xmlschemas.c

xml开源解析代码.版本为libxml2-2.6.29,可支持GB3212.网络消息发送XML时很有用.

/*
 * schemas.c : implementation of the XML Schema handling and
 *             schema validity checking
 *
 * See Copyright for the status of this software.
 *
 * Daniel Veillard <[e-mail]>
 */

/*
 * TODO:
 *   - when types are redefined in includes, check that all
 *     types in the redef list are equal
 *     -> need a type equality operation.
 *   - if we don't intend to use the schema for schemas, we
 *     need to validate all schema attributes (ref, type, name)
 *     against their types.
 *   - Eliminate item creation for: ??
 *
 * URGENT TODO:
 *   - For xsi-driven schema acquisition, augment the IDCs after every
 *     acquisition episode (xmlSchemaAugmentIDC).
 *
 * NOTES:
 *   - Elimated item creation for: <restriction>, <extension>,
 *     <simpleContent>, <complexContent>, <list>, <union>
 *
 * PROBLEMS:
 *   - http://lists.w3.org/Archives/Public/www-xml-schema-comments/2005JulSep/0337.html
 *     IDC XPath expression and chameleon includes: the targetNamespace is changed, so
 *     XPath will have trouble to resolve to this namespace, since not known.
 *
 *
 * CONSTRAINTS:
 *
 * Schema Component Constraint:
 *   All Group Limited (cos-all-limited)
 *   Status: complete
 *   (1.2)
 *     In xmlSchemaGroupDefReferenceTermFixup() and
 *   (2)
 *     In xmlSchemaParseModelGroup()
 *     TODO: Actually this should go to component-level checks,
 *     but is done here due to performance. Move it to an other layer
 *     is schema construction via an API is implemented.
 */
#define IN_LIBXML
#include "libxml.h"

#ifdef LIBXML_SCHEMAS_ENABLED

#include <string.h>
#include <libxml/xmlmemory.h>
#include <libxml/parser.h>
#include <libxml/parserInternals.h>
#include <libxml/hash.h>
#include <libxml/uri.h>
#include <libxml/xmlschemas.h>
#include <libxml/schemasInternals.h>
#include <libxml/xmlschemastypes.h>
#include <libxml/xmlautomata.h>
#include <libxml/xmlregexp.h>
#include <libxml/dict.h>
#include <libxml/encoding.h>
#include <libxml/xmlIO.h>
#ifdef LIBXML_PATTERN_ENABLED
#include <libxml/pattern.h>
#endif
#ifdef LIBXML_READER_ENABLED
#include <libxml/xmlreader.h>
#endif

/* #define DEBUG 1 */

/* #define DEBUG_CONTENT 1 */

/* #define DEBUG_TYPE 1 */

/* #define DEBUG_CONTENT_REGEXP 1 */

/* #define DEBUG_AUTOMATA 1 */

/* #define DEBUG_IDC */

/* #define DEBUG_IDC_NODE_TABLE */

/* #define WXS_ELEM_DECL_CONS_ENABLED */

#ifdef DEBUG_IDC
 #ifndef DEBUG_IDC_NODE_TABLE
  #define DEBUG_IDC_NODE_TABLE
 #endif
#endif   

/* #define ENABLE_PARTICLE_RESTRICTION 1 */

#define ENABLE_REDEFINE

/* #define ENABLE_NAMED_LOCALS */

/* #define ENABLE_IDC_NODE_TABLES_TEST */

#define DUMP_CONTENT_MODEL

#ifdef LIBXML_READER_ENABLED
/* #define XML_SCHEMA_READER_ENABLED */
#endif

#define UNBOUNDED (1 << 30)
#define TODO 								\
    xmlGenericError(xmlGenericErrorContext,				\
	    "Unimplemented block at %s:%d\n",				\
            __FILE__, __LINE__);

#define XML_SCHEMAS_NO_NAMESPACE (const xmlChar *) "##"

/*
 * The XML Schemas namespaces
 */
static const xmlChar *xmlSchemaNs = (const xmlChar *)
    "http://www.w3.org/2001/XMLSchema";

static const xmlChar *xmlSchemaInstanceNs = (const xmlChar *)
    "http://www.w3.org/2001/XMLSchema-instance";

static const xmlChar *xmlNamespaceNs = (const xmlChar *)
    "http://www.w3.org/2000/xmlns/";

/*
* Come casting macros.
*/
#define ACTXT_CAST (xmlSchemaAbstractCtxtPtr)
#define PCTXT_CAST (xmlSchemaParserCtxtPtr)
#define VCTXT_CAST (xmlSchemaValidCtxtPtr)
#define WXS_BASIC_CAST (xmlSchemaBasicItemPtr)
#define WXS_TREE_CAST (xmlSchemaTreeItemPtr)
#define WXS_PTC_CAST (xmlSchemaParticlePtr)
#define WXS_TYPE_CAST (xmlSchemaTypePtr)
#define WXS_ELEM_CAST (xmlSchemaElementPtr)
#define WXS_ATTR_GROUP_CAST (xmlSchemaAttributeGroupPtr)
#define WXS_ATTR_CAST (xmlSchemaAttributePtr)
#define WXS_ATTR_USE_CAST (xmlSchemaAttributeUsePtr)
#define WXS_ATTR_PROHIB_CAST (xmlSchemaAttributeUseProhibPtr)
#define WXS_MODEL_GROUPDEF_CAST (xmlSchemaModelGroupDefPtr)
#define WXS_MODEL_GROUP_CAST (xmlSchemaModelGroupPtr)
#define WXS_IDC_CAST (xmlSchemaIDCPtr)
#define WXS_QNAME_CAST (xmlSchemaQNameRefPtr)
#define WXS_LIST_CAST (xmlSchemaItemListPtr)

/*
* Macros to query common properties of components.
*/
#define WXS_ITEM_NODE(i) xmlSchemaGetComponentNode(WXS_BASIC_CAST (i))

#define WXS_ITEM_TYPE_NAME(i) xmlSchemaGetComponentTypeStr(WXS_BASIC_CAST (i))
/*
* Macros for element declarations.
*/
#define WXS_ELEM_TYPEDEF(e) (e)->subtypes

#define WXS_SUBST_HEAD(item) (item)->refDecl
/*
* Macros for attribute declarations.
*/
#define WXS_ATTR_TYPEDEF(a) (a)->subtypes
/*
* Macros for attribute uses.
*/
#define WXS_ATTRUSE_DECL(au) WXS_ATTR_CAST (WXS_ATTR_USE_CAST (au))->attrDecl

#define WXS_ATTRUSE_TYPEDEF(au) WXS_ATTR_TYPEDEF(WXS_ATTRUSE_DECL( WXS_ATTR_USE_CAST au))

#define WXS_ATTRUSE_DECL_NAME(au) (WXS_ATTRUSE_DECL(au))->name

#define WXS_ATTRUSE_DECL_TNS(au) (WXS_ATTRUSE_DECL(au))->targetNamespace
/*
* Macros for attribute groups.
*/
#define WXS_ATTR_GROUP_HAS_REFS(ag) ((WXS_ATTR_GROUP_CAST (ag))->flags & XML_SCHEMAS_ATTRGROUP_HAS_REFS)
#define WXS_ATTR_GROUP_EXPANDED(ag) ((WXS_ATTR_GROUP_CAST (ag))->flags & XML_SCHEMAS_ATTRGROUP_WILDCARD_BUILDED)
/*
* Macros for particles.
*/
#define WXS_PARTICLE(p) WXS_PTC_CAST (p)

#define WXS_PARTICLE_TERM(p) (WXS_PARTICLE(p))->children

#define WXS_PARTICLE_TERM_AS_ELEM(p) (WXS_ELEM_CAST WXS_PARTICLE_TERM(p))

#define WXS_PARTICLE_MODEL(p) WXS_MODEL_GROUP_CAST WXS_PARTICLE(p)->children
/*
* Macros for model groups definitions.
*/
#define WXS_MODELGROUPDEF_MODEL(mgd) (WXS_MODEL_GROUP_CAST (mgd))->children
/*
* Macros for model groups.
*/
#define WXS_IS_MODEL_GROUP(i) \
    (((i)->type == XML_SCHEMA_TYPE_SEQUENCE) || \
     ((i)->type == XML_SCHEMA_TYPE_CHOICE) || \
     ((i)->type == XML_SCHEMA_TYPE_ALL))

#define WXS_MODELGROUP_PARTICLE(mg) WXS_PTC_CAST (mg)->children
/*
* Macros for schema buckets.
*/
#define WXS_IS_BUCKET_INCREDEF(t) (((t) == XML_SCHEMA_SCHEMA_INCLUDE) || \
    ((t) == XML_SCHEMA_SCHEMA_REDEFINE))

#define WXS_IS_BUCKET_IMPMAIN(t) (((t) == XML_SCHEMA_SCHEMA_MAIN) || \
    ((t) == XML_SCHEMA_SCHEMA_IMPORT))

#define WXS_IMPBUCKET(b) ((xmlSchemaImportPtr) (b))

#define WXS_INCBUCKET(b) ((xmlSchemaIncludePtr) (b))
/*
* Macros for complex/simple types.
*/
#define WXS_IS_ANYTYPE(i) \
     (( (i)->type == XML_SCHEMA_TYPE_BASIC) && \
      ( (WXS_TYPE_CAST (i))->builtInType == XML_SCHEMAS_ANYTYPE))

#define WXS_IS_COMPLEX(i) \
    (((i)->type == XML_SCHEMA_TYPE_COMPLEX) || \
     ((i)->builtInType == XML_SCHEMAS_ANYTYPE))

#define WXS_IS_SIMPLE(item) \
    ((item->type == XML_SCHEMA_TYPE_SIMPLE) || \
     ((item->type == XML_SCHEMA_TYPE_BASIC) && \
      (item->builtInType != XML_SCHEMAS_ANYTYPE)))

#define WXS_IS_ANY_SIMPLE_TYPE(i) \
    (((i)->type == XML_SCHEMA_TYPE_BASIC) && \
      ((i)->builtInType == XML_SCHEMAS_ANYSIMPLETYPE))

#define WXS_IS_RESTRICTION(t) \
    ((t)->flags & XML_SCHEMAS_TYPE_DERIVATION_METHOD_RESTRICTION)

#define WXS_IS_EXTENSION(t) \
    ((t)->flags & XML_SCHEMAS_TYPE_DERIVATION_METHOD_EXTENSION)

#define WXS_IS_TYPE_NOT_FIXED(i) \
    (((i)->type != XML_SCHEMA_TYPE_BASIC) && \
     (((i)->flags & XML_SCHEMAS_TYPE_INTERNAL_RESOLVED) == 0))

#define WXS_IS_TYPE_NOT_FIXED_1(item) \
    (((item)->type != XML_SCHEMA_TYPE_BASIC) && \
     (((item)->flags & XML_SCHEMAS_TYPE_FIXUP_1) == 0))

#define WXS_TYPE_IS_GLOBAL(t) ((t)->flags & XML_SCHEMAS_TYPE_GLOBAL)

#define WXS_TYPE_IS_LOCAL(t) (((t)->flags & XML_SCHEMAS_TYPE_GLOBAL) == 0)
/*
* Macros for exclusively for complex types.
*/
#define WXS_HAS_COMPLEX_CONTENT(item) \
    ((item->contentType == XML_SCHEMA_CONTENT_MIXED) || \
     (item->contentType == XML_SCHEMA_CONTENT_EMPTY) || \
     (item->contentType == XML_SCHEMA_CONTENT_ELEMENTS))

#define WXS_HAS_SIMPLE_CONTENT(item) \
    ((item->contentType == XML_SCHEMA_CONTENT_SIMPLE) || \
     (item->contentType == XML_SCHEMA_CONTENT_BASIC))

#define WXS_HAS_MIXED_CONTENT(item) \
    (item->contentType == XML_SCHEMA_CONTENT_MIXED)

#define WXS_EMPTIABLE(t) \
    (xmlSchemaIsParticleEmptiable(WXS_PTC_CAST (t)->subtypes))

#define WXS_TYPE_CONTENTTYPE(t) (t)->subtypes

#define WXS_TYPE_PARTICLE(t) WXS_PTC_CAST (t)->subtypes

#define WXS_TYPE_PARTICLE_TERM(t) WXS_PARTICLE_TERM(WXS_TYPE_PARTICLE(t))
/*
* Macros for exclusively for simple types.
*/
#define WXS_LIST_ITEMTYPE(t) (t)->subtypes

#define WXS_IS_ATOMIC(t) (t->flags & XML_SCHEMAS_TYPE_VARIETY_ATOMIC)

#define WXS_IS_LIST(t) (t->flags & XML_SCHEMAS_TYPE_VARIETY_LIST)

#define WXS_IS_UNION(t) (t->flags & XML_SCHEMAS_TYPE_VARIETY_UNION)
/*
* Misc parser context macros.
*/
#define WXS_CONSTRUCTOR(ctx) (ctx)->constructor

#define WXS_HAS_BUCKETS(ctx) \
( (WXS_CONSTRUCTOR((ctx))->buckets != NULL) && \
(WXS_CONSTRUCTOR((ctx))->buckets->nbItems > 0) )

#define WXS_SUBST_GROUPS(ctx) WXS_CONSTRUCTOR((ctx))->substGroups

#define WXS_BUCKET(ctx) WXS_CONSTRUCTOR((ctx))->bucket

#define WXS_SCHEMA(ctx) (ctx)->schema

#define WXS_ADD_LOCAL(ctx, item) \
    xmlSchemaAddItemSize(&(WXS_BUCKET(ctx)->locals), 10, item)

#define WXS_ADD_GLOBAL(ctx, item) \
    xmlSchemaAddItemSize(&(WXS_BUCKET(ctx)->globals), 5, item)

#define WXS_ADD_PENDING(ctx, item) \
    xmlSchemaAddItemSize(&((ctx)->constructor->pending), 10, item)
/*
* xmlSchemaItemList macros.
*/
#define WXS_ILIST_IS_EMPTY(l) ((l == NULL) || ((l)->nbItems == 0))
/*
* Misc macros.
*/
#define IS_SCHEMA(node, type) \
   ((node != NULL) && (node->ns != NULL) && \
    (xmlStrEqual(node->name, (const xmlChar *) type)) && \
    (xmlStrEqual(node->ns->href, xmlSchemaNs)))

#define FREE_AND_NULL(str) if ((str) != NULL) { xmlFree((xmlChar *) (str)); str = NULL; }

/*
* Since we put the default/fixed values into the dict, we can
* use pointer comparison for those values.
* REMOVED: (xmlStrEqual((v1), (v2)))
*/
#define WXS_ARE_DEFAULT_STR_EQUAL(v1, v2) ((v1) == (v2))

#define INODE_NILLED(item) (item->flags & XML_SCHEMA_ELEM_INFO_NILLED)

#define CAN_PARSE_SCHEMA(b) (((b)->doc != NULL) && ((b)->parsed == 0))

#define HFAILURE if (res == -1) goto exit_failure;

#define HERROR if (res != 0) goto exit_error;

#define HSTOP(ctx) if ((ctx)->stop) goto exit;
/*
* Some flags used for various schema constraints.
*/
#define SUBSET_RESTRICTION  1<<0
#define SUBSET_EXTENSION    1<<1
#define SUBSET_SUBSTITUTION 1<<2
#define SUBSET_LIST         1<<3
#define SUBSET_UNION        1<<4

typedef struct _xmlSchemaNodeInfo xmlSchemaNodeInfo;
typedef xmlSchemaNodeInfo *xmlSchemaNodeInfoPtr;

typedef struct _xmlSchemaItemList xmlSchemaItemList;
typedef xmlSchemaItemList *xmlSchemaItemListPtr;
struct _xmlSchemaItemList {
    void **items;  /* used for dynamic addition of schemata */
    int nbItems; /* used for dynamic addition of schemata */
    int sizeItems; /* used for dynamic addition of schemata */
};

#define XML_SCHEMA_CTXT_PARSER 1
#define XML_SCHEMA_CTXT_VALIDATOR 2

typedef struct _xmlSchemaAbstractCtxt xmlSchemaAbstractCtxt;
typedef xmlSchemaAbstractCtxt *xmlSchemaAbstractCtxtPtr;
struct _xmlSchemaAbstractCtxt {
    int type; /* E.g. XML_SCHEMA_CTXT_VALIDATOR */
};

typedef struct _xmlSchemaBucket xmlSchemaBucket;
typedef xmlSchemaBucket *xmlSchemaBucketPtr;

#define XML_SCHEMA_SCHEMA_MAIN 0
#define XML_SCHEMA_SCHEMA_IMPORT 1
#define XML_SCHEMA_SCHEMA_INCLUDE 2
#define XML_SCHEMA_SCHEMA_REDEFINE 3

/**
 * xmlSchemaSchemaRelation: 
 *
 * Used to create a graph of schema relationships.
 */
typedef struct _xmlSchemaSchemaRelation xmlSchemaSchemaRelation;
typedef xmlSchemaSchemaRelation *xmlSchemaSchemaRelationPtr;
struct _xmlSchemaSchemaRelation {
    xmlSchemaSchemaRelationPtr next;
    int type; /* E.g. XML_SCHEMA_SCHEMA_IMPORT */
    const xmlChar *importNamespace;
    xmlSchemaBucketPtr bucket;
};

#define XML_SCHEMA_BUCKET_MARKED 1<<0
#define XML_SCHEMA_BUCKET_COMPS_ADDED 1<<1

struct _xmlSchemaBucket {
    int type;
    int flags;
    const xmlChar *schemaLocation;
    const xmlChar *origTargetNamespace;
    const xmlChar *targetNamespace;
    xmlDocPtr doc;
    xmlSchemaSchemaRelationPtr relations;
    int located;
    int parsed;
    int imported;
    int preserveDoc;
    xmlSchemaItemListPtr globals; /* Global components. */ 
    xmlSchemaItemListPtr locals; /* Local components. */
};

/**
 * xmlSchemaImport: 
 * (extends xmlSchemaBucket)
 *
 * Reflects a schema. Holds some information
 * about the schema and its toplevel components. Duplicate
 * toplevel components are not checked at this level.
 */
typedef struct _xmlSchemaImport xmlSchemaImport;
typedef xmlSchemaImport *xmlSchemaImportPtr;
struct _xmlSchemaImport {
    int type; /* Main OR import OR include. */
    int flags;
    const xmlChar *schemaLocation; /* The URI of the schema document. */
    /* For chameleon includes, @origTargetNamespace will be NULL */
    const xmlChar *origTargetNamespace;
    /* 
    * For chameleon includes, @targetNamespace will be the
    * targetNamespace of the including schema. 
    */
    const xmlChar *targetNamespace;
    xmlDocPtr doc; /* The schema node-tree. */
    /* @relations will hold any included/imported/redefined schemas. */
    xmlSchemaSchemaRelationPtr relations;
    int located;
    int parsed;
    int imported;
    int preserveDoc;
    xmlSchemaItemListPtr globals;
    xmlSchemaItemListPtr locals;
    /* The imported schema. */
    xmlSchemaPtr schema;
};

/*