libxml.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.

    return (Py_None);
}

PyObject *
libxml_htmlSAXParseFile(ATTRIBUTE_UNUSED PyObject * self, PyObject * args)
{
#ifdef LIBXML_HTML_ENABLED
    const char *URI;
    const char *encoding;
    PyObject *pyobj_SAX = NULL;
    xmlSAXHandlerPtr SAX = NULL;

    if (!PyArg_ParseTuple
        (args, (char *) "Osz:htmlSAXParseFile", &pyobj_SAX, &URI,
         &encoding))
        return (NULL);

#ifdef DEBUG
    printf("libxml_htmlSAXParseFile(%p, %s, %s) called\n",
           pyobj_SAX, URI, encoding);
#endif
    if (pyobj_SAX == Py_None) {
        Py_INCREF(Py_None);
        return (Py_None);
    }
    SAX = &pythonSaxHandler;
    Py_INCREF(pyobj_SAX);
    /* The reference is released in pythonEndDocument() */
    htmlSAXParseFile(URI, encoding, SAX, pyobj_SAX);
    Py_INCREF(Py_None);
    return (Py_None);
#else
    Py_INCREF(Py_None);
    return (Py_None);
#endif /* LIBXML_HTML_ENABLED */
}

/************************************************************************
 *									*
 *			Error message callback				*
 *									*
 ************************************************************************/

static PyObject *libxml_xmlPythonErrorFuncHandler = NULL;
static PyObject *libxml_xmlPythonErrorFuncCtxt = NULL;

/* helper to build a xmlMalloc'ed string from a format and va_list */
/* 
 * disabled the loop, the repeated call to vsnprintf without reset of ap
 * in case the initial buffer was too small segfaulted on x86_64
 * we now directly vsnprintf on a large buffer.
 */
static char *
libxml_buildMessage(const char *msg, va_list ap)
{
    int chars;
    char *str;

    str = (char *) xmlMalloc(1000);
    if (str == NULL)
        return NULL;

    chars = vsnprintf(str, 999, msg, ap);
    if (chars >= 998)
        str[999] = 0;

    return str;
}

static void
libxml_xmlErrorFuncHandler(ATTRIBUTE_UNUSED void *ctx, const char *msg,
                           ...)
{
    va_list ap;
    PyObject *list;
    PyObject *message;
    PyObject *result;
    char str[1000];

#ifdef DEBUG_ERROR
    printf("libxml_xmlErrorFuncHandler(%p, %s, ...) called\n", ctx, msg);
#endif


    if (libxml_xmlPythonErrorFuncHandler == NULL) {
        va_start(ap, msg);
        vfprintf(stderr, msg, ap);
        va_end(ap);
    } else {
        va_start(ap, msg);
        if (vsnprintf(str, 999, msg, ap) >= 998)
	    str[999] = 0;
        va_end(ap);

        list = PyTuple_New(2);
        PyTuple_SetItem(list, 0, libxml_xmlPythonErrorFuncCtxt);
        Py_XINCREF(libxml_xmlPythonErrorFuncCtxt);
        message = libxml_charPtrConstWrap(str);
        PyTuple_SetItem(list, 1, message);
        result = PyEval_CallObject(libxml_xmlPythonErrorFuncHandler, list);
        Py_XDECREF(list);
        Py_XDECREF(result);
    }
}

static void
libxml_xmlErrorInitialize(void)
{
#ifdef DEBUG_ERROR
    printf("libxml_xmlErrorInitialize() called\n");
#endif
    xmlSetGenericErrorFunc(NULL, libxml_xmlErrorFuncHandler);
    xmlThrDefSetGenericErrorFunc(NULL, libxml_xmlErrorFuncHandler);
}

static PyObject *
libxml_xmlRegisterErrorHandler(ATTRIBUTE_UNUSED PyObject * self,
                               PyObject * args)
{
    PyObject *py_retval;
    PyObject *pyobj_f;
    PyObject *pyobj_ctx;

    if (!PyArg_ParseTuple
        (args, (char *) "OO:xmlRegisterErrorHandler", &pyobj_f,
         &pyobj_ctx))
        return (NULL);

#ifdef DEBUG_ERROR
    printf("libxml_xmlRegisterErrorHandler(%p, %p) called\n", pyobj_ctx,
           pyobj_f);
#endif

    if (libxml_xmlPythonErrorFuncHandler != NULL) {
        Py_XDECREF(libxml_xmlPythonErrorFuncHandler);
    }
    if (libxml_xmlPythonErrorFuncCtxt != NULL) {
        Py_XDECREF(libxml_xmlPythonErrorFuncCtxt);
    }

    Py_XINCREF(pyobj_ctx);
    Py_XINCREF(pyobj_f);

    /* TODO: check f is a function ! */
    libxml_xmlPythonErrorFuncHandler = pyobj_f;
    libxml_xmlPythonErrorFuncCtxt = pyobj_ctx;

    py_retval = libxml_intWrap(1);
    return (py_retval);
}


/************************************************************************
 *									*
 *                      Per parserCtxt error handler                    *
 *									*
 ************************************************************************/

typedef struct 
{
    PyObject *f;
    PyObject *arg;
} xmlParserCtxtPyCtxt;
typedef xmlParserCtxtPyCtxt *xmlParserCtxtPyCtxtPtr;

static void
libxml_xmlParserCtxtGenericErrorFuncHandler(void *ctx, int severity, char *str) 
{
    PyObject *list;
    PyObject *result;
    xmlParserCtxtPtr ctxt;
    xmlParserCtxtPyCtxtPtr pyCtxt;
    
#ifdef DEBUG_ERROR
    printf("libxml_xmlParserCtxtGenericErrorFuncHandler(%p, %s, ...) called\n", ctx, str);
#endif

    ctxt = (xmlParserCtxtPtr)ctx;
    pyCtxt = (xmlParserCtxtPyCtxtPtr)ctxt->_private;

    list = PyTuple_New(4);
    PyTuple_SetItem(list, 0, pyCtxt->arg);
    Py_XINCREF(pyCtxt->arg);
    PyTuple_SetItem(list, 1, libxml_charPtrWrap(str));
    PyTuple_SetItem(list, 2, libxml_intWrap(severity));
    PyTuple_SetItem(list, 3, Py_None);
    Py_INCREF(Py_None);
    result = PyEval_CallObject(pyCtxt->f, list);
    if (result == NULL) 
    {
	/* TODO: manage for the exception to be propagated... */
	PyErr_Print();
    }
    Py_XDECREF(list);
    Py_XDECREF(result);
}

static void 
libxml_xmlParserCtxtErrorFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlParserCtxtGenericErrorFuncHandler(ctx,XML_PARSER_SEVERITY_ERROR,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static void 
libxml_xmlParserCtxtWarningFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlParserCtxtGenericErrorFuncHandler(ctx,XML_PARSER_SEVERITY_WARNING,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static void 
libxml_xmlParserCtxtValidityErrorFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlParserCtxtGenericErrorFuncHandler(ctx,XML_PARSER_SEVERITY_VALIDITY_ERROR,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static void 
libxml_xmlParserCtxtValidityWarningFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlParserCtxtGenericErrorFuncHandler(ctx,XML_PARSER_SEVERITY_VALIDITY_WARNING,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static PyObject *
libxml_xmlParserCtxtSetErrorHandler(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) 
{
    PyObject *py_retval;
    xmlParserCtxtPtr ctxt;
    xmlParserCtxtPyCtxtPtr pyCtxt;
    PyObject *pyobj_ctxt;
    PyObject *pyobj_f;
    PyObject *pyobj_arg;

    if (!PyArg_ParseTuple(args, (char *)"OOO:xmlParserCtxtSetErrorHandler",
		          &pyobj_ctxt, &pyobj_f, &pyobj_arg))
        return(NULL);
    ctxt = (xmlParserCtxtPtr) PyparserCtxt_Get(pyobj_ctxt);
    if (ctxt->_private == NULL) {
	pyCtxt = xmlMalloc(sizeof(xmlParserCtxtPyCtxt));
	if (pyCtxt == NULL) {
	    py_retval = libxml_intWrap(-1);
	    return(py_retval);
	}
	memset(pyCtxt,0,sizeof(xmlParserCtxtPyCtxt));
	ctxt->_private = pyCtxt;
    }
    else {
	pyCtxt = (xmlParserCtxtPyCtxtPtr)ctxt->_private;
    }
    /* TODO: check f is a function ! */
    Py_XDECREF(pyCtxt->f);
    Py_XINCREF(pyobj_f);
    pyCtxt->f = pyobj_f;
    Py_XDECREF(pyCtxt->arg);
    Py_XINCREF(pyobj_arg);
    pyCtxt->arg = pyobj_arg;

    if (pyobj_f != Py_None) {
	ctxt->sax->error = libxml_xmlParserCtxtErrorFuncHandler;
	ctxt->sax->warning = libxml_xmlParserCtxtWarningFuncHandler;
	ctxt->vctxt.error = libxml_xmlParserCtxtValidityErrorFuncHandler;
	ctxt->vctxt.warning = libxml_xmlParserCtxtValidityWarningFuncHandler;
    }
    else {
	ctxt->sax->error = xmlParserError;
	ctxt->vctxt.error = xmlParserValidityError;
	ctxt->sax->warning = xmlParserWarning;
	ctxt->vctxt.warning = xmlParserValidityWarning;
    }

    py_retval = libxml_intWrap(1);
    return(py_retval);
}

static PyObject *
libxml_xmlParserCtxtGetErrorHandler(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) 
{
    PyObject *py_retval;
    xmlParserCtxtPtr ctxt;
    xmlParserCtxtPyCtxtPtr pyCtxt;
    PyObject *pyobj_ctxt;

    if (!PyArg_ParseTuple(args, (char *)"O:xmlParserCtxtGetErrorHandler",
		          &pyobj_ctxt))
        return(NULL);
    ctxt = (xmlParserCtxtPtr) PyparserCtxt_Get(pyobj_ctxt);
    py_retval = PyTuple_New(2);
    if (ctxt->_private != NULL) {
	pyCtxt = (xmlParserCtxtPyCtxtPtr)ctxt->_private;

	PyTuple_SetItem(py_retval, 0, pyCtxt->f);
	Py_XINCREF(pyCtxt->f);
	PyTuple_SetItem(py_retval, 1, pyCtxt->arg);
	Py_XINCREF(pyCtxt->arg);
    }
    else {
	/* no python error handler registered */
	PyTuple_SetItem(py_retval, 0, Py_None);
	Py_XINCREF(Py_None);
	PyTuple_SetItem(py_retval, 1, Py_None);
	Py_XINCREF(Py_None);
    }
    return(py_retval);
}

static PyObject *
libxml_xmlFreeParserCtxt(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) {
    xmlParserCtxtPtr ctxt;
    PyObject *pyobj_ctxt;
    xmlParserCtxtPyCtxtPtr pyCtxt;

    if (!PyArg_ParseTuple(args, (char *)"O:xmlFreeParserCtxt", &pyobj_ctxt))
        return(NULL);
    ctxt = (xmlParserCtxtPtr) PyparserCtxt_Get(pyobj_ctxt);

    if (ctxt != NULL) {
	pyCtxt = (xmlParserCtxtPyCtxtPtr)((xmlParserCtxtPtr)ctxt)->_private;
	if (pyCtxt) {
	    Py_XDECREF(pyCtxt->f);
	    Py_XDECREF(pyCtxt->arg);
	    xmlFree(pyCtxt);
	}
	xmlFreeParserCtxt(ctxt);
    }

    Py_INCREF(Py_None);
    return(Py_None);
}

/***
 * xmlValidCtxt stuff
 */

typedef struct 
{
    PyObject *warn;
    PyObject *error;
    PyObject *arg;
} xmlValidCtxtPyCtxt;
typedef xmlValidCtxtPyCtxt *xmlValidCtxtPyCtxtPtr;

static void
libxml_xmlValidCtxtGenericErrorFuncHandler(void *ctx, int severity, char *str) 
{
    PyObject *list;
    PyObject *result;
    xmlValidCtxtPyCtxtPtr pyCtxt;
    
#ifdef DEBUG_ERROR
    printf("libxml_xmlValidCtxtGenericErrorFuncHandler(%p, %d, %s, ...) called\n", ctx, severity, str);
#endif

    pyCtxt = (xmlValidCtxtPyCtxtPtr)ctx;
    
    list = PyTuple_New(2);
    PyTuple_SetItem(list, 0, libxml_charPtrWrap(str));
    PyTuple_SetItem(list, 1, pyCtxt->arg);
    Py_XINCREF(pyCtxt->arg);
    result = PyEval_CallObject(pyCtxt->error, list);
    if (result == NULL) 
    {
	/* TODO: manage for the exception to be propagated... */
	PyErr_Print();
    }
    Py_XDECREF(list);
    Py_XDECREF(result);
}

static void
libxml_xmlValidCtxtGenericWarningFuncHandler(void *ctx, int severity, char *str) 
{
    PyObject *list;
    PyObject *result;
    xmlValidCtxtPyCtxtPtr pyCtxt;
    
#ifdef DEBUG_ERROR
    printf("libxml_xmlValidCtxtGenericWarningFuncHandler(%p, %d, %s, ...) called\n", ctx, severity, str);
#endif

    pyCtxt = (xmlValidCtxtPyCtxtPtr)ctx;

    list = PyTuple_New(2);
    PyTuple_SetItem(list, 0, libxml_charPtrWrap(str));
    PyTuple_SetItem(list, 1, pyCtxt->arg);
    Py_XINCREF(pyCtxt->arg);
    result = PyEval_CallObject(pyCtxt->warn, list);
    if (result == NULL) 
    {
	/* TODO: manage for the exception to be propagated... */
	PyErr_Print();
    }
    Py_XDECREF(list);
    Py_XDECREF(result);
}

static void 
libxml_xmlValidCtxtErrorFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlValidCtxtGenericErrorFuncHandler(ctx,XML_PARSER_SEVERITY_VALIDITY_ERROR,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static void 
libxml_xmlValidCtxtWarningFuncHandler(void *ctx, const char *msg, ...) 
{
    va_list ap;

    va_start(ap, msg);
    libxml_xmlValidCtxtGenericWarningFuncHandler(ctx,XML_PARSER_SEVERITY_VALIDITY_WARNING,libxml_buildMessage(msg,ap));
    va_end(ap);
}

static PyObject *
libxml_xmlSetValidErrors(ATTRIBUTE_UNUSED PyObject * self, PyObject * args)
{
    PyObject *py_retval;
    PyObject *pyobj_error;
    PyObject *pyobj_warn;
    PyObject *pyobj_ctx;
    PyObject *pyobj_arg = Py_None;
    xmlValidCtxtPtr ctxt;
    xmlValidCtxtPyCtxtPtr pyCtxt;

    if (!PyArg_ParseTuple
        (args, (char *) "OOO|O:xmlSetValidErrors", &pyobj_ctx, &pyobj_error, &pyobj_warn, &pyobj_arg))
        return (NULL);

#ifdef DEBUG_ERROR
    printf("libxml_xmlSetValidErrors(%p, %p, %p) called\n", pyobj_ctx, pyobj_error, pyobj_warn);
#endif

    ctxt = PyValidCtxt_Get(pyobj_ctx);
    pyCtxt = xmlMalloc(sizeof(xmlValidCtxtPyCtxt));
    if (pyCtxt == NULL) {
            py_retval = libxml_intWrap(-1);
            return(py_retval);
    }
    memset(pyCtxt, 0, sizeof(xmlValidCtxtPyCtxt));

    
    /* TODO: check warn and error is a function ! */
    Py_XDECREF(pyCtxt->error);
    Py_XINCREF(pyobj_error);
    pyCtxt->error = pyobj_error;
    
    Py_XDECREF(pyCtxt->warn);