xmlio.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.

/*
 * xmlIO.c : implementation of the I/O interfaces used by the parser
 *
 * See Copyright for the status of this software.
 *
 * [e-mail]
 *
 * 14 Nov 2000 ht - for VMS, truncated name of long functions to under 32 char
 */

#define IN_LIBXML
#include "libxml.h"

#include <string.h>
#ifdef HAVE_ERRNO_H
#include <errno.h>
#endif


#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#ifdef HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#ifdef HAVE_STDLIB_H
#include <stdlib.h>
#endif
#ifdef HAVE_ZLIB_H
#include <zlib.h>
#endif

/* Figure a portable way to know if a file is a directory. */
#ifndef HAVE_STAT
#  ifdef HAVE__STAT
     /* MS C library seems to define stat and _stat. The definition
        is identical. Still, mapping them to each other causes a warning. */
#    ifndef _MSC_VER
#      define stat(x,y) _stat(x,y)
#    endif
#    define HAVE_STAT
#  endif
#endif
#ifdef HAVE_STAT
#  ifndef S_ISDIR
#    ifdef _S_ISDIR
#      define S_ISDIR(x) _S_ISDIR(x)
#    else
#      ifdef S_IFDIR
#        ifndef S_IFMT
#          ifdef _S_IFMT
#            define S_IFMT _S_IFMT
#          endif
#        endif
#        ifdef S_IFMT
#          define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR)
#        endif
#      endif
#    endif
#  endif
#endif

#include <libxml/xmlmemory.h>
#include <libxml/parser.h>
#include <libxml/parserInternals.h>
#include <libxml/xmlIO.h>
#include <libxml/uri.h>
#include <libxml/nanohttp.h>
#include <libxml/nanoftp.h>
#include <libxml/xmlerror.h>
#ifdef LIBXML_CATALOG_ENABLED
#include <libxml/catalog.h>
#endif
#include <libxml/globals.h>

/* #define VERBOSE_FAILURE */
/* #define DEBUG_EXTERNAL_ENTITIES */
/* #define DEBUG_INPUT */

#ifdef DEBUG_INPUT
#define MINLEN 40
#else
#define MINLEN 4000
#endif

/*
 * Input I/O callback sets
 */
typedef struct _xmlInputCallback {
    xmlInputMatchCallback matchcallback;
    xmlInputOpenCallback opencallback;
    xmlInputReadCallback readcallback;
    xmlInputCloseCallback closecallback;
} xmlInputCallback;

#define MAX_INPUT_CALLBACK 15

static xmlInputCallback xmlInputCallbackTable[MAX_INPUT_CALLBACK];
static int xmlInputCallbackNr = 0;
static int xmlInputCallbackInitialized = 0;

#ifdef LIBXML_OUTPUT_ENABLED
/*
 * Output I/O callback sets
 */
typedef struct _xmlOutputCallback {
    xmlOutputMatchCallback matchcallback;
    xmlOutputOpenCallback opencallback;
    xmlOutputWriteCallback writecallback;
    xmlOutputCloseCallback closecallback;
} xmlOutputCallback;

#define MAX_OUTPUT_CALLBACK 15

static xmlOutputCallback xmlOutputCallbackTable[MAX_OUTPUT_CALLBACK];
static int xmlOutputCallbackNr = 0;
static int xmlOutputCallbackInitialized = 0;
#endif /* LIBXML_OUTPUT_ENABLED */

/************************************************************************
 *									*
 * 		Tree memory error handler				*
 *									*
 ************************************************************************/

static const char *IOerr[] = {
    "Unknown IO error",         /* UNKNOWN */
    "Permission denied",	/* EACCES */
    "Resource temporarily unavailable",/* EAGAIN */
    "Bad file descriptor",	/* EBADF */
    "Bad message",		/* EBADMSG */
    "Resource busy",		/* EBUSY */
    "Operation canceled",	/* ECANCELED */
    "No child processes",	/* ECHILD */
    "Resource deadlock avoided",/* EDEADLK */
    "Domain error",		/* EDOM */
    "File exists",		/* EEXIST */
    "Bad address",		/* EFAULT */
    "File too large",		/* EFBIG */
    "Operation in progress",	/* EINPROGRESS */
    "Interrupted function call",/* EINTR */
    "Invalid argument",		/* EINVAL */
    "Input/output error",	/* EIO */
    "Is a directory",		/* EISDIR */
    "Too many open files",	/* EMFILE */
    "Too many links",		/* EMLINK */
    "Inappropriate message buffer length",/* EMSGSIZE */
    "Filename too long",	/* ENAMETOOLONG */
    "Too many open files in system",/* ENFILE */
    "No such device",		/* ENODEV */
    "No such file or directory",/* ENOENT */
    "Exec format error",	/* ENOEXEC */
    "No locks available",	/* ENOLCK */
    "Not enough space",		/* ENOMEM */
    "No space left on device",	/* ENOSPC */
    "Function not implemented",	/* ENOSYS */
    "Not a directory",		/* ENOTDIR */
    "Directory not empty",	/* ENOTEMPTY */
    "Not supported",		/* ENOTSUP */
    "Inappropriate I/O control operation",/* ENOTTY */
    "No such device or address",/* ENXIO */
    "Operation not permitted",	/* EPERM */
    "Broken pipe",		/* EPIPE */
    "Result too large",		/* ERANGE */
    "Read-only file system",	/* EROFS */
    "Invalid seek",		/* ESPIPE */
    "No such process",		/* ESRCH */
    "Operation timed out",	/* ETIMEDOUT */
    "Improper link",		/* EXDEV */
    "Attempt to load network entity %s", /* XML_IO_NETWORK_ATTEMPT */
    "encoder error",		/* XML_IO_ENCODER */
    "flush error",
    "write error",
    "no input",
    "buffer full",
    "loading error",
    "not a socket",		/* ENOTSOCK */
    "already connected",	/* EISCONN */
    "connection refused",	/* ECONNREFUSED */
    "unreachable network",	/* ENETUNREACH */
    "adddress in use",		/* EADDRINUSE */
    "already in use",		/* EALREADY */
    "unknown address familly",	/* EAFNOSUPPORT */
};

/**
 * xmlIOErrMemory:
 * @extra:  extra informations
 *
 * Handle an out of memory condition
 */
static void
xmlIOErrMemory(const char *extra)
{
    __xmlSimpleError(XML_FROM_IO, XML_ERR_NO_MEMORY, NULL, NULL, extra);
}

/**
 * __xmlIOErr:
 * @code:  the error number
 * @
 * @extra:  extra informations
 *
 * Handle an I/O error
 */
void
__xmlIOErr(int domain, int code, const char *extra)
{
    unsigned int idx;

    if (code == 0) {
#ifdef HAVE_ERRNO_H
	if (errno == 0) code = 0;
#ifdef EACCES
        else if (errno == EACCES) code = XML_IO_EACCES;
#endif
#ifdef EAGAIN
        else if (errno == EAGAIN) code = XML_IO_EAGAIN;
#endif
#ifdef EBADF
        else if (errno == EBADF) code = XML_IO_EBADF;
#endif
#ifdef EBADMSG
        else if (errno == EBADMSG) code = XML_IO_EBADMSG;
#endif
#ifdef EBUSY
        else if (errno == EBUSY) code = XML_IO_EBUSY;
#endif
#ifdef ECANCELED
        else if (errno == ECANCELED) code = XML_IO_ECANCELED;
#endif
#ifdef ECHILD
        else if (errno == ECHILD) code = XML_IO_ECHILD;
#endif
#ifdef EDEADLK
        else if (errno == EDEADLK) code = XML_IO_EDEADLK;
#endif
#ifdef EDOM
        else if (errno == EDOM) code = XML_IO_EDOM;
#endif
#ifdef EEXIST
        else if (errno == EEXIST) code = XML_IO_EEXIST;
#endif
#ifdef EFAULT
        else if (errno == EFAULT) code = XML_IO_EFAULT;
#endif
#ifdef EFBIG
        else if (errno == EFBIG) code = XML_IO_EFBIG;
#endif
#ifdef EINPROGRESS
        else if (errno == EINPROGRESS) code = XML_IO_EINPROGRESS;
#endif
#ifdef EINTR
        else if (errno == EINTR) code = XML_IO_EINTR;
#endif
#ifdef EINVAL
        else if (errno == EINVAL) code = XML_IO_EINVAL;
#endif
#ifdef EIO
        else if (errno == EIO) code = XML_IO_EIO;
#endif
#ifdef EISDIR
        else if (errno == EISDIR) code = XML_IO_EISDIR;
#endif
#ifdef EMFILE
        else if (errno == EMFILE) code = XML_IO_EMFILE;
#endif
#ifdef EMLINK
        else if (errno == EMLINK) code = XML_IO_EMLINK;
#endif
#ifdef EMSGSIZE
        else if (errno == EMSGSIZE) code = XML_IO_EMSGSIZE;
#endif
#ifdef ENAMETOOLONG
        else if (errno == ENAMETOOLONG) code = XML_IO_ENAMETOOLONG;
#endif
#ifdef ENFILE
        else if (errno == ENFILE) code = XML_IO_ENFILE;
#endif
#ifdef ENODEV
        else if (errno == ENODEV) code = XML_IO_ENODEV;
#endif
#ifdef ENOENT
        else if (errno == ENOENT) code = XML_IO_ENOENT;
#endif
#ifdef ENOEXEC
        else if (errno == ENOEXEC) code = XML_IO_ENOEXEC;
#endif
#ifdef ENOLCK
        else if (errno == ENOLCK) code = XML_IO_ENOLCK;
#endif
#ifdef ENOMEM
        else if (errno == ENOMEM) code = XML_IO_ENOMEM;
#endif
#ifdef ENOSPC
        else if (errno == ENOSPC) code = XML_IO_ENOSPC;
#endif
#ifdef ENOSYS
        else if (errno == ENOSYS) code = XML_IO_ENOSYS;
#endif
#ifdef ENOTDIR
        else if (errno == ENOTDIR) code = XML_IO_ENOTDIR;
#endif
#ifdef ENOTEMPTY
        else if (errno == ENOTEMPTY) code = XML_IO_ENOTEMPTY;
#endif
#ifdef ENOTSUP
        else if (errno == ENOTSUP) code = XML_IO_ENOTSUP;
#endif
#ifdef ENOTTY
        else if (errno == ENOTTY) code = XML_IO_ENOTTY;
#endif
#ifdef ENXIO
        else if (errno == ENXIO) code = XML_IO_ENXIO;
#endif
#ifdef EPERM
        else if (errno == EPERM) code = XML_IO_EPERM;
#endif
#ifdef EPIPE
        else if (errno == EPIPE) code = XML_IO_EPIPE;
#endif
#ifdef ERANGE
        else if (errno == ERANGE) code = XML_IO_ERANGE;
#endif
#ifdef EROFS
        else if (errno == EROFS) code = XML_IO_EROFS;
#endif
#ifdef ESPIPE
        else if (errno == ESPIPE) code = XML_IO_ESPIPE;
#endif
#ifdef ESRCH
        else if (errno == ESRCH) code = XML_IO_ESRCH;
#endif
#ifdef ETIMEDOUT
        else if (errno == ETIMEDOUT) code = XML_IO_ETIMEDOUT;
#endif
#ifdef EXDEV
        else if (errno == EXDEV) code = XML_IO_EXDEV;
#endif
#ifdef ENOTSOCK
        else if (errno == ENOTSOCK) code = XML_IO_ENOTSOCK;
#endif
#ifdef EISCONN
        else if (errno == EISCONN) code = XML_IO_EISCONN;
#endif
#ifdef ECONNREFUSED
        else if (errno == ECONNREFUSED) code = XML_IO_ECONNREFUSED;
#endif
#ifdef ETIMEDOUT
        else if (errno == ETIMEDOUT) code = XML_IO_ETIMEDOUT;
#endif
#ifdef ENETUNREACH
        else if (errno == ENETUNREACH) code = XML_IO_ENETUNREACH;
#endif
#ifdef EADDRINUSE
        else if (errno == EADDRINUSE) code = XML_IO_EADDRINUSE;
#endif
#ifdef EINPROGRESS
        else if (errno == EINPROGRESS) code = XML_IO_EINPROGRESS;
#endif
#ifdef EALREADY
        else if (errno == EALREADY) code = XML_IO_EALREADY;
#endif
#ifdef EAFNOSUPPORT
        else if (errno == EAFNOSUPPORT) code = XML_IO_EAFNOSUPPORT;
#endif
        else code = XML_IO_UNKNOWN;
#endif /* HAVE_ERRNO_H */
    }
    idx = 0;
    if (code >= XML_IO_UNKNOWN) idx = code - XML_IO_UNKNOWN;
    if (idx >= (sizeof(IOerr) / sizeof(IOerr[0]))) idx = 0;
    
    __xmlSimpleError(domain, code, NULL, IOerr[idx], extra);
}

/**
 * xmlIOErr:
 * @code:  the error number
 * @extra:  extra informations
 *
 * Handle an I/O error
 */
static void
xmlIOErr(int code, const char *extra)
{
    __xmlIOErr(XML_FROM_IO, code, extra);
}

/**
 * __xmlLoaderErr:
 * @ctx: the parser context
 * @extra:  extra informations
 *
 * Handle a resource access error
 */
void
__xmlLoaderErr(void *ctx, const char *msg, const char *filename)
{
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
    xmlStructuredErrorFunc schannel = NULL;
    xmlGenericErrorFunc channel = NULL;
    void *data = NULL;
    xmlErrorLevel level = XML_ERR_ERROR;

    if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
        (ctxt->instate == XML_PARSER_EOF))
	return;
    if ((ctxt != NULL) && (ctxt->sax != NULL)) {
        if (ctxt->validate) {
	    channel = ctxt->sax->error;
	    level = XML_ERR_ERROR;
	} else {
	    channel = ctxt->sax->warning;
	    level = XML_ERR_WARNING;
	}
	if (ctxt->sax->initialized == XML_SAX2_MAGIC)
	    schannel = ctxt->sax->serror;
	data = ctxt->userData;
    }
    __xmlRaiseError(schannel, channel, data, ctxt, NULL, XML_FROM_IO,
                    XML_IO_LOAD_ERROR, level, NULL, 0,
		    filename, NULL, NULL, 0, 0,
		    msg, filename);
                    
}

/************************************************************************
 *									*
 * 		Tree memory error handler				*
 *									*
 ************************************************************************/
/**
 * xmlNormalizeWindowsPath:
 * @path: the input file path
 *
 * This function is obsolete. Please see xmlURIFromPath in uri.c for
 * a better solution.
 *
 * Returns a canonicalized version of the path
 */
xmlChar *
xmlNormalizeWindowsPath(const xmlChar *path)
{
    return xmlCanonicPath(path);
}

/**
 * xmlCleanupInputCallbacks:
 *
 * clears the entire input callback table. this includes the
 * compiled-in I/O. 
 */
void
xmlCleanupInputCallbacks(void)
{
    int i;

    if (!xmlInputCallbackInitialized)
        return;

    for (i = xmlInputCallbackNr - 1; i >= 0; i--) {
        xmlInputCallbackTable[i].matchcallback = NULL;
        xmlInputCallbackTable[i].opencallback = NULL;
        xmlInputCallbackTable[i].readcallback = NULL;
        xmlInputCallbackTable[i].closecallback = NULL;
    }

    xmlInputCallbackNr = 0;
    xmlInputCallbackInitialized = 0;
}

/**
 * xmlPopInputCallbacks:
 *
 * Clear the top input callback from the input stack. this includes the
 * compiled-in I/O. 
 *
 * Returns the number of input callback registered or -1 in case of error.
 */
int
xmlPopInputCallbacks(void)
{
    if (!xmlInputCallbackInitialized)
        return(-1);

    if (xmlInputCallbackNr <= 0)
        return(-1);
        
    xmlInputCallbackNr--;
    xmlInputCallbackTable[xmlInputCallbackNr].matchcallback = NULL;
    xmlInputCallbackTable[xmlInputCallbackNr].opencallback = NULL;
    xmlInputCallbackTable[xmlInputCallbackNr].readcallback = NULL;
    xmlInputCallbackTable[xmlInputCallbackNr].closecallback = NULL;

    return(xmlInputCallbackNr);
}

#ifdef LIBXML_OUTPUT_ENABLED
/**
 * xmlCleanupOutputCallbacks:
 *
 * clears the entire output callback table. this includes the
 * compiled-in I/O callbacks. 
 */
void
xmlCleanupOutputCallbacks(void)
{
    int i;

    if (!xmlOutputCallbackInitialized)
        return;

    for (i = xmlOutputCallbackNr - 1; i >= 0; i--) {
        xmlOutputCallbackTable[i].matchcallback = NULL;
        xmlOutputCallbackTable[i].opencallback = NULL;