uconfig.c

来自「网络爬虫程序」· C语言 代码 · 共 1,533 行 · 第 1/3 页

int xget_cfg_values_lim(void)
{
  struct tm ftm;
  char *p = NULL;
  bool_t b;
  struct stat estat;

  LOCK_GCFG;
  GET_HTML_TAGS;

  GET_PATH(gui_cfg.subdir_label, p);
  if(p && !((stat(p, &estat) >= 0) && S_ISDIR(estat.st_mode)))
  {
    _free(p);
    xprintf(0, gettext("Bad parameter - Working subdirectory"));
    _MT_CFGSTAMP;
    UNLOCK_GCFG;
    return -1;
  }
  else
  {
    _free(cfg.subdir);
    if(p)
      cfg.subdir = get_abs_file_path_oss(p);
    _free(p);
  }

  GET_BOOL(gui_cfg.ftp_html, cfg.ftp_html);

  GET_BOOL(gui_cfg.leaves_sw, cfg.condition.dont_leave_site);

  GET_BOOL(gui_cfg.leaved_sw, cfg.condition.dont_leave_dir);

  GET_BOOL(gui_cfg.dont_leave_site_dir, cfg.condition.dont_leave_site_dir);

  GET_BOOL(gui_cfg.singlepage, cfg.singlepage);

  GET_BOOL(gui_cfg.limit_inlines, cfg.condition.limit_inlines);

  GET_BOOL(gui_cfg.cgi_sw, cfg.condition.cgi);

  GET_BOOL(gui_cfg.ftpd_sw, cfg.condition.ftpdir);

  GET_BOOL(gui_cfg.allow_ports, cfg.condition.allow_ports);

  GET_BOOL(gui_cfg.hosts_sw, cfg.condition.allow_site);

  GET_BOOL(gui_cfg.domain_sw, cfg.condition.allow_domain);

  GET_BOOL(gui_cfg.mime_sw, cfg.condition.allow_mime);

  GET_BOOL(gui_cfg.prefix_sw, cfg.condition.allow_prefix);

  GET_BOOL(gui_cfg.sufix_sw, cfg.condition.allow_sufix);

  GET_BOOL(gui_cfg.robots_sw, cfg.condition.allow_robots);

  GET_NUMLIST(gui_cfg.ports, cfg.condition.ports);

  GET_PAT(gui_cfg.pattern_label, cfg.condition.pattern);
  GET_PAT(gui_cfg.skip_pattern, cfg.condition.skip_pattern);
  GET_PAT(gui_cfg.url_pattern_label, cfg.condition.url_pattern);
  GET_PAT(gui_cfg.skip_url_pattern, cfg.condition.skip_url_pattern);

#ifdef HAVE_REGEX
  GET_REPAT(gui_cfg.rpattern, cfg.condition.rpattern);
  GET_REPAT(gui_cfg.skip_rpattern, cfg.condition.rskip_pattern);
  GET_REPAT(gui_cfg.url_rpattern, cfg.condition.rurl_pattern);
  GET_REPAT(gui_cfg.url_skip_rpattern, cfg.condition.rskip_url_pattern);
  GET_REPAT(gui_cfg.aip, cfg.condition.aip);
  GET_REPAT(gui_cfg.skipip, cfg.condition.skipip);

#endif

  GET_STR(gui_cfg.en_uexit, cfg.condition.uexit);
  GET_STR(gui_cfg.follow_cmd, cfg.condition.follow_cmd);

  GET_LIST(gui_cfg.hosts_list, cfg.condition.sites);

  GET_LIST(gui_cfg.amimelist, cfg.condition.mime);

  GET_LIST(gui_cfg.domain_list, cfg.condition.domains);

  GET_NUM(gui_cfg.maxlev_label, cfg.condition.max_levels);

  GET_NUM(gui_cfg.leave_level, cfg.condition.leave_level);

  GET_NUM(gui_cfg.site_level, cfg.condition.site_level);

  GET_NUM(gui_cfg.maxdoc_label, cfg.condition.max_documents);

  GET_NUM(gui_cfg.maxsize_label, cfg.condition.max_size);
  GET_NUM(gui_cfg.min_size, cfg.condition.min_size);

  GET_LIST(gui_cfg.sufixlist, cfg.condition.sufix);

  GET_LIST(gui_cfg.prefixlist, cfg.condition.dir_prefix);

  GET_DOUBLE(gui_cfg.max_time, cfg.max_time);

  GET_BOOL(gui_cfg.btime_sw, b);
  if(b)
  {
    GET_DATE(gui_cfg.btime_cal, ftm);
    ftm.tm_sec = 0;
    ftm.tm_isdst = -1;
    GET_NUM(gui_cfg.btime_h_entry, ftm.tm_hour);
    GET_NUM(gui_cfg.btime_min_entry, ftm.tm_min);
    cfg.condition.btime = mktime(&ftm);
  }

  else
    cfg.condition.btime = (time_t) 0;

  GET_BOOL(gui_cfg.etime_sw, b);
  if(b)
  {
    GET_DATE(gui_cfg.etime_cal, ftm);
    ftm.tm_sec = 0;
    ftm.tm_isdst = -1;
    GET_NUM(gui_cfg.etime_h_entry, ftm.tm_hour);
    GET_NUM(gui_cfg.etime_min_entry, ftm.tm_min);
    cfg.condition.etime = mktime(&ftm);
  }
  else
    cfg.condition.etime = (time_t) 0;

  _MT_CFGSTAMP;
  UNLOCK_GCFG;

  return 0;
}

int xget_cfg_values_comm(void)
{
  char *tp, *p = NULL;
  struct stat estat;
  dllist *ptr;

  LOCK_GCFG;
  GET_PATH(gui_cfg.cdir_label, p);
  if(p && !((stat(p, &estat) >= 0) && S_ISDIR(estat.st_mode)))
  {
    xprintf(0, gettext("Bad parameter - Cache directory"));
    _MT_CFGSTAMP;
    UNLOCK_GCFG;
    return -1;
  }
  else
  {
    _free(cfg.cache_dir);
    if(p)
      cfg.cache_dir = get_abs_file_path_oss(p);
    _free(p);
  }

  GET_PATH(gui_cfg.info_dir, cfg.info_dir);

#ifdef HAVE_BDB_18x
  GET_PATH(gui_cfg.ns_cache_dir, cfg.ns_cache_dir);
  GET_PATH(gui_cfg.moz_cache_dir, cfg.moz_cache_dir);
#endif

  p = cfg.cookie_file;
  GET_S_PATH(gui_cfg.en_cookie_file, cfg.cookie_file);
  if(!((cfg.cookie_file && p && !strcmp(p, cfg.cookie_file))
      || !cfg.cookie_file))
  {
    cookie_read_file(cfg.cookie_file);
  }
  _free(p);

  GET_STR(gui_cfg.default_prefix, cfg.default_prefix);

  GET_BOOL(gui_cfg.check_size, cfg.check_size);
  GET_BOOL(gui_cfg.store_index, cfg.store_index);

  GET_BOOL(gui_cfg.unique_doc, cfg.unique_doc);
  GET_BOOL(gui_cfg.del_after, cfg.del_after);

  GET_BOOL(gui_cfg.preserve_perm, cfg.preserve_perm);
  GET_BOOL(gui_cfg.preserve_links, cfg.preserve_links);
  GET_BOOL(gui_cfg.retrieve_slink, cfg.retrieve_slink);
  GET_BOOL(gui_cfg.ftp_list, cfg.ftp_list);
  GET_BOOL(gui_cfg.fix_wuftpd, cfg.fix_wuftpd);
  GET_BOOL(gui_cfg.use_http11, cfg.use_http11);

  GETFTPHS_LIST(gui_cfg.ftp_login_hs, cfg.ftp_login_hs);

  GET_STR(gui_cfg.index_name, cfg.index_name);
  GET_STR(gui_cfg.store_name, cfg.store_name);

  GET_BOOL(gui_cfg.sw_cookie_send, cfg.send_cookies);
  GET_BOOL(gui_cfg.sw_cookie_recv, cfg.recv_cookies);
  GET_BOOL(gui_cfg.sw_cookie_update, cfg.update_cookies);
  GET_BOOL(gui_cfg.cookie_check_domain, cfg.cookie_check_domain);

  GET_BOOL(gui_cfg.ptime, cfg.preserve_time);

  GET_BOOL(gui_cfg.cache_sw, cfg.cache);

  GET_BOOL(gui_cfg.oldrm_sw, cfg.remove_old);

  GET_BOOL(gui_cfg.ftp_sw, cfg.condition.ftp);


  GET_BOOL(gui_cfg.http_sw, cfg.condition.http);

#ifdef USE_SSL
  GET_BOOL(gui_cfg.https_sw, cfg.condition.https);
  GET_BOOL(gui_cfg.ftps_sw, cfg.condition.ftps);
#endif
  GET_BOOL(gui_cfg.gopher_sw, cfg.condition.gopher);

  GET_BOOL(gui_cfg.gopher_httpgw, cfg.gopher_via_http);

  GET_BOOL(gui_cfg.ftp_httpgw, cfg.ftp_via_http);

  GET_BOOL(gui_cfg.ftp_dirtyp, cfg.ftp_dirtyp);

  GET_BOOL(gui_cfg.freget_sw, cfg.freget);

  GET_BOOL(gui_cfg.noreloc_sw, cfg.rewrite_links);

  GET_BOOL(gui_cfg.all_to_local, cfg.all_to_local);
  GET_BOOL(gui_cfg.all_to_remote, cfg.all_to_remote);
  GET_BOOL(gui_cfg.sel_to_local, cfg.sel_to_local);
  GET_BOOL(gui_cfg.post_update, cfg.post_update);

  GET_BOOL(gui_cfg.enable_js, cfg.enable_js);

  GET_BOOL(gui_cfg.auto_referer, cfg.auto_referer);
  GET_BOOL(gui_cfg.referer, cfg.referer);

  GET_BOOL(gui_cfg.enable_info, cfg.enable_info);

  GET_BOOL(gui_cfg.enc_sw, cfg.use_enc);

  GET_BOOL(gui_cfg.auth_reuse_nonce, cfg.auth_reuse_nonce);
  GET_BOOL(gui_cfg.auth_reuse_proxy_nonce, cfg.auth_reuse_proxy_nonce);

  GET_BOOL(gui_cfg.send_if_range, cfg.send_if_range);

  GET_BOOL(gui_cfg.show_time, cfg.show_time);

  GET_ENUM(gui_cfg.http_auth_scheme, cfg.auth_scheme);
  GET_ENUM(gui_cfg.http_proxy_auth_scheme, cfg.proxy_auth_scheme);

#ifdef ENABLE_NTLM
  GET_STR(gui_cfg.auth_ntlm_domain, cfg.auth_ntlm_domain);
  GET_STR(gui_cfg.auth_proxy_ntlm_domain, cfg.auth_proxy_ntlm_domain);
#endif

  GET_C_EXTRAS;
  GET_SSL_VER;

#ifdef WITH_TREE
  GET_STR(gui_cfg.browser_label, cfg.browser);
#endif

  GET_STR(gui_cfg.remind_cmd, cfg.remind_cmd);
  GET_STR(gui_cfg.post_cmd, cfg.post_cmd);
  GET_STR(gui_cfg.identity, cfg.identity);

  GET_STR(gui_cfg.ftp_proxyh_label, cfg.ftp_proxy);
  GET_NUM(gui_cfg.ftp_proxyp_label, cfg.ftp_proxy_port);

  GET_STR(gui_cfg.gopher_proxyh_label, cfg.gopher_proxy);
  GET_NUM(gui_cfg.gopher_proxyp_label, cfg.gopher_proxy_port);

  GET_BOOL(gui_cfg.read_css, cfg.read_css);
#ifdef USE_SSL
  GET_STR(gui_cfg.ssl_proxyh_label, cfg.ssl_proxy);
  GET_NUM(gui_cfg.ssl_proxyp_label, cfg.ssl_proxy_port);
  GET_STR(gui_cfg.ssl_cert_passwd_en, cfg.ssl_cert_passwd);
  GET_STR(gui_cfg.ssl_cipher_list, cfg.ssl_cipher_list);
  GET_BOOL(gui_cfg.unique_sslid, cfg.unique_sslid);

#ifdef USE_SSL_IMPL_OPENSSL
  GET_PATH(gui_cfg.ssl_cert_file_en, cfg.ssl_cert_file);
  GET_PATH(gui_cfg.ssl_key_file_en, cfg.ssl_key_file);

#ifdef HAVE_RAND_EGD
  GET_PATH(gui_cfg.egd_socket, cfg.egd_socket);
#endif
#endif
#ifdef USE_SSL_IMPL_NSS
  GET_PATH(gui_cfg.nss_cert_dir, cfg.nss_cert_dir);
  GET_BOOL(gui_cfg.nss_accept_unknown_cert, cfg.nss_accept_unknown_cert);
  GET_BOOL(gui_cfg.nss_domestic_policy, cfg.nss_domestic_policy);
#endif
#endif

  GET_STR(gui_cfg.local_ip, cfg.local_ip);
  memset(&cfg.local_ip_addr, '\0', sizeof(cfg.local_ip_addr));
  if(cfg.local_ip && net_host_to_in_addr(cfg.local_ip, &cfg.local_ip_addr))
  {
    xherror(cfg.local_ip);
  }

  GET_S_STR(gui_cfg.http_headers, tp);
  p = tp;
  while(cfg.http_headers)
  {
    httphdr_free((httphdr *) cfg.http_headers->data);
    cfg.http_headers =
      dllist_remove_entry(cfg.http_headers, cfg.http_headers);
  }
  if(p && *p)
  {
    while((p = get_1qstr(p)))
    {
      httphdr *hdr = httphdr_parse(p);
      if(hdr)
        cfg.http_headers = dllist_append(cfg.http_headers, (dllist_t) hdr);
      else
        xprintf(0, "Bad HTTP headers entry - %s\n", p);
      p = NULL;
    }
  }
  _free(tp);

  GET_STR(gui_cfg.ftp_list_options, cfg.ftp_list_options);

  GET_NUM(gui_cfg.en_cookie_max, cfg.cookies_max);

  GET_NUM(gui_cfg.base_level_label, cfg.base_level);

  GET_DOUBLE(gui_cfg.timeout_label, cfg.ctimeout);

  GET_NUM(gui_cfg.retry_label, cfg.nretry);

  GET_NUM(gui_cfg.redir_label, cfg.nredir);

  GET_NUM(gui_cfg.reget_label, cfg.nreget);

  GET_NUM(gui_cfg.sleep_label, cfg.sleep);
  GET_BOOL(gui_cfg.rsleep, cfg.rsleep);

  GET_NUM(gui_cfg.xloglen_label, cfg.xlogsize);

  GET_NUM(gui_cfg.bufsize, cfg.bufsize);

#ifdef HAVE_MT
  GET_NUM(gui_cfg.nthr, cfg.nthr);
#endif

#ifdef HAVE_FSTATFS
  GET_NUM(gui_cfg.fs_quota, cfg.fs_quota);
#endif

  GET_NUM(gui_cfg.file_quota, cfg.file_quota);

  GET_NUM(gui_cfg.trans_quota, cfg.trans_quota);

  GET_DOUBLE(gui_cfg.maxrate, cfg.maxrate);

  GET_DOUBLE(gui_cfg.minrate, cfg.minrate);

  GET_PATH(gui_cfg.slog_label, cfg.short_logfile);

  GET_BOOL(gui_cfg.gen_logname, cfg.gen_logname);

  GET_PATH(gui_cfg.log_label, cfg.logfile);
  log_start(cfg.logfile);

  GET_STR(gui_cfg.auth_label, cfg.name_auth);

  GET_STR(gui_cfg.pass_label, cfg.passwd_auth);

  GET_STR(gui_cfg.proxy_auth_label, cfg.http_proxy_user);
  GET_STR(gui_cfg.proxy_pass_label, cfg.http_proxy_pass);

  GET_STR(gui_cfg.ftp_proxy_user, cfg.ftp_proxy_user);
  GET_STR(gui_cfg.ftp_proxy_pass, cfg.ftp_proxy_pass);

  GET_BOOL(gui_cfg.send_from, cfg.send_from);

  GET_STR(gui_cfg.from_label, cfg.from);

  GET_NUM(gui_cfg.rollback_label, cfg.rollback);

  GET_NUM(gui_cfg.ddays_label, cfg.ddays);

  GET_NUM(gui_cfg.hash_size, cfg.hash_size);
  dlhash_resize(cfg.url_hash_tbl, cfg.hash_size);
  dlhash_resize(cfg.fn_hash_tbl, cfg.hash_size);

  GET_LIST(gui_cfg.alanglist, cfg.accept_lang);
  GET_LIST(gui_cfg.acharset_list, cfg.accept_chars);

  GET_LIST(gui_cfg.cookie_domain_list, cfg.cookies_disabled_domains);

  GET_STR(gui_cfg.tr_del_chr, cfg.tr_del_chr);
  GET_STR(gui_cfg.tr_str_s1, cfg.tr_str_s1);
  GET_STR(gui_cfg.tr_str_s2, cfg.tr_str_s2);
  GET_STR(gui_cfg.tr_chr_s1, cfg.tr_chr_s1);
  GET_STR(gui_cfg.tr_chr_s2, cfg.tr_chr_s2);

  GET_COMBO_IDX(gui_cfg.scheduling_strategie, cfg.scheduling_strategie);

  GET_PAT(gui_cfg.dont_touch_url_pattern, cfg.dont_touch_url_pattern);

#ifdef HAVE_REGEX
  GET_REPAT(gui_cfg.dont_touch_url_rpattern, cfg.dont_touch_url_rpattern);
  GET_REPAT(gui_cfg.dont_touch_tag_rpattern, cfg.dont_touch_tag_rpattern);
  GET_BOOL(gui_cfg.remove_adv, cfg.remove_adv);

  ptr = cfg.advert_res;

  while(ptr)
  {
    re_free((re_entry *) ptr->data);
    ptr = dllist_remove_entry(ptr, ptr);
  }

  cfg.advert_res = NULL;

  GET_DLLIST(gui_cfg.advert_res, ptr);

  while(ptr)
  {
    re_entry *ree;

    ree = re_make((char *) ptr->data);
    if(ree)
      cfg.advert_res = dllist_append(cfg.advert_res, (dllist_t) ree);
    if(ptr->data) free((char *)ptr->data);
    ptr = dllist_remove_entry(ptr, ptr);
  }


  ptr = cfg.js_patterns;

  while(ptr)
  {
    re_free((re_entry *) ptr->data);
    ptr = dllist_remove_entry(ptr, ptr);
  }

  cfg.advert_res = NULL;

  GET_DLLIST(gui_cfg.js_patterns, ptr);

  while(ptr)
  {
    re_entry *ree;

    ree = re_make((char *) ptr->data);
    if(ree)
      cfg.js_patterns = dllist_append(cfg.js_patterns, (dllist_t) ree);
    if(ptr->data) free((char *)ptr->data);
    ptr = dllist_remove_entry(ptr, ptr);
  }
#endif


#ifdef GTK_FACE
  {
    int i;

    while(cfg.lfnames)
    {
      lfname_free((lfname *) cfg.lfnames->data);
      cfg.lfnames = dllist_remove_entry(cfg.lfnames, cfg.lfnames);
    }

    for(i = 0; i < GTK_CLIST(gui_cfg.rules_list)->rows; i++)
    {
      lfname *lfn;
      lfname_type tp;
      char *ld[3];

      gtk_clist_get_text(GTK_CLIST(gui_cfg.rules_list), i, 0, &ld[0]);
      gtk_clist_get_text(GTK_CLIST(gui_cfg.rules_list), i, 1, &ld[1]);
      gtk_clist_get_text(GTK_CLIST(gui_cfg.rules_list), i, 2, &ld[2]);
#ifdef HAVE_REGEX
      tp = (*ld[0] == 'F') ? LFNAME_FNMATCH : LFNAME_REGEX;
#else
      tp = LFNAME_FNMATCH;
#endif
      if((lfn = lfname_new(tp, ld[1], ld[2])))
        cfg.lfnames = dllist_append(cfg.lfnames, (dllist_t) lfn);
      else
      {
        _MT_CFGSTAMP;
        UNLOCK_GCFG;
        return -1;
      }
    }
  }

#endif

  GET_URLLIST(gui_cfg.url_list, cfg.request);

  GET_URLLIST(gui_cfg.formdata_list, cfg.formdata);

  GET_PROXYLIST(gui_cfg.http_proxy_list, cfg.http_proxy);

  GET_JST(gui_cfg.js_transform, cfg.js_transform);

  _MT_CFGSTAMP;
  UNLOCK_GCFG;

  return 0;
}

#endif /* I_FACE */