mcs/class/System.XML/System.Xml.Schema/XmlSchemaInference.cs

   1 //
   2 // XmlSchemaInference.cs
   3 //
   4 // Author:
   5 //      Atsushi Enomoto <atsushi@ximian.com>
   6 //
   7 // Copyright (C)2004 Novell Inc.
   8 //
   9
  10 //
  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:
  18 //
  19 // The above copyright notice and this permission notice shall be
  20 // included in all copies or substantial portions of the Software.
  21 //
  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.
  29 //
  30
  31 #if NET_2_0
  32
  33 using System;
  34 using System.Collections;
  35 using System.Xml;
  36 using System.Xml.Schema;
  37
  38 using QName = System.Xml.XmlQualifiedName;
  39 using Form = System.Xml.Schema.XmlSchemaForm;
  40 using Use = System.Xml.Schema.XmlSchemaUse;
  41 using SOMList = System.Xml.Schema.XmlSchemaObjectCollection;
  42 using SOMObject = System.Xml.Schema.XmlSchemaObject;
  43 using Element = System.Xml.Schema.XmlSchemaElement;
  44 using Attr = System.Xml.Schema.XmlSchemaAttribute;
  45 using AttrGroup = System.Xml.Schema.XmlSchemaAttributeGroup;
  46 using AttrGroupRef = System.Xml.Schema.XmlSchemaAttributeGroupRef;
  47 using SimpleType = System.Xml.Schema.XmlSchemaSimpleType;
  48 using ComplexType = System.Xml.Schema.XmlSchemaComplexType;
  49 using SimpleModel = System.Xml.Schema.XmlSchemaSimpleContent;
  50 using SimpleExt = System.Xml.Schema.XmlSchemaSimpleContentExtension;
  51 using SimpleRst = System.Xml.Schema.XmlSchemaSimpleContentRestriction;
  52 using ComplexModel = System.Xml.Schema.XmlSchemaComplexContent;
  53 using ComplexExt = System.Xml.Schema.XmlSchemaComplexContentExtension;
  54 using ComplexRst = System.Xml.Schema.XmlSchemaComplexContentRestriction;
  55 using SimpleTypeRst = System.Xml.Schema.XmlSchemaSimpleTypeRestriction;
  56 using SimpleList = System.Xml.Schema.XmlSchemaSimpleTypeList;
  57 using SimpleUnion = System.Xml.Schema.XmlSchemaSimpleTypeUnion;
  58 using SchemaFacet = System.Xml.Schema.XmlSchemaFacet;
  59 using LengthFacet = System.Xml.Schema.XmlSchemaLengthFacet;
  60 using MinLengthFacet = System.Xml.Schema.XmlSchemaMinLengthFacet;
  61 using Particle = System.Xml.Schema.XmlSchemaParticle;
  62 using Sequence = System.Xml.Schema.XmlSchemaSequence;
  63 using Choice = System.Xml.Schema.XmlSchemaChoice;
  64
  65
  66 namespace System.Xml.Schema
  67 {
  68         [MonoTODO ("merge primitive types; infer gYearMonth too; in some cases sequence should contain element whose minOccurs=0 (no obvious rules right now); reject some non-supported schema components")]
  69         public class XmlSchemaInference
  70         {
  71                 public enum InferenceOption {
  72                         Restricted,
  73                         Relaxed,
  74                 }
  75
  76                 InferenceOption occurrence = InferenceOption.Restricted;
  77                 InferenceOption typeInference = InferenceOption.Restricted;
  78
  79                 public XmlSchemaInference ()
  80                 {
  81                 }
  82
  83                 public InferenceOption Occurrence {
  84                         get { return occurrence; }
  85                         set { occurrence = value; }
  86                 }
  87
  88                 public InferenceOption TypeInference {
  89                         get { return TypeInference; }
  90                         set { typeInference = value; }
  91                 }
  92
  93                 public XmlSchemaSet InferSchema (XmlReader xmlReader)
  94                 {
  95                         return InferSchema (xmlReader, new XmlSchemaSet ());
  96                 }
  97
  98                 public XmlSchemaSet InferSchema (XmlReader xmlReader,
  99                         XmlSchemaSet schemas)
 100                 {
 101                         return XsdInference.Process (xmlReader, schemas,
 102                                 occurrence == InferenceOption.Relaxed,
 103                                 typeInference == InferenceOption.Relaxed);
 104                 }
 105         }
 106
 107         class XsdInference
 108         {
 109                 public static XmlSchemaSet Process (XmlReader xmlReader,
 110                         XmlSchemaSet schemas,
 111                         bool laxOccurence,
 112                         bool laxTypeInference)
 113                 {
 114                         XsdInference impl = new XsdInference (xmlReader,
 115                                 schemas, laxOccurence, laxTypeInference);
 116                         impl.Run ();
 117                         return impl.schemas;
 118                 }
 119
 120                 public const string NamespaceXml =
 121                         "http://www.w3.org/XML/1998/namespace";
 122
 123                 public const string NamespaceXmlns =
 124                         "http://www.w3.org/2000/xmlns/";
 125
 126                 public const string XdtNamespace =
 127                         "http://www.w3.org/2003/11/xpath-datatypes";
 128
 129                 static readonly QName QNameString = new QName (
 130                         "string", XmlSchema.Namespace);
 131
 132                 static readonly QName QNameBoolean = new QName (
 133                         "boolean", XmlSchema.Namespace);
 134
 135                 static readonly QName QNameAnyType = new QName (
 136                         "anyType", XmlSchema.Namespace);
 137
 138                 static readonly QName QNameByte = new QName (
 139                         "byte", XmlSchema.Namespace);
 140
 141                 static readonly QName QNameUByte = new QName (
 142                         "unsignedByte", XmlSchema.Namespace);
 143
 144                 static readonly QName QNameShort = new QName (
 145                         "short", XmlSchema.Namespace);
 146
 147                 static readonly QName QNameUShort = new QName (
 148                         "unsignedShort", XmlSchema.Namespace);
 149
 150                 static readonly QName QNameInt = new QName (
 151                         "int", XmlSchema.Namespace);
 152
 153                 static readonly QName QNameUInt = new QName (
 154                         "unsignedInt", XmlSchema.Namespace);
 155
 156                 static readonly QName QNameLong = new QName (
 157                         "long", XmlSchema.Namespace);
 158
 159                 static readonly QName QNameULong = new QName (
 160                         "unsignedLong", XmlSchema.Namespace);
 161
 162                 static readonly QName QNameDecimal = new QName (
 163                         "decimal", XmlSchema.Namespace);
 164
 165                 static readonly QName QNameUDecimal = new QName (
 166                         "unsignedDecimal", XmlSchema.Namespace);
 167
 168                 static readonly QName QNameDouble = new QName (
 169                         "double", XmlSchema.Namespace);
 170
 171                 static readonly QName QNameFloat = new QName (
 172                         "float", XmlSchema.Namespace);
 173
 174                 static readonly QName QNameDateTime = new QName (
 175                         "dateTime", XmlSchema.Namespace);
 176
 177                 static readonly QName QNameDuration = new QName (
 178                         "duration", XmlSchema.Namespace);
 179
 180                 XmlReader source;
 181                 XmlSchemaSet schemas;
 182                 bool laxOccurence;
 183                 bool laxTypeInference;
 184
 185                 Hashtable newElements = new Hashtable ();
 186                 Hashtable newAttributes = new Hashtable ();
 187
 188                 private XsdInference (XmlReader xmlReader,
 189                         XmlSchemaSet schemas,
 190                         bool laxOccurence,
 191                         bool laxTypeInference)
 192                 {
 193                         this.source = xmlReader;
 194                         this.schemas = schemas;
 195                         this.laxOccurence = laxOccurence;
 196                         this.laxTypeInference = laxTypeInference;
 197                 }
 198
 199                 private void Run ()
 200                 {
 201                         // XmlSchemaSet need to be compiled.
 202                         schemas.Compile ();
 203
 204                         // move to top-level element
 205                         source.MoveToContent ();
 206                         int depth = source.Depth;
 207                         if (source.NodeType != XmlNodeType.Element)
 208                                 throw new ArgumentException ("Argument XmlReader content is expected to be an element.");
 209
 210                         QName qname = new QName (source.LocalName,
 211                                 source.NamespaceURI);
 212                         Element el = GetGlobalElement (qname);
 213                         if (el == null) {
 214                                 el = CreateGlobalElement (qname);
 215                                 InferElement (el, qname.Namespace, true);
 216                         }
 217                         else
 218                                 InferElement (el, qname.Namespace, false);
 219
 220                         // finally compile again.
 221                         schemas.Compile ();
 222                 }
 223
 224                 private void IncludeXmlAttributes ()
 225                 {
 226                         if (schemas.Schemas (NamespaceXml).Count == 0)
 227                                 // FIXME: do it from resources.
 228                                 schemas.Add (NamespaceXml,
 229                                         "http://www.w3.org/2001/xml.xsd");
 230                 }
 231
 232                 private void InferElement (Element el, string ns, bool isNew)
 233                 {
 234                         // Quick check for reference to another definition
 235                         // (i.e. element ref='...' that should be redirected)
 236                         if (el.RefName != QName.Empty) {
 237                                 Element body = GetGlobalElement (el.RefName);
 238                                 if (body == null) {
 239                                         body = CreateElement (el.RefName);
 240                                         InferElement (body, ns, true);
 241                                 }
 242                                 else
 243                                         InferElement (body, ns, isNew);
 244                                 return;
 245                         }
 246
 247                         // Attributes
 248                         if (source.MoveToFirstAttribute ()) {
 249                                 InferAttributes (el, ns, isNew);
 250                                 source.MoveToElement ();
 251                         }
 252
 253                         // Content
 254                         if (source.IsEmptyElement) {
 255                                 InferAsEmptyElement (el, ns, isNew);
 256                                 source.Read ();
 257                                 source.MoveToContent ();
 258                         }
 259                         else {
 260                                 InferContent (el, ns, isNew);
 261                                 source.ReadEndElement ();
 262                         }
 263                         if (el.SchemaType == null &&
 264                                 el.SchemaTypeName == QName.Empty)
 265                                 el.SchemaTypeName = QNameString;
 266                 }
 267
 268                 #region Attribute Inference
 269
 270                 private Hashtable CollectAttrTable (SOMList attList)
 271                 {
 272                         // get attribute definition table.
 273                         Hashtable table = new Hashtable ();
 274                         foreach (XmlSchemaObject obj in attList) {
 275                                 Attr attr = obj as Attr;
 276                                 if (attr == null)
 277                                         throw Error (obj, String.Format ("Attribute inference only supports direct attribute definition. {0} is not supported.", obj.GetType ()));
 278                                 if (attr.RefName != QName.Empty)
 279                                         table.Add (attr.RefName, attr);
 280                                 else
 281                                         table.Add (new QName (attr.Name, ""),
 282                                                 attr);
 283                         }
 284                         return table;
 285                 }
 286
 287                 private void InferAttributes (Element el, string ns, bool isNew)
 288                 {
 289                         // Now this element is going to have complexType.
 290                         // It currently not, then we have to replace it.
 291                         ComplexType ct = null;
 292                         SOMList attList = null;
 293                         Hashtable table = null;
 294
 295                         do {
 296                                 switch (source.NamespaceURI) {
 297                                 case NamespaceXml:
 298                                         if (schemas.Schemas (
 299                                                 NamespaceXml) .Count == 0)
 300                                                 IncludeXmlAttributes ();
 301                                         break;
 302                                 case XmlSchema.InstanceNamespace:
 303                                         if (source.LocalName == "nil")
 304                                                 el.IsNillable = true;
 305                                         // all other xsi:* atts are ignored
 306                                         continue;
 307                                 case NamespaceXmlns:
 308                                         continue;
 309                                 }
 310                                 if (ct == null) {
 311                                         ct = ToComplexType (el);
 312                                         attList = GetAttributes (ct);
 313                                         table = CollectAttrTable (attList);
 314                                 }
 315                                 QName attrName = new QName (
 316                                         source.LocalName, source.NamespaceURI);
 317                                 Attr attr = table [attrName] as Attr;
 318                                 if (attr == null) {
 319                                         attList.Add (InferNewAttribute (
 320                                                 attrName, isNew));
 321                                 } else {
 322                                         table.Remove (attrName);
 323                                         if (attr.RefName != null &&
 324                                                 attr.RefName != QName.Empty)
 325                                                 continue; // just a reference
 326                                         InferMergedAttribute (attr);
 327                                 }
 328                         } while (source.MoveToNextAttribute ());
 329
 330                         // mark all attr definitions that did not appear
 331                         // as optional.
 332                         if (table != null)
 333                                 foreach (Attr attr in table.Values)
 334                                         attr.Use = Use.Optional;
 335                 }
 336
 337                 private XmlSchemaAttribute InferNewAttribute (
 338                         QName attrName, bool isNewTypeDefinition)
 339                 {
 340                         Attr attr = null;
 341                         bool mergedRequired = false;
 342                         if (attrName.Namespace.Length > 0) {
 343                                 // global attribute; might be already defined.
 344                                 attr = GetGlobalAttribute (attrName) as Attr;
 345                                 if (attr == null) {
 346                                         attr = CreateGlobalAttribute (attrName);
 347                                         attr.SchemaTypeName =
 348                                                 InferSimpleType (source.Value);
 349                                 } else {
 350                                         InferMergedAttribute (attr);
 351                                         mergedRequired =
 352                                                 attr.Use == Use.Required;
 353                                 }
 354                                 attr = new Attr ();
 355                                 attr.RefName = attrName;
 356                         } else {
 357                                 // local attribute
 358                                 attr = new Attr ();
 359                                 attr.Name = attrName.Name;
 360                                 attr.SchemaTypeName =
 361                                         InferSimpleType (source.Value);
 362                         }
 363                         if (!laxOccurence &&
 364                                 (isNewTypeDefinition || mergedRequired))
 365                                 attr.Use = Use.Required;
 366                         else
 367                                 attr.Use = Use.Optional;
 368
 369                         return attr;
 370                 }
 371
 372                 // validate string value agains attr and
 373                 // if invalid, then relax the type.
 374                 private void InferMergedAttribute (Attr attr)
 375                 {
 376                         attr.SchemaTypeName = InferMergedType (source.Value,
 377                                 attr.SchemaTypeName);
 378                         attr.SchemaType = null;
 379                 }
 380
 381                 private QName InferMergedType (string value, QName typeName)
 382                 {
 383                         // examine value against specified type and
 384                         // if unacceptable, then return a relaxed type.
 385
 386                         SimpleType st = XmlSchemaType.GetBuiltInSimpleType (
 387                                 typeName);
 388                         if (st == null) // non-primitive type => see above.
 389                                 return QNameString;
 390                         do {
 391                                 try {
 392                                         st.Datatype.ParseValue (value,
 393                                                 source.NameTable,
 394                                                 source as IXmlNamespaceResolver);
 395                                         return typeName;
 396                                 } catch {
 397                                         st = st.BaseXmlSchemaType as XmlSchemaSimpleType;
 398                                         typeName = st != null ? st.QualifiedName : QName.Empty;
 399                                 }
 400                         } while (typeName != QName.Empty);
 401                         return QNameString;
 402                 }
 403
 404                 private SOMList GetAttributes (ComplexType ct)
 405                 {
 406                         if (ct.ContentModel == null)
 407                                 return ct.Attributes;
 408
 409                         SimpleModel sc = ct.ContentModel as SimpleModel;
 410                         if (sc != null) {
 411                                 SimpleExt sce = sc.Content as SimpleExt;
 412                                 if (sce != null)
 413                                         return sce.Attributes;
 414                                 SimpleRst scr = sc.Content as SimpleRst;
 415                                 if (scr != null)
 416                                         return scr.Attributes;
 417                                 else
 418                                         throw Error (sc, "Invalid simple content model.");
 419                         }
 420                         ComplexModel cc = ct.ContentModel as ComplexModel;
 421                         if (cc != null) {
 422                                 ComplexExt cce = cc.Content as ComplexExt;
 423                                 if (cce != null)
 424                                         return cce.Attributes;
 425                                 ComplexRst ccr = cc.Content as ComplexRst;
 426                                 if (ccr != null)
 427                                         return ccr.Attributes;
 428                                 else
 429                                         throw Error (cc, "Invalid simple content model.");
 430                         }
 431                         throw Error (cc, "Invalid complexType. Should not happen.");
 432                 }
 433
 434                 private ComplexType ToComplexType (Element el)
 435                 {
 436                         QName name = el.SchemaTypeName;
 437                         XmlSchemaType type = el.SchemaType;
 438
 439                         // 1. element type is complex.
 440                         ComplexType ct = type as ComplexType;
 441                         if (ct != null)
 442                                 return ct;
 443
 444                         // 2. reference to global complexType.
 445                         XmlSchemaType globalType = schemas.GlobalTypes [name]
 446                                 as XmlSchemaType;
 447                         ct = globalType as ComplexType;
 448                         if (ct != null)
 449                                 return ct;
 450
 451                         ct = new ComplexType ();
 452                         el.SchemaType = ct;
 453                         el.SchemaTypeName = QName.Empty;
 454
 455                         // 3. base type name is xs:anyType or no specification.
 456                         // <xs:complexType />
 457                         if (name == QNameAnyType)
 458                                 return ct;
 459                         else if (type == null && name == QName.Empty)
 460                                 return ct;
 461
 462                         SimpleModel sc = new SimpleModel ();
 463                         ct.ContentModel = sc;
 464
 465                         // 4. type is simpleType
 466                         //    -> extension of existing simple type.
 467                         SimpleType st = type as SimpleType;
 468                         if (st != null) {
 469                                 SimpleRst scr = new SimpleRst ();
 470                                 scr.BaseType = st;
 471                                 sc.Content = scr;
 472                                 return ct;
 473                         }
 474
 475                         SimpleExt sce = new SimpleExt ();
 476                         sc.Content = sce;
 477
 478                         // 5. type name points to primitive type
 479                         //    -> simple extension of a primitive type
 480                         st = XmlSchemaType.GetBuiltInSimpleType (name);
 481                         if (st != null) {
 482                                 sce.BaseTypeName = name;
 483                                 return ct;
 484                         }
 485
 486                         // 6. type name points to global simpleType.
 487                         st = globalType as SimpleType;
 488                         if (st != null) {
 489                                 sce.BaseTypeName = name;
 490                                 return ct;
 491                         }
 492
 493                         throw Error (el, "Unexpected schema component that contains simpleTypeName that could not be resolved.");
 494                 }
 495
 496                 #endregion
 497
 498                 #region Element Type
 499
 500                 private void InferAsEmptyElement (Element el, string ns,
 501                         bool isNew)
 502                 {
 503                         ComplexType ct = el.SchemaType as ComplexType;
 504                         if (ct != null) {
 505                                 SimpleModel sm =
 506                                         ct.ContentModel as SimpleModel;
 507                                 if (sm != null) {
 508                                         ToEmptiableSimpleContent (sm, isNew);
 509                                         return;
 510                                 }
 511
 512                                 ComplexModel cm = ct.ContentModel
 513                                         as ComplexModel;
 514                                 if (cm != null) {
 515                                         ToEmptiableComplexContent (cm, isNew);
 516                                         return;
 517                                 }
 518
 519                                 if (ct.Particle != null)
 520                                         ct.Particle.MinOccurs = 0;
 521                                 return;
 522                         }
 523                         SimpleType st = el.SchemaType as SimpleType;
 524                         if (st != null) {
 525                                 st = MakeBaseTypeAsEmptiable (st);
 526                                 switch (st.QualifiedName.Namespace) {
 527                                 case XmlSchema.Namespace:
 528                                 case XdtNamespace:
 529                                         el.SchemaTypeName = st.QualifiedName;
 530                                         break;
 531                                 default:
 532                                         el.SchemaType =st;
 533                                         break;
 534                                 }
 535                         }
 536                 }
 537
 538                 private SimpleType MakeBaseTypeAsEmptiable (SimpleType st)
 539                 {
 540                         switch (st.QualifiedName.Namespace) {
 541                         case XmlSchema.Namespace:
 542                         case XdtNamespace:
 543                                 // If a primitive type
 544                                 return XmlSchemaType.GetBuiltInSimpleType (
 545                                         XmlTypeCode.String);
 546                         }
 547                         SimpleTypeRst str = st.Content as SimpleTypeRst;
 548                         if (str != null) {
 549                                 ArrayList al = null;
 550                                 foreach (SchemaFacet f in str.Facets) {
 551                                         if (f is LengthFacet ||
 552                                                 f is MinLengthFacet) {
 553                                                 if (al == null)
 554                                                         al = new ArrayList ();
 555                                                 al.Add (f);
 556                                         }
 557                                 }
 558                                 foreach (SchemaFacet f in al)
 559                                         str.Facets.Remove (f);
 560                                 if (str.BaseType != null)
 561                                         str.BaseType =
 562                                                 MakeBaseTypeAsEmptiable (st);
 563                                 else
 564                                         // It might have a reference to an
 565                                         // external simple type, but there is
 566                                         // no assurance that any of those
 567                                         // external types allow an empty
 568                                         // string. So just set base type as
 569                                         // xs:string.
 570                                         str.BaseTypeName = QNameString;
 571                         } // union/list can have empty string value.
 572
 573                         return st;
 574                 }
 575
 576                 private void ToEmptiableSimpleContent (
 577                         SimpleModel sm, bool isNew)
 578                 {
 579                         SimpleExt se = sm.Content as SimpleExt;
 580                         if (se != null)
 581                                 se.BaseTypeName = QNameString;
 582                         else {
 583                                 SimpleRst sr = sm.Content
 584                                         as SimpleRst;
 585                                 if (sr == null)
 586                                         throw Error (sm, "Invalid simple content model was passed.");
 587                                 sr.BaseTypeName = QNameString;
 588                                 sr.BaseType = null;
 589                         }
 590                 }
 591
 592                 private void ToEmptiableComplexContent (
 593                         ComplexModel cm, bool isNew)
 594                 {
 595                         ComplexExt ce = cm.Content
 596                                 as ComplexExt;
 597                         if (ce != null) {
 598                                 if (ce.Particle != null)
 599                                         ce.Particle.MinOccurs = 0;
 600                                 else if (ce.BaseTypeName != null &&
 601                                         ce.BaseTypeName != QName.Empty &&
 602                                         ce.BaseTypeName != QNameAnyType)
 603                                         throw Error (ce, "Complex type content extension has a reference to an external component that is not supported.");
 604                         }
 605                         else {
 606                                 ComplexRst cr = cm.Content
 607                                         as ComplexRst;
 608                                 if (cr == null)
 609                                         throw Error (cm, "Invalid complex content model was passed.");
 610                                 if (cr.Particle != null)
 611                                         cr.Particle.MinOccurs = 0;
 612                                 else if (cr.BaseTypeName != null &&
 613                                         cr.BaseTypeName != QName.Empty &&
 614                                         cr.BaseTypeName != QNameAnyType)
 615                                         throw Error (cr, "Complex type content extension has a reference to an external component that is not supported.");
 616                         }
 617                 }
 618
 619                 private void InferContent (Element el, string ns, bool isNew)
 620                 {
 621                         source.Read ();
 622                         source.MoveToContent ();
 623                         switch (source.NodeType) {
 624                         case XmlNodeType.EndElement:
 625                                 InferAsEmptyElement (el, ns, isNew);
 626                                 break;
 627                         case XmlNodeType.Element:
 628                                 InferComplexContent (el, ns, isNew);
 629                                 break;
 630                         case XmlNodeType.Text:
 631                         case XmlNodeType.CDATA:
 632                         case XmlNodeType.SignificantWhitespace:
 633                                 InferTextContent (el, isNew);
 634                                 source.MoveToContent ();
 635                                 if (source.NodeType == XmlNodeType.Element)
 636                                         goto case XmlNodeType.Element;
 637                                 break;
 638                         case XmlNodeType.Whitespace:
 639                                 InferContent (el, ns, isNew); // skip and retry
 640                                 break;
 641                         }
 642                 }
 643
 644                 private void InferComplexContent (Element el, string ns,
 645                         bool isNew)
 646                 {
 647                         ComplexType ct = ToComplexType (el);
 648                         ToComplexContentType (ct);
 649
 650                         int position = 0;
 651                         bool consumed = false;
 652
 653                         do {
 654                                 switch (source.NodeType) {
 655                                 case XmlNodeType.Element:
 656                                         Sequence s = PopulateSequence (ct);
 657                                         Choice c = s.Items.Count > 0 ?
 658                                                 s.Items [0] as Choice :
 659                                                 null;
 660                                         if (c != null)
 661                                                 ProcessLax (c, ns);
 662                                         else
 663                                                 ProcessSequence (ct, s, ns,
 664                                                         ref position,
 665                                                         ref consumed,
 666                                                         isNew);
 667                                         source.MoveToContent ();
 668                                         break;
 669                                 case XmlNodeType.Text:
 670                                 case XmlNodeType.CDATA:
 671                                 case XmlNodeType.SignificantWhitespace:
 672                                         MarkAsMixed (ct);
 673                                         source.ReadString ();
 674                                         source.MoveToContent ();
 675                                         break;
 676                                 case XmlNodeType.EndElement:
 677                                         return; // finished
 678                                 case XmlNodeType.None:
 679                                         throw new NotImplementedException ("Internal Error: Should not happen.");
 680                                 }
 681                         } while (true);
 682                 }
 683
 684                 private void InferTextContent (Element el, bool isNew)
 685                 {
 686                         string value = source.ReadString ();
 687                         if (el.SchemaType == null) {
 688                                 if (el.SchemaTypeName == QName.Empty) {
 689                                         // no type information -> infer type
 690                                         if (isNew)
 691                                                 el.SchemaTypeName =
 692                                                         InferSimpleType (
 693                                                         value);
 694                                         else
 695                                                 el.SchemaTypeName =
 696                                                         QNameString;
 697                                         return;
 698                                 }
 699                                 switch (el.SchemaTypeName.Namespace) {
 700                                 case XmlSchema.Namespace:
 701                                 case XdtNamespace:
 702                                         // existing primitive type
 703                                         el.SchemaTypeName = InferMergedType (
 704                                                 value, el.SchemaTypeName);
 705                                         break;
 706                                 default:
 707                                         ComplexType ct = schemas.GlobalTypes [
 708                                                 el.SchemaTypeName]
 709                                                 as ComplexType;
 710                                         // If it is complex, then just set
 711                                         // mixed='true' (type cannot be set.)
 712                                         // If it is simple, then we cannot
 713                                         // make sure that string value is
 714                                         // valid. So just set as xs:string.
 715                                         if (ct != null)
 716                                                 MarkAsMixed (ct);
 717                                         else
 718                                                 el.SchemaTypeName = QNameString;
 719                                         break;
 720                                 }
 721                                 return;
 722                         }
 723                         // simpleType
 724                         SimpleType st = el.SchemaType as SimpleType;
 725                         if (st != null) {
 726                                 // If simple, then (described above)
 727                                 el.SchemaType = null;
 728                                 el.SchemaTypeName = QNameString;
 729                                 return;
 730                         }
 731
 732                         // complexType
 733                         ComplexType ect = el.SchemaType as ComplexType;
 734
 735                         SimpleModel sm = ect.ContentModel as SimpleModel;
 736                         if (sm == null) {
 737                                 // - ComplexContent
 738                                 MarkAsMixed (ect);
 739                                 return;
 740                         }
 741
 742                         // - SimpleContent
 743                         SimpleExt se = sm.Content as SimpleExt;
 744                         if (se != null)
 745                                 se.BaseTypeName = InferMergedType (value,
 746                                         se.BaseTypeName);
 747                         SimpleRst sr = sm.Content as SimpleRst;
 748                         if (sr != null) {
 749                                 sr.BaseTypeName = InferMergedType (value,
 750                                         sr.BaseTypeName);
 751                                 sr.BaseType = null;
 752                         }
 753                 }
 754
 755                 private void MarkAsMixed (ComplexType ct)
 756                 {
 757                         ComplexModel cm = ct.ContentModel as ComplexModel;
 758                         if (cm != null)
 759                                 cm.IsMixed = true;
 760                         else
 761                                 ct.IsMixed = true;
 762                 }
 763
 764                 #endregion
 765
 766                 #region Particles
 767
 768                 private void ProcessLax (Choice c, string ns)
 769                 {
 770                         foreach (Particle p in c.Items) {
 771                                 Element el = p as Element;
 772                                 if (el == null)
 773                                         throw Error (c, String.Format ("Target schema item contains unacceptable particle {0}. Only element is allowed here."));
 774                                 if (ElementMatches (el, ns)) {
 775                                         InferElement (el, ns, false);
 776                                         return;
 777                                 }
 778                         }
 779                         // append a new element particle to lax term.
 780                         Element nel = new Element ();
 781                         if (source.NamespaceURI == ns)
 782                                 nel.Name = source.LocalName;
 783                         else
 784                                 nel.RefName = new QName (source.LocalName,
 785                                         source.NamespaceURI);
 786                         InferElement (nel, source.NamespaceURI, true);
 787                         c.Items.Add (nel);
 788                 }
 789
 790                 private bool ElementMatches (Element el, string ns)
 791                 {
 792                         bool matches = false;
 793                         if (el.RefName != QName.Empty) {
 794                                 if (el.RefName.Name == source.LocalName &&
 795                                         el.RefName.Namespace ==
 796                                         source.NamespaceURI)
 797                                         matches = true;
 798                         }
 799                         else if (el.Name == source.LocalName &&
 800                                 ns == source.NamespaceURI)
 801                                         matches = true;
 802                         return matches;
 803                 }
 804
 805                 private void ProcessSequence (ComplexType ct, Sequence s,
 806                         string ns, ref int position, ref bool consumed,
 807                         bool isNew)
 808                 {
 809                         for (int i = 0; i < position; i++) {
 810                                 Element iel = s.Items [i] as Element;
 811                                 if (ElementMatches (iel, ns)) {
 812                                         // Sequence element type violation
 813                                         // might happen (might not, but we
 814                                         // cannot backtrack here). So switch
 815                                         // to sequence of choice* here.
 816                                         ProcessLax (ToSequenceOfChoice (s), ns);
 817                                         return;
 818                                 }
 819                         }
 820
 821                         if (s.Items.Count <= position) {
 822                                 QName name = new QName (source.LocalName,
 823                                         source.NamespaceURI);
 824                                 Element nel = CreateElement (name);
 825                                 InferElement (nel, ns, true);
 826                                 if (ns == name.Namespace)
 827                                         s.Items.Add (nel);
 828                                 else {
 829                                         Element re = new Element ();
 830                                         re.RefName = name;
 831                                         s.Items.Add (re);
 832                                 }
 833                                 consumed = true;
 834                                 return;
 835                         }
 836                         Element el = s.Items [position] as Element;
 837                         if (el == null)
 838                                 throw Error (s, String.Format ("Target complex type content sequence has an unacceptable type of particle {0}", s.Items [position]));
 839                         bool matches = ElementMatches (el, ns);
 840                         if (matches) {
 841                                 if (consumed)
 842                                         el.MaxOccursString = "unbounded";
 843                                 InferElement (el, source.NamespaceURI, false);
 844                                 source.MoveToContent ();
 845                                 switch (source.NodeType) {
 846                                 case XmlNodeType.None:
 847                                         if (source.NodeType ==
 848                                                 XmlNodeType.Element)
 849                                                 goto case XmlNodeType.Element;
 850                                         else if (source.NodeType ==
 851                                                 XmlNodeType.EndElement)
 852                                                 goto case XmlNodeType.EndElement;
 853                                         break;
 854                                 case XmlNodeType.Element:
 855                                         ProcessSequence (ct, s, ns, ref position,
 856                                                 ref consumed, isNew);
 857                                         break;
 858                                 case XmlNodeType.Text:
 859                                 case XmlNodeType.CDATA:
 860                                 case XmlNodeType.SignificantWhitespace:
 861                                         MarkAsMixed (ct);
 862                                         source.ReadString ();
 863                                         goto case XmlNodeType.None;
 864                                 case XmlNodeType.Whitespace:
 865                                         source.ReadString ();
 866                                         goto case XmlNodeType.None;
 867                                 case XmlNodeType.EndElement:
 868                                         return;
 869                                 default:
 870                                         source.Read ();
 871                                         break;
 872                                 }
 873                         }
 874                         else {
 875                                 if (consumed) {
 876                                         position++;
 877                                         consumed = false;
 878                                         ProcessSequence (ct, s, ns,
 879                                                 ref position, ref consumed,
 880                                                 isNew);
 881                                 }
 882                                 else
 883                                         ProcessLax (ToSequenceOfChoice (s), ns);
 884                         }
 885                 }
 886
 887                 // Note that it does not return the changed sequence.
 888                 private Choice ToSequenceOfChoice (Sequence s)
 889                 {
 890                         Choice c = new Choice ();
 891                         if (laxOccurence)
 892                                 c.MinOccurs = 0;
 893                         c.MaxOccursString = "unbounded";
 894                         foreach (Particle p in s.Items)
 895                                 c.Items.Add (p);
 896                         s.Items.Clear ();
 897                         s.Items.Add (c);
 898                         return c;
 899                 }
 900
 901                 // It makes complexType not to have Simple content model.
 902                 private void ToComplexContentType (ComplexType type)
 903                 {
 904                         SimpleModel sm = type.ContentModel as SimpleModel;
 905                         if (sm == null)
 906                                 return;
 907
 908                         SOMList atts = GetAttributes (type);
 909                         foreach (SOMObject o in atts)
 910                                 type.Attributes.Add (o);
 911                         // FIXME: need to copy AnyAttribute.
 912                         // (though not considered right now)
 913                         type.ContentModel = null;
 914                         type.IsMixed = true;
 915                 }
 916
 917                 private Sequence PopulateSequence (ComplexType ct)
 918                 {
 919                         Particle p = PopulateParticle (ct);
 920                         Sequence s = p as Sequence;
 921                         if (s != null)
 922                                 return s;
 923                         else
 924                                 throw Error (ct, String.Format ("Target complexType contains unacceptable type of particle {0}", p));
 925                 }
 926
 927                 private Sequence CreateSequence ()
 928                 {
 929                         Sequence s = new Sequence ();
 930                         if (laxOccurence)
 931                                 s.MinOccurs = 0;
 932                         return s;
 933                 }
 934
 935                 private Particle PopulateParticle (ComplexType ct)
 936                 {
 937                         if (ct.ContentModel == null) {
 938                                 if (ct.Particle == null)
 939                                         ct.Particle = CreateSequence ();
 940                                 return ct.Particle;
 941                         }
 942                         ComplexModel cm = ct.ContentModel as ComplexModel;
 943                         if (cm != null) {
 944                                 ComplexExt  ce = cm.Content as ComplexExt;
 945                                 if (ce != null) {
 946                                         if (ce.Particle == null)
 947                                                 ce.Particle = CreateSequence ();
 948                                         return ce.Particle;
 949                                 }
 950                                 ComplexRst cr = cm.Content as ComplexRst;
 951                                 if (cr != null) {
 952                                         if (cr.Particle == null)
 953                                                 cr.Particle = CreateSequence ();
 954                                         return cr.Particle;
 955                                 }
 956                         }
 957                         throw Error (ct, "Schema inference internal error. The complexType should have been converted to have a complex content.");
 958                 }
 959
 960                 #endregion
 961
 962                 #region String Value
 963
 964                 // primitive type inference.
 965                 // When running lax type inference, it just returns xs:string.
 966                 private QName InferSimpleType (string value)
 967                 {
 968                         if (laxTypeInference)
 969                                 return QNameString;
 970
 971                         switch (value) {
 972                         // 0 and 1 are not infered as byte unlike MS.XSDInfer
 973 //                      case "0":
 974 //                      case "1":
 975                         case "true":
 976                         case "false":
 977                                 return QNameBoolean;
 978                         }
 979                         try {
 980                                 long dec = XmlConvert.ToInt64 (value);
 981                                 if (byte.MinValue <= dec && dec <= byte.MaxValue)
 982                                         return QNameUByte;
 983                                 if (sbyte.MinValue <= dec && dec <= sbyte.MaxValue)
 984                                         return QNameByte;
 985                                 if (ushort.MinValue <= dec && dec <= ushort.MaxValue)
 986                                         return QNameUShort;
 987                                 if (short.MinValue <= dec && dec <= short.MaxValue)
 988                                         return QNameShort;
 989                                 if (uint.MinValue <= dec && dec <= uint.MaxValue)
 990                                         return QNameUInt;
 991                                 if (int.MinValue <= dec && dec <= int.MaxValue)
 992                                         return QNameInt;
 993                                 return QNameLong;
 994                         } catch (Exception) {
 995                         }
 996                         try {
 997                                 XmlConvert.ToUInt64 (value);
 998                                 return QNameULong;
 999                         } catch (Exception) {
1000                         }
1001                         try {
1002                                 XmlConvert.ToDecimal (value);
1003                                 return QNameDecimal;
1004                         } catch (Exception) {
1005                         }
1006                         try {
1007                                 double dbl = XmlConvert.ToDouble (value);
1008                                 if (float.MinValue <= dbl &&
1009                                         dbl <= float.MaxValue)
1010                                         return QNameFloat;
1011                                 else
1012                                         return QNameDouble;
1013                         } catch (Exception) {
1014                         }
1015                         try {
1016                                 // FIXME: also try DateTimeSerializationMode
1017                                 // and gYearMonth
1018                                 XmlConvert.ToDateTime (value);
1019                                 return QNameDateTime;
1020                         } catch (Exception) {
1021                         }
1022                         try {
1023                                 XmlConvert.ToTimeSpan (value);
1024                                 return QNameDuration;
1025                         } catch (Exception) {
1026                         }
1027
1028                         // xs:string
1029                         return QNameString;
1030                 }
1031
1032                 #endregion
1033
1034                 #region Utilities
1035
1036                 private Element GetGlobalElement (QName name)
1037                 {
1038                         Element el = newElements [name] as Element;
1039                         if (el == null)
1040                                 el = schemas.GlobalElements [name] as Element;
1041                         return el;
1042                 }
1043
1044                 private Attr GetGlobalAttribute (QName name)
1045                 {
1046                         Attr a = newElements [name] as Attr;
1047                         if (a == null)
1048                                 a = schemas.GlobalAttributes [name] as Attr;
1049                         return a;
1050                 }
1051
1052                 private Element CreateElement (QName name)
1053                 {
1054                         Element el = new Element ();
1055                         el.Name = name.Name;
1056                         return el;
1057                 }
1058
1059                 private Element CreateGlobalElement (QName name)
1060                 {
1061                         Element el = CreateElement (name);
1062                         XmlSchema schema = PopulateSchema (name.Namespace);
1063                         schema.Items.Add (el);
1064                         newElements.Add (name, el);
1065                         return el;
1066                 }
1067
1068                 private Attr CreateGlobalAttribute (QName name)
1069                 {
1070                         Attr attr = new Attr ();
1071                         XmlSchema schema = PopulateSchema (name.Namespace);
1072                         attr.Name = name.Name;
1073                         schema.Items.Add (attr);
1074                         newAttributes.Add (name, attr);
1075                         return attr;
1076                 }
1077
1078                 // Note that the return value never assures that all the
1079                 // components in the parameter ns must reside in it.
1080                 private XmlSchema PopulateSchema (string ns)
1081                 {
1082                         ICollection list = schemas.Schemas (ns);
1083                         if (list.Count > 0) {
1084                                 IEnumerator e = list.GetEnumerator ();
1085                                 e.MoveNext ();
1086                                 return (XmlSchema) e.Current;
1087                         }
1088                         XmlSchema s = new XmlSchema ();
1089                         if (ns != null && ns.Length > 0)
1090                                 s.TargetNamespace = ns;
1091                         s.ElementFormDefault = Form.Qualified;
1092                         s.AttributeFormDefault = Form.Unqualified;
1093                         schemas.Add (s);
1094                         return s;
1095                 }
1096
1097                 private XmlSchemaInferenceException Error (
1098                         XmlSchemaObject sourceObj,
1099                         string message)
1100                 {
1101                         // This override is mainly for schema component error.
1102                         return Error (sourceObj, false, message);
1103                 }
1104
1105                 private XmlSchemaInferenceException Error (
1106                         XmlSchemaObject sourceObj,
1107                         bool useReader,
1108                         string message)
1109                 {
1110                         string msg = String.Concat (
1111                                 message,
1112                                 sourceObj != null ?
1113                                         String.Format (". Related schema component is {0}",
1114                                                 sourceObj.SourceUri,
1115                                                 sourceObj.LineNumber,
1116                                                 sourceObj.LinePosition) :
1117                                         String.Empty,
1118                                 useReader ?
1119                                         String.Format (". {0}", source.BaseURI) :
1120                                         String.Empty);
1121
1122                         IXmlLineInfo li = source as IXmlLineInfo;
1123                         if (useReader && li != null)
1124                                 return new XmlSchemaInferenceException (
1125                                         msg, null, li.LineNumber,
1126                                         li.LinePosition);
1127                         else
1128                                 return new XmlSchemaInferenceException (msg);
1129                 }
1130
1131                 #endregion
1132         }
1133 }
1134
1135 #endif