xmlschemas.c

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

/*
 * 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: ??
 *
 * NOTES:
 *   - Elimated item creation for: <restriction>, <extension>, 
 *     <simpleContent>, <complexContent>, <list>, <union>
 *     
 */
#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>
#ifdef LIBXML_PATTERN_ENABLED
#include <libxml/pattern.h>
#endif

/* #define DEBUG 1 */

/* #define DEBUG_CONTENT 1 */

/* #define DEBUG_TYPE 1 */

/* #define DEBUG_CONTENT_REGEXP 1 */

/* #define DEBUG_AUTOMATA 1 */

/* #define DEBUG_ATTR_VALIDATION 1 */

/* #define DEBUG_UNION_VALIDATION 1 */

/* #define DEBUG_IDC 1 */

/* #define DEBUG_INCLUDES 1 */

#define DUMP_CONTENT_MODEL


#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 *xmlSchemaElemDesElemDecl = (const xmlChar *)
    "Element decl.";
static const xmlChar *xmlSchemaElemDesElemRef = (const xmlChar *)
    "Element ref.";
static const xmlChar *xmlSchemaElemDesAttrDecl = (const xmlChar *)
    "Attribute decl.";
static const xmlChar *xmlSchemaElemDesAttrRef = (const xmlChar *)
    "Attribute ref.";
static const xmlChar *xmlSchemaElemDesST = (const xmlChar *)
    "simple type";
static const xmlChar *xmlSchemaElemDesCT = (const xmlChar *)
    "complex type";
static const xmlChar *xmlSchemaElemModelGrDef = (const xmlChar *)
    "Model group";
#if 0
static const xmlChar *xmlSchemaElemModelGrRef = (const xmlChar *)
    "Model group ref.";
#endif

#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(str);							\
	str = NULL;							\
    }

#define IS_ANYTYPE(item)                           \
     ((item->type == XML_SCHEMA_TYPE_BASIC) &&     \
      (item->builtInType == XML_SCHEMAS_ANYTYPE))

#define IS_COMPLEX_TYPE(item)                      \
    ((item->type == XML_SCHEMA_TYPE_COMPLEX) ||    \
     (item->builtInType == XML_SCHEMAS_ANYTYPE))

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

#define IS_ANY_SIMPLE_TYPE(item)                   \
    ((item->type == XML_SCHEMA_TYPE_BASIC) &&      \
      (item->builtInType == XML_SCHEMAS_ANYSIMPLETYPE)) 

#define IS_NOT_TYPEFIXED(item)                      \
    ((item->type != XML_SCHEMA_TYPE_BASIC) &&       \
     ((item->flags & XML_SCHEMAS_TYPE_INTERNAL_RESOLVED) == 0))

#define HAS_COMPLEX_CONTENT(item)			 \
    ((item->contentType == XML_SCHEMA_CONTENT_MIXED) ||  \
     (item->contentType == XML_SCHEMA_CONTENT_EMPTY) ||  \
     (item->contentType == XML_SCHEMA_CONTENT_ELEMENTS))

#define GET_NODE(item) xmlSchemaGetComponentNode((xmlSchemaBasicItemPtr) item)

#define IS_MODEL_GROUP(item)                     \
    ((item->type == XML_SCHEMA_TYPE_SEQUENCE) || \
     (item->type == XML_SCHEMA_TYPE_CHOICE) ||   \
     (item->type == XML_SCHEMA_TYPE_ALL))

#if 0
#define WXS_GET_NEXT(item) xmlSchemaGetNextComponent((xmlSchemaBasicItemPtr) item)
#endif

/*
#define XML_SCHEMAS_VAL_WTSP_PRESERVE 0
#define XML_SCHEMAS_VAL_WTSP_REPLACE  1
#define XML_SCHEMAS_VAL_WTSP_COLLAPSE 2
*/

#define XML_SCHEMAS_PARSE_ERROR		1

#define SCHEMAS_PARSE_OPTIONS XML_PARSE_NOENT


/*
* XML_SCHEMA_VAL_LOCATE_BY_NSNAME = 1<<2
* locate schemata to be imported
* using the namespace name; otherwise
* the location URI will be used */

/*
* xmlSchemaParserOption:
*
* This is the set of XML Schema parser options.
*
typedef enum {
    XML_SCHEMA_PAR_LOCATE_BY_NSNAME	= 1<<0
	* locate schemata to be imported
	* using the namespace name; otherwise
	* the location URI will be used *
} xmlSchemaParserOption;
*/

typedef struct _xmlSchemaItemList xmlSchemaAssemble;
typedef xmlSchemaAssemble *xmlSchemaAssemblePtr;

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 */
};

struct _xmlSchemaParserCtxt {
    void *userData;             /* user specific data block */
    xmlSchemaValidityErrorFunc error;   /* the callback in case of errors */
    xmlSchemaValidityWarningFunc warning;       /* the callback in case of warning */
    xmlSchemaValidError err;
    int nberrors;
    xmlStructuredErrorFunc serror;

    xmlSchemaPtr topschema;	/* The main schema */
    xmlHashTablePtr namespaces;	/* Hash table of namespaces to schemas */

    xmlSchemaPtr schema;        /* The schema in use */
    const xmlChar *container;   /* the current element, group, ... */
    int counter;

    const xmlChar *URL;
    xmlDocPtr doc;
    int preserve;		/* Whether the doc should be freed  */

    const char *buffer;
    int size;

    /*
     * Used to build complex element content models
     */
    xmlAutomataPtr am;
    xmlAutomataStatePtr start;
    xmlAutomataStatePtr end;
    xmlAutomataStatePtr state;

    xmlDictPtr dict;		/* dictionnary for interned string names */
    int        includes;	/* the inclusion level, 0 for root or imports */
    xmlSchemaTypePtr ctxtType; /* The current context simple/complex type */
    xmlSchemaTypePtr parentItem; /* The current parent schema item */
    xmlSchemaAssemblePtr assemble;
    int options;
    xmlSchemaValidCtxtPtr vctxt;
    const xmlChar **localImports; /* list of locally imported namespaces */
    int sizeLocalImports;
    int nbLocalImports;
};


#define XML_SCHEMAS_ATTR_UNKNOWN 1
#define XML_SCHEMAS_ATTR_CHECKED 2
#define XML_SCHEMAS_ATTR_PROHIBITED 3
#define XML_SCHEMAS_ATTR_MISSING 4
#define XML_SCHEMAS_ATTR_INVALID_VALUE 5
#define XML_SCHEMAS_ATTR_TYPE_NOT_RESOLVED 6
#define XML_SCHEMAS_ATTR_INVALID_FIXED_VALUE 7
#define XML_SCHEMAS_ATTR_DEFAULT 8
#define XML_SCHEMAS_ATTR_VALIDATE_VALUE 9
#define XML_SCHEMAS_ATTR_WILD_NO_DECL 10

typedef struct _xmlSchemaAttrState xmlSchemaAttrState;
typedef xmlSchemaAttrState *xmlSchemaAttrStatePtr;
struct _xmlSchemaAttrState {
    xmlSchemaAttrStatePtr next;
    xmlAttrPtr attr;
    int state;
    xmlSchemaAttributePtr decl;
    const xmlChar *value;
};

/**
 * xmlSchemaBasicItem:
 *
 * The abstract base type for schema components.
 */
typedef struct _xmlSchemaBasicItem xmlSchemaBasicItem;
typedef xmlSchemaBasicItem *xmlSchemaBasicItemPtr;
struct _xmlSchemaBasicItem {
    xmlSchemaTypeType type;
};

/**
 * xmlSchemaAnnotItem:
 *
 * The abstract base type for annotated schema components.
 * (Extends xmlSchemaBasicItem)
 */
typedef struct _xmlSchemaAnnotItem xmlSchemaAnnotItem;
typedef xmlSchemaAnnotItem *xmlSchemaAnnotItemPtr;
struct _xmlSchemaAnnotItem {
    xmlSchemaTypeType type;
    xmlSchemaAnnotPtr annot;
};

/**
 * xmlSchemaTreeItem:
 *
 * The abstract base type for tree-like structured schema components.
 * (Extends xmlSchemaAnnotItem)
 */
typedef struct _xmlSchemaTreeItem xmlSchemaTreeItem;
typedef xmlSchemaTreeItem *xmlSchemaTreeItemPtr;
struct _xmlSchemaTreeItem {
    xmlSchemaTypeType type;
    xmlSchemaAnnotPtr annot;
    xmlSchemaTreeItemPtr next;
    xmlSchemaTreeItemPtr children;
};
    
/**
 * xmlSchemaQNameRef:
 *
 * A component reference item (not a schema component)
 * (Extends xmlSchemaBasicItem)
 */
typedef struct _xmlSchemaQNameRef xmlSchemaQNameRef;
typedef xmlSchemaQNameRef *xmlSchemaQNameRefPtr;
struct _xmlSchemaQNameRef {
    xmlSchemaTypeType type;
    xmlSchemaBasicItemPtr item;
    xmlSchemaTypeType itemType;
    const xmlChar *name;
    const xmlChar *targetNamespace;
};

/**
 * xmlSchemaParticle:
 *
 * A particle component.
 * (Extends xmlSchemaTreeItem)
 */
typedef struct _xmlSchemaParticle xmlSchemaParticle;
typedef xmlSchemaParticle *xmlSchemaParticlePtr;
struct _xmlSchemaParticle {
    xmlSchemaTypeType type;
    xmlSchemaAnnotPtr annot;
    xmlSchemaTreeItemPtr next; /* next particle (OR "element decl" OR "wildcard") */
    xmlSchemaTreeItemPtr children; /* the "term" ("model group" OR "group definition") */
    int minOccurs;
    int maxOccurs;  
    xmlNodePtr node;
};

/**
 * xmlSchemaModelGroup:
 *
 * A model group component.
 * (Extends xmlSchemaTreeItem)
 */
typedef struct _xmlSchemaModelGroup xmlSchemaModelGroup;
typedef xmlSchemaModelGroup *xmlSchemaModelGroupPtr;
struct _xmlSchemaModelGroup {
    xmlSchemaTypeType type; /* XML_SCHEMA_TYPE_SEQUENCE, XML_SCHEMA_TYPE_CHOICE, XML_SCHEMA_TYPE_ALL */
    xmlSchemaAnnotPtr annot;
    xmlSchemaTreeItemPtr next; /* not used */
    xmlSchemaTreeItemPtr children; /* first particle (OR "element decl" OR "wildcard") */
    xmlNodePtr node;
};

/**
 * xmlSchemaModelGroupDef:
 *
 * A model group definition component.
 * (Extends xmlSchemaTreeItem)
 */
typedef struct _xmlSchemaModelGroupDef xmlSchemaModelGroupDef;
typedef xmlSchemaModelGroupDef *xmlSchemaModelGroupDefPtr;
struct _xmlSchemaModelGroupDef {
    xmlSchemaTypeType type; /* XML_SCHEMA_TYPE_GROUP */
    xmlSchemaAnnotPtr annot;
    xmlSchemaTreeItemPtr next; /* not used */
    xmlSchemaTreeItemPtr children; /* the "model group" */
    const xmlChar *name;
    const xmlChar *targetNamespace;
    xmlNodePtr node;
};

typedef struct _xmlSchemaIDC xmlSchemaIDC;
typedef xmlSchemaIDC *xmlSchemaIDCPtr;

/**
 * xmlSchemaIDCSelect:
 *
 * The identity-constraint "field" and "selector" item, holding the
 * XPath expression.
 */
typedef struct _xmlSchemaIDCSelect xmlSchemaIDCSelect;
typedef xmlSchemaIDCSelect *xmlSchemaIDCSelectPtr;
struct _xmlSchemaIDCSelect {    
    xmlSchemaIDCSelectPtr next;
    xmlSchemaIDCPtr idc;
    int index; /* an index position if significant for IDC key-sequences */
    const xmlChar *xpath; /* the XPath expression */
    void *xpathComp; /* the compiled XPath expression */
};

/**
 * xmlSchemaIDC:
 *
 * The identity-constraint definition component.
 * (Extends xmlSchemaAnnotItem)
 */

struct _xmlSchemaIDC {
    xmlSchemaTypeType type;
    xmlSchemaAnnotPtr annot;
    xmlSchemaIDCPtr next;
    xmlNodePtr node;
    const xmlChar *name;    
    const xmlChar *targetNamespace;
    xmlSchemaIDCSelectPtr selector;
    xmlSchemaIDCSelectPtr fields;
    int nbFields;
    xmlSchemaQNameRefPtr ref;
};

/**
 * xmlSchemaIDCAug:
 *
 * The augmented IDC information used for validation.
 */
typedef struct _xmlSchemaIDCAug xmlSchemaIDCAug;
typedef xmlSchemaIDCAug *xmlSchemaIDCAugPtr;
struct _xmlSchemaIDCAug {
    xmlSchemaIDCAugPtr next; /* next in a list */
    xmlSchemaIDCPtr def; /* the IDC definition */
    int bubbleDepth; /* the lowest tree level to which IDC 
                        tables need to be bubbled upwards */
};

/**
 * xmlSchemaPSVIIDCKeySequence:
 *
 * The key sequence of a node table item.
 */
typedef struct _xmlSchemaPSVIIDCKey xmlSchemaPSVIIDCKey;
typedef xmlSchemaPSVIIDCKey *xmlSchemaPSVIIDCKeyPtr;
struct _xmlSchemaPSVIIDCKey {
    xmlSchemaTypePtr type;
    xmlSchemaValPtr compValue;
};

/**
 * xmlSchemaPSVIIDCNode:
 *
 * The node table item of a node table.
 */
typedef struct _xmlSchemaPSVIIDCNode xmlSchemaPSVIIDCNode;
typedef xmlSchemaPSVIIDCNode *xmlSchemaPSVIIDCNodePtr;
struct _xmlSchemaPSVIIDCNode {
    xmlNodePtr node;
    xmlSchemaPSVIIDCKeyPtr *keys;
};