sax2.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.

/*
 * SAX2.c : Default SAX2 handler to build a tree.
 *
 * See Copyright for the status of this software.
 *
 * Daniel Veillard <[e-mail]>
 */


#define IN_LIBXML
#include "libxml.h"
#include <stdlib.h>
#include <string.h>
#include <libxml/xmlmemory.h>
#include <libxml/tree.h>
#include <libxml/parser.h>
#include <libxml/parserInternals.h>
#include <libxml/valid.h>
#include <libxml/entities.h>
#include <libxml/xmlerror.h>
#include <libxml/debugXML.h>
#include <libxml/xmlIO.h>
#include <libxml/SAX.h>
#include <libxml/uri.h>
#include <libxml/valid.h>
#include <libxml/HTMLtree.h>
#include <libxml/globals.h>

/* #define DEBUG_SAX2 */
/* #define DEBUG_SAX2_TREE */

/**
 * TODO:
 *
 * macro to flag unimplemented blocks
 * XML_CATALOG_PREFER user env to select between system/public prefered
 * option. C.f. Richard Tobin <[e-mail]>
 *> Just FYI, I am using an environment variable XML_CATALOG_PREFER with
 *> values "system" and "public".  I have made the default be "system" to
 *> match yours.
 */
#define TODO 								\
    xmlGenericError(xmlGenericErrorContext,				\
	    "Unimplemented block at %s:%d\n",				\
            __FILE__, __LINE__);

/*
 * xmlSAX2ErrMemory:
 * @ctxt:  an XML validation parser context
 * @msg:   a string to accompany the error message
 */
static void
xmlSAX2ErrMemory(xmlParserCtxtPtr ctxt, const char *msg) {
    if (ctxt != NULL) {
	if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL))
	    ctxt->sax->error(ctxt->userData, "%s: out of memory\n", msg);
	ctxt->errNo = XML_ERR_NO_MEMORY;
	ctxt->instate = XML_PARSER_EOF;
	ctxt->disableSAX = 1;
    }
}

/**
 * xmlValidError:
 * @ctxt:  an XML validation parser context
 * @error:  the error number
 * @msg:  the error message
 * @str1:  extra data
 * @str2:  extra data
 *
 * Handle a validation error
 */
static void
xmlErrValid(xmlParserCtxtPtr ctxt, xmlParserErrors error,
            const char *msg, const char *str1, const char *str2)
{
    xmlStructuredErrorFunc schannel = NULL;

    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if (ctxt != NULL) {
	ctxt->errNo = error;
	if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC))
	    schannel = ctxt->sax->serror;
	__xmlRaiseError(schannel,
			ctxt->vctxt.error, ctxt->vctxt.userData,
			ctxt, NULL, XML_FROM_DTD, error,
			XML_ERR_ERROR, NULL, 0, (const char *) str1,
			(const char *) str2, NULL, 0, 0,
			msg, (const char *) str1, (const char *) str2);
	ctxt->valid = 0;
    } else {
	__xmlRaiseError(schannel,
			NULL, NULL,
			ctxt, NULL, XML_FROM_DTD, error,
			XML_ERR_ERROR, NULL, 0, (const char *) str1,
			(const char *) str2, NULL, 0, 0,
			msg, (const char *) str1, (const char *) str2);
    }
}

/**
 * xmlFatalErrMsg:
 * @ctxt:  an XML parser context
 * @error:  the error number
 * @msg:  the error message
 * @str1:  an error string
 * @str2:  an error string
 *
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
 */
static void
xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
               const char *msg, const xmlChar *str1, const xmlChar *str2)
{
    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if (ctxt != NULL)
	ctxt->errNo = error;
    __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error,
                    XML_ERR_FATAL, NULL, 0, 
		    (const char *) str1, (const char *) str2,
		    NULL, 0, 0, msg, str1, str2);
    if (ctxt != NULL) {
	ctxt->wellFormed = 0;
	ctxt->valid = 0;
	if (ctxt->recovery == 0)
	    ctxt->disableSAX = 1;
    }
}

/**
 * xmlWarnMsg:
 * @ctxt:  an XML parser context
 * @error:  the error number
 * @msg:  the error message
 * @str1:  an error string
 * @str2:  an error string
 *
 * Handle a parser warning
 */
static void
xmlWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
               const char *msg, const xmlChar *str1)
{
    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if (ctxt != NULL)
	ctxt->errNo = error;
    __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error,
                    XML_ERR_WARNING, NULL, 0, 
		    (const char *) str1, NULL,
		    NULL, 0, 0, msg, str1);
}

/**
 * xmlNsErrMsg:
 * @ctxt:  an XML parser context
 * @error:  the error number
 * @msg:  the error message
 * @str1:  an error string
 * @str2:  an error string
 *
 * Handle a namespace error
 */
static void
xmlNsErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
            const char *msg, const xmlChar *str1, const xmlChar *str2)
{
    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if (ctxt != NULL)
	ctxt->errNo = error;
    __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error,
                    XML_ERR_ERROR, NULL, 0, 
		    (const char *) str1, (const char *) str2,
		    NULL, 0, 0, msg, str1, str2);
}

/**
 * xmlNsWarnMsg:
 * @ctxt:  an XML parser context
 * @error:  the error number
 * @msg:  the error message
 * @str1:  an error string
 *
 * Handle a namespace warning
 */
static void
xmlNsWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
             const char *msg, const xmlChar *str1, const xmlChar *str2)
{
    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if (ctxt != NULL)
	ctxt->errNo = error;
    __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error,
                    XML_ERR_WARNING, NULL, 0, 
		    (const char *) str1, (const char *) str2,
		    NULL, 0, 0, msg, str1, str2);
}

/**
 * xmlSAX2GetPublicId:
 * @ctx: the user data (XML parser context)
 *
 * Provides the public ID e.g. "-//SGMLSOURCE//DTD DEMO//EN"
 *
 * Returns a xmlChar *
 */
const xmlChar *
xmlSAX2GetPublicId(void *ctx ATTRIBUTE_UNUSED)
{
    /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
    return(NULL);
}

/**
 * xmlSAX2GetSystemId:
 * @ctx: the user data (XML parser context)
 *
 * Provides the system ID, basically URL or filename e.g.
 * http://www.sgmlsource.com/dtds/memo.dtd
 *
 * Returns a xmlChar *
 */
const xmlChar *
xmlSAX2GetSystemId(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctx == NULL) || (ctxt->input == NULL)) return(NULL);
    return((const xmlChar *) ctxt->input->filename); 
}

/**
 * xmlSAX2GetLineNumber:
 * @ctx: the user data (XML parser context)
 *
 * Provide the line number of the current parsing point.
 *
 * Returns an int
 */
int
xmlSAX2GetLineNumber(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctx == NULL) || (ctxt->input == NULL)) return(0);
    return(ctxt->input->line);
}

/**
 * xmlSAX2GetColumnNumber:
 * @ctx: the user data (XML parser context)
 *
 * Provide the column number of the current parsing point.
 *
 * Returns an int
 */
int
xmlSAX2GetColumnNumber(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctx == NULL) || (ctxt->input == NULL)) return(0);
    return(ctxt->input->col);
}

/**
 * xmlSAX2IsStandalone:
 * @ctx: the user data (XML parser context)
 *
 * Is this document tagged standalone ?
 *
 * Returns 1 if true
 */
int
xmlSAX2IsStandalone(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctx == NULL) || (ctxt->myDoc == NULL)) return(0);
    return(ctxt->myDoc->standalone == 1);
}

/**
 * xmlSAX2HasInternalSubset:
 * @ctx: the user data (XML parser context)
 *
 * Does this document has an internal subset
 *
 * Returns 1 if true
 */
int
xmlSAX2HasInternalSubset(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0);
    return(ctxt->myDoc->intSubset != NULL);
}

/**
 * xmlSAX2HasExternalSubset:
 * @ctx: the user data (XML parser context)
 *
 * Does this document has an external subset
 *
 * Returns 1 if true
 */
int
xmlSAX2HasExternalSubset(void *ctx)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0);
    return(ctxt->myDoc->extSubset != NULL);
}

/**
 * xmlSAX2InternalSubset:
 * @ctx:  the user data (XML parser context)
 * @name:  the root element name
 * @ExternalID:  the external ID
 * @SystemID:  the SYSTEM ID (e.g. filename or URL)
 *
 * Callback on internal subset declaration.
 */
void
xmlSAX2InternalSubset(void *ctx, const xmlChar *name,
	       const xmlChar *ExternalID, const xmlChar *SystemID)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    xmlDtdPtr dtd;
    if (ctx == NULL) return;
#ifdef DEBUG_SAX
    xmlGenericError(xmlGenericErrorContext,
	    "SAX.xmlSAX2InternalSubset(%s, %s, %s)\n",
            name, ExternalID, SystemID);
#endif

    if (ctxt->myDoc == NULL)
	return;
    dtd = xmlGetIntSubset(ctxt->myDoc);
    if (dtd != NULL) {
	if (ctxt->html)
	    return;
	xmlUnlinkNode((xmlNodePtr) dtd);
	xmlFreeDtd(dtd);
	ctxt->myDoc->intSubset = NULL;
    }
    ctxt->myDoc->intSubset = 
	xmlCreateIntSubset(ctxt->myDoc, name, ExternalID, SystemID);
    if (ctxt->myDoc->intSubset == NULL)
        xmlSAX2ErrMemory(ctxt, "xmlSAX2InternalSubset");
}

/**
 * xmlSAX2ExternalSubset:
 * @ctx: the user data (XML parser context)
 * @name:  the root element name
 * @ExternalID:  the external ID
 * @SystemID:  the SYSTEM ID (e.g. filename or URL)
 *
 * Callback on external subset declaration.
 */
void
xmlSAX2ExternalSubset(void *ctx, const xmlChar *name,
	       const xmlChar *ExternalID, const xmlChar *SystemID)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    if (ctx == NULL) return;
#ifdef DEBUG_SAX
    xmlGenericError(xmlGenericErrorContext,
	    "SAX.xmlSAX2ExternalSubset(%s, %s, %s)\n",
            name, ExternalID, SystemID);
#endif
    if (((ExternalID != NULL) || (SystemID != NULL)) &&
        (((ctxt->validate) || (ctxt->loadsubset != 0)) &&
	 (ctxt->wellFormed && ctxt->myDoc))) {
	/*
	 * Try to fetch and parse the external subset.
	 */
	xmlParserInputPtr oldinput;
	int oldinputNr;
	int oldinputMax;
	xmlParserInputPtr *oldinputTab;
	xmlParserInputPtr input = NULL;
	xmlCharEncoding enc;
	int oldcharset;

	/*
	 * Ask the Entity resolver to load the damn thing
	 */
	if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
	    input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID,
	                                        SystemID);
	if (input == NULL) {
	    return;
	}

	xmlNewDtd(ctxt->myDoc, name, ExternalID, SystemID);

	/*
	 * make sure we won't destroy the main document context
	 */
	oldinput = ctxt->input;
	oldinputNr = ctxt->inputNr;
	oldinputMax = ctxt->inputMax;
	oldinputTab = ctxt->inputTab;
	oldcharset = ctxt->charset;

	ctxt->inputTab = (xmlParserInputPtr *)
	                 xmlMalloc(5 * sizeof(xmlParserInputPtr));
	if (ctxt->inputTab == NULL) {
	    xmlSAX2ErrMemory(ctxt, "xmlSAX2ExternalSubset");
	    ctxt->input = oldinput;
	    ctxt->inputNr = oldinputNr;
	    ctxt->inputMax = oldinputMax;
	    ctxt->inputTab = oldinputTab;
	    ctxt->charset = oldcharset;
	    return;
	}
	ctxt->inputNr = 0;
	ctxt->inputMax = 5;
	ctxt->input = NULL;
	xmlPushInput(ctxt, input);

	/*
	 * On the fly encoding conversion if needed
	 */
	if (ctxt->input->length >= 4) {
	    enc = xmlDetectCharEncoding(ctxt->input->cur, 4);
	    xmlSwitchEncoding(ctxt, enc);
	}

	if (input->filename == NULL)
	    input->filename = (char *) xmlCanonicPath(SystemID);
	input->line = 1;
	input->col = 1;
	input->base = ctxt->input->cur;
	input->cur = ctxt->input->cur;
	input->free = NULL;

	/*
	 * let's parse that entity knowing it's an external subset.
	 */
	xmlParseExternalSubset(ctxt, ExternalID, SystemID);

        /*
	 * Free up the external entities
	 */

	while (ctxt->inputNr > 1)
	    xmlPopInput(ctxt);
	xmlFreeInputStream(ctxt->input);
        xmlFree(ctxt->inputTab);

	/*
	 * Restore the parsing context of the main entity
	 */
	ctxt->input = oldinput;
	ctxt->inputNr = oldinputNr;
	ctxt->inputMax = oldinputMax;
	ctxt->inputTab = oldinputTab;
	ctxt->charset = oldcharset;
	/* ctxt->wellFormed = oldwellFormed; */
    }
}