parserinternals.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.

 *
 * change the input functions when discovering the character encoding
 * of a given entity.
 *
 * Returns 0 in case of success, -1 otherwise
 */
int
xmlSwitchToEncoding(xmlParserCtxtPtr ctxt, xmlCharEncodingHandlerPtr handler) 
{
    if (handler != NULL) {
        if (ctxt->input != NULL) {
	    xmlSwitchInputEncoding(ctxt, ctxt->input, handler);
	} else {
	    xmlErrInternal(ctxt, "xmlSwitchToEncoding : no input\n",
	                   NULL);
	    return(-1);
	}
	/*
	 * The parsing is now done in UTF8 natively
	 */
	ctxt->charset = XML_CHAR_ENCODING_UTF8;
    } else 
	return(-1);
    return(0);
}

/************************************************************************
 *									*
 *	Commodity functions to handle entities processing		*
 *									*
 ************************************************************************/

/**
 * xmlFreeInputStream:
 * @input:  an xmlParserInputPtr
 *
 * Free up an input stream.
 */
void
xmlFreeInputStream(xmlParserInputPtr input) {
    if (input == NULL) return;

    if (input->filename != NULL) xmlFree((char *) input->filename);
    if (input->directory != NULL) xmlFree((char *) input->directory);
    if (input->encoding != NULL) xmlFree((char *) input->encoding);
    if (input->version != NULL) xmlFree((char *) input->version);
    if ((input->free != NULL) && (input->base != NULL))
        input->free((xmlChar *) input->base);
    if (input->buf != NULL) 
        xmlFreeParserInputBuffer(input->buf);
    xmlFree(input);
}

/**
 * xmlNewInputStream:
 * @ctxt:  an XML parser context
 *
 * Create a new input stream structure
 * Returns the new input stream or NULL
 */
xmlParserInputPtr
xmlNewInputStream(xmlParserCtxtPtr ctxt) {
    xmlParserInputPtr input;
    static int id = 0;

    input = (xmlParserInputPtr) xmlMalloc(sizeof(xmlParserInput));
    if (input == NULL) {
        xmlErrMemory(ctxt,  "couldn't allocate a new input stream\n");
	return(NULL);
    }
    memset(input, 0, sizeof(xmlParserInput));
    input->line = 1;
    input->col = 1;
    input->standalone = -1;
    /*
     * we don't care about thread reentrancy unicity for a single
     * parser context (and hence thread) is sufficient.
     */
    input->id = id++;
    return(input);
}

/**
 * xmlNewIOInputStream:
 * @ctxt:  an XML parser context
 * @input:  an I/O Input
 * @enc:  the charset encoding if known
 *
 * Create a new input stream structure encapsulating the @input into
 * a stream suitable for the parser.
 *
 * Returns the new input stream or NULL
 */
xmlParserInputPtr
xmlNewIOInputStream(xmlParserCtxtPtr ctxt, xmlParserInputBufferPtr input,
	            xmlCharEncoding enc) {
    xmlParserInputPtr inputStream;

    if (input == NULL) return(NULL);
    if (xmlParserDebugEntities)
	xmlGenericError(xmlGenericErrorContext, "new input from I/O\n");
    inputStream = xmlNewInputStream(ctxt);
    if (inputStream == NULL) {
	return(NULL);
    }
    inputStream->filename = NULL;
    inputStream->buf = input;
    inputStream->base = inputStream->buf->buffer->content;
    inputStream->cur = inputStream->buf->buffer->content;
    inputStream->end = &inputStream->base[inputStream->buf->buffer->use];
    if (enc != XML_CHAR_ENCODING_NONE) {
        xmlSwitchEncoding(ctxt, enc);
    }

    return(inputStream);
}

/**
 * xmlNewEntityInputStream:
 * @ctxt:  an XML parser context
 * @entity:  an Entity pointer
 *
 * Create a new input stream based on an xmlEntityPtr
 *
 * Returns the new input stream or NULL
 */
xmlParserInputPtr
xmlNewEntityInputStream(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
    xmlParserInputPtr input;

    if (entity == NULL) {
        xmlErrInternal(ctxt, "xmlNewEntityInputStream entity = NULL\n",
	               NULL);
	return(NULL);
    }
    if (xmlParserDebugEntities)
	xmlGenericError(xmlGenericErrorContext,
		"new input from entity: %s\n", entity->name);
    if (entity->content == NULL) {
	switch (entity->etype) {
            case XML_EXTERNAL_GENERAL_UNPARSED_ENTITY:
	        xmlErrInternal(ctxt, "Cannot parse entity %s\n",
		               entity->name);
                break;
            case XML_EXTERNAL_GENERAL_PARSED_ENTITY:
            case XML_EXTERNAL_PARAMETER_ENTITY:
		return(xmlLoadExternalEntity((char *) entity->URI,
		       (char *) entity->ExternalID, ctxt));
            case XML_INTERNAL_GENERAL_ENTITY:
	        xmlErrInternal(ctxt,
		      "Internal entity %s without content !\n",
		               entity->name);
                break;
            case XML_INTERNAL_PARAMETER_ENTITY:
	        xmlErrInternal(ctxt,
		      "Internal parameter entity %s without content !\n",
		               entity->name);
                break;
            case XML_INTERNAL_PREDEFINED_ENTITY:
	        xmlErrInternal(ctxt,
		      "Predefined entity %s without content !\n",
		               entity->name);
                break;
	}
	return(NULL);
    }
    input = xmlNewInputStream(ctxt);
    if (input == NULL) {
	return(NULL);
    }
    input->filename = (char *) entity->URI;
    input->base = entity->content;
    input->cur = entity->content;
    input->length = entity->length;
    input->end = &entity->content[input->length];
    return(input);
}

/**
 * xmlNewStringInputStream:
 * @ctxt:  an XML parser context
 * @buffer:  an memory buffer
 *
 * Create a new input stream based on a memory buffer.
 * Returns the new input stream
 */
xmlParserInputPtr
xmlNewStringInputStream(xmlParserCtxtPtr ctxt, const xmlChar *buffer) {
    xmlParserInputPtr input;

    if (buffer == NULL) {
        xmlErrInternal(ctxt, "xmlNewStringInputStream string = NULL\n",
	               NULL);
	return(NULL);
    }
    if (xmlParserDebugEntities)
	xmlGenericError(xmlGenericErrorContext,
		"new fixed input: %.30s\n", buffer);
    input = xmlNewInputStream(ctxt);
    if (input == NULL) {
        xmlErrMemory(ctxt,  "couldn't allocate a new input stream\n");
	return(NULL);
    }
    input->base = buffer;
    input->cur = buffer;
    input->length = xmlStrlen(buffer);
    input->end = &buffer[input->length];
    return(input);
}

/**
 * xmlNewInputFromFile:
 * @ctxt:  an XML parser context
 * @filename:  the filename to use as entity
 *
 * Create a new input stream based on a file or an URL.
 *
 * Returns the new input stream or NULL in case of error
 */
xmlParserInputPtr
xmlNewInputFromFile(xmlParserCtxtPtr ctxt, const char *filename) {
    xmlParserInputBufferPtr buf;
    xmlParserInputPtr inputStream;
    char *directory = NULL;
    xmlChar *URI = NULL;

    if (xmlParserDebugEntities)
	xmlGenericError(xmlGenericErrorContext,
		"new input from file: %s\n", filename);
    if (ctxt == NULL) return(NULL);
    buf = xmlParserInputBufferCreateFilename(filename, XML_CHAR_ENCODING_NONE);
    if (buf == NULL) {
        __xmlLoaderErr(ctxt, "failed to load external entity \"%s\"\n",
                     (const char *) filename);
	return(NULL);
    }

    inputStream = xmlNewInputStream(ctxt);
    if (inputStream == NULL) {
	if (directory != NULL) xmlFree((char *) directory);
	if (URI != NULL) xmlFree((char *) URI);
	return(NULL);
    }
    inputStream->buf = buf;
    inputStream = xmlCheckHTTPInput(ctxt, inputStream);
    if (inputStream == NULL)
        return(NULL);
    
    if (inputStream->filename == NULL)
	URI = xmlStrdup((xmlChar *) filename);
    else
	URI = xmlStrdup((xmlChar *) inputStream->filename);
    directory = xmlParserGetDirectory((const char *) URI);
    inputStream->filename = (char *) xmlCanonicPath((const xmlChar *) URI);
    if (URI != NULL) xmlFree((char *) URI);
    inputStream->directory = directory;

    inputStream->base = inputStream->buf->buffer->content;
    inputStream->cur = inputStream->buf->buffer->content;
    inputStream->end = &inputStream->base[inputStream->buf->buffer->use];
    if ((ctxt->directory == NULL) && (directory != NULL))
        ctxt->directory = (char *) xmlStrdup((const xmlChar *) directory);
    return(inputStream);
}

/************************************************************************
 *									*
 *		Commodity functions to handle parser contexts		*
 *									*
 ************************************************************************/

/**
 * xmlInitParserCtxt:
 * @ctxt:  an XML parser context
 *
 * Initialize a parser context
 *
 * Returns 0 in case of success and -1 in case of error
 */

int
xmlInitParserCtxt(xmlParserCtxtPtr ctxt)
{
    xmlParserInputPtr input;

    if(ctxt==NULL) {
        xmlErrInternal(NULL, "Got NULL parser context\n", NULL);
        return(-1);
    }

    xmlDefaultSAXHandlerInit();

    if (ctxt->dict == NULL)
	ctxt->dict = xmlDictCreate();
    if (ctxt->dict == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	return(-1);
    }
    if (ctxt->sax == NULL)
	ctxt->sax = (xmlSAXHandler *) xmlMalloc(sizeof(xmlSAXHandler));
    if (ctxt->sax == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	return(-1);
    }
    else
        xmlSAXVersion(ctxt->sax, 2);

    ctxt->maxatts = 0;
    ctxt->atts = NULL;
    /* Allocate the Input stack */
    if (ctxt->inputTab == NULL) {
	ctxt->inputTab = (xmlParserInputPtr *)
		    xmlMalloc(5 * sizeof(xmlParserInputPtr));
	ctxt->inputMax = 5;
    }
    if (ctxt->inputTab == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	ctxt->inputNr = 0;
	ctxt->inputMax = 0;
	ctxt->input = NULL;
	return(-1);
    }
    while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */
        xmlFreeInputStream(input);
    }
    ctxt->inputNr = 0;
    ctxt->input = NULL;

    ctxt->version = NULL;
    ctxt->encoding = NULL;
    ctxt->standalone = -1;
    ctxt->hasExternalSubset = 0;
    ctxt->hasPErefs = 0;
    ctxt->html = 0;
    ctxt->external = 0;
    ctxt->instate = XML_PARSER_START;
    ctxt->token = 0;
    ctxt->directory = NULL;

    /* Allocate the Node stack */
    if (ctxt->nodeTab == NULL) {
	ctxt->nodeTab = (xmlNodePtr *) xmlMalloc(10 * sizeof(xmlNodePtr));
	ctxt->nodeMax = 10;
    }
    if (ctxt->nodeTab == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	ctxt->nodeNr = 0;
	ctxt->nodeMax = 0;
	ctxt->node = NULL;
	ctxt->inputNr = 0;
	ctxt->inputMax = 0;
	ctxt->input = NULL;
	return(-1);
    }
    ctxt->nodeNr = 0;
    ctxt->node = NULL;

    /* Allocate the Name stack */
    if (ctxt->nameTab == NULL) {
	ctxt->nameTab = (const xmlChar **) xmlMalloc(10 * sizeof(xmlChar *));
	ctxt->nameMax = 10;
    }
    if (ctxt->nameTab == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	ctxt->nodeNr = 0;
	ctxt->nodeMax = 0;
	ctxt->node = NULL;
	ctxt->inputNr = 0;
	ctxt->inputMax = 0;
	ctxt->input = NULL;
	ctxt->nameNr = 0;
	ctxt->nameMax = 0;
	ctxt->name = NULL;
	return(-1);
    }
    ctxt->nameNr = 0;
    ctxt->name = NULL;

    /* Allocate the space stack */
    if (ctxt->spaceTab == NULL) {
	ctxt->spaceTab = (int *) xmlMalloc(10 * sizeof(int));
	ctxt->spaceMax = 10;
    }
    if (ctxt->spaceTab == NULL) {
        xmlErrMemory(NULL, "cannot initialize parser context\n");
	ctxt->nodeNr = 0;
	ctxt->nodeMax = 0;
	ctxt->node = NULL;
	ctxt->inputNr = 0;
	ctxt->inputMax = 0;
	ctxt->input = NULL;
	ctxt->nameNr = 0;
	ctxt->nameMax = 0;
	ctxt->name = NULL;
	ctxt->spaceNr = 0;
	ctxt->spaceMax = 0;
	ctxt->space = NULL;
	return(-1);
    }
    ctxt->spaceNr = 1;
    ctxt->spaceMax = 10;
    ctxt->spaceTab[0] = -1;
    ctxt->space = &ctxt->spaceTab[0];
    ctxt->userData = ctxt;
    ctxt->myDoc = NULL;
    ctxt->wellFormed = 1;
    ctxt->nsWellFormed = 1;
    ctxt->valid = 1;
    ctxt->loadsubset = xmlLoadExtDtdDefaultValue;
    ctxt->validate = xmlDoValidityCheckingDefaultValue;
    ctxt->pedantic = xmlPedanticParserDefaultValue;
    ctxt->linenumbers = xmlLineNumbersDefaultValue;
    ctxt->keepBlanks = xmlKeepBlanksDefaultValue;
    if (ctxt->keepBlanks == 0)
	ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace;

    ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_0;
    ctxt->vctxt.userData = ctxt;
    ctxt->vctxt.error = xmlParserValidityError;
    ctxt->vctxt.warning = xmlParserValidityWarning;