2 // System.Security.Cryptography.MiniParser: Internal XML parser implementation
7 // Copyright (c) 2001, 2002 Wild West Software
8 // Copyright (c) 2002 Sergey Chaban
9 // Copyright (C) 2004 Novell, Inc (http://www.novell.com)
11 // Permission is hereby granted, free of charge, to any person obtaining
12 // a copy of this software and associated documentation files (the
13 // "Software"), to deal in the Software without restriction, including
14 // without limitation the rights to use, copy, modify, merge, publish,
15 // distribute, sublicense, and/or sell copies of the Software, and to
16 // permit persons to whom the Software is furnished to do so, subject to
17 // the following conditions:
19 // The above copyright notice and this permission notice shall be
20 // included in all copies or substantial portions of the Software.
22 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
23 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
24 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
25 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
26 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
27 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
28 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
33 using System.Collections;
34 using System.Globalization;
38 internal class MiniParser {
40 public interface IReader {
44 public interface IAttrList {
47 string GetName(int i);
48 string GetValue(int i);
49 string GetValue(string name);
50 void ChangeValue(string name, string newValue);
52 string[] Values {get;}
55 public interface IMutableAttrList : IAttrList {
57 void Add(string name, string value);
58 void CopyFrom(IAttrList attrs);
60 void Remove(string name);
63 public interface IHandler {
64 void OnStartParsing(MiniParser parser);
65 void OnStartElement(string name, IAttrList attrs);
66 void OnEndElement(string name);
67 void OnChars(string ch);
68 void OnEndParsing(MiniParser parser);
71 public class HandlerAdapter : IHandler {
72 public HandlerAdapter() {}
73 public void OnStartParsing(MiniParser parser) {}
74 public void OnStartElement(string name, IAttrList attrs) {}
75 public void OnEndElement(string name) {}
76 public void OnChars(string ch) {}
77 public void OnEndParsing(MiniParser parser) {}
80 private enum CharKind : byte {
100 private enum ActionCode : byte {
111 FLUSH_CHARS_STATE_CHANGE = 0xA,
112 ACC_CHARS_STATE_CHANGE = 0xB,
118 public class AttrListImpl : IMutableAttrList {
119 protected ArrayList names;
120 protected ArrayList values;
122 public AttrListImpl() : this(0) {}
124 public AttrListImpl(int initialCapacity) {
125 if (initialCapacity <= 0) {
126 names = new ArrayList();
127 values = new ArrayList();
129 names = new ArrayList(initialCapacity);
130 values = new ArrayList(initialCapacity);
134 public AttrListImpl(IAttrList attrs)
135 : this(attrs != null ? attrs.Length : 0) {
136 if (attrs != null) this.CopyFrom(attrs);
140 get {return names.Count;}
143 public bool IsEmpty {
144 get {return this.Length != 0;}
147 public string GetName(int i) {
149 if (i >= 0 && i < this.Length) {
150 res = names[i] as string;
155 public string GetValue(int i) {
157 if (i >= 0 && i < this.Length) {
158 res = values[i] as string;
163 public string GetValue(string name) {
164 return this.GetValue(names.IndexOf(name));
167 public void ChangeValue(string name, string newValue) {
168 int i = names.IndexOf(name);
169 if (i >= 0 && i < this.Length) {
170 values[i] = newValue;
174 public string[] Names {
175 get {return names.ToArray(typeof(string)) as string[];}
178 public string[] Values {
179 get {return values.ToArray(typeof(string)) as string[];}
182 public void Clear() {
187 public void Add(string name, string value) {
192 public void Remove(int i) {
199 public void Remove(string name) {
200 this.Remove(names.IndexOf(name));
203 public void CopyFrom(IAttrList attrs) {
204 if (attrs != null && ((object)this == (object)attrs)) {
206 int n = attrs.Length;
207 for (int i = 0; i < n; i++) {
208 this.Add(attrs.GetName(i), attrs.GetValue(i));
214 public class XMLError : Exception {
215 protected string descr;
216 protected int line, column;
217 public XMLError() : this("Unknown") {}
218 public XMLError(string descr) : this(descr, -1, -1) {}
219 public XMLError(string descr, int line, int column)
223 this.column = column;
225 public int Line {get {return line;}}
226 public int Column {get {return column;}}
227 public override string ToString() {
228 return (String.Format("{0} @ (line = {1}, col = {2})", descr, line, column));
232 private static readonly int INPUT_RANGE = 13;
233 private static readonly ushort[] tbl = {
234 (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 1), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 0), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 128), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 128),
235 (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 2), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 133), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 16), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.FLUSH_CHARS_STATE_CHANGE << 8) | 4),
236 (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.END_ELEM << 8) | 0), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 2), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 2), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129),
237 (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 5), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3),
238 (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 4), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.END_NAME << 8) | 6), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.END_NAME << 8) | 7), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.END_NAME << 8) | 8), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129),
239 (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 0), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 3),
240 (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 0), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ERROR << 8) | 129),
241 (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.FLUSH_CHARS_STATE_CHANGE << 8) | 1), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.PROC_CHAR_REF << 8) | 10), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 7), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10),
242 (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 9), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.START_ELEM << 8) | 6), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.START_ELEM << 8) | 7), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 8), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129),
243 (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 9), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.SET_ATTR_NAME << 8) | 11), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 12), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 130),
244 (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 13), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.PROC_CHAR_REF << 8) | 10), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 10),
245 (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 11), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 132), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ERROR << 8) | 132),
246 (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.SET_ATTR_NAME << 8) | 11), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 12), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 130), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ERROR << 8) | 130),
247 (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.SEND_CHARS << 8) | 2), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 16), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 134), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ERROR << 8) | 134),
248 (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.SET_ATTR_VAL << 8) | 17), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.PROC_CHAR_REF << 8) | 14), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 14),
249 (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.SET_ATTR_VAL << 8) | 17), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.PROC_CHAR_REF << 8) | 15), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 15),
250 (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.START_CDATA << 8) | 18), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 0), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.START_CDATA << 8) | 19),
251 (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.START_ELEM << 8) | 6), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.START_ELEM << 8) | 7), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.STATE_CHANGE << 8) | 17), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CHARS_STATE_CHANGE << 8) | 9), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ERROR << 8) | 129), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ERROR << 8) | 129),
252 (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.END_CDATA << 8) | 10), (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 18),
253 (ushort)(((ushort)CharKind.LEFT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SLASH << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.RIGHT_BR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.PI_MARK << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.EQ << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.AMP << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.BANG << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.LEFT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.SPACE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.RIGHT_SQBR << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.DQUOTE << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19), (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.ACC_CDATA << 8) | 19),
254 (ushort)(((ushort)CharKind.CHARS << 12) | ((ushort)ActionCode.UNKNOWN << 8) | 255),
258 protected static string[] errors = {
259 /* 0 */ "Expected element",
260 /* 1 */ "Invalid character in tag",
262 /* 3 */ "Invalid character entity",
263 /* 4 */ "Invalid attr value",
265 /* 6 */ "No end tag",
266 /* 7 */ "Bad entity ref"
271 protected int[] twoCharBuff;
272 protected bool splitCData;
274 public MiniParser() {
275 twoCharBuff = new int[2];
280 public void Reset() {
285 protected static bool StrEquals(string str, StringBuilder sb, int sbStart, int len) {
286 if (len != str.Length) return false;
287 for (int i = 0; i < len; i++) {
288 if (str[i] != sb[sbStart + i]) return false;
293 protected void FatalErr(string descr) {
294 throw new XMLError(descr, this.line, this.col);
297 protected static int Xlat(int charCode, int state) {
298 int p = state * INPUT_RANGE;
299 int n = System.Math.Min(tbl.Length - p, INPUT_RANGE);
301 ushort code = tbl[p];
302 if (charCode == (code >> 12)) return (code & 0xFFF);
308 public void Parse(IReader reader, IHandler handler) {
309 if (reader == null) throw new ArgumentNullException("reader");
310 if (handler == null) handler = new HandlerAdapter();
312 AttrListImpl attrList = new AttrListImpl();
313 string lastAttrName = null;
314 Stack tagStack = new Stack();
315 string elementName = null;
320 StringBuilder sbChars = new StringBuilder();
321 bool seenCData = false;
322 bool isComment = false;
324 int bracketSwitch = 0;
326 handler.OnStartParsing(this);
331 currCh = reader.Read();
334 if (stateCode != 0) {
335 FatalErr("Unexpected EOF");
340 int charCode = "<>/?=&'\"![ ]\t\r\n".IndexOf((char)currCh) & 0xF;
341 if (charCode == (int)CharKind.CR) continue; // ignore
342 // whitepace ::= (#x20 | #x9 | #xd | #xa)+
343 if (charCode == (int)CharKind.TAB) charCode = (int)CharKind.SPACE; // tab == space
344 if (charCode == (int)CharKind.EOL) {
347 charCode = (int)CharKind.SPACE;
350 int actionCode = MiniParser.Xlat(charCode, stateCode);
351 stateCode = actionCode & 0xFF;
352 // Ignore newline inside attribute value.
353 if (currCh == '\n' && (stateCode == 0xE || stateCode == 0xF)) continue;
356 if (stateCode >= 0x80) {
357 if (stateCode == 0xFF) {
358 FatalErr("State dispatch error.");
360 FatalErr(errors[stateCode ^ 0x80]);
364 switch (actionCode) {
365 case (int)ActionCode.START_ELEM:
366 handler.OnStartElement(elementName, attrList);
368 tagStack.Push(elementName);
370 handler.OnEndElement(elementName);
375 case (int)ActionCode.END_ELEM:
376 elementName = sbChars.ToString();
377 sbChars = new StringBuilder();
378 string endName = null;
379 if (tagStack.Count == 0 ||
380 elementName != (endName = tagStack.Pop() as string)) {
381 if (endName == null) {
382 FatalErr("Tag stack underflow");
384 FatalErr(String.Format("Expected end tag '{0}' but found '{1}'", elementName, endName));
387 handler.OnEndElement(elementName);
390 case (int)ActionCode.END_NAME:
391 elementName = sbChars.ToString();
392 sbChars = new StringBuilder();
393 if (currCh != '/' && currCh != '>') break;
394 goto case (int)ActionCode.START_ELEM;
396 case (int)ActionCode.SET_ATTR_NAME:
397 lastAttrName = sbChars.ToString();
398 sbChars = new StringBuilder();
401 case (int)ActionCode.SET_ATTR_VAL:
402 if (lastAttrName == null) FatalErr("Internal error.");
403 attrList.Add(lastAttrName, sbChars.ToString());
404 sbChars = new StringBuilder();
408 case (int)ActionCode.SEND_CHARS:
409 handler.OnChars(sbChars.ToString());
410 sbChars = new StringBuilder();
413 case (int)ActionCode.START_CDATA:
414 string cdata = "CDATA[";
419 currCh = reader.Read();
421 if (currCh != '-') FatalErr("Invalid comment");
434 for (int i = 0; i < cdata.Length; i++) {
435 if (reader.Read() != cdata[i]) {
440 this.col += cdata.Length;
445 case (int)ActionCode.END_CDATA:
449 while (currCh == ']') {
450 currCh = reader.Read();
455 for (int i = 0; i < n; i++) sbChars.Append(']');
456 sbChars.Append((char)currCh);
459 for (int i = 0; i < n-2; i++) sbChars.Append(']');
466 case (int)ActionCode.ERROR:
467 FatalErr(String.Format("Error {0}", stateCode));
470 case (int)ActionCode.STATE_CHANGE:
473 case (int)ActionCode.FLUSH_CHARS_STATE_CHANGE:
474 sbChars = new StringBuilder();
475 if (currCh != '<') goto case (int)ActionCode.ACC_CHARS_STATE_CHANGE;
478 case (int)ActionCode.ACC_CHARS_STATE_CHANGE:
479 sbChars.Append((char)currCh);
482 case (int)ActionCode.ACC_CDATA:
485 && twoCharBuff[0] == '-'
486 && twoCharBuff[1] == '-') {
490 twoCharBuff[0] = twoCharBuff[1];
491 twoCharBuff[1] = currCh;
494 if (currCh == '<' || currCh == '>') bracketSwitch ^= 1;
495 if (currCh == '>' && bracketSwitch != 0) {
501 && sbChars.Length > 0
503 handler.OnChars(sbChars.ToString());
504 sbChars = new StringBuilder();
507 sbChars.Append((char)currCh);
511 case (int)ActionCode.PROC_CHAR_REF:
512 currCh = reader.Read();
513 int cl = this.col + 1;
514 if (currCh == '#') { // character reference
518 currCh = reader.Read();
522 currCh = reader.Read();
527 NumberStyles style = r == 16 ? NumberStyles.HexNumber : NumberStyles.Integer;
531 if (Char.IsNumber((char)currCh) || "abcdef".IndexOf(Char.ToLower((char)currCh)) != -1) {
533 x = Int32.Parse(new string((char)currCh, 1), style);
534 } catch (FormatException) {x = -1;}
540 currCh = reader.Read();
544 if (currCh == ';' && nDigits > 0) {
545 sbChars.Append((char)chCode);
547 FatalErr("Bad char ref");
551 string entityRefChars = "aglmopqstu"; // amp | apos | quot | gt | lt
552 string entities = "&'\"><";
558 int sbLen = sbChars.Length;
561 if (pos != 0xF) pos = entityRefChars.IndexOf((char)currCh) & 0xF;
562 if (pos == 0xF) FatalErr(errors[7]);
563 sbChars.Append((char)currCh);
565 int path = "\uFF35\u3F8F\u4F8F\u0F5F\uFF78\uE1F4\u2299\uEEFF\uEEFF\uFF4F"[pos];
566 int lBr = (path >> 4) & 0xF;
567 int rBr = path & 0xF;
568 int lPred = path >> 12;
569 int rPred = (path >> 8) & 0xF;
570 currCh = reader.Read();
573 if (lBr != 0xF && currCh == entityRefChars[lBr]) {
574 if (lPred < 0xE) entIdx = lPred;
576 predShift = 12; // left
577 } else if (rBr != 0xF && currCh == entityRefChars[rBr]) {
578 if (rPred < 0xE) entIdx = rPred;
580 predShift = 8; // right
581 } else if (currCh == ';') {
584 && ((path >> predShift) & 0xF) == 0xE) break;
585 continue; // pos == 0xF
592 int l = cl - this.col - 1;
595 &&(StrEquals("amp", sbChars, sbLen, l)
596 || StrEquals("apos", sbChars, sbLen, l)
597 || StrEquals("quot", sbChars, sbLen, l)
598 || StrEquals("lt", sbChars, sbLen, l)
599 || StrEquals("gt", sbChars, sbLen, l))
601 sbChars.Length = sbLen;
602 sbChars.Append(entities[entIdx]);
603 } else FatalErr(errors[7]);
610 FatalErr(String.Format("Unexpected action code - {0}.", actionCode));
615 handler.OnEndParsing(this);