simple question XML parser C

**RoxPro** · 08-09-2011

Originally Posted by Salem

This is your code, with a few additions
- a character handler
- showing how to pass user data around, to record bits of information and state as we go.

Code:

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <expat.h>

#if defined(__amigaos__) && defined(__USE_INLINE__)
#include <proto/expat.h>
#endif

#ifdef XML_LARGE_SIZE
#if defined(XML_USE_MSC_EXTENSIONS) && _MSC_VER < 1400
#define XML_FMT_INT_MOD "I64"
#else
#define XML_FMT_INT_MOD "ll"
#endif
#else
#define XML_FMT_INT_MOD "l"
#endif

#define BUFFSIZE        8192

char Buff[BUFFSIZE];

int Depth;

typedef enum {
  S_NONE,
  S_IFNAME,
  S_PROTO,
  S_IPADDR,
  S_NETMASK,
} state_et;

typedef struct {
  state_et  state;
  char  interface[100];
  char  ifname[100];
  char  proto[100];
  char  ipaddr[100];
  char  netmask[100];
} info_st;

static void XMLCALL
chardata(void *userData, const XML_Char *s, int len)
{
  info_st   *info = userData;
  switch ( info->state ) {
    case S_IFNAME:
      strncpy(info->ifname,s,len);
      info->ifname[len] = '\0';
      break;
    case S_PROTO:
      strncpy(info->proto,s,len);
      info->proto[len] = '\0';
      break;
    case S_IPADDR:
      strncpy(info->ipaddr,s,len);
      info->ipaddr[len] = '\0';
      break;
    case S_NETMASK:
      strncpy(info->netmask,s,len);
      info->netmask[len] = '\0';
      break;
  }
  info->state = S_NONE;
//  printf(">-%.*s-<",len,s);
}

static void XMLCALL
start(void *data, const char *el, const char **attr)
{
  info_st   *info = data;
  int i;

  {
    char temp[100];
    if ( sscanf( el, "config_interface_%s", temp) == 1 ) {
      strcpy(info->interface,temp);
    }
    if ( strcmp(el,"ifname")==0) info->state = S_IFNAME;
    if ( strcmp(el,"proto")==0) info->state = S_PROTO;
    if ( strcmp(el,"ipaddr")==0) info->state = S_IPADDR;
    if ( strcmp(el,"netmask")==0) info->state = S_NETMASK;
  }
  for (i = 0; i < Depth; i++)
    printf("  ", data);

  printf("%s", el);

  for (i = 0; attr[i]; i += 2) {
    printf(" %s='%s'", attr[i], attr[i + 1]);
  }

  printf("\n");
  Depth++;
}

static void XMLCALL
end(void *data, const char *el)
{
  info_st   *info = data;
  {
    char temp[100];
    if ( sscanf( el, "config_interface_%s", temp) == 1 ) {
      printf("DATA=%s %s %s %s %s\n",
             info->interface,
             info->ifname,
             info->proto,
             info->ipaddr,
             info->netmask);
    }
  }
  Depth--;
}

int
main(int argc, char *argv[])
{
  info_st   info = { 0 };
  XML_Parser p = XML_ParserCreate(NULL);
  if (! p) {
    fprintf(stderr, "Couldn't allocate memory for parser\n");
    exit(-1);
  }

  XML_SetElementHandler(p, start, end);
  XML_SetCharacterDataHandler(p,chardata);
  XML_SetUserData(p, &info);

  for (;;) {
    int done;
    int len;

    len = (int)fread(Buff, 1, BUFFSIZE, stdin);
    if (ferror(stdin)) {
      fprintf(stderr, "Read error\n");
      exit(-1);
    }
    done = feof(stdin);

    if (XML_Parse(p, Buff, len, done) == XML_STATUS_ERROR) {
      fprintf(stderr, "Parse error at line %" XML_FMT_INT_MOD "u:\n%s\n",
              XML_GetCurrentLineNumber(p),
              XML_ErrorString(XML_GetErrorCode(p)));
      exit(-1);
    }

    if (done)
      break;
  }
  XML_ParserFree(p);
  return 0;
}

The extra information printed looks like this
DATA=loopback lo static 127.0.0.1 255.0.0.0
DATA=lan eth0 static 192.168.1.1 255.255.255.0
DATA=wifi eth0 static 192.168.2.1 255.255.255.0

Could I ask another question please?Thanks!
If I just want to parse the contents of the element Network, what should I do? I tried configure out the XML_setelementhandler, but it dosen't work... So Could you please take a few minutes to answer the question please? Thanks!

**Salem** · 08-09-2011

> If I just want to parse the contents of the element Network, what should I do?
In the start() function, you add

Code:

if ( strcmp(el,"network") ) info->parsingNetwork = true;

In the end function, you add

Code:

if ( strcmp(el,"network") ) {
    info->parsingNetwork = true;
    info->done = true;
}

If parsingNetwork is true, then you do the other bits.

If done becomes true, then you basically stop doing anything at all in start()

Or perhaps there is some other API you can call from the end() handler to indicate that parsing has been completed.

**RoxPro** · 08-10-2011

Originally Posted by Salem

> If I just want to parse the contents of the element Network, what should I do?
In the start() function, you add

Code:

if ( strcmp(el,"network") ) info->parsingNetwork = true;

In the end function, you add

Code:

if ( strcmp(el,"network") ) {
    info->parsingNetwork = true;
    info->done = true;
}

If parsingNetwork is true, then you do the other bits.

If done becomes true, then you basically stop doing anything at all in start()

Or perhaps there is some other API you can call from the end() handler to indicate that parsing has been completed.

Thanks for your reply, if I still have questions, I'll ask it after. Thanks a lot!

**RoxPro** · 08-22-2011

Originally Posted by Salem

> If I just want to parse the contents of the element Network, what should I do?
In the start() function, you add

Code:

if ( strcmp(el,"network") ) info->parsingNetwork = true;

In the end function, you add

Code:

if ( strcmp(el,"network") ) {
    info->parsingNetwork = true;
    info->done = true;
}

If parsingNetwork is true, then you do the other bits.

If done becomes true, then you basically stop doing anything at all in start()

Or perhaps there is some other API you can call from the end() handler to indicate that parsing has been completed.

That's the new code I've wrote, but there is a problem of Info, I tried to configure it out, but It saids info is redefined... so I post the code, and I let you remind me~ Thanks!

Code:

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <expat.h>
#include <stdbool.h>

#define bool _Bool
#define false 0
#define true 1
#define __bool_true_false_are_defined 1

#if defined(__amigaos__) && defined(__USE_INLINE__)
#include <proto/expat.h>
#endif

#ifdef XML_LARGE_SIZE
#if defined(XML_USE_MSC_EXTENSIONS) && _MSC_VER < 1400
#define XML_FMT_INT_MOD "I64"
#else
#define XML_FMT_INT_MOD "ll"
#endif
#else
#define XML_FMT_INT_MOD "l"
#endif

#define BUFFSIZE        8192

char Buff[BUFFSIZE];

int Depth;

typedef enum {
  S_NONE,
  S_IFNAME,
  S_PROTO,
  S_IPADDR,
  S_NETMASK,
} state_et;

typedef struct {
  state_et  state;
  char  interface[100];
  char  ifname[100];
  char  proto[100];
  char  ipaddr[100];
  char  netmask[100];
  bool  parsingNetwork ;
  bool  done;
} info_st;

static void XMLCALL
chardata(void *userData, const XML_Char *s, int len)
{
  info_st   *info = userData;
  switch ( info->state ) {
    case S_IFNAME:
      strncpy(info->ifname,s,len);
      info->ifname[len] = '\0';
      break;
    case S_PROTO:
      strncpy(info->proto,s,len);
      info->proto[len] = '\0';
      break;
    case S_IPADDR:
      strncpy(info->ipaddr,s,len);
      info->ipaddr[len] = '\0';
      break;
    case S_NETMASK:
      strncpy(info->netmask,s,len);
      info->netmask[len] = '\0';
      break;
  }
  info->state = S_NONE;
//  printf(">-%.*s-<",len,s);
}

static void XMLCALL
start(void *data, const char *el, const char **attr)
{ 
  info_st *info = data;//加上这句
  {
    char temp[100];
    if ( sscanf( el, "config_interface_%s", temp) == 1 ) {
      strcpy(info->interface,temp);
    }
    if ( strcmp(el,"ifname")==0) info->state = S_IFNAME;
    if ( strcmp(el,"proto")==0) info->state = S_PROTO;
    if ( strcmp(el,"ipaddr")==0) info->state = S_IPADDR;
    if ( strcmp(el,"netmask")==0) info->state = S_NETMASK;
  }
  for (i = 0; i < Depth; i++)
    printf("  ", data);//这句话是不是漏了个通配符？？？
 
  if ( strcmp(el,"network") ) info->parsingNetwork = true;
  //info_st   *info = data;
  int i;

  printf("%s", el);

  for (i = 0; attr[i]; i += 2) {
    printf(" %s='%s'", attr[i], attr[i + 1]);
  }

  printf("\n");
  Depth++;
 
   
}

static void XMLCALL
end(void *data, const char *el)
{
 info_st info = data;
  if ( strcmp(el,"network") )
 {
    info->parsingNetwork = true;
    info->done = true;}

  //info_st   *info = data;
  {
    char temp[100];
    if ( sscanf( el, "config_interface_%s", temp) == 1 ) {
      printf("DATA=%s %s %s %s %s\n",
             info->interface,
             info->ifname,
             info->proto,
             info->ipaddr,
             info->netmask);
    }
  }
  Depth--;
  
  
}


int
main(int argc, char *argv[])
{
  info_st   *info = data;
  XML_Parser p = XML_ParserCreate(NULL);
  if (! p) {
    fprintf(stderr, "Couldn't allocate memory for parser\n");
    exit(-1);
  }

  XML_SetElementHandler(p, start, end);
  XML_SetCharacterDataHandler(p,chardata);
  XML_SetUserData(p, &info);

  for (;;) {
    int done;
    int len;

    len = (int)fread(Buff, 1, BUFFSIZE, stdin);
    if (ferror(stdin)) {
      fprintf(stderr, "Read error\n");
      exit(-1);
    }
    done = feof(stdin);

    if (XML_Parse(p, Buff, len, done) == XML_STATUS_ERROR) {
      fprintf(stderr, "Parse error at line %" XML_FMT_INT_MOD "u:\n%s\n",
              XML_GetCurrentLineNumber(p),
              XML_ErrorString(XML_GetErrorCode(p)));
      exit(-1);
    }

    if (done)
      break;
  }
  XML_ParserFree(p);
  return 0;
}

**Salem** · 08-22-2011

The diffs between your code, and the thing I posted 2 weeks ago.

Code:

$ diff -b xx1.txt xx2.txt 
4a5,10
> #include <stdbool.h>
> 
> #define bool _Bool
> #define false 0
> #define true 1
> #define __bool_true_false_are_defined 1
40a47,48
>   bool  parsingNetwork ;
>   bool  done;
72,74c80
<   info_st   *info = data;
<   int i;
< 
---
>   info_st *info = data;//加上这句
86c92,96
<     printf("  ", data);
---
>     printf("  ", data);//这句话是不是漏了个通配符？？？
>  
>   if ( strcmp(el,"network") ) info->parsingNetwork = true;
>   //info_st   *info = data;
>   int i;
95a106,107
>  
>    
101c113,119
<   info_st   *info = data;
---
>  info_st info = data;
>   if ( strcmp(el,"network") )
>  {
>     info->parsingNetwork = true;
>     info->done = true;}
> 
>   //info_st   *info = data;
113a132,133
>   
>   
115a136
> 
119c140
<   info_st   info = { 0 };
---
>   info_st   *info = data;

The bits in blue are what you need to change back to the original.

The 'info' in main has to be a real structure, not just a pointer.
EVERY callback function gets a pointer to this structure instance.

Thread: simple question XML parser C

Thread Tools

Search Thread

Display

Similar Threads

A simple C parser problem

Simple parser

Simple Parser Program

very simple html parser

question about parser??