mcs/class/System.Security/Mono.Xml/XmlCanonicalizer.cs

   1 //
   2 // XmlCanonicalizer.cs - C14N implementation for XML Signature
   3 // http://www.w3.org/TR/xml-c14n
   4 //
   5 // Author:
   6 //      Aleksey Sanin (aleksey@aleksey.com)
   7 //
   8 // (C) 2003 Aleksey Sanin (aleksey@aleksey.com)
   9 //
  10
  11 using System.Collections;
  12 using System.IO;
  13 using System.Text;
  14 using System.Xml;
  15
  16 namespace Mono.Xml {
  17
  18         internal class XmlCanonicalizer {
  19
  20                 private enum XmlCanonicalizerState
  21                 {
  22                         BeforeDocElement,
  23                         InsideDocElement,
  24                         AfterDocElement
  25                 }
  26
  27                 // c14n parameters
  28                 private bool comments;
  29                 private bool exclusive;
  30
  31                 // input/output
  32                 private XmlNodeList xnl;
  33                 private StringBuilder res;
  34
  35                 // namespaces rendering stack
  36                 private XmlCanonicalizerState state;
  37                 private ArrayList visibleNamespaces;
  38                 private int prevVisibleNamespacesStart;
  39                 private int prevVisibleNamespacesEnd;
  40
  41                 public XmlCanonicalizer (bool withComments, bool excC14N)
  42                 {
  43                         res = new StringBuilder ();
  44                         comments = withComments;
  45                         exclusive = excC14N;
  46                         state = XmlCanonicalizerState.BeforeDocElement;
  47                         visibleNamespaces = new ArrayList ();
  48                         prevVisibleNamespacesStart = 0;
  49                         prevVisibleNamespacesEnd = 0;
  50                 }
  51
  52                 public Stream Canonicalize (XmlDocument doc)
  53                 {
  54                         WriteDocumentNode (doc);
  55
  56                         UTF8Encoding utf8 = new UTF8Encoding ();
  57                         byte[] data = utf8.GetBytes (res.ToString ());
  58                         return new MemoryStream (data);
  59                 }
  60
  61                 public Stream Canonicalize (XmlNodeList nodes)
  62                 {
  63                         xnl = nodes;
  64                         if (nodes == null || nodes.Count < 1)
  65                             return null;
  66                         return Canonicalize (nodes[0].OwnerDocument);
  67                 }
  68
  69                 private void WriteNode (XmlNode node)
  70                 {
  71                         // Console.WriteLine ("C14N Debug: node=" + node.Name);
  72
  73                         bool visible = IsNodeVisible (node);
  74                         switch (node.NodeType) {
  75                         case XmlNodeType.Document:
  76                         case XmlNodeType.DocumentFragment:
  77                                 WriteDocumentNode (node);
  78                                 break;
  79                         case XmlNodeType.Element:
  80                                 WriteElementNode (node, visible);
  81                                 break;
  82                         case XmlNodeType.CDATA:
  83                         case XmlNodeType.SignificantWhitespace:
  84                         case XmlNodeType.Text:
  85                                 // CDATA sections are processed as text nodes
  86                                 WriteTextNode (node, visible);
  87                                 break;
  88                         case XmlNodeType.Whitespace:
  89                                 if (state == XmlCanonicalizerState.InsideDocElement)
  90                                         WriteTextNode (node, visible);
  91                                 break;
  92                         case XmlNodeType.Comment:
  93                                 WriteCommentNode (node, visible);
  94                                 break;
  95                         case XmlNodeType.ProcessingInstruction:
  96                                 WriteProcessingInstructionNode (node, visible);
  97                                 break;
  98                         case XmlNodeType.EntityReference:
  99 //                              throw new XmlException ("Entity references should be resolved by parser", null);
 100                                 for (int i = 0; i < node.ChildNodes.Count; i++)
 101                                         WriteNode (node.ChildNodes [i]);
 102                                 break;
 103                         case XmlNodeType.Attribute:
 104                                 throw new XmlException ("Attribute node is impossible here", null);
 105                         case XmlNodeType.EndElement:
 106                                 throw new XmlException ("EndElement node is impossible here", null);
 107                         case XmlNodeType.EndEntity:
 108                                 throw new XmlException ("EndEntity node is impossible here", null);
 109                         case XmlNodeType.DocumentType:
 110                         case XmlNodeType.Entity:
 111                         case XmlNodeType.Notation:
 112                         case XmlNodeType.XmlDeclaration:
 113                                 // just do nothing
 114                                 break;
 115                         }
 116                 }
 117
 118                 private void WriteDocumentNode (XmlNode node)
 119                 {
 120                         state = XmlCanonicalizerState.BeforeDocElement;
 121                         for (XmlNode child = node.FirstChild; child != null; child = child.NextSibling)
 122                                 WriteNode (child);
 123                 }
 124
 125                 // Element Nodes
 126                 // If the element is not in the node-set, then the result is obtained
 127                 // by processing the namespace axis, then the attribute axis, then
 128                 // processing the child nodes of the element that are in the node-set
 129                 // (in document order). If the element is inthe node-set, then the result
 130                 // is an open angle bracket (<), the element QName, the result of
 131                 // processing the namespace axis, the result of processing the attribute
 132                 // axis, a close angle bracket (>), the result of processing the child
 133                 // nodes of the element that are in the node-set (in document order), an
 134                 // open angle bracket, a forward slash (/), the element QName, and a close
 135                 // angle bracket.
 136                 private void WriteElementNode (XmlNode node, bool visible)
 137                 {
 138                         // Console.WriteLine ("Debug: element node");
 139
 140                         // remember current state
 141                         int savedPrevVisibleNamespacesStart = prevVisibleNamespacesStart;
 142                         int savedPrevVisibleNamespacesEnd = prevVisibleNamespacesEnd;
 143                         int savedVisibleNamespacesSize = visibleNamespaces.Count;
 144                         XmlCanonicalizerState s = state;
 145                         if (visible && state == XmlCanonicalizerState.BeforeDocElement)
 146                                 state = XmlCanonicalizerState.InsideDocElement;
 147
 148                         // write start tag
 149                         if (visible) {
 150                                 res.Append ("<");
 151                                 res.Append (node.Name);
 152                         }
 153
 154                         // this is odd but you can select namespaces
 155                         // and attributes even if node itself is not visible
 156                         WriteNamespacesAxis (node, visible);
 157                         WriteAttributesAxis (node);
 158
 159                         if (visible)
 160                                 res.Append (">");
 161
 162                         // write children
 163                         for (XmlNode child = node.FirstChild; child != null; child = child.NextSibling)
 164                                 WriteNode (child);
 165
 166                         // write end tag
 167                         if (visible) {
 168                                 res.Append ("</");
 169                                 res.Append (node.Name);
 170                                 res.Append (">");
 171                         }
 172
 173                         // restore state
 174                         if (visible && s == XmlCanonicalizerState.BeforeDocElement)
 175                                 state = XmlCanonicalizerState.AfterDocElement;
 176                         prevVisibleNamespacesStart = savedPrevVisibleNamespacesStart;
 177                         prevVisibleNamespacesEnd = savedPrevVisibleNamespacesEnd;
 178                         if (visibleNamespaces.Count > savedVisibleNamespacesSize) {
 179                                 visibleNamespaces.RemoveRange (savedVisibleNamespacesSize,
 180                                         visibleNamespaces.Count - savedVisibleNamespacesSize);
 181                         }
 182                 }
 183
 184                 // Namespace Axis
 185                 // Consider a list L containing only namespace nodes in the
 186                 // axis and in the node-set in lexicographic order (ascending). To begin
 187                 // processing L, if the first node is not the default namespace node (a node
 188                 // with no namespace URI and no local name), then generate a space followed
 189                 // by xmlns="" if and only if the following conditions are met:
 190                 //    - the element E that owns the axis is in the node-set
 191                 //    - The nearest ancestor element of E in the node-set has a default
 192                 //          namespace node in the node-set (default namespace nodes always
 193                 //      have non-empty values in XPath)
 194                 // The latter condition eliminates unnecessary occurrences of xmlns="" in
 195                 // the canonical form since an element only receives an xmlns="" if its
 196                 // default namespace is empty and if it has an immediate parent in the
 197                 // canonical form that has a non-empty default namespace. To finish
 198                 // processing  L, simply process every namespace node in L, except omit
 199                 // namespace node with local name xml, which defines the xml prefix,
 200                 // if its string value is http://www.w3.org/XML/1998/namespace.
 201                 private void WriteNamespacesAxis (XmlNode node, bool visible)
 202                 {
 203                         // Console.WriteLine ("Debug: namespaces");
 204
 205                         XmlDocument doc = node.OwnerDocument;
 206                         bool has_empty_namespace = false;
 207                         ArrayList list = new ArrayList ();
 208                         for (XmlNode cur = node; cur != null && cur != doc; cur = cur.ParentNode) {
 209                                 foreach (XmlNode attribute in cur.Attributes) {
 210                                         if (!IsNamespaceNode (attribute))
 211                                                 continue;
 212
 213                                         // get namespace prefix
 214                                         string prefix = string.Empty;
 215                                         if (attribute.Prefix == "xmlns")
 216                                                 prefix = attribute.LocalName;
 217
 218                                         // check if it is "xml" namespace
 219                                         if (prefix == "xml" && attribute.Value == "http://www.w3.org/XML/1998/namespace")
 220                                                 continue;
 221
 222                                         // make sure that this is an active namespace
 223                                         // for our node
 224                                         string ns = node.GetNamespaceOfPrefix (prefix);
 225                                         if (ns != attribute.Value)
 226                                                 continue;
 227
 228                                         // check that it is selected with XPath
 229                                         if (!IsNodeVisible (attribute))
 230                                                 continue;
 231
 232                                         // check that we have not rendered it yet
 233                                         bool rendered = IsNamespaceRendered (prefix, attribute.Value);
 234
 235                                         // add to the visible namespaces stack
 236                                         if (visible)
 237                                                 visibleNamespaces.Add (attribute);
 238
 239                                         if (!rendered)
 240                                                 list.Add (attribute);
 241
 242                                         if (prefix == string.Empty)
 243                                                 has_empty_namespace = true;
 244                                 }
 245                         }
 246
 247                         // add empty namespace if needed
 248                         if (visible && !has_empty_namespace && !IsNamespaceRendered (string.Empty, string.Empty))
 249                                 res.Append (" xmlns=\"\"");
 250
 251                         list.Sort (new XmlDsigC14NTransformNamespacesComparer ());
 252                         foreach (object obj in list) {
 253                                 XmlNode attribute = (obj as XmlNode);
 254                                 if (attribute != null) {
 255                                         res.Append (" ");
 256                                         res.Append (attribute.Name);
 257                                         res.Append ("=\"");
 258                                         res.Append (attribute.Value);
 259                                         res.Append ("\"");
 260                                 }
 261                         }
 262
 263                         // move the rendered namespaces stack
 264                         if (visible) {
 265                                 prevVisibleNamespacesStart = prevVisibleNamespacesEnd;
 266                                 prevVisibleNamespacesEnd = visibleNamespaces.Count;
 267                         }
 268                 }
 269
 270                 // Attribute Axis
 271                 // In lexicographic order (ascending), process each node that
 272                 // is in the element's attribute axis and in the node-set.
 273                 //
 274                 // The processing of an element node E MUST be modified slightly
 275                 // when an XPath node-set is given as input and the element's
 276                 // parent is omitted from the node-set.
 277                 private void WriteAttributesAxis (XmlNode node)
 278                 {
 279                         // Console.WriteLine ("Debug: attributes");
 280
 281                         ArrayList list = new ArrayList ();
 282                         foreach (XmlNode attribute in node.Attributes) {
 283                                 if (!IsNamespaceNode (attribute) && IsNodeVisible (attribute))
 284                                         list.Add (attribute);
 285                         }
 286
 287                         // Add attributes from "xml" namespace for "inclusive" c14n only:
 288                         //
 289                         // The method for processing the attribute axis of an element E
 290                         // in the node-set is enhanced. All element nodes along E's
 291                         // ancestor axis are examined for nearest occurrences of
 292                         // attributes in the xml namespace, such as xml:lang and
 293                         // xml:space (whether or not they are in the node-set).
 294                         // From this list of attributes, remove any that are in E's
 295                         // attribute axis (whether or not they are in the node-set).
 296                         // Then, lexicographically merge this attribute list with the
 297                         // nodes of E's attribute axis that are in the node-set. The
 298                         // result of visiting the attribute axis is computed by
 299                         // processing the attribute nodes in this merged attribute list.
 300                         if (!exclusive && node.ParentNode != null && !IsNodeVisible (node.ParentNode.ParentNode)) {
 301                                 // if we have whole document then the node.ParentNode.ParentNode
 302                                 // is always visible
 303                                 for (XmlNode cur = node.ParentNode; cur != null; cur = cur.ParentNode) {
 304                                         foreach (XmlNode attribute in cur.Attributes) {
 305                                                 // we are looking for "xml:*" attributes
 306                                                 if (attribute.Prefix != "xml")
 307                                                         continue;
 308
 309                                                 // exclude ones that are in the node's attributes axis
 310                                                 if (node.Attributes.GetNamedItem (attribute.LocalName, attribute.NamespaceURI) != null)
 311                                                         continue;
 312
 313                                                 // finally check that we don't have the same attribute in our list
 314                                                 bool found = false;
 315                                                 foreach (object obj in list) {
 316                                                         XmlNode n = (obj as XmlNode);
 317                                                         if (n.Prefix == "xml" && n.LocalName == attribute.LocalName) {
 318                                                                 found = true;
 319                                                                 break;
 320                                                         }
 321                                                 }
 322
 323                                                 if (found)
 324                                                         continue;
 325
 326                                                 // now we can add this attribute to our list
 327                                                 list.Add (attribute);
 328                                         }
 329                                 }
 330                         }
 331
 332                         // sort namespaces and write results
 333                         list.Sort (new XmlDsigC14NTransformAttributesComparer ());
 334                         foreach (object obj in list) {
 335                                 XmlNode attribute = (obj as XmlNode);
 336                                 if (attribute != null) {
 337                                         res.Append (" ");
 338                                         res.Append (attribute.Name);
 339                                         res.Append ("=\"");
 340                                         res.Append (NormalizeString (attribute.Value, XmlNodeType.Attribute));
 341                                         res.Append ("\"");
 342                                 }
 343                         }
 344                 }
 345
 346                 // Text Nodes
 347                 // the string value, except all ampersands are replaced
 348                 // by &amp;, all open angle brackets (<) are replaced by &lt;, all closing
 349                 // angle brackets (>) are replaced by &gt;, and all #xD characters are
 350                 // replaced by &#xD;.
 351                 private void WriteTextNode (XmlNode node, bool visible)
 352                 {
 353                         // Console.WriteLine ("Debug: text node");
 354                         if (visible)
 355                                 res.Append (NormalizeString (node.Value, XmlNodeType.Text));
 356                 }
 357
 358                 // Comment Nodes
 359                 // Nothing if generating canonical XML without comments. For
 360                 // canonical XML with comments, generate the opening comment
 361                 // symbol (<!--), the string value of the node, and the
 362                 // closing comment symbol (-->). Also, a trailing #xA is rendered
 363                 // after the closing comment symbol for comment children of the
 364                 // root node with a lesser document order than the document
 365                 // element, and a leading #xA is rendered before the opening
 366                 // comment symbol of comment children of the root node with a
 367                 // greater document order than the document element. (Comment
 368                 // children of the root node represent comments outside of the
 369                 // top-level document element and outside of the document type
 370                 // declaration).
 371                 private void WriteCommentNode (XmlNode node, bool visible)
 372                 {
 373                         // Console.WriteLine ("Debug: comment node");
 374                         if (visible && comments) {
 375                             if (state == XmlCanonicalizerState.AfterDocElement)
 376                                     res.Append ("\x0A<!--");
 377                             else
 378                                     res.Append ("<!--");
 379
 380                             res.Append (NormalizeString (node.Value, XmlNodeType.Comment));
 381
 382                             if (state == XmlCanonicalizerState.BeforeDocElement)
 383                                     res.Append ("-->\x0A");
 384                             else
 385                                     res.Append ("-->");
 386                         }
 387                 }
 388
 389                 // Processing Instruction (PI) Nodes-
 390                 // The opening PI symbol (<?), the PI target name of the node,
 391                 // a leading space and the string value if it is not empty, and
 392                 // the closing PI symbol (?>). If the string value is empty,
 393                 // then the leading space is not added. Also, a trailing #xA is
 394                 // rendered after the closing PI symbol for PI children of the
 395                 // root node with a lesser document order than the document
 396                 // element, and a leading #xA is rendered before the opening PI
 397                 // symbol of PI children of the root node with a greater document
 398                 // order than the document element.
 399                 private void WriteProcessingInstructionNode (XmlNode node, bool visible)
 400                 {
 401                         // Console.WriteLine ("Debug: PI node");
 402
 403                         if (visible) {
 404                                 if (state == XmlCanonicalizerState.AfterDocElement)
 405                                         res.Append ("\x0A<?");
 406                                 else
 407                                         res.Append ("<?");
 408
 409                                 res.Append (node.Name);
 410                                 if (node.Value.Length > 0) {
 411                                         res.Append (" ");
 412                                         res.Append (NormalizeString (node.Value, XmlNodeType.ProcessingInstruction));
 413                                 }
 414
 415                                 if (state == XmlCanonicalizerState.BeforeDocElement)
 416                                         res.Append ("?>\x0A");
 417                                 else
 418                                         res.Append ("?>");
 419                         }
 420                 }
 421
 422                 private bool IsNodeVisible (XmlNode node)
 423                 {
 424                         // if node list is empty then we process whole document
 425                         if (xnl == null)
 426                                 return true;
 427
 428                         // walk thru the list
 429                         foreach (XmlNode xn in xnl) {
 430                                 if (node.Equals (xn))
 431                                         return true;
 432                         }
 433
 434                         return false;
 435                 }
 436
 437                 private bool IsNamespaceRendered (string prefix, string uri)
 438                 {
 439                         // if the default namespace xmlns="" is not re-defined yet
 440                         // then we do not want to print it out
 441                         bool IsEmptyNs = prefix == string.Empty && uri == string.Empty;
 442                         int start = (IsEmptyNs) ? 0 : prevVisibleNamespacesStart;
 443                         for (int i = visibleNamespaces.Count - 1; i >= start; i--) {
 444                                 XmlNode node = (visibleNamespaces[i] as XmlNode);
 445                                 if (node != null) {
 446                                         // get namespace prefix
 447                                         string p = string.Empty;
 448                                         if (node.Prefix == "xmlns")
 449                                                 p = node.LocalName;
 450                                         if (p == prefix)
 451                                                 return node.Value == uri;
 452                                 }
 453                         }
 454
 455                         return IsEmptyNs;
 456                 }
 457
 458                 private bool IsNamespaceNode (XmlNode node)
 459                 {
 460                         if (node == null || node.NodeType != XmlNodeType.Attribute)
 461                                 return false;
 462                         return node.NamespaceURI == "http://www.w3.org/2000/xmlns/";
 463                 }
 464
 465                 private string NormalizeString (string input, XmlNodeType type)
 466                 {
 467                         StringBuilder sb = new StringBuilder ();
 468                         for (int i = 0; i < input.Length; i++) {
 469                                 char ch = input[i];
 470                                 if (ch == '<' && (type == XmlNodeType.Attribute || type == XmlNodeType.Text))
 471                                         sb.Append ("&lt;");
 472                                 else if (ch == '>' && type == XmlNodeType.Text)
 473                                         sb.Append ("&gt;");
 474                                 else if (ch == '&' && (type == XmlNodeType.Attribute || type == XmlNodeType.Text))
 475                                         sb.Append ("&amp;");
 476                                 else if (ch == '\"' && type == XmlNodeType.Attribute)
 477                                         sb.Append ("&quot;");
 478                                 else if (ch == '\x09' && type == XmlNodeType.Attribute)
 479                                         sb.Append ("&#x9;");
 480                                 else if (ch == '\x0A' && type == XmlNodeType.Attribute)
 481                                         sb.Append ("&#xA;");
 482                                 else if (ch == '\x0D' && (type == XmlNodeType.Attribute ||
 483                                                           type == XmlNodeType.Text ||
 484                                                           type == XmlNodeType.Comment ||
 485                                                           type == XmlNodeType.ProcessingInstruction))
 486                                         sb.Append ("&#xD;");
 487                                 else
 488                                         sb.Append (ch);
 489                         }
 490
 491                         return sb.ToString ();
 492                 }
 493         }
 494
 495         internal class XmlDsigC14NTransformAttributesComparer : IComparer
 496         {
 497                 public int Compare (object x, object y)
 498                 {
 499                         XmlNode n1 = (x as XmlNode);
 500                         XmlNode n2 = (y as XmlNode);
 501
 502                         // simple cases
 503                         if (n1 == n2)
 504                                 return 0;
 505                         else if (n1 == null)
 506                                 return -1;
 507                         else if (n2 == null)
 508                                 return 1;
 509                         else if (n1.Prefix == n2.Prefix)
 510                                 return string.Compare (n1.LocalName, n2.LocalName);
 511
 512                         // Attributes in the default namespace are first
 513                         // because the default namespace is not applied to
 514                         // unqualified attributes
 515                         if (n1.Prefix == string.Empty)
 516                                 return -1;
 517                         else if (n2.Prefix == string.Empty)
 518                                 return 1;
 519
 520                         int ret = string.Compare (n1.NamespaceURI, n2.NamespaceURI);
 521                         if (ret == 0)
 522                                 ret = string.Compare (n1.LocalName, n2.LocalName);
 523                         return ret;
 524                 }
 525         }
 526
 527         internal class XmlDsigC14NTransformNamespacesComparer : IComparer
 528         {
 529                 public int Compare (object x, object y)
 530                 {
 531                         XmlNode n1 = (x as XmlNode);
 532                         XmlNode n2 = (y as XmlNode);
 533
 534                         // simple cases
 535                         if (n1 == n2)
 536                                 return 0;
 537                         else if (n1 == null)
 538                                 return -1;
 539                         else if (n2 == null)
 540                                 return 1;
 541                         else if (n1.Prefix == string.Empty)
 542                                 return -1;
 543                         else if (n2.Prefix == string.Empty)
 544                                 return 1;
 545
 546                         return string.Compare (n1.LocalName, n2.LocalName);
 547                 }
 548         }
 549 }
 550