| author | wenzelm | 
| Mon, 29 Dec 2008 18:27:33 +0100 | |
| changeset 29200 | 787ba47201c7 | 
| parent 29180 | 62513d4d34c2 | 
| child 29521 | 736bf7117153 | 
| permissions | -rw-r--r-- | 
| 27930 | 1 | /* Title: Pure/General/yxml.scala | 
| 2 | Author: Makarius | |
| 3 | ||
| 4 | Efficient text representation of XML trees. | |
| 5 | */ | |
| 6 | ||
| 7 | package isabelle | |
| 8 | ||
| 9 | import java.util.regex.Pattern | |
| 10 | ||
| 11 | ||
| 12 | object YXML {
 | |
| 13 | ||
| 27943 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 14 | /* chunk markers */ | 
| 27930 | 15 | |
| 16 | private val X = '\5' | |
| 17 | private val Y = '\6' | |
| 27960 | 18 | private val X_string = X.toString | 
| 27945 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 19 | private val Y_string = Y.toString | 
| 27930 | 20 | |
| 21 |   def detect(source: CharSequence) = {
 | |
| 22 | source.length >= 2 && | |
| 23 | source.charAt(0) == X && | |
| 24 | source.charAt(1) == Y | |
| 25 | } | |
| 26 | ||
| 27 | ||
| 27943 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 28 | /* iterate over chunks (resembles space_explode in ML) */ | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 29 | |
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 30 |   private def chunks(sep: Char, source: CharSequence) = new Iterator[CharSequence] {
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 31 | private val end = source.length | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 32 | private var state = if (end == 0) None else get_chunk(-1) | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 33 |     private def get_chunk(i: Int) = {
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 34 |       if (i < end) {
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 35 | var j = i; do j += 1 while (j < end && source.charAt(j) != sep) | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 36 | Some((source.subSequence(i + 1, j), j)) | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 37 | } | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 38 | else None | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 39 | } | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 40 | |
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 41 | def hasNext() = state.isDefined | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 42 |     def next() = state match {
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 43 |       case Some((s, i)) => { state = get_chunk(i); s }
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 44 |       case None => throw new NoSuchElementException("next on empty iterator")
 | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 45 | } | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 46 | } | 
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 47 | |
| 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 48 | |
| 27930 | 49 | /* parsing */ | 
| 50 | ||
| 27993 
6dd90ef9f927
simplified exceptions: use plain error function / RuntimeException;
 wenzelm parents: 
27971diff
changeset | 51 |   private def err(msg: String) = error("Malformed YXML: " + msg)
 | 
| 27930 | 52 |   private def err_attribute() = err("bad attribute")
 | 
| 53 |   private def err_element() = err("bad element")
 | |
| 54 | private def err_unbalanced(name: String) = | |
| 55 |     if (name == "") err("unbalanced element")
 | |
| 56 |     else err("unbalanced element \"" + name + "\"")
 | |
| 57 | ||
| 27944 
2bf3f30558ed
parse_attrib: more efficient due to indexOf('=');
 wenzelm parents: 
27943diff
changeset | 58 |   private def parse_attrib(source: CharSequence) = {
 | 
| 
2bf3f30558ed
parse_attrib: more efficient due to indexOf('=');
 wenzelm parents: 
27943diff
changeset | 59 | val s = source.toString | 
| 
2bf3f30558ed
parse_attrib: more efficient due to indexOf('=');
 wenzelm parents: 
27943diff
changeset | 60 |     val i = s.indexOf('=')
 | 
| 
2bf3f30558ed
parse_attrib: more efficient due to indexOf('=');
 wenzelm parents: 
27943diff
changeset | 61 | if (i <= 0) err_attribute() | 
| 27946 | 62 | (s.substring(0, i), s.substring(i + 1)) | 
| 27944 
2bf3f30558ed
parse_attrib: more efficient due to indexOf('=');
 wenzelm parents: 
27943diff
changeset | 63 | } | 
| 27930 | 64 | |
| 65 | ||
| 66 |   def parse_body(source: CharSequence) = {
 | |
| 67 | ||
| 68 | /* stack operations */ | |
| 69 | ||
| 70 | var stack: List[((String, XML.Attributes), List[XML.Tree])] = null | |
| 71 | ||
| 29180 | 72 |     def add(x: XML.Tree) = (stack: @unchecked) match {
 | 
| 27930 | 73 | case ((elem, body) :: pending) => stack = (elem, x :: body) :: pending | 
| 74 | } | |
| 75 | ||
| 76 | def push(name: String, atts: XML.Attributes) = | |
| 77 | if (name == "") err_element() | |
| 78 | else stack = ((name, atts), Nil) :: stack | |
| 79 | ||
| 29180 | 80 |     def pop() = (stack: @unchecked) match {
 | 
| 27930 | 81 |       case ((("", _), _) :: _) => err_unbalanced("")
 | 
| 82 | case (((name, atts), body) :: pending) => | |
| 83 | stack = pending; add(XML.Elem(name, atts, body.reverse)) | |
| 84 | } | |
| 85 | ||
| 86 | ||
| 87 | /* parse chunks */ | |
| 88 | ||
| 89 |     stack = List((("", Nil), Nil))
 | |
| 27943 
f34ff5e7728f
replaced Pattern.split by chunks iterator (more efficient, resembles ML version more closely);
 wenzelm parents: 
27930diff
changeset | 90 |     for (chunk <- chunks(X, source) if chunk != "") {
 | 
| 27945 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 91 | if (chunk == Y_string) pop() | 
| 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 92 |       else {
 | 
| 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 93 |         chunks(Y, chunk).toList match {
 | 
| 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 94 | case "" :: name :: atts => push(name.toString, atts.map(parse_attrib)) | 
| 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 95 | case txts => for (txt <- txts) add(XML.Text(txt.toString)) | 
| 
d2dc5a1903e8
tuned parse performance: avoid splitting terminal Y chunk;
 wenzelm parents: 
27944diff
changeset | 96 | } | 
| 27930 | 97 | } | 
| 98 | } | |
| 99 |     stack match {
 | |
| 100 |       case List((("", _), result)) => result.reverse
 | |
| 101 | case ((name, _), _) :: _ => err_unbalanced(name) | |
| 102 | } | |
| 103 | } | |
| 104 | ||
| 105 | def parse(source: CharSequence) = | |
| 106 |     parse_body(source) match {
 | |
| 107 | case List(result) => result | |
| 108 |       case Nil => XML.Text("")
 | |
| 109 |       case _ => err("multiple results")
 | |
| 110 | } | |
| 27960 | 111 | |
| 112 |   def parse_failsafe(source: CharSequence) = {
 | |
| 113 |     try { parse(source) }
 | |
| 114 |     catch {
 | |
| 27993 
6dd90ef9f927
simplified exceptions: use plain error function / RuntimeException;
 wenzelm parents: 
27971diff
changeset | 115 | case _: RuntimeException => XML.Elem (Markup.BAD, Nil, | 
| 27960 | 116 | List(XML.Text(source.toString.replace(X_string, "\\<^X>").replace(Y_string, "\\<^Y>")))) | 
| 117 | } | |
| 118 | } | |
| 119 | ||
| 27930 | 120 | } |