gui_common.c

网络爬虫程序

    "xterm -e lynx",
    "rxvt -e links",
    NULL,
  };
  const char *sample_identity[] = {
    "",
    PACKAGE "/" VERSION " (" HOSTTYPE ")",
    "Mozilla/4.0 (compatible; " PACKAGE " " VERSION
      "; Linux 2.2.14 i486; X11)",
    "Mozilla/4.7 [en] (X11; Linux 2.2.14 i486; I)",
    "Mozilla/4.0 (compatible; MSIE 5.0; Windows NT; DigExt)",
    "Lynx/2.7 libWWW-FM/2.14",
    "NCSA_Mosaic/2.7b5 (X11;Linux 2.0.31 i486)  libwww/2.12 modified",
    "Opera/3.0",
    NULL,
  };

  tbox = gtk_vbox_new(FALSE, 0);
  gtk_widget_show(tbox);
  label = gtk_label_new(gettext("Grabber I"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), tbox, label);

  box = gtk_table_new(2, 11, FALSE);
  gtk_box_pack_start(GTK_BOX(tbox), box, FALSE, FALSE, 0);
  gtk_widget_show(box);

  gui_cfg.cdir_label = guitl_tab_add_path_entry(box,
    gettext("Cache Directory: "), 0, 0, TRUE);

  gui_cfg.default_prefix = guitl_tab_add_entry(box,
    gettext("Default URL prefix: "), 0, 1, FALSE);

  gui_cfg.info_dir = guitl_tab_add_path_entry(box,
    gettext("Separate info directory: "), 0, 2, TRUE);

  gui_cfg.index_name = guitl_tab_add_enum(box,
    gettext("Index file name: "), 0, 3, sample_index_names, TRUE);

  gui_cfg.store_name = guitl_tab_add_entry(box,
    gettext("Store file name: "), 0, 4, FALSE);

  gui_cfg.identity = guitl_tab_add_enum(box,
    gettext("Identity string: "), 0, 5, sample_identity, TRUE);

#ifdef HAVE_BDB_18x
  gui_cfg.ns_cache_dir = guitl_tab_add_path_entry(box,
    gettext("Netscape browser cache directory: "), 0, 6, TRUE);

  gui_cfg.moz_cache_dir = guitl_tab_add_path_entry(box,
    gettext("Mozilla browser cache directory: "), 0, 7, TRUE);
#endif

#ifdef WITH_TREE
  gui_cfg.browser_label = guitl_tab_add_enum(box,
    gettext("Browser: "), 0, 8, sample_browsers, FALSE);
#endif

  gui_cfg.remind_cmd = guitl_tab_add_entry(box,
    gettext("Reminder command: "), 0, 9, FALSE);

  gui_cfg.post_cmd = guitl_tab_add_entry(box,
    gettext("Post command: "), 0, 10, FALSE);

/*********/
  box = gtk_table_new(1, 2, TRUE);
  gtk_box_pack_start(GTK_BOX(tbox), box, FALSE, FALSE, 0);
  gtk_widget_show(box);

  frame = gtk_frame_new(NULL);
  gtk_table_attach(GTK_TABLE(box), frame, 0, 1, 0, 1, GTK_FILL, GTK_FILL, 1,
    1);
  gtk_widget_show(frame);

  ptab = gtk_table_new(6, 3, FALSE);
  gtk_container_add(GTK_CONTAINER(frame), ptab);
  gtk_widget_show(ptab);

  gui_cfg.retry_label = guitl_tab_add_numentry(ptab,
    gettext("How many times retry on fail: "), 0, 0, USHRT_MAX);

  gui_cfg.redir_label = guitl_tab_add_numentry(ptab,
    gettext("How many moved links to follow: "), 0, 1, USHRT_MAX);

  gui_cfg.reget_label = guitl_tab_add_numentry(ptab,
    gettext("How many times to reget file: "), 0, 2, USHRT_MAX);

  gui_cfg.ddays_label = guitl_tab_add_numentry(ptab,
    gettext("Document age before syncing with server: "), 0, 3, USHRT_MAX);

  label = gtk_label_new(gettext(" days "));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 3, 4,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

  gui_cfg.bufsize = guitl_tab_add_numentry(ptab,
    gettext("Read buffer size: "), 0, 4, 10000);
  gtk_adjustment_clamp_page(GTK_ADJUSTMENT(GTK_SPIN_BUTTON(gui_cfg.bufsize)->
      adjustment), 1.0, 1024.0);

  label = gtk_label_new(gettext(" kB "));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 4, 5,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

  gui_cfg.hash_size = guitl_tab_add_numentry(ptab,
    gettext("Hash tables size: "), 0, 5, SHRT_MAX);

  label = gtk_label_new(gettext(" entries "));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 5, 6,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);
/*****/
  frame = gtk_frame_new(NULL);
  gtk_table_attach(GTK_TABLE(box), frame, 1, 2, 0, 1, GTK_FILL, GTK_FILL, 1,
    1);
  gtk_widget_show(frame);

  ptab = gtk_table_new(6, 4, FALSE);
  gtk_container_add(GTK_CONTAINER(frame), ptab);
  gtk_widget_show(ptab);

  gui_cfg.sleep_label = guitl_tab_add_numentry(ptab,
    gettext("Sleep time between transfers: "), 0, 0, INT_MAX);

  label = gtk_label_new(gettext(" sec."));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 0, 1,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

  gui_cfg.rsleep = gtk_check_button_new_with_label(gettext("randomize"));
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.rsleep, 3, 4, 0, 1,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(gui_cfg.rsleep);

  gui_cfg.rollback_label = guitl_tab_add_numentry(ptab,
    gettext("Rollback amount on reget: "), 0, 1, INT_MAX);

  label = gtk_label_new(gettext(" bytes"));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 1, 2,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

  gui_cfg.trans_quota = guitl_tab_add_numentry(ptab,
    gettext("Transfer quota: "), 0, 2, INT_MAX);

  label = gtk_label_new(gettext(" kB"));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 2, 3,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

  gui_cfg.file_quota = guitl_tab_add_numentry(ptab,
    gettext("File size quota: "), 0, 3, INT_MAX);

  label = gtk_label_new(gettext(" kB"));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 3, 4,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);

#if defined HAVE_FSTATFS || HAVE_FSTATVFS
  gui_cfg.fs_quota = guitl_tab_add_numentry(ptab,
    gettext("Filesystem freespace quota: "), 0, 4, INT_MAX);

  label = gtk_label_new(gettext(" kB"));
  gtk_table_attach(GTK_TABLE(ptab), label, 2, 3, 4, 5,
    GTK_SHRINK, GTK_FILL, 2, 2);
  gtk_widget_show(label);
#endif

#ifdef HAVE_MT
  {
    int n = 61;

#ifdef PTHREAD_THREADS_MAX
    n = PTHREAD_THREADS_MAX - 3;
#elif defined _POSIX_THREADS_THREAD_MAX
    n = _POSIX_THREADS_THREAD_MAX - 3;
#endif
    gui_cfg.nthr = guitl_tab_add_numentry(ptab,
      gettext("Number of downloading threads: "), 0, 5, n);
  }
#endif
}

/*** GRABER II ***/
static void cfgtab_graberII(GtkWidget * notebook)
{
  GtkWidget *brow, *label, *frame, *ptab;
  GtkWidget *menu, *mi;
  int i;

  brow = gtk_hbox_new(FALSE, 5);
  gtk_widget_show(brow);
  label = gtk_label_new(gettext("Grabber II"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), brow, label);

  frame = gtk_frame_new(gettext("Misc settings"));
  gtk_widget_show(frame);
  gtk_box_pack_start(GTK_BOX(brow), frame, FALSE, TRUE, 1);

  brow = gtk_vbox_new(0, 1);
  gtk_widget_show(brow);
  gtk_container_add(GTK_CONTAINER(frame), brow);

  gui_cfg.unique_doc =
    gtk_check_button_new_with_label(gettext
    ("Always generate unique name for document"));
  gtk_widget_show(gui_cfg.unique_doc);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.unique_doc, FALSE, TRUE, 1);

  gui_cfg.del_after =
    gtk_check_button_new_with_label(gettext
    ("Delete FTP document after succesful download"));
  gtk_widget_show(gui_cfg.del_after);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.del_after, FALSE, TRUE, 1);

  gui_cfg.ptime =
    gtk_check_button_new_with_label(gettext
    ("Preserve document modification time"));
  gtk_widget_show(gui_cfg.ptime);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.ptime, FALSE, TRUE, 1);

  gui_cfg.preserve_perm =
    gtk_check_button_new_with_label(gettext
    ("Preserve FTP document permissions"));
  gtk_widget_show(gui_cfg.preserve_perm);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.preserve_perm, FALSE, TRUE, 1);

  gui_cfg.preserve_links =
    gtk_check_button_new_with_label(gettext
    ("Preserve FTP symbolic links paths"));
  gtk_widget_show(gui_cfg.preserve_links);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.preserve_links, FALSE, TRUE, 1);

  gui_cfg.retrieve_slink =
    gtk_check_button_new_with_label(gettext
    ("Retrieve FTP symbolic links like files"));
  gtk_widget_show(gui_cfg.retrieve_slink);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.retrieve_slink, FALSE, TRUE, 1);


  gui_cfg.freget_sw =
    gtk_check_button_new_with_label(gettext
    ("Whole reget when partial not supported"));
  gtk_widget_show(gui_cfg.freget_sw);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.freget_sw, FALSE, TRUE, 1);


  gui_cfg.enc_sw =
    gtk_check_button_new_with_label(gettext
    ("Use gzip encoding for transfer"));
  gtk_widget_show(gui_cfg.enc_sw);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.enc_sw, FALSE, TRUE, 1);

  gui_cfg.oldrm_sw =
    gtk_check_button_new_with_label(gettext("Remove improper documents"));
  gtk_widget_show(gui_cfg.oldrm_sw);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.oldrm_sw, FALSE, TRUE, 1);

  gui_cfg.check_size =
    gtk_check_button_new_with_label(gettext
    ("Check transferred size of document"));
  gtk_widget_show(gui_cfg.check_size);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.check_size, FALSE, TRUE, 1);

  gui_cfg.store_index =
    gtk_check_button_new_with_label(gettext
    ("Store directory URLs as index files"));
  gtk_widget_show(gui_cfg.store_index);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.store_index, FALSE, TRUE, 1);

  gui_cfg.enable_info =
    gtk_check_button_new_with_label(gettext
    ("Store info files with each document"));
  gtk_widget_show(gui_cfg.enable_info);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.enable_info, FALSE, TRUE, 1);

  gui_cfg.auto_referer =
    gtk_check_button_new_with_label(gettext
    ("Send self URL as Referer for starting URLs"));
  gtk_widget_show(gui_cfg.auto_referer);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.auto_referer, FALSE, TRUE, 1);

  gui_cfg.referer =
    gtk_check_button_new_with_label(gettext
    ("Send Referer with HTTP requests"));
  gtk_widget_show(gui_cfg.referer);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.referer, FALSE, TRUE, 1);

  gui_cfg.read_css =
    gtk_check_button_new_with_label(gettext
    ("Fetch objects mentioned in style sheets"));
  gtk_widget_show(gui_cfg.read_css);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.read_css, FALSE, TRUE, 1);

  gui_cfg.send_if_range =
    gtk_check_button_new_with_label(gettext
    ("Send If-Range header field when regeting"));
  gtk_widget_show(gui_cfg.send_if_range);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.send_if_range, FALSE, TRUE, 1);

  gui_cfg.show_time =
    gtk_check_button_new_with_label(gettext
    ("Show time of start and end of downloading"));
  gtk_widget_show(gui_cfg.show_time);
  gtk_box_pack_start(GTK_BOX(brow), gui_cfg.show_time, FALSE, TRUE, 1);

  ptab = gtk_table_new(2, 1, FALSE);
  gtk_box_pack_start(GTK_BOX(brow), ptab, FALSE, FALSE, 1);
  gtk_widget_show(ptab);

  label = gtk_label_new(gettext("URL scheduling strategy: "));
  gtk_table_attach(GTK_TABLE(ptab), label, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 1, 1);
  gtk_widget_show(label);

  gui_cfg.scheduling_strategie = gtk_option_menu_new();

  menu = gtk_menu_new();

  for(i = 0; i < SSTRAT_LAST; i++)
  {
    mi = gtk_menu_item_new_with_label(get_strategie_label(i));
    gtk_object_set_user_data(GTK_OBJECT(mi), (gpointer) i);
    gtk_menu_append(GTK_MENU(menu), mi);
    gtk_widget_show(mi);
  }

  gtk_option_menu_set_menu(GTK_OPTION_MENU(gui_cfg.scheduling_strategie),
    menu);

  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.scheduling_strategie,
    1, 2, 0, 1, GTK_FILL, GTK_FILL, 1, 1);
  gtk_widget_show(gui_cfg.scheduling_strategie);
}

/*** HTML ***/
static void HtmlswSens(GtkWidget * w, gpointer data)
{
  gboolean act = GTK_TOGGLE_BUTTON(w)->active;

  gtk_widget_set_sensitive(gui_cfg.all_to_local, act);
  gtk_widget_set_sensitive(gui_cfg.sel_to_local, act);
  gtk_widget_set_sensitive(gui_cfg.url_to_local, act);
  gtk_widget_set_sensitive(gui_cfg.all_to_remote, act);
  gtk_widget_set_sensitive(gui_cfg.post_update, act);
}

static void cfgtab_html(GtkWidget * notebook)
{
  GtkWidget *brow, *label, *frame, *ptab, *brow2;
  GSList *rg;

  brow = gtk_table_new(2, 2, FALSE);
  gtk_widget_show(brow);
  label = gtk_label_new(gettext("HTML"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), brow, label);

  frame = gtk_frame_new(gettext("HTML document URL rewriting rules"));
  gtk_widget_show(frame);
  gtk_table_attach(GTK_TABLE(brow), frame, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 2, 2);

  brow2 = gtk_vbox_new(0, 1);
  gtk_widget_show(brow2);
  gtk_container_add(GTK_CONTAINER(frame), brow2);

  gui_cfg.noreloc_sw =
    gtk_check_button_new_with_label(gettext("Rewrite URLs inside HTML doc."));
  gtk_widget_show(gui_cfg.noreloc_sw);
  gtk_box_pack_start(GTK_BOX(brow2), gui_cfg.noreloc_sw, FALSE, TRUE, 2);
  gtk_signal_connect(GTK_OBJECT(gui_cfg.noreloc_sw), "toggled",
    GTK_SIGNAL_FUNC(HtmlswSens), NULL);

  rg = NULL;
  gui_cfg.url_to_local = gtk_radio_button_new_with_label(rg,
    gettext("Rewrite URLs to local when stored locally"));
  rg = gtk_radio_button_group(GTK_RADIO_BUTTON(gui_cfg.url_to_local));
  gtk_widget_show(gui_cfg.url_to_local);
  gtk_box_pack_start(GTK_BOX(brow2), gui_cfg.url_to_local, FALSE, TRUE, 1);

  gui_cfg.sel_to_local = gtk_radio_button_new_with_label(rg,
    gettext("Rewrite all suitable URLs to local"));
  rg = gtk_radio_button_group(GTK_RADIO_BUTTON(gui_cfg.sel_to_local));
  gtk_widget_show(gui_cfg.sel_to_local);
  gtk_box_pack_start(GTK_BOX(brow2), gui_cfg.sel_to_local, FALSE, TRUE, 1);

  gui_cfg.all_to_local = gtk_radio_button_new_with_label(rg,
    gettext("Rewrite all URLs to local immediately"));
  rg = gtk_radio_button_group(GTK_RADIO_BUTTON(gui_cfg.all_to_local));
  gtk_widget_show(gui_cfg.all_to_local);
  gtk_box_pack_start(GTK_BOX(brow2), gui_cfg.all_to_local, FALSE, TRUE, 1);

  gui_cfg.all_to_remote = gtk_radio_button_new_with_label(rg,
    gettext("Rewrite all URLs to remote immediately"));
  rg = gtk_radio_button_group(GTK_RADIO_BUTTON(gui_cfg.all_to_remote));
  gtk_widget_show(gui_cfg.all_to_remote);
  gtk_box_pack_start(GTK_BOX(brow2), gui_cfg.all_to_remote, FALSE, TRUE, 1);

  gui_cfg.post_update = gtk_radio_button_new_with_label(rg,
    gettext("Rewrite only one currently download URL"));