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.

                                    (char *) "ignorableWhitespace",
                                    (char *) "s#", ch, len);
        else if (type == 2)
            result =
                PyObject_CallMethod(handler, (char *) "data",
                                    (char *) "s#", ch, len);
        Py_XDECREF(result);
    }
}

static void
pythonProcessingInstruction(void *user_data,
                            const xmlChar * target, const xmlChar * data)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonProcessingInstruction(%s, %s) called\n", target, data);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "processingInstruction")) {
        result = PyObject_CallMethod(handler, (char *)
                                     "processingInstruction",
                                     (char *) "ss", target, data);
        Py_XDECREF(result);
    }
}

static void
pythonComment(void *user_data, const xmlChar * value)
{
    PyObject *handler;
    PyObject *result;

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

static void
pythonWarning(void *user_data, const char *msg, ...)
{
    PyObject *handler;
    PyObject *result;
    va_list args;
    char buf[1024];

#ifdef DEBUG_SAX
    printf("pythonWarning(%s) called\n", msg);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "warning")) {
        va_start(args, msg);
        vsnprintf(buf, 1023, msg, args);
        va_end(args);
        buf[1023] = 0;
        result =
            PyObject_CallMethod(handler, (char *) "warning", (char *) "s",
                                buf);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonError(void *user_data, const char *msg, ...)
{
    PyObject *handler;
    PyObject *result;
    va_list args;
    char buf[1024];

#ifdef DEBUG_SAX
    printf("pythonError(%s) called\n", msg);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "error")) {
        va_start(args, msg);
        vsnprintf(buf, 1023, msg, args);
        va_end(args);
        buf[1023] = 0;
        result =
            PyObject_CallMethod(handler, (char *) "error", (char *) "s",
                                buf);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonFatalError(void *user_data, const char *msg, ...)
{
    PyObject *handler;
    PyObject *result;
    va_list args;
    char buf[1024];

#ifdef DEBUG_SAX
    printf("pythonFatalError(%s) called\n", msg);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "fatalError")) {
        va_start(args, msg);
        vsnprintf(buf, 1023, msg, args);
        va_end(args);
        buf[1023] = 0;
        result =
            PyObject_CallMethod(handler, (char *) "fatalError",
                                (char *) "s", buf);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

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

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

static void
pythonExternalSubset(void *user_data,
                     const xmlChar * name,
                     const xmlChar * externalID, const xmlChar * systemID)
{
    PyObject *handler;
    PyObject *result;

#ifdef DEBUG_SAX
    printf("pythonExternalSubset(%s, %s, %s) called\n",
           name, externalID, systemID);
#endif
    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "externalSubset")) {
        result =
            PyObject_CallMethod(handler, (char *) "externalSubset",
                                (char *) "sss", name, externalID,
                                systemID);
        Py_XDECREF(result);
    }
}

static void
pythonEntityDecl(void *user_data,
                 const xmlChar * name,
                 int type,
                 const xmlChar * publicId,
                 const xmlChar * systemId, xmlChar * content)
{
    PyObject *handler;
    PyObject *result;

    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "entityDecl")) {
        result = PyObject_CallMethod(handler, (char *) "entityDecl",
                                     (char *) "sisss", name, type,
                                     publicId, systemId, content);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}



static void

pythonNotationDecl(void *user_data,
                   const xmlChar * name,
                   const xmlChar * publicId, const xmlChar * systemId)
{
    PyObject *handler;
    PyObject *result;

    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "notationDecl")) {
        result = PyObject_CallMethod(handler, (char *) "notationDecl",
                                     (char *) "sss", name, publicId,
                                     systemId);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonAttributeDecl(void *user_data,
                    const xmlChar * elem,
                    const xmlChar * name,
                    int type,
                    int def,
                    const xmlChar * defaultValue, xmlEnumerationPtr tree)
{
    PyObject *handler;
    PyObject *nameList;
    PyObject *newName;
    xmlEnumerationPtr node;
    PyObject *result;
    int count;

    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "attributeDecl")) {
        count = 0;
        for (node = tree; node != NULL; node = node->next) {
            count++;
        }
        nameList = PyList_New(count);
        count = 0;
        for (node = tree; node != NULL; node = node->next) {
            newName = PyString_FromString((char *) node->name);
            PyList_SetItem(nameList, count, newName);
	    Py_DECREF(newName);
            count++;
        }
        result = PyObject_CallMethod(handler, (char *) "attributeDecl",
                                     (char *) "ssiisO", elem, name, type,
                                     def, defaultValue, nameList);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(nameList);
        Py_XDECREF(result);
    }
}

static void
pythonElementDecl(void *user_data,
                  const xmlChar * name,
                  int type, ATTRIBUTE_UNUSED xmlElementContentPtr content)
{
    PyObject *handler;
    PyObject *obj;
    PyObject *result;

    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "elementDecl")) {
        /* TODO: wrap in an elementContent object */
        printf
            ("pythonElementDecl: xmlElementContentPtr wrapper missing !\n");
        obj = Py_None;
        /* Py_XINCREF(Py_None); isn't the reference just borrowed ??? */
        result = PyObject_CallMethod(handler, (char *) "elementDecl",
                                     (char *) "siO", name, type, obj);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonUnparsedEntityDecl(void *user_data,
                         const xmlChar * name,
                         const xmlChar * publicId,
                         const xmlChar * systemId,
                         const xmlChar * notationName)
{
    PyObject *handler;
    PyObject *result;

    handler = (PyObject *) user_data;
    if (PyObject_HasAttrString(handler, (char *) "unparsedEntityDecl")) {
        result =
            PyObject_CallMethod(handler, (char *) "unparsedEntityDecl",
                                (char *) "ssss", name, publicId, systemId,
                                notationName);
        if (PyErr_Occurred())
            PyErr_Print();
        Py_XDECREF(result);
    }
}

static void
pythonInternalSubset(void *user_data, const xmlChar * name,
                     const xmlChar * ExternalID, const xmlChar * SystemID)
{
    PyObject *handler;
    PyObject *result;

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

static xmlSAXHandler pythonSaxHandler = {
    pythonInternalSubset,
    NULL,                       /* TODO pythonIsStandalone, */
    NULL,                       /* TODO pythonHasInternalSubset, */
    NULL,                       /* TODO pythonHasExternalSubset, */
    NULL,                       /* TODO pythonResolveEntity, */
    NULL,                       /* TODO pythonGetEntity, */
    pythonEntityDecl,
    pythonNotationDecl,
    pythonAttributeDecl,
    pythonElementDecl,
    pythonUnparsedEntityDecl,
    NULL,                       /* OBSOLETED pythonSetDocumentLocator, */
    pythonStartDocument,
    pythonEndDocument,
    pythonStartElement,
    pythonEndElement,
    pythonReference,
    pythonCharacters,
    pythonIgnorableWhitespace,
    pythonProcessingInstruction,
    pythonComment,
    pythonWarning,
    pythonError,
    pythonFatalError,
    NULL,                       /* TODO pythonGetParameterEntity, */
    pythonCdataBlock,
    pythonExternalSubset,
    1,
    NULL,			/* TODO mograte to SAX2 */
    NULL,
    NULL,
    NULL
};

/************************************************************************
 *									*
 *		Handling of specific parser context			*
 *									*
 ************************************************************************/

PyObject *
libxml_xmlCreatePushParser(ATTRIBUTE_UNUSED PyObject * self,
                           PyObject * args)
{
    const char *chunk;
    int size;
    const char *URI;
    PyObject *pyobj_SAX = NULL;
    xmlSAXHandlerPtr SAX = NULL;
    xmlParserCtxtPtr ret;
    PyObject *pyret;

    if (!PyArg_ParseTuple
        (args, (char *) "Oziz:xmlCreatePushParser", &pyobj_SAX, &chunk,
         &size, &URI))
        return (NULL);

#ifdef DEBUG
    printf("libxml_xmlCreatePushParser(%p, %s, %d, %s) called\n",
           pyobj_SAX, chunk, size, URI);
#endif
    if (pyobj_SAX != Py_None) {
        SAX = &pythonSaxHandler;
        Py_INCREF(pyobj_SAX);
        /* The reference is released in pythonEndDocument() */
    }
    ret = xmlCreatePushParserCtxt(SAX, pyobj_SAX, chunk, size, URI);
    pyret = libxml_xmlParserCtxtPtrWrap(ret);
    return (pyret);
}

PyObject *
libxml_htmlCreatePushParser(ATTRIBUTE_UNUSED PyObject * self,
                            PyObject * args)
{
#ifdef LIBXML_HTML_ENABLED
    const char *chunk;
    int size;
    const char *URI;
    PyObject *pyobj_SAX = NULL;
    xmlSAXHandlerPtr SAX = NULL;
    xmlParserCtxtPtr ret;
    PyObject *pyret;

    if (!PyArg_ParseTuple
        (args, (char *) "Oziz:htmlCreatePushParser", &pyobj_SAX, &chunk,
         &size, &URI))
        return (NULL);

#ifdef DEBUG
    printf("libxml_htmlCreatePushParser(%p, %s, %d, %s) called\n",
           pyobj_SAX, chunk, size, URI);
#endif
    if (pyobj_SAX != Py_None) {
        SAX = &pythonSaxHandler;
        Py_INCREF(pyobj_SAX);
        /* The reference is released in pythonEndDocument() */
    }
    ret = htmlCreatePushParserCtxt(SAX, pyobj_SAX, chunk, size, URI,
                                   XML_CHAR_ENCODING_NONE);
    pyret = libxml_xmlParserCtxtPtrWrap(ret);
    return (pyret);
#else
    Py_INCREF(Py_None);
    return (Py_None);
#endif /* LIBXML_HTML_ENABLED */
}

PyObject *
libxml_xmlSAXParseFile(ATTRIBUTE_UNUSED PyObject * self, PyObject * args)
{
    int recover;
    const char *URI;
    PyObject *pyobj_SAX = NULL;
    xmlSAXHandlerPtr SAX = NULL;

    if (!PyArg_ParseTuple(args, (char *) "Osi:xmlSAXParseFile", &pyobj_SAX,
                          &URI, &recover))
        return (NULL);

#ifdef DEBUG
    printf("libxml_xmlSAXParseFile(%p, %s, %d) called\n",
           pyobj_SAX, URI, recover);
#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() */
    xmlSAXUserParseFile(SAX, pyobj_SAX, URI);
    Py_INCREF(Py_None);