libxml.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.

/**
 * libxml_outputBufferGetPythonFile:
 * @buffer:  the I/O buffer
 *
 * read the Python I/O from the CObject
 *
 * Returns the new parser output or NULL
 */
static PyObject *
libxml_outputBufferGetPythonFile(ATTRIBUTE_UNUSED PyObject *self,
                                    PyObject *args) {
    PyObject *buffer;
    PyObject *file;
    xmlOutputBufferPtr obj;

    if (!PyArg_ParseTuple(args, (char *)"O:outputBufferGetPythonFile",
			  &buffer))
	return(NULL);

    obj = PyoutputBuffer_Get(buffer);
    if (obj == NULL) {
	fprintf(stderr,
	        "outputBufferGetPythonFile: obj == NULL\n");
	Py_INCREF(Py_None);
	return(Py_None);
    }
    if (obj->closecallback != xmlPythonFileClose) {
	fprintf(stderr,
	        "outputBufferGetPythonFile: not a python file wrapper\n");
	Py_INCREF(Py_None);
	return(Py_None);
    }
    file = (PyObject *) obj->context;
    if (file == NULL) {
	Py_INCREF(Py_None);
	return(Py_None);
    }
    Py_INCREF(file);
    return(file);
}

static PyObject *
libxml_xmlOutputBufferClose(PyObject *self ATTRIBUTE_UNUSED, PyObject *args) {
    PyObject *py_retval;
    int c_retval;
    xmlOutputBufferPtr out;
    PyObject *pyobj_out;

    if (!PyArg_ParseTuple(args, (char *)"O:xmlOutputBufferClose", &pyobj_out))
        return(NULL);
    out = (xmlOutputBufferPtr) PyoutputBuffer_Get(pyobj_out);
    /* Buffer may already have been destroyed elsewhere. This is harmless. */
    if (out == NULL) {
	Py_INCREF(Py_None);
	return(Py_None);
    }

    c_retval = xmlOutputBufferClose(out);
    py_retval = libxml_intWrap((int) c_retval);
    return(py_retval);
}

static PyObject *
libxml_xmlOutputBufferFlush(PyObject *self ATTRIBUTE_UNUSED, PyObject *args) {
    PyObject *py_retval;
    int c_retval;
    xmlOutputBufferPtr out;
    PyObject *pyobj_out;

    if (!PyArg_ParseTuple(args, (char *)"O:xmlOutputBufferFlush", &pyobj_out))
        return(NULL);
    out = (xmlOutputBufferPtr) PyoutputBuffer_Get(pyobj_out);

    c_retval = xmlOutputBufferFlush(out);
    py_retval = libxml_intWrap((int) c_retval);
    return(py_retval);
}

static PyObject *
libxml_xmlSaveFileTo(PyObject *self ATTRIBUTE_UNUSED, PyObject *args) {
    PyObject *py_retval;
    int c_retval;
    xmlOutputBufferPtr buf;
    PyObject *pyobj_buf;
    xmlDocPtr cur;
    PyObject *pyobj_cur;
    char * encoding;

    if (!PyArg_ParseTuple(args, (char *)"OOz:xmlSaveFileTo", &pyobj_buf, &pyobj_cur, &encoding))
        return(NULL);
    buf = (xmlOutputBufferPtr) PyoutputBuffer_Get(pyobj_buf);
    cur = (xmlDocPtr) PyxmlNode_Get(pyobj_cur);

    c_retval = xmlSaveFileTo(buf, cur, encoding);
	/* xmlSaveTo() freed the memory pointed to by buf, so record that in the
	 * Python object. */
    ((PyoutputBuffer_Object *)(pyobj_buf))->obj = NULL;
    py_retval = libxml_intWrap((int) c_retval);
    return(py_retval);
}

static PyObject *
libxml_xmlSaveFormatFileTo(PyObject *self ATTRIBUTE_UNUSED, PyObject *args) {
    PyObject *py_retval;
    int c_retval;
    xmlOutputBufferPtr buf;
    PyObject *pyobj_buf;
    xmlDocPtr cur;
    PyObject *pyobj_cur;
    char * encoding;
    int format;

    if (!PyArg_ParseTuple(args, (char *)"OOzi:xmlSaveFormatFileTo", &pyobj_buf, &pyobj_cur, &encoding, &format))
        return(NULL);
    buf = (xmlOutputBufferPtr) PyoutputBuffer_Get(pyobj_buf);
    cur = (xmlDocPtr) PyxmlNode_Get(pyobj_cur);

    c_retval = xmlSaveFormatFileTo(buf, cur, encoding, format);
	/* xmlSaveFormatFileTo() freed the memory pointed to by buf, so record that
	 * in the Python object */
	((PyoutputBuffer_Object *)(pyobj_buf))->obj = NULL;
    py_retval = libxml_intWrap((int) c_retval);
    return(py_retval);
}
#endif /* LIBXML_OUTPUT_ENABLED */


/**
 * xmlParserInputBufferCreatePythonFile:
 * @file:  a PyFile_Type
 * @encoder:  the encoding converter or NULL
 *
 * Create a buffered output for the progressive saving to a PyFile_Type
 * buffered C I/O
 *
 * Returns the new parser output or NULL
 */
static xmlParserInputBufferPtr
xmlParserInputBufferCreatePythonFile(PyObject *file, 
	                        xmlCharEncoding encoding) {
    xmlParserInputBufferPtr ret;

    if (file == NULL) return(NULL);

    ret = xmlAllocParserInputBuffer(encoding);
    if (ret != NULL) {
        ret->context = file;
	/* Py_INCREF(file); */
	ret->readcallback = xmlPythonFileRead;
	ret->closecallback = xmlPythonFileClose;
    }

    return(ret);
}

PyObject *
libxml_xmlCreateInputBuffer(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) {
    PyObject *py_retval;
    PyObject *file;
    xmlChar  *encoding;
    xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
    xmlParserInputBufferPtr buffer;


    if (!PyArg_ParseTuple(args, (char *)"Oz:xmlParserInputBufferCreate",
		&file, &encoding))
	return(NULL);
    if ((encoding != NULL) && (encoding[0] != 0)) {
	enc = xmlParseCharEncoding((const char *) encoding);
    }
    buffer = xmlParserInputBufferCreatePythonFile(file, enc);
    if (buffer == NULL)
	printf("libxml_xmlParserInputBufferCreate: buffer == NULL\n");
    py_retval = libxml_xmlParserInputBufferPtrWrap(buffer);
    return(py_retval);
}

/************************************************************************
 *									*
 *		Providing the resolver at the Python level		*
 *									*
 ************************************************************************/

static xmlExternalEntityLoader defaultExternalEntityLoader = NULL;
static PyObject *pythonExternalEntityLoaderObjext;

static xmlParserInputPtr
pythonExternalEntityLoader(const char *URL, const char *ID,
			   xmlParserCtxtPtr ctxt) {
    xmlParserInputPtr result = NULL;
    if (pythonExternalEntityLoaderObjext != NULL) {
	PyObject *ret;
	PyObject *ctxtobj;

	ctxtobj = libxml_xmlParserCtxtPtrWrap(ctxt);
#ifdef DEBUG_LOADER
	printf("pythonExternalEntityLoader: ready to call\n");
#endif

	ret = PyObject_CallFunction(pythonExternalEntityLoaderObjext,
		      (char *) "(ssO)", URL, ID, ctxtobj);
	Py_XDECREF(ctxtobj);
#ifdef DEBUG_LOADER
	printf("pythonExternalEntityLoader: result ");
	PyObject_Print(ret, stderr, 0);
	printf("\n");
#endif

	if (ret != NULL) {
	    if (PyObject_HasAttrString(ret, (char *) "read")) {
		xmlParserInputBufferPtr buf;

		buf = xmlAllocParserInputBuffer(XML_CHAR_ENCODING_NONE);
		if (buf != NULL) {
		    buf->context = ret;
		    buf->readcallback = xmlPythonFileReadRaw;
		    buf->closecallback = xmlPythonFileCloseRaw;
		    result = xmlNewIOInputStream(ctxt, buf,
			                         XML_CHAR_ENCODING_NONE);
		}
	    } else {
		printf("pythonExternalEntityLoader: can't read\n");
	    }
	    if (result == NULL) {
		Py_DECREF(ret);
	    } else if (URL != NULL) {
		result->filename = (char *) xmlStrdup((const xmlChar *)URL);
		result->directory = xmlParserGetDirectory((const char *) URL);
	    }
	}
    }
    if ((result == NULL) && (defaultExternalEntityLoader != NULL)) {
	result = defaultExternalEntityLoader(URL, ID, ctxt);
    }
    return(result);
}

PyObject *
libxml_xmlSetEntityLoader(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) {
    PyObject *py_retval;
    PyObject *loader;

    if (!PyArg_ParseTuple(args, (char *)"O:libxml_xmlSetEntityLoader",
		&loader))
	return(NULL);

#ifdef DEBUG_LOADER
    printf("libxml_xmlSetEntityLoader\n");
#endif
    if (defaultExternalEntityLoader == NULL) 
	defaultExternalEntityLoader = xmlGetExternalEntityLoader();

    pythonExternalEntityLoaderObjext = loader;
    xmlSetExternalEntityLoader(pythonExternalEntityLoader);

    py_retval = PyInt_FromLong(0);
    return(py_retval);
}


/************************************************************************
 *									*
 *		Handling SAX/xmllib/sgmlop callback interfaces		*
 *									*
 ************************************************************************/

static void
pythonStartElement(void *user_data, const xmlChar * name,
                   const xmlChar ** attrs)
{
    int i;
    PyObject *handler;
    PyObject *dict;
    PyObject *attrname;
    PyObject *attrvalue;
    PyObject *result = NULL;
    int type = 0;

#ifdef DEBUG_SAX
    printf("pythonStartElement(%s) called\n", name);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "startElement"))
        type = 1;
    else if (PyObject_HasAttrString(handler, (char *) "start"))
        type = 2;
    if (type != 0) {
        /*
         * the xmllib interface always generate a dictionnary,
         * possibly empty
         */
        if ((attrs == NULL) && (type == 1)) {
            Py_XINCREF(Py_None);
            dict = Py_None;
        } else if (attrs == NULL) {
            dict = PyDict_New();
        } else {
            dict = PyDict_New();
            for (i = 0; attrs[i] != NULL; i++) {
                attrname = PyString_FromString((char *) attrs[i]);
                i++;
                if (attrs[i] != NULL) {
                    attrvalue = PyString_FromString((char *) attrs[i]);
                } else {
                    Py_XINCREF(Py_None);
                    attrvalue = Py_None;
                }
                PyDict_SetItem(dict, attrname, attrvalue);
            }
        }

        if (type == 1)
            result = PyObject_CallMethod(handler, (char *) "startElement",
                                         (char *) "sO", name, dict);
        else if (type == 2)
            result = PyObject_CallMethod(handler, (char *) "start",
                                         (char *) "sO", name, dict);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(dict);
        Py_XDECREF(result);
    }
}

static void
pythonStartDocument(void *user_data)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonStartDocument() called\n");
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "startDocument")) {
        result =
            PyObject_CallMethod(handler, (char *) "startDocument", NULL);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonEndDocument(void *user_data)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonEndDocument() called\n");
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "endDocument")) {
        result =
            PyObject_CallMethod(handler, (char *) "endDocument", NULL);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
    /*
     * The reference to the handler is released there
     */
    Py_XDECREF(handler);
}

static void
pythonEndElement(void *user_data, const xmlChar * name)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonEndElement(%s) called\n", name);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "endElement")) {
        result = PyObject_CallMethod(handler, (char *) "endElement",
                                     (char *) "s", name);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    } else if (PyObject_HasAttrString(handler, (char *) "end")) {
        result = PyObject_CallMethod(handler, (char *) "end",
                                     (char *) "s", name);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonReference(void *user_data, const xmlChar * name)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonReference(%s) called\n", name);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "reference")) {
        result = PyObject_CallMethod(handler, (char *) "reference",
                                     (char *) "s", name);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonCharacters(void *user_data, const xmlChar * ch, int len)
{
    PyObject *handler;
    PyObject *result = NULL;
    int type = 0;

#ifdef DEBUG_SAX
    printf("pythonCharacters(%s, %d) called\n", ch, len);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "characters"))
        type = 1;
    else if (PyObject_HasAttrString(handler, (char *) "data"))
        type = 2;
    if (type != 0) {
        if (type == 1)
            result = PyObject_CallMethod(handler, (char *) "characters",
                                         (char *) "s#", ch, len);
        else if (type == 2)
            result = PyObject_CallMethod(handler, (char *) "data",
                                         (char *) "s#", ch, len);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonIgnorableWhitespace(void *user_data, const xmlChar * ch, int len)
{
    PyObject *handler;
    PyObject *result = NULL;
    int type = 0;

#ifdef DEBUG_SAX
    printf("pythonIgnorableWhitespace(%s, %d) called\n", ch, len);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "ignorableWhitespace"))
        type = 1;
    else if (PyObject_HasAttrString(handler, (char *) "data"))
        type = 2;
    if (type != 0) {
        if (type == 1)
            result =
                PyObject_CallMethod(handler,
                                    (char *) "ignorableWhitespace",
                                    (char *) "s#", ch, len);
        else if (type == 2)
            result =