parserinternals.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

		/*
		 * We used to keep the internal content in the
		 * document encoding however this turns being unmaintainable
		 * So xmlGetCharEncodingHandler() will return non-null
		 * values for this now.
		 */
		if ((ctxt->inputNr == 1) &&
		    (ctxt->encoding == NULL) &&
		    (ctxt->input->encoding != NULL)) {
		    ctxt->encoding = xmlStrdup(ctxt->input->encoding);
		}
		ctxt->charset = enc;
		return(0);
	    case XML_CHAR_ENCODING_2022_JP:
		__xmlErrEncoding(ctxt, XML_ERR_UNSUPPORTED_ENCODING,
			       "encoding not supported %s\n",
			       BAD_CAST "ISO-2022-JP", NULL);
		break;
	    case XML_CHAR_ENCODING_SHIFT_JIS:
		__xmlErrEncoding(ctxt, XML_ERR_UNSUPPORTED_ENCODING,
			       "encoding not supported %s\n",
			       BAD_CAST "Shift_JIS", NULL);
		break;
	    case XML_CHAR_ENCODING_EUC_JP:
		__xmlErrEncoding(ctxt, XML_ERR_UNSUPPORTED_ENCODING,
			       "encoding not supported %s\n",
			       BAD_CAST "EUC-JP", NULL);
		break;
	}
    }
    if (handler == NULL)
	return(-1);
    ctxt->charset = XML_CHAR_ENCODING_UTF8;
    return(xmlSwitchToEncoding(ctxt, handler));
}

/**
 * xmlSwitchInputEncoding:
 * @ctxt:  the parser context
 * @input:  the input stream
 * @handler:  the encoding handler
 *
 * change the input functions when discovering the character encoding
 * of a given entity.
 *
 * Returns 0 in case of success, -1 otherwise
 */
int
xmlSwitchInputEncoding(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
                       xmlCharEncodingHandlerPtr handler)
{
    int nbchars;

    if (handler == NULL)
        return (-1);
    if (input == NULL)
        return (-1);
    if (input->buf != NULL) {
        if (input->buf->encoder != NULL) {
            /*
             * Check in case the auto encoding detetection triggered
             * in already.
             */
            if (input->buf->encoder == handler)
                return (0);

            /*
             * "UTF-16" can be used for both LE and BE
             if ((!xmlStrncmp(BAD_CAST input->buf->encoder->name,
             BAD_CAST "UTF-16", 6)) &&
             (!xmlStrncmp(BAD_CAST handler->name,
             BAD_CAST "UTF-16", 6))) {
             return(0);
             }
             */

            /*
             * Note: this is a bit dangerous, but that's what it
             * takes to use nearly compatible signature for different
             * encodings.
             */
            xmlCharEncCloseFunc(input->buf->encoder);
            input->buf->encoder = handler;
            return (0);
        }
        input->buf->encoder = handler;

        /*
         * Is there already some content down the pipe to convert ?
         */
        if ((input->buf->buffer != NULL) && (input->buf->buffer->use > 0)) {
            int processed;
	    unsigned int use;

            /*
             * Specific handling of the Byte Order Mark for 
             * UTF-16
             */
            if ((handler->name != NULL) &&
                (!strcmp(handler->name, "UTF-16LE")) &&
                (input->cur[0] == 0xFF) && (input->cur[1] == 0xFE)) {
                input->cur += 2;
            }
            if ((handler->name != NULL) &&
                (!strcmp(handler->name, "UTF-16BE")) &&
                (input->cur[0] == 0xFE) && (input->cur[1] == 0xFF)) {
                input->cur += 2;
            }
            /*
             * Errata on XML-1.0 June 20 2001
             * Specific handling of the Byte Order Mark for
             * UTF-8
             */
            if ((handler->name != NULL) &&
                (!strcmp(handler->name, "UTF-8")) &&
                (input->cur[0] == 0xEF) &&
                (input->cur[1] == 0xBB) && (input->cur[2] == 0xBF)) {
                input->cur += 3;
            }

            /*
             * Shrink the current input buffer.
             * Move it as the raw buffer and create a new input buffer
             */
            processed = input->cur - input->base;
            xmlBufferShrink(input->buf->buffer, processed);
            input->buf->raw = input->buf->buffer;
            input->buf->buffer = xmlBufferCreate();
	    input->buf->rawconsumed = processed;
	    use = input->buf->raw->use;

            if (ctxt->html) {
                /*
                 * convert as much as possible of the buffer
                 */
                nbchars = xmlCharEncInFunc(input->buf->encoder,
                                           input->buf->buffer,
                                           input->buf->raw);
            } else {
                /*
                 * convert just enough to get
                 * '<?xml version="1.0" encoding="xxx"?>'
                 * parsed with the autodetected encoding
                 * into the parser reading buffer.
                 */
                nbchars = xmlCharEncFirstLine(input->buf->encoder,
                                              input->buf->buffer,
                                              input->buf->raw);
            }
            if (nbchars < 0) {
                xmlErrInternal(ctxt,
                               "switching encoding: encoder error\n",
                               NULL);
                return (-1);
            }
	    input->buf->rawconsumed += use - input->buf->raw->use;
            input->base = input->cur = input->buf->buffer->content;
            input->end = &input->base[input->buf->buffer->use];

        }
        return (0);
    } else {
        if ((input->length == 0) || (input->buf == NULL)) {
            /*
             * When parsing a static memory array one must know the
             * size to be able to convert the buffer.
             */
            xmlErrInternal(ctxt, "switching encoding : no input\n", NULL);
            return (-1);
        } else {
            int processed;

            /*
             * Shrink the current input buffer.
             * Move it as the raw buffer and create a new input buffer
             */
            processed = input->cur - input->base;

            input->buf->raw = xmlBufferCreate();
            xmlBufferAdd(input->buf->raw, input->cur,
                         input->length - processed);
            input->buf->buffer = xmlBufferCreate();

            /*
             * convert as much as possible of the raw input
             * to the parser reading buffer.
             */
            nbchars = xmlCharEncInFunc(input->buf->encoder,
                                       input->buf->buffer,
                                       input->buf->raw);
            if (nbchars < 0) {
                xmlErrInternal(ctxt,
                               "switching encoding: encoder error\n",
                               NULL);
                return (-1);
            }

            /*
             * Conversion succeeded, get rid of the old buffer
             */
            if ((input->free != NULL) && (input->base != NULL))
                input->free((xmlChar *) input->base);
            input->base = input->cur = input->buf->buffer->content;
            input->end = &input->base[input->buf->buffer->use];
        }
    }
    return (0);
}

/**
 * xmlSwitchToEncoding:
 * @ctxt:  the parser context
 * @handler:  the encoding handler
 *
 * 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 (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)
{
    if(ctxt==NULL) {
        xmlErrInternal(NULL, "Got NULL parser context\n", NULL);
        return(-1);
    }

    xmlDefaultSAXHandlerInit();

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