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.

/*
 * libxml.c: this modules implements the main part of the glue of the
 *           libxml2 library and the Python interpreter. It provides the
 *           entry points where an automatically generated stub is either
 *           unpractical or would not match cleanly the Python model.
 *
 * If compiled with MERGED_MODULES, the entry point will be used to
 * initialize both the libxml2 and the libxslt wrappers
 *
 * See Copyright for the status of this software.
 *
 * [e-mail]
 */
#include <Python.h>
#include <fileobject.h>
/* #include "config.h" */
#include <libxml/xmlmemory.h>
#include <libxml/parser.h>
#include <libxml/tree.h>
#include <libxml/xpath.h>
#include <libxml/xmlerror.h>
#include <libxml/xpathInternals.h>
#include <libxml/xmlmemory.h>
#include <libxml/xmlIO.h>
#include <libxml/c14n.h>
#include "libxml_wrap.h"
#include "libxml2-py.h"

#if (defined(_MSC_VER) || defined(__MINGW32__)) && !defined(vsnprintf)
#define vsnprintf(b,c,f,a) _vsnprintf(b,c,f,a)
#elif defined(WITH_TRIO)
#include "trio.h"
#define vsnprintf trio_vsnprintf
#endif

/* #define DEBUG */
/* #define DEBUG_SAX */
/* #define DEBUG_XPATH */
/* #define DEBUG_ERROR */
/* #define DEBUG_MEMORY */
/* #define DEBUG_FILES */
/* #define DEBUG_LOADER */

void initlibxml2mod(void);

/**
 * TODO:
 *
 * macro to flag unimplemented blocks
 */
#define TODO 								\
    xmlGenericError(xmlGenericErrorContext,				\
	    "Unimplemented block at %s:%d\n",				\
            __FILE__, __LINE__);
/*
 * the following vars are used for XPath extensions, but
 * are also referenced within the parser cleanup routine.
 */
static int libxml_xpathCallbacksInitialized = 0;

typedef struct libxml_xpathCallback {
    xmlXPathContextPtr ctx;
    xmlChar *name;
    xmlChar *ns_uri;
    PyObject *function;
} libxml_xpathCallback, *libxml_xpathCallbackPtr;
typedef libxml_xpathCallback libxml_xpathCallbackArray[];
static int libxml_xpathCallbacksAllocd = 10;
static libxml_xpathCallbackArray *libxml_xpathCallbacks = NULL;
static int libxml_xpathCallbacksNb = 0;

/************************************************************************
 *									*
 *		Memory debug interface					*
 *									*
 ************************************************************************/

#if 0
extern void xmlMemFree(void *ptr);
extern void *xmlMemMalloc(size_t size);
extern void *xmlMemRealloc(void *ptr, size_t size);
extern char *xmlMemoryStrdup(const char *str);
#endif

static int libxmlMemoryDebugActivated = 0;
static long libxmlMemoryAllocatedBase = 0;

static int libxmlMemoryDebug = 0;
static xmlFreeFunc freeFunc = NULL;
static xmlMallocFunc mallocFunc = NULL;
static xmlReallocFunc reallocFunc = NULL;
static xmlStrdupFunc strdupFunc = NULL;

static void
libxml_xmlErrorInitialize(void); /* forward declare */

PyObject *
libxml_xmlMemoryUsed(PyObject * self ATTRIBUTE_UNUSED, 
        PyObject * args ATTRIBUTE_UNUSED)
{
    long ret;
    PyObject *py_retval;

    ret = xmlMemUsed();

    py_retval = libxml_longWrap(ret);
    return (py_retval);
}

PyObject *
libxml_xmlDebugMemory(PyObject * self ATTRIBUTE_UNUSED, PyObject * args)
{
    int activate;
    PyObject *py_retval;
    long ret;

    if (!PyArg_ParseTuple(args, (char *) "i:xmlDebugMemory", &activate))
        return (NULL);

#ifdef DEBUG_MEMORY
    printf("libxml_xmlDebugMemory(%d) called\n", activate);
#endif

    if (activate != 0) {
        if (libxmlMemoryDebug == 0) {
            /*
             * First initialize the library and grab the old memory handlers
             * and switch the library to memory debugging
             */
            xmlMemGet((xmlFreeFunc *) & freeFunc,
                      (xmlMallocFunc *) & mallocFunc,
                      (xmlReallocFunc *) & reallocFunc,
                      (xmlStrdupFunc *) & strdupFunc);
            if ((freeFunc == xmlMemFree) && (mallocFunc == xmlMemMalloc) &&
                (reallocFunc == xmlMemRealloc) &&
                (strdupFunc == xmlMemoryStrdup)) {
                libxmlMemoryAllocatedBase = xmlMemUsed();
            } else {
                /* 
                 * cleanup first, because some memory has been
                 * allocated with the non-debug malloc in xmlInitParser
                 * when the python module was imported
                 */
                xmlCleanupParser();
                ret = (long) xmlMemSetup(xmlMemFree, xmlMemMalloc,
                                         xmlMemRealloc, xmlMemoryStrdup);
                if (ret < 0)
                    goto error;
                libxmlMemoryAllocatedBase = xmlMemUsed();
                /* reinitialize */
                xmlInitParser();
                libxml_xmlErrorInitialize();
            }
            ret = 0;
        } else if (libxmlMemoryDebugActivated == 0) {
            libxmlMemoryAllocatedBase = xmlMemUsed();
            ret = 0;
        } else {
            ret = xmlMemUsed() - libxmlMemoryAllocatedBase;
        }
        libxmlMemoryDebug = 1;
        libxmlMemoryDebugActivated = 1;
    } else {
        if (libxmlMemoryDebugActivated == 1)
            ret = xmlMemUsed() - libxmlMemoryAllocatedBase;
        else
            ret = 0;
        libxmlMemoryDebugActivated = 0;
    }
  error:
    py_retval = libxml_longWrap(ret);
    return (py_retval);
}

PyObject *
libxml_xmlPythonCleanupParser(PyObject *self ATTRIBUTE_UNUSED,
                              PyObject *args ATTRIBUTE_UNUSED) {

    int ix;
    long freed = -1;

    if (libxmlMemoryDebug) {
        freed = xmlMemUsed();
    }

    xmlCleanupParser();
    /*
     * Need to confirm whether we really want to do this (required for
     * memcheck) in all cases...
     */
   
    if (libxml_xpathCallbacks != NULL) {	/* if ext funcs declared */
	for (ix=0; ix<libxml_xpathCallbacksNb; ix++) {
	    if ((*libxml_xpathCallbacks)[ix].name != NULL)
	        xmlFree((*libxml_xpathCallbacks)[ix].name);
	    if ((*libxml_xpathCallbacks)[ix].ns_uri != NULL)
	        xmlFree((*libxml_xpathCallbacks)[ix].ns_uri);
	}
	libxml_xpathCallbacksNb = 0;
        xmlFree(libxml_xpathCallbacks);
	libxml_xpathCallbacks = NULL;
    }

    if (libxmlMemoryDebug) {
        freed -= xmlMemUsed();
	libxmlMemoryAllocatedBase -= freed;
	if (libxmlMemoryAllocatedBase < 0)
	    libxmlMemoryAllocatedBase = 0;
    }

    Py_INCREF(Py_None);
    return(Py_None);
}

PyObject *
libxml_xmlDumpMemory(ATTRIBUTE_UNUSED PyObject * self,
                     ATTRIBUTE_UNUSED PyObject * args)
{

    if (libxmlMemoryDebug != 0)
        xmlMemoryDump();
    Py_INCREF(Py_None);
    return (Py_None);
}

/************************************************************************
 *									*
 *		Handling Python FILE I/O at the C level			*
 *	The raw I/O attack diectly the File objects, while the		*
 *	other routines address the ioWrapper instance instead		*
 *									*
 ************************************************************************/

/**
 * xmlPythonFileCloseUnref:
 * @context:  the I/O context
 *
 * Close an I/O channel
 */
static int
xmlPythonFileCloseRaw (void * context) {
    PyObject *file, *ret;

#ifdef DEBUG_FILES
    printf("xmlPythonFileCloseUnref\n");
#endif
    file = (PyObject *) context;
    if (file == NULL) return(-1);
    ret = PyEval_CallMethod(file, (char *) "close", (char *) "()");
    if (ret != NULL) {
	Py_DECREF(ret);
    }
    Py_DECREF(file);
    return(0);
}

/**
 * xmlPythonFileReadRaw:
 * @context:  the I/O context
 * @buffer:  where to drop data
 * @len:  number of bytes to write
 *
 * Read @len bytes to @buffer from the Python file in the I/O channel
 *
 * Returns the number of bytes read
 */
static int
xmlPythonFileReadRaw (void * context, char * buffer, int len) {
    PyObject *file;
    PyObject *ret;
    int lenread = -1;
    char *data;

#ifdef DEBUG_FILES
    printf("xmlPythonFileReadRaw: %d\n", len);
#endif
    file = (PyObject *) context;
    if (file == NULL) return(-1);
    ret = PyEval_CallMethod(file, (char *) "read", (char *) "(i)", len);
    if (ret == NULL) {
	printf("xmlPythonFileReadRaw: result is NULL\n");
	return(-1);
    } else if (PyString_Check(ret)) {
	lenread = PyString_Size(ret);
	data = PyString_AsString(ret);
	if (lenread > len)
	    memcpy(buffer, data, len);
	else
	    memcpy(buffer, data, lenread);
	Py_DECREF(ret);
    } else {
	printf("xmlPythonFileReadRaw: result is not a String\n");
	Py_DECREF(ret);
    }
    return(lenread);
}

/**
 * xmlPythonFileRead:
 * @context:  the I/O context
 * @buffer:  where to drop data
 * @len:  number of bytes to write
 *
 * Read @len bytes to @buffer from the I/O channel.
 *
 * Returns the number of bytes read
 */
static int
xmlPythonFileRead (void * context, char * buffer, int len) {
    PyObject *file;
    PyObject *ret;
    int lenread = -1;
    char *data;

#ifdef DEBUG_FILES
    printf("xmlPythonFileRead: %d\n", len);
#endif
    file = (PyObject *) context;
    if (file == NULL) return(-1);
    ret = PyEval_CallMethod(file, (char *) "io_read", (char *) "(i)", len);
    if (ret == NULL) {
	printf("xmlPythonFileRead: result is NULL\n");
	return(-1);
    } else if (PyString_Check(ret)) {
	lenread = PyString_Size(ret);
	data = PyString_AsString(ret);
	if (lenread > len)
	    memcpy(buffer, data, len);
	else
	    memcpy(buffer, data, lenread);
	Py_DECREF(ret);
    } else {
	printf("xmlPythonFileRead: result is not a String\n");
	Py_DECREF(ret);
    }
    return(lenread);
}

/**
 * xmlFileWrite:
 * @context:  the I/O context
 * @buffer:  where to drop data
 * @len:  number of bytes to write
 *
 * Write @len bytes from @buffer to the I/O channel.
 *
 * Returns the number of bytes written
 */
static int
xmlPythonFileWrite (void * context, const char * buffer, int len) {
    PyObject *file;
    PyObject *string;
    PyObject *ret = NULL;
    int written = -1;

#ifdef DEBUG_FILES
    printf("xmlPythonFileWrite: %d\n", len);
#endif
    file = (PyObject *) context;
    if (file == NULL) return(-1);
    string = PyString_FromStringAndSize(buffer, len);
    if (string == NULL) return(-1);
    if (PyObject_HasAttrString(file, (char *) "io_write")) {
        ret = PyEval_CallMethod(file, (char *) "io_write", (char *) "(O)",
	                        string);
    } else if (PyObject_HasAttrString(file, (char *) "write")) {
        ret = PyEval_CallMethod(file, (char *) "write", (char *) "(O)",
	                        string);
    }
    Py_DECREF(string);
    if (ret == NULL) {
	printf("xmlPythonFileWrite: result is NULL\n");
	return(-1);
    } else if (PyInt_Check(ret)) {
	written = (int) PyInt_AsLong(ret);
	Py_DECREF(ret);
    } else if (ret == Py_None) {
	written = len;
	Py_DECREF(ret);
    } else {
	printf("xmlPythonFileWrite: result is not an Int nor None\n");
	Py_DECREF(ret);
    }
    return(written);
}

/**
 * xmlPythonFileClose:
 * @context:  the I/O context
 *
 * Close an I/O channel
 */
static int
xmlPythonFileClose (void * context) {
    PyObject *file, *ret = NULL;

#ifdef DEBUG_FILES
    printf("xmlPythonFileClose\n");
#endif
    file = (PyObject *) context;
    if (file == NULL) return(-1);
    if (PyObject_HasAttrString(file, (char *) "io_close")) {
        ret = PyEval_CallMethod(file, (char *) "io_close", (char *) "()");
    } else if (PyObject_HasAttrString(file, (char *) "flush")) {
        ret = PyEval_CallMethod(file, (char *) "flush", (char *) "()");
    }
    if (ret != NULL) {
	Py_DECREF(ret);
    }
    return(0);
}

#ifdef LIBXML_OUTPUT_ENABLED
/**
 * xmlOutputBufferCreatePythonFile:
 * @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 xmlOutputBufferPtr
xmlOutputBufferCreatePythonFile(PyObject *file, 
	                        xmlCharEncodingHandlerPtr encoder) {
    xmlOutputBufferPtr ret;

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

    ret = xmlAllocOutputBuffer(encoder);
    if (ret != NULL) {
        ret->context = file;
	/* Py_INCREF(file); */
	ret->writecallback = xmlPythonFileWrite;
	ret->closecallback = xmlPythonFileClose;
    }

    return(ret);
}

PyObject *
libxml_xmlCreateOutputBuffer(ATTRIBUTE_UNUSED PyObject *self, PyObject *args) {
    PyObject *py_retval;
    PyObject *file;
    xmlChar  *encoding;
    xmlCharEncodingHandlerPtr handler = NULL;
    xmlOutputBufferPtr buffer;


    if (!PyArg_ParseTuple(args, (char *)"Oz:xmlOutputBufferCreate",
		&file, &encoding))
	return(NULL);
    if ((encoding != NULL) && (encoding[0] != 0)) {
	handler = xmlFindCharEncodingHandler((const char *) encoding);
    }
    buffer = xmlOutputBufferCreatePythonFile(file, handler);
    if (buffer == NULL)
	printf("libxml_xmlCreateOutputBuffer: buffer == NULL\n");
    py_retval = libxml_xmlOutputBufferPtrWrap(buffer);
    return(py_retval);
}