
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    15
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    16
signature YXML =
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    17
sig
34095
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    18
  val binary_text: string -> string
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    19
  val output_markup: Markup.T -> string * string
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    20
  val element: string -> XML.attributes -> string list -> string
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    21
  val string_of: XML.tree -> string
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    22
  val parse_body: string -> XML.tree list
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    23
  val parse: string -> XML.tree
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    24
end;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    25
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    26
structure YXML: YXML =
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    27
struct
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    28
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    29
(** string representation **)
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    30
34095
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    31
(* binary_text -- idempotent recoding *)
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    32
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    33
fun pseudo_utf8 c =
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    34
  if Symbol.is_ascii_control c
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    35
  then chr 192 ^ chr (128 + ord c)
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    36
  else c;
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    37
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    38
fun binary_text str =
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    39
  if exists_string Symbol.is_ascii_control str
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    40
  then translate_string pseudo_utf8 str
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    41
  else str;
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    42
c2f176a38448 robust representation of low ASCII control characters within XML/YXML text;
wenzelm
parents: 31469
diff changeset
    43
26547
1112375f6a69 tuned comments;
wenzelm
parents: 26540
diff changeset
    44
(* markers *)
1112375f6a69 tuned comments;
wenzelm
parents: 26540
diff changeset
    45
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    46
val X = Symbol.ENQ;
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    47
val Y = Symbol.ACK;
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    48
val XY = X ^ Y;
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    49
val XYX = XY ^ X;
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    50
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    51
26547
1112375f6a69 tuned comments;
wenzelm
parents: 26540
diff changeset
    52
(* output *)
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    53
27884
10c927e4abf5 output_markup: check Markup.is_none;
wenzelm
parents: 27798
diff changeset
    54
fun output_markup (markup as (name, atts)) =
29325
a205acc94356 Markup.no_output;
wenzelm
parents: 28033
diff changeset
    55
  if Markup.is_none markup then Markup.no_output
27884
10c927e4abf5 output_markup: check Markup.is_none;
wenzelm
parents: 27798
diff changeset
    56
  else (XY ^ name ^ implode (map (fn (a, x) => Y ^ a ^ "=" ^ x) atts) ^ X, XYX);
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    57
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    58
fun element name atts body =
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    59
  let val (pre, post) = output_markup (name, atts)
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    60
  in pre ^ implode body ^ post end;
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    61
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    62
fun string_of t =
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    63
  let
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    64
    fun attrib (a, x) =
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    65
      Buffer.add Y #>
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    66
      Buffer.add a #> Buffer.add "=" #> Buffer.add x;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    67
    fun tree (XML.Elem (name, atts, ts)) =
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    68
          Buffer.add XY #> Buffer.add name #> fold attrib atts #> Buffer.add X #>
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    69
          fold tree ts #>
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    70
          Buffer.add XYX
28033
f03b5856f286 removed obsolete XML.Output workaround;
wenzelm
parents: 28025
diff changeset
    71
      | tree (XML.Text s) = Buffer.add s;
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    72
  in Buffer.empty |> tree t |> Buffer.content end;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    73
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    74
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    75
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    76
(** efficient YXML parsing **)
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    77
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    78
local
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    79
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    80
(* splitting *)
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    81
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    82
fun is_char s c = ord s = Char.ord c;
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    83
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    84
val split_string =
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    85
  Substring.full #>
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    86
  Substring.tokens (is_char X) #>
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    87
  map (Substring.fields (is_char Y) #> map Substring.string);
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    88
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    89
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
    90
(* structural errors *)
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    91
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    92
fun err msg = raise Fail ("Malformed YXML encoding: " ^ msg);
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    93
fun err_attribute () = err "bad attribute";
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    94
fun err_element () = err "bad element";
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    95
fun err_unbalanced "" = err "unbalanced element"
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    96
  | err_unbalanced name = err ("unbalanced element " ^ quote name);
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    97
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    98
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
    99
(* stack operations *)
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   100
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   101
fun add x ((elem, body) :: pending) = (elem, x :: body) :: pending;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   102
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   103
fun push "" _ _ = err_element ()
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   104
  | push name atts pending = ((name, atts), []) :: pending;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   105
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   106
fun pop ((("", _), _) :: _) = err_unbalanced ""
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   107
  | pop (((name, atts), body) :: pending) = add (XML.Elem (name, atts, rev body)) pending;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   108
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   109
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
   110
(* parsing *)
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   111
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   112
fun parse_attrib s =
28025
d9fcab768496 replaced find_substring by first_field;
wenzelm
parents: 28023
diff changeset
   113
  (case first_field "=" s of
28023
92dd3ad302b7 simplified parse_attrib (find_substring instead of space_explode);
wenzelm
parents: 27932
diff changeset
   114
    NONE => err_attribute ()
28025
d9fcab768496 replaced find_substring by first_field;
wenzelm
parents: 28023
diff changeset
   115
  | SOME ("", _) => err_attribute ()
d9fcab768496 replaced find_substring by first_field;
wenzelm
parents: 28023
diff changeset
   116
  | SOME att => att);
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   117
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
   118
fun parse_chunk ["", ""] = pop
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
   119
  | parse_chunk ("" :: name :: atts) = push name (map parse_attrib atts)
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
   120
  | parse_chunk txts = fold (add o XML.Text) txts;
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   121
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   122
in
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   123
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   124
fun parse_body source =
26540
173d548ce9d2 replaced ETX/EOT by ENQ/ACK, which are less likely to be interpreted by tty etc.;
wenzelm
parents: 26528
diff changeset
   125
  (case fold parse_chunk (split_string source) [(("", []), [])] of
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   126
    [(("", _), result)] => rev result
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   127
  | ((name, _), _) :: _ => err_unbalanced name);
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   128
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   129
fun parse source =
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   130
  (case parse_body source of
27798
b96c73f11a23 YXML.parse: allow text without markup, potentially empty;
wenzelm
parents: 26684
diff changeset
   131
    [result] => result
b96c73f11a23 YXML.parse: allow text without markup, potentially empty;
wenzelm
parents: 26684
diff changeset
   132
  | [] => XML.Text ""
b96c73f11a23 YXML.parse: allow text without markup, potentially empty;
wenzelm
parents: 26684
diff changeset
   133
  | _ => err "multiple results");
26528
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   134
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   135
end;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   136
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   137
end;
944f9bf26d2d Why XML notation?
wenzelm
parents:
diff changeset
   138