Exemplo n.º 1
0
/* 
module entry point 
*/
EXTERNAL_FUNCTION int hts_plug(httrackp *opt, const char* argv) {
  const char *arg = strchr(argv, ',');
  if (arg != NULL)
    arg++;

  /* Check args */
  fprintf(stderr, "Plugged..\n");
  if (arg == NULL || *arg == '\0' || strlen(arg) >= HTS_URLMAXSIZE / 2) {
    fprintf(stderr, "** callback error: arguments expected or bad arguments\n");
    fprintf(stderr, "usage: httrack --wrapper modulename,base\n");
    fprintf(stderr, "example: httrack --wrapper callback,http://www.example.com/\n");
    return 0;  /* failed */
  } else {
    char *callbacks_userdef = strdup(arg);      /* userdef */

    /* Plug callback functions */
    CHAIN_FUNCTION(opt, check_html, process_file, callbacks_userdef);
    CHAIN_FUNCTION(opt, linkdetected, check_detectedlink, callbacks_userdef);
    CHAIN_FUNCTION(opt, end, check_detectedlink_end, callbacks_userdef);

    fprintf(stderr, "Using root '%s'\n", callbacks_userdef);
  }

  return 1;  /* success */
}
Exemplo n.º 2
0
EXTERNAL_FUNCTION int hts_plug(httrackp * opt, const char *argv) {
  /* Plug callback functions */
  CHAIN_FUNCTION(opt, detect, hts_detect_java, NULL);
  CHAIN_FUNCTION(opt, parse, hts_parse_java, NULL);

  return 1;                     /* success */
}
Exemplo n.º 3
0
/*
module entry point
the function name and prototype MUST match this prototype
*/
EXTERNAL_FUNCTION int hts_plug(httrackp *opt, const char* argv) {
  /* optional argument passed in the commandline we won't be using here */
  const char *arg = strchr(argv, ',');
  if (arg != NULL)
    arg++;

  /* plug callback functions */
  CHAIN_FUNCTION(opt, check_html, process_file, /*optional user-defined arg*/NULL);
  CHAIN_FUNCTION(opt, end, end_of_mirror, /*optional user-defined arg*/NULL);

  return 1;  /* success */
}
/* 
module entry point 
*/
EXTERNAL_FUNCTION int hts_plug(httrackp * opt, const char *argv) {
  const char *arg = strchr(argv, ',');

  if (arg != NULL)
    arg++;

  /* Check args */
  if (arg == NULL || *arg == '\0') {
    fprintf(stderr, "** callback error: arguments expected or bad arguments\n");
    fprintf(stderr,
            "usage: httrack --wrapper callback,stringtofind,stringtofind..\n");
    fprintf(stderr, "example: httrack --wrapper callback,apple,orange,lemon\n");
    return 0;
  } else {
    t_my_userdef *userdef = (t_my_userdef *) malloc(sizeof(t_my_userdef));      /* userdef */
    char *const stringfilter = userdef->stringfilter;
    char **const stringfilters = userdef->stringfilters;

    /* */
    char *a = stringfilter;
    int i = 0;

    fprintf(stderr, "** info: wrapper_init(%s) called!\n", arg);
    fprintf(stderr,
            "** callback example: crawling pages only if specific keywords are found\n");

    /* stringfilters = split(arg, ','); */
    strcpy(stringfilter, arg);
    while(a != NULL) {
      stringfilters[i] = a;
      a = strchr(a, ',');
      if (a != NULL) {
        *a = '\0';
        a++;
      }
      fprintf(stderr, "** callback info: will crawl pages with '%s' in them\n",
              stringfilters[i]);
      i++;
    }
    stringfilters[i++] = NULL;

    /* Plug callback functions */
    CHAIN_FUNCTION(opt, check_html, process, userdef);
    CHAIN_FUNCTION(opt, end, end, userdef);
  }

  return 1;                     /* success */
}
Exemplo n.º 5
0
/*
module entry point
the function name and prototype MUST match this prototype
*/
EXTERNAL_FUNCTION int hts_plug(httrackp *opt, const char* argv) {
  /* optional argument passed in the commandline we won't be using here */
  const char *arg = strchr(argv, ',');
  if (arg != NULL)
    arg++;

  /* plug callback functions */
  if (arg == NULL)
    arg = "log-wrapper-info";
  hts_log(opt, arg, "* plugging functions");
  CHAIN_FUNCTION(opt, check_html, process_file, (char*) arg);
  CHAIN_FUNCTION(opt, start, start_of_mirror, (char*) arg);
  CHAIN_FUNCTION(opt, end, end_of_mirror, (char*) arg);

  hts_log(opt, arg, "* module successfully plugged");
  return 1;  /* success */
}
/* 
module entry point 
*/
EXTERNAL_FUNCTION int hts_plug(httrackp * opt, const char *argv) {
  const char *arg = strchr(argv, ',');

  if (arg != NULL)
    arg++;
  CHAIN_FUNCTION(opt, savename, mysavename, NULL);
  return 1;                     /* success */
}
Exemplo n.º 7
0
/* 
module entry point 
*/
EXTERNAL_FUNCTION int hts_plug(httrackp *opt, const char* argv) {
  t_my_userdef *userdef;
  /* */
  const char *arg = strchr(argv, ',');
  if (arg != NULL)
    arg++;

  /* Create user-defined structure */
  userdef = (t_my_userdef*) malloc(sizeof(t_my_userdef));    /* userdef */
  userdef->currentURLBeingParsed[0] = '\0';

  /* Plug callback functions */
  CHAIN_FUNCTION(opt, check_html, process_file, userdef);
  CHAIN_FUNCTION(opt, end, end, userdef);
  CHAIN_FUNCTION(opt, linkdetected, check_detectedlink, userdef);
  CHAIN_FUNCTION(opt, loop, check_loop, userdef);

  return 1;  /* success */
}
Exemplo n.º 8
0
void start_dl(proj) {

	arg_init(&proj);

	CHAIN_FUNCTION(opt, init, htsshow_init, &proj);
	CHAIN_FUNCTION(opt, uninit, htsshow_uninit, &proj);
	CHAIN_FUNCTION(opt, start, htsshow_start, &proj);
	CHAIN_FUNCTION(opt, end, htsshow_end, &proj);
	CHAIN_FUNCTION(opt, chopt, htsshow_chopt, &proj);
	CHAIN_FUNCTION(opt, preprocess, htsshow_preprocesshtml, &proj);
	CHAIN_FUNCTION(opt, postprocess, htsshow_postprocesshtml, &proj);
	CHAIN_FUNCTION(opt, check_html, htsshow_checkhtml, &proj);
	CHAIN_FUNCTION(opt, query, htsshow_query, &proj);
	CHAIN_FUNCTION(opt, query2, htsshow_query2, &proj);
	CHAIN_FUNCTION(opt, query3, htsshow_query3, &proj);
	CHAIN_FUNCTION(opt, loop, htsshow_loop, &proj);
	CHAIN_FUNCTION(opt, check_link, htsshow_check, &proj);
	CHAIN_FUNCTION(opt, check_mime, htsshow_check_mime, &proj);
	CHAIN_FUNCTION(opt, pause, htsshow_pause, &proj);
	CHAIN_FUNCTION(opt, filesave, htsshow_filesave, &proj);
	CHAIN_FUNCTION(opt, filesave2, htsshow_filesave2, &proj);
	CHAIN_FUNCTION(opt, linkdetected, htsshow_linkdetected, &proj);
	CHAIN_FUNCTION(opt, linkdetected2, htsshow_linkdetected2, &proj);
	CHAIN_FUNCTION(opt, xfrstatus, htsshow_xfrstatus, &proj);
	CHAIN_FUNCTION(opt, savename, htsshow_savename, &proj);
	CHAIN_FUNCTION(opt, sendhead, htsshow_sendheader, &proj);
	CHAIN_FUNCTION(opt, receivehead, htsshow_receiveheader, &proj);

	if (hts_main2(proj.argc, proj.argv, opt)) {
		fprintf(stderr, "hts_main2 with error %s\n", hts_errmsg(opt));
	}
}
Exemplo n.º 9
0
int main(int argc, char **argv) {
	int ret = 0;
	httrackp *opt;
	signal_handlers();
	hts_init();
	opt = global_opt = hts_create_opt();

	CHAIN_FUNCTION(opt, init, htsshow_init, NULL);
	CHAIN_FUNCTION(opt, uninit, htsshow_uninit, NULL);
	CHAIN_FUNCTION(opt, start, htsshow_start, NULL);
	CHAIN_FUNCTION(opt, end, htsshow_end, NULL);
	CHAIN_FUNCTION(opt, chopt, htsshow_chopt, NULL);
	CHAIN_FUNCTION(opt, preprocess, htsshow_preprocesshtml, NULL);
	CHAIN_FUNCTION(opt, postprocess, htsshow_postprocesshtml, NULL);
	CHAIN_FUNCTION(opt, check_html, htsshow_checkhtml, NULL);
	CHAIN_FUNCTION(opt, query, htsshow_query, NULL);
	CHAIN_FUNCTION(opt, query2, htsshow_query2, NULL);
	CHAIN_FUNCTION(opt, query3, htsshow_query3, NULL);
	CHAIN_FUNCTION(opt, loop, htsshow_loop, NULL);
	CHAIN_FUNCTION(opt, check_link, htsshow_check, NULL);
	CHAIN_FUNCTION(opt, check_mime, htsshow_check_mime, NULL);
	CHAIN_FUNCTION(opt, pause, htsshow_pause, NULL);
	CHAIN_FUNCTION(opt, filesave, htsshow_filesave, NULL);
	CHAIN_FUNCTION(opt, filesave2, htsshow_filesave2, NULL);
	CHAIN_FUNCTION(opt, linkdetected, htsshow_linkdetected, NULL);
	CHAIN_FUNCTION(opt, linkdetected2, htsshow_linkdetected2, NULL);
	CHAIN_FUNCTION(opt, xfrstatus, htsshow_xfrstatus, NULL);
	CHAIN_FUNCTION(opt, savename, htsshow_savename, NULL);
	CHAIN_FUNCTION(opt, sendhead, htsshow_sendheader, NULL);
	CHAIN_FUNCTION(opt, receivehead, htsshow_receiveheader, NULL);

	ret = hts_main2(argc, argv, opt);
	if (ret) {
		fprintf(stderr, "* %s\n", hts_errmsg(opt));
	}
	global_opt = NULL;
	hts_free_opt(opt);
	htsthread_wait(); /* wait for pending threads */
	hts_uninit();
	return ret;
}
Exemplo n.º 10
0
int main(int argc, char **argv) {
  int ret = 0;
  httrackp *opt;

#ifdef _WIN32
  {
    WORD wVersionRequested;     // requested version WinSock API
    WSADATA wsadata;            // Windows Sockets API data
    int stat;

    wVersionRequested = 0x0101;
    stat = WSAStartup(wVersionRequested, &wsadata);
    if (stat != 0) {
      printf("Winsock not found!\n");
      return;
    } else if (LOBYTE(wsadata.wVersion) != 1 && HIBYTE(wsadata.wVersion) != 1) {
      printf("WINSOCK.DLL does not support version 1.1\n");
      WSACleanup();
      return;
    }
  }
#endif

  signal_handlers();
  hts_init();

  // Check version compatibility
  if (hts_sizeof_opt() != sizeof(httrackp)) {
    fprintf(stderr,
      "incompatible current httrack library version %s, expected version %s",
      hts_version(), HTTRACK_VERSIONID);
    abortLog("incompatible httrack library version, please update both httrack and its library");
  }

  opt = global_opt = hts_create_opt();
  assert(opt->size_httrackp == sizeof(httrackp));

  CHAIN_FUNCTION(opt, init, htsshow_init, NULL);
  CHAIN_FUNCTION(opt, uninit, htsshow_uninit, NULL);
  CHAIN_FUNCTION(opt, start, htsshow_start, NULL);
  CHAIN_FUNCTION(opt, end, htsshow_end, NULL);
  CHAIN_FUNCTION(opt, chopt, htsshow_chopt, NULL);
  CHAIN_FUNCTION(opt, preprocess, htsshow_preprocesshtml, NULL);
  CHAIN_FUNCTION(opt, postprocess, htsshow_postprocesshtml, NULL);
  CHAIN_FUNCTION(opt, check_html, htsshow_checkhtml, NULL);
  CHAIN_FUNCTION(opt, query, htsshow_query, NULL);
  CHAIN_FUNCTION(opt, query2, htsshow_query2, NULL);
  CHAIN_FUNCTION(opt, query3, htsshow_query3, NULL);
  CHAIN_FUNCTION(opt, loop, htsshow_loop, NULL);
  CHAIN_FUNCTION(opt, check_link, htsshow_check, NULL);
  CHAIN_FUNCTION(opt, check_mime, htsshow_check_mime, NULL);
  CHAIN_FUNCTION(opt, pause, htsshow_pause, NULL);
  CHAIN_FUNCTION(opt, filesave, htsshow_filesave, NULL);
  CHAIN_FUNCTION(opt, filesave2, htsshow_filesave2, NULL);
  CHAIN_FUNCTION(opt, linkdetected, htsshow_linkdetected, NULL);
  CHAIN_FUNCTION(opt, linkdetected2, htsshow_linkdetected2, NULL);
  CHAIN_FUNCTION(opt, xfrstatus, htsshow_xfrstatus, NULL);
  CHAIN_FUNCTION(opt, savename, htsshow_savename, NULL);
  CHAIN_FUNCTION(opt, sendhead, htsshow_sendheader, NULL);
  CHAIN_FUNCTION(opt, receivehead, htsshow_receiveheader, NULL);

  ret = hts_main2(argc, argv, opt);
  if (ret) {
    fprintf(stderr, "* %s\n", hts_errmsg(opt));
  }
  global_opt = NULL;
  hts_free_opt(opt);
  htsthread_wait();             /* wait for pending threads */
  hts_uninit();

#ifdef _WIN32
  WSACleanup();
#endif

  return ret;
}
Exemplo n.º 11
0
static int webhttrack_runmain(httrackp *opt, int argc, char** argv) {
	int ret;

  CHAIN_FUNCTION(opt, init, htsshow_init, NULL);
  CHAIN_FUNCTION(opt, uninit, htsshow_uninit, NULL);
  CHAIN_FUNCTION(opt, start, htsshow_start, NULL);
  CHAIN_FUNCTION(opt, end, htsshow_end, NULL);
  CHAIN_FUNCTION(opt, chopt, htsshow_chopt, NULL);
  CHAIN_FUNCTION(opt, preprocess, htsshow_preprocesshtml, NULL);
  CHAIN_FUNCTION(opt, postprocess, htsshow_postprocesshtml, NULL);
  CHAIN_FUNCTION(opt, check_html, htsshow_checkhtml, NULL);
  CHAIN_FUNCTION(opt, query, htsshow_query, NULL);
  CHAIN_FUNCTION(opt, query2, htsshow_query2, NULL);
  CHAIN_FUNCTION(opt, query3, htsshow_query3, NULL);
  CHAIN_FUNCTION(opt, loop, htsshow_loop, NULL);
  CHAIN_FUNCTION(opt, check_link, htsshow_check, NULL);
  CHAIN_FUNCTION(opt, check_mime, htsshow_check_mime, NULL);
  CHAIN_FUNCTION(opt, pause, htsshow_pause, NULL);
  CHAIN_FUNCTION(opt, filesave, htsshow_filesave, NULL);
  CHAIN_FUNCTION(opt, filesave2, htsshow_filesave2, NULL);
  CHAIN_FUNCTION(opt, linkdetected, htsshow_linkdetected, NULL);
  CHAIN_FUNCTION(opt, linkdetected2, htsshow_linkdetected2, NULL);
  CHAIN_FUNCTION(opt, xfrstatus, htsshow_xfrstatus, NULL);
  CHAIN_FUNCTION(opt, savename, htsshow_savename, NULL);
  CHAIN_FUNCTION(opt, sendhead, htsshow_sendheader, NULL);
  CHAIN_FUNCTION(opt, receivehead, htsshow_receiveheader, NULL);

	ret = hts_main2(argc, argv, opt);
  htsthread_wait_n(1);
	
	return ret;
}
Exemplo n.º 12
0
/*
 * Name:           main
 * Description:    main() function
 * Parameters:     None
 * Should return:  error status
*/
int main(void) {
  /* 
     First, ask for an URL 
     Note: For the test, option r2 (mirror max depth=1) and --testscan (no index, no cache, do not store, no log files)
  */
  char _argv[][256] = {"httrack_test" , "<URL>" , "-r3" , "--testscan" , ""  };
  char* argv[]      = {NULL           , NULL    , NULL  , NULL        , NULL};
  int         argc = 0;
  httrackp *opt;
	int ret;
  while(strlen(_argv[argc])) {
    argv[argc]=_argv[argc];
    argc++;
  }
  argv[argc]=NULL;
  printf("HTTrackLib test program\n");
  printf("Enter URL (example: www.foobar.com/index.html) :");
  scanf("%s",argv[1]);
  printf("Test: 1 depth\n");

	/* Initialize the library */
#ifdef _WIN32
  {
    WORD   wVersionRequested;   // requested version WinSock API
    WSADATA wsadata;            // Windows Sockets API data
    int stat;
    wVersionRequested = 0x0101;
    stat = WSAStartup( wVersionRequested, &wsadata );
    if (stat != 0) {
      printf("Winsock not found!\n");
      return;
    } else if (LOBYTE(wsadata.wVersion) != 1  && HIBYTE(wsadata.wVersion) != 1) {
      printf("WINSOCK.DLL does not support version 1.1\n");
      WSACleanup();
      return;
    }
  }
#endif
  hts_init();

	/* Create option settings and set callbacks (wrappers) */
	opt = hts_create_opt();

  CHAIN_FUNCTION(opt, init, httrack_wrapper_init, NULL);
  CHAIN_FUNCTION(opt, uninit, httrack_wrapper_uninit, NULL);
  CHAIN_FUNCTION(opt, start, httrack_wrapper_start, NULL);
  CHAIN_FUNCTION(opt, end, httrack_wrapper_end, NULL);
  CHAIN_FUNCTION(opt, chopt, httrack_wrapper_chopt, NULL);
  CHAIN_FUNCTION(opt, preprocess, httrack_wrapper_preprocesshtml, NULL);
  CHAIN_FUNCTION(opt, postprocess, httrack_wrapper_postprocesshtml, NULL);
  CHAIN_FUNCTION(opt, check_html, httrack_wrapper_checkhtml, NULL);
  CHAIN_FUNCTION(opt, query, httrack_wrapper_query, NULL);
  CHAIN_FUNCTION(opt, query2, httrack_wrapper_query2, NULL);
  CHAIN_FUNCTION(opt, query3, httrack_wrapper_query3, NULL);
  CHAIN_FUNCTION(opt, loop, httrack_wrapper_loop, NULL);
  CHAIN_FUNCTION(opt, check_link, httrack_wrapper_check, NULL);
  CHAIN_FUNCTION(opt, check_mime, httrack_wrapper_check_mime, NULL);
  CHAIN_FUNCTION(opt, pause, httrack_wrapper_pause, NULL);
  CHAIN_FUNCTION(opt, filesave, httrack_wrapper_filesave, NULL);
  CHAIN_FUNCTION(opt, filesave2, httrack_wrapper_filesave2, NULL);
  CHAIN_FUNCTION(opt, linkdetected, httrack_wrapper_linkdetected, NULL);
  CHAIN_FUNCTION(opt, linkdetected2, httrack_wrapper_linkdetected2, NULL);
  CHAIN_FUNCTION(opt, xfrstatus, httrack_wrapper_xfrstatus, NULL);
  CHAIN_FUNCTION(opt, savename, httrack_wrapper_savename, NULL);
  CHAIN_FUNCTION(opt, sendhead, httrack_wrapper_sendheader, NULL);
  CHAIN_FUNCTION(opt, receivehead, httrack_wrapper_receiveheader, NULL);

  /* Then, launch the mirror */
	ret = hts_main2(argc, argv, opt);

  /* Wait for a key */
  printf("\nPress ENTER key to exit\n");
  scanf("%s",argv[1]);

	/* Clear option state */
	hts_free_opt(opt);
	hts_uninit();
#ifdef _WIN32
  WSACleanup();
#endif

  /* That's all! */
  return 0;
}