valid.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

/*
 * valid.c : part of the code use to do the DTD handling and the validity
 *           checking
 *
 * See Copyright for the status of this software.
 *
 * [e-mail]
 */

#define IN_LIBXML
#include "libxml.h"

#include <string.h>

#ifdef HAVE_STDLIB_H
#include <stdlib.h>
#endif

#include <libxml/xmlmemory.h>
#include <libxml/hash.h>
#include <libxml/valid.h>
#include <libxml/parser.h>
#include <libxml/parserInternals.h>
#include <libxml/xmlerror.h>
#include <libxml/list.h>
#include <libxml/globals.h>

static xmlElementPtr xmlGetDtdElementDesc2(xmlDtdPtr dtd, const xmlChar *name,
	                           int create);
/* #define DEBUG_VALID_ALGO */
/* #define DEBUG_REGEXP_ALGO */

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

/************************************************************************
 *									*
 *			Error handling routines				*
 *									*
 ************************************************************************/

/**
 * xmlVErrMemory:
 * @ctxt:  an XML validation parser context
 * @extra:  extra informations
 *
 * Handle an out of memory error
 */
static void
xmlVErrMemory(xmlValidCtxtPtr ctxt, const char *extra)
{
    xmlGenericErrorFunc channel = NULL;
    xmlParserCtxtPtr pctxt = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        channel = ctxt->error;
        data = ctxt->userData;
	pctxt = ctxt->userData;
    }
    if (extra)
        __xmlRaiseError(NULL, channel, data,
                        pctxt, NULL, XML_FROM_VALID, XML_ERR_NO_MEMORY,
                        XML_ERR_FATAL, NULL, 0, extra, NULL, NULL, 0, 0,
                        "Memory allocation failed : %s\n", extra);
    else
        __xmlRaiseError(NULL, channel, data,
                        pctxt, NULL, XML_FROM_VALID, XML_ERR_NO_MEMORY,
                        XML_ERR_FATAL, NULL, 0, NULL, NULL, NULL, 0, 0,
                        "Memory allocation failed\n");
}

/**
 * xmlErrValid:
 * @ctxt:  an XML validation parser context
 * @error:  the error number
 * @extra:  extra informations
 *
 * Handle a validation error
 */
static void
xmlErrValid(xmlValidCtxtPtr ctxt, xmlParserErrors error,
            const char *msg, const char *extra)
{
    xmlGenericErrorFunc channel = NULL;
    xmlParserCtxtPtr pctxt = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        channel = ctxt->error;
        data = ctxt->userData;
	pctxt = ctxt->userData;
    }
    if (extra)
        __xmlRaiseError(NULL, channel, data,
                        pctxt, NULL, XML_FROM_VALID, error,
                        XML_ERR_ERROR, NULL, 0, extra, NULL, NULL, 0, 0,
                        msg, extra);
    else
        __xmlRaiseError(NULL, channel, data,
                        pctxt, NULL, XML_FROM_VALID, error,
                        XML_ERR_ERROR, NULL, 0, NULL, NULL, NULL, 0, 0,
                        msg);
}

#if defined(LIBXML_VALID_ENABLED) || defined(LIBXML_SCHEMAS_ENABLED)
/**
 * xmlErrValidNode:
 * @ctxt:  an XML validation parser context
 * @node:  the node raising the error
 * @error:  the error number
 * @str1:  extra informations
 * @str2:  extra informations
 * @str3:  extra informations
 *
 * Handle a validation error, provide contextual informations
 */
static void
xmlErrValidNode(xmlValidCtxtPtr ctxt,
                xmlNodePtr node, xmlParserErrors error,
                const char *msg, const xmlChar * str1,
                const xmlChar * str2, const xmlChar * str3)
{
    xmlStructuredErrorFunc schannel = NULL;
    xmlGenericErrorFunc channel = NULL;
    xmlParserCtxtPtr pctxt = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        channel = ctxt->error;
        data = ctxt->userData;
	pctxt = ctxt->userData;
    }
    __xmlRaiseError(schannel, channel, data, pctxt, node, XML_FROM_VALID, error,
                    XML_ERR_ERROR, NULL, 0,
                    (const char *) str1,
                    (const char *) str1,
                    (const char *) str3, 0, 0, msg, str1, str2, str3);
}
#endif /* LIBXML_VALID_ENABLED or LIBXML_SCHEMAS_ENABLED */

#ifdef LIBXML_VALID_ENABLED
/**
 * xmlErrValidNodeNr:
 * @ctxt:  an XML validation parser context
 * @node:  the node raising the error
 * @error:  the error number
 * @str1:  extra informations
 * @int2:  extra informations
 * @str3:  extra informations
 *
 * Handle a validation error, provide contextual informations
 */
static void
xmlErrValidNodeNr(xmlValidCtxtPtr ctxt,
                xmlNodePtr node, xmlParserErrors error,
                const char *msg, const xmlChar * str1,
                int int2, const xmlChar * str3)
{
    xmlStructuredErrorFunc schannel = NULL;
    xmlGenericErrorFunc channel = NULL;
    xmlParserCtxtPtr pctxt = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        channel = ctxt->error;
        data = ctxt->userData;
	pctxt = ctxt->userData;
    }
    __xmlRaiseError(schannel, channel, data, pctxt, node, XML_FROM_VALID, error,
                    XML_ERR_ERROR, NULL, 0,
                    (const char *) str1,
                    (const char *) str3,
                    NULL, int2, 0, msg, str1, int2, str3);
}

/**
 * xmlErrValidWarning:
 * @ctxt:  an XML validation parser context
 * @node:  the node raising the error
 * @error:  the error number
 * @str1:  extra information
 * @str2:  extra information
 * @str3:  extra information
 *
 * Handle a validation error, provide contextual information
 */
static void
xmlErrValidWarning(xmlValidCtxtPtr ctxt,
                xmlNodePtr node, xmlParserErrors error,
                const char *msg, const xmlChar * str1,
                const xmlChar * str2, const xmlChar * str3)
{
    xmlStructuredErrorFunc schannel = NULL;
    xmlGenericErrorFunc channel = NULL;
    xmlParserCtxtPtr pctxt = NULL;
    void *data = NULL;

    if (ctxt != NULL) {
        channel = ctxt->error;
        data = ctxt->userData;
	pctxt = ctxt->userData;
    }
    __xmlRaiseError(schannel, channel, data, pctxt, node, XML_FROM_VALID, error,
                    XML_ERR_WARNING, NULL, 0,
                    (const char *) str1,
                    (const char *) str1,
                    (const char *) str3, 0, 0, msg, str1, str2, str3);
}



#ifdef LIBXML_REGEXP_ENABLED
/*
 * If regexp are enabled we can do continuous validation without the
 * need of a tree to validate the content model. this is done in each
 * callbacks.
 * Each xmlValidState represent the validation state associated to the
 * set of nodes currently open from the document root to the current element.
 */


typedef struct _xmlValidState {
    xmlElementPtr	 elemDecl;	/* pointer to the content model */
    xmlNodePtr           node;		/* pointer to the current node */
    xmlRegExecCtxtPtr    exec;		/* regexp runtime */
} _xmlValidState;


static int
vstateVPush(xmlValidCtxtPtr ctxt, xmlElementPtr elemDecl, xmlNodePtr node) {
    if ((ctxt->vstateMax == 0) || (ctxt->vstateTab == NULL)) {
	ctxt->vstateMax = 10;
	ctxt->vstateTab = (xmlValidState *) xmlMalloc(ctxt->vstateMax *
		              sizeof(ctxt->vstateTab[0]));
        if (ctxt->vstateTab == NULL) {
	    xmlVErrMemory(ctxt, "malloc failed");
	    return(-1);
	}
    }

    if (ctxt->vstateNr >= ctxt->vstateMax) {
        xmlValidState *tmp;

	tmp = (xmlValidState *) xmlRealloc(ctxt->vstateTab,
	             2 * ctxt->vstateMax * sizeof(ctxt->vstateTab[0]));
        if (tmp == NULL) {
	    xmlVErrMemory(ctxt, "realloc failed");
	    return(-1);
	}
	ctxt->vstateMax *= 2;
	ctxt->vstateTab = tmp;
    }
    ctxt->vstate = &ctxt->vstateTab[ctxt->vstateNr];
    ctxt->vstateTab[ctxt->vstateNr].elemDecl = elemDecl;
    ctxt->vstateTab[ctxt->vstateNr].node = node;
    if ((elemDecl != NULL) && (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)) {
	if (elemDecl->contModel == NULL)
	    xmlValidBuildContentModel(ctxt, elemDecl);
	if (elemDecl->contModel != NULL) {
	    ctxt->vstateTab[ctxt->vstateNr].exec = 
		xmlRegNewExecCtxt(elemDecl->contModel, NULL, NULL);
	} else {
	    ctxt->vstateTab[ctxt->vstateNr].exec = NULL;
	    xmlErrValidNode(ctxt, (xmlNodePtr) elemDecl,
	                    XML_ERR_INTERNAL_ERROR,
			    "Failed to build content model regexp for %s\n",
			    node->name, NULL, NULL);
	}
    }
    return(ctxt->vstateNr++);
}

static int
vstateVPop(xmlValidCtxtPtr ctxt) {
    xmlElementPtr elemDecl;

    if (ctxt->vstateNr < 1) return(-1);
    ctxt->vstateNr--;
    elemDecl = ctxt->vstateTab[ctxt->vstateNr].elemDecl;
    ctxt->vstateTab[ctxt->vstateNr].elemDecl = NULL;
    ctxt->vstateTab[ctxt->vstateNr].node = NULL;
    if ((elemDecl != NULL) && (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)) {
	xmlRegFreeExecCtxt(ctxt->vstateTab[ctxt->vstateNr].exec);
    }
    ctxt->vstateTab[ctxt->vstateNr].exec = NULL;
    if (ctxt->vstateNr >= 1)
	ctxt->vstate = &ctxt->vstateTab[ctxt->vstateNr - 1];
    else
	ctxt->vstate = NULL;
    return(ctxt->vstateNr);
}

#else /* not LIBXML_REGEXP_ENABLED */
/*
 * If regexp are not enabled, it uses a home made algorithm less
 * complex and easier to
 * debug/maintain than a generic NFA -> DFA state based algo. The
 * only restriction is on the deepness of the tree limited by the
 * size of the occurs bitfield
 *
 * this is the content of a saved state for rollbacks
 */

#define ROLLBACK_OR	0
#define ROLLBACK_PARENT	1

typedef struct _xmlValidState {
    xmlElementContentPtr cont;	/* pointer to the content model subtree */
    xmlNodePtr           node;	/* pointer to the current node in the list */
    long                 occurs;/* bitfield for multiple occurrences */
    unsigned char        depth; /* current depth in the overall tree */
    unsigned char        state; /* ROLLBACK_XXX */
} _xmlValidState;

#define MAX_RECURSE 25000
#define MAX_DEPTH ((sizeof(_xmlValidState.occurs)) * 8)
#define CONT ctxt->vstate->cont
#define NODE ctxt->vstate->node
#define DEPTH ctxt->vstate->depth
#define OCCURS ctxt->vstate->occurs
#define STATE ctxt->vstate->state

#define OCCURRENCE (ctxt->vstate->occurs & (1 << DEPTH))
#define PARENT_OCCURRENCE (ctxt->vstate->occurs & ((1 << DEPTH) - 1))

#define SET_OCCURRENCE ctxt->vstate->occurs |= (1 << DEPTH)
#define RESET_OCCURRENCE ctxt->vstate->occurs &= ((1 << DEPTH) - 1)

static int
vstateVPush(xmlValidCtxtPtr ctxt, xmlElementContentPtr cont,
	    xmlNodePtr node, unsigned char depth, long occurs,
	    unsigned char state) {
    int i = ctxt->vstateNr - 1;

    if (ctxt->vstateNr > MAX_RECURSE) {
	return(-1);
    }
    if (ctxt->vstateTab == NULL) {
	ctxt->vstateMax = 8;
	ctxt->vstateTab = (xmlValidState *) xmlMalloc(
		     ctxt->vstateMax * sizeof(ctxt->vstateTab[0]));
	if (ctxt->vstateTab == NULL) {
	    xmlVErrMemory(ctxt, "malloc failed");
	    return(-1);
	}
    }
    if (ctxt->vstateNr >= ctxt->vstateMax) {
        xmlValidState *tmp;

        tmp = (xmlValidState *) xmlRealloc(ctxt->vstateTab,
	             2 * ctxt->vstateMax * sizeof(ctxt->vstateTab[0]));
        if (tmp == NULL) {
	    xmlVErrMemory(ctxt, "malloc failed");
	    return(-1);
	}
	ctxt->vstateMax *= 2;
	ctxt->vstateTab = tmp;
	ctxt->vstate = &ctxt->vstateTab[0];
    }
    /*
     * Don't push on the stack a state already here
     */
    if ((i >= 0) && (ctxt->vstateTab[i].cont == cont) &&
	(ctxt->vstateTab[i].node == node) &&
	(ctxt->vstateTab[i].depth == depth) &&
	(ctxt->vstateTab[i].occurs == occurs) &&
	(ctxt->vstateTab[i].state == state))
	return(ctxt->vstateNr);
    ctxt->vstateTab[ctxt->vstateNr].cont = cont;
    ctxt->vstateTab[ctxt->vstateNr].node = node;
    ctxt->vstateTab[ctxt->vstateNr].depth = depth;
    ctxt->vstateTab[ctxt->vstateNr].occurs = occurs;
    ctxt->vstateTab[ctxt->vstateNr].state = state;
    return(ctxt->vstateNr++);
}

static int
vstateVPop(xmlValidCtxtPtr ctxt) {
    if (ctxt->vstateNr <= 1) return(-1);
    ctxt->vstateNr--;
    ctxt->vstate = &ctxt->vstateTab[0];
    ctxt->vstate->cont =  ctxt->vstateTab[ctxt->vstateNr].cont;
    ctxt->vstate->node = ctxt->vstateTab[ctxt->vstateNr].node;
    ctxt->vstate->depth = ctxt->vstateTab[ctxt->vstateNr].depth;
    ctxt->vstate->occurs = ctxt->vstateTab[ctxt->vstateNr].occurs;
    ctxt->vstate->state = ctxt->vstateTab[ctxt->vstateNr].state;
    return(ctxt->vstateNr);
}

#endif /* LIBXML_REGEXP_ENABLED */

static int
nodeVPush(xmlValidCtxtPtr ctxt, xmlNodePtr value)
{
    if (ctxt->nodeMax <= 0) {
        ctxt->nodeMax = 4;
        ctxt->nodeTab =
            (xmlNodePtr *) xmlMalloc(ctxt->nodeMax *
                                     sizeof(ctxt->nodeTab[0]));
        if (ctxt->nodeTab == NULL) {
	    xmlVErrMemory(ctxt, "malloc failed");
            ctxt->nodeMax = 0;
            return (0);
        }
    }
    if (ctxt->nodeNr >= ctxt->nodeMax) {
        xmlNodePtr *tmp;
        tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab,
			      ctxt->nodeMax * 2 * sizeof(ctxt->nodeTab[0]));
        if (tmp == NULL) {
	    xmlVErrMemory(ctxt, "realloc failed");
            return (0);
        }
        ctxt->nodeMax *= 2;
	ctxt->nodeTab = tmp;
    }
    ctxt->nodeTab[ctxt->nodeNr] = value;
    ctxt->node = value;
    return (ctxt->nodeNr++);
}
static xmlNodePtr
nodeVPop(xmlValidCtxtPtr ctxt)
{
    xmlNodePtr ret;

    if (ctxt->nodeNr <= 0)
        return (0);
    ctxt->nodeNr--;
    if (ctxt->nodeNr > 0)
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
    else
        ctxt->node = NULL;
    ret = ctxt->nodeTab[ctxt->nodeNr];
    ctxt->nodeTab[ctxt->nodeNr] = 0;
    return (ret);
}

#ifdef DEBUG_VALID_ALGO
static void
xmlValidPrintNode(xmlNodePtr cur) {
    if (cur == NULL) {
	xmlGenericError(xmlGenericErrorContext, "null");
	return;
    }
    switch (cur->type) {
	case XML_ELEMENT_NODE:
	    xmlGenericError(xmlGenericErrorContext, "%s ", cur->name);
	    break;
	case XML_TEXT_NODE:
	    xmlGenericError(xmlGenericErrorContext, "text ");
	    break;
	case XML_CDATA_SECTION_NODE:
	    xmlGenericError(xmlGenericErrorContext, "cdata ");
	    break;
	case XML_ENTITY_REF_NODE:
	    xmlGenericError(xmlGenericErrorContext, "&%s; ", cur->name);
	    break;
	case XML_PI_NODE:
	    xmlGenericError(xmlGenericErrorContext, "pi(%s) ", cur->name);
	    break;
	case XML_COMMENT_NODE:
	    xmlGenericError(xmlGenericErrorContext, "comment ");
	    break;
	case XML_ATTRIBUTE_NODE:
	    xmlGenericError(xmlGenericErrorContext, "?attr? ");
	    break;
	case XML_ENTITY_NODE:
	    xmlGenericError(xmlGenericErrorContext, "?ent? ");
	    break;
	case XML_DOCUMENT_NODE:
	    xmlGenericError(xmlGenericErrorContext, "?doc? ");
	    break;
	case XML_DOCUMENT_TYPE_NODE:
	    xmlGenericError(xmlGenericErrorContext, "?doctype? ");