gui_limits.c

网络爬虫程序

/***************************************************************************/
/*    This code is part of WWW grabber called pavuk                        */
/*    Copyright (c) 1997 - 2001 Stefan Ondrejicka                          */
/*    Distributed under GPL 2 or later                                     */
/***************************************************************************/

#include "config.h"
#include "gui.h"

#ifdef GTK_FACE
#include <stdio.h>
#include <gdk/gdkkeysyms.h>

#include "gtkmulticol.h"
#include "html.h"
#include "uconfig.h"
#include "mimetype.h"
#include "tag_pattern.h"

#include "icons/append.xpm"
#include "icons/modify.xpm"
#include "icons/clear.xpm"
#include "icons/delete.xpm"
#include "icons/ok.xpm"
#include "icons/apply.xpm"
#include "icons/common.xpm"
#include "icons/cancel.xpm"

static void CfgLimits(GtkObject * object, gpointer func_data)
{
  if(!xget_cfg_values_lim())
  {
    if(cfg.use_prefs)
      cfg_dump_pref();
  }
}

static void limtab_tree(GtkWidget * notebook)
{
  GtkWidget *label, *box, *ptab, *frame;

  box = gtk_vbox_new(FALSE, 2);
  gtk_widget_show(box);
  label = gtk_label_new(gettext("Tree"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), box, label);

  frame = gtk_frame_new(NULL);
  gtk_box_pack_start(GTK_BOX(box), frame, FALSE, FALSE, 2);
  gtk_widget_show(frame);

  ptab = gtk_table_new(2, 4, FALSE);
  gtk_container_add(GTK_CONTAINER(frame), ptab);
  gtk_widget_show(ptab);

  gui_cfg.cgi_sw =
    gtk_check_button_new_with_label(gettext("Download cgi-generated pages"));
  gtk_widget_show(gui_cfg.cgi_sw);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.cgi_sw, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.ftpd_sw =
    gtk_check_button_new_with_label(gettext("Recurse through FTP directory"));
  gtk_widget_show(gui_cfg.ftpd_sw);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.ftpd_sw, 0, 1, 1, 2,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.robots_sw =
    gtk_check_button_new_with_label(gettext("Allow \"robots.txt\""));
  gtk_widget_show(gui_cfg.robots_sw);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.robots_sw, 0, 1, 2, 3,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.ftp_html =
    gtk_check_button_new_with_label(gettext
    ("Process HTML files downloaded over FTP"));
  gtk_widget_show(gui_cfg.ftp_html);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.ftp_html, 0, 1, 3, 4,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.leaves_sw =
    gtk_check_button_new_with_label(gettext("Don't leave starting site"));
  gtk_widget_show(gui_cfg.leaves_sw);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.leaves_sw, 1, 2, 0, 1,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.leaved_sw =
    gtk_check_button_new_with_label(gettext
    ("Don't leave starting directory"));
  gtk_widget_show(gui_cfg.leaved_sw);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.leaved_sw, 1, 2, 1, 2,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.dont_leave_site_dir =
    gtk_check_button_new_with_label(gettext
    ("Don't leave site enter directory"));
  gtk_widget_show(gui_cfg.dont_leave_site_dir);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.dont_leave_site_dir, 1, 2, 2, 3,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.singlepage =
    gtk_check_button_new_with_label(gettext("Download just single page"));
  gtk_widget_show(gui_cfg.singlepage);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.singlepage, 1, 2, 3, 4,
    GTK_FILL, GTK_FILL, 5, 0);

  gui_cfg.limit_inlines =
    gtk_check_button_new_with_label(gettext
    ("Apply limiting options on inline objects"));
  gtk_widget_show(gui_cfg.limit_inlines);
  gtk_table_attach(GTK_TABLE(ptab), gui_cfg.limit_inlines, 0, 1, 4, 5,
    GTK_FILL, GTK_FILL, 5, 0);


  frame = gtk_frame_new(NULL);
  gtk_box_pack_start(GTK_BOX(box), frame, FALSE, FALSE, 2);
  gtk_widget_show(frame);

  ptab = gtk_table_new(4, 3, FALSE);
  gtk_container_add(GTK_CONTAINER(frame), ptab);
  gtk_widget_show(ptab);

  gui_cfg.maxdoc_label = guitl_tab_add_numentry(ptab,
    gettext("Max. count of documents: "), 0, 4, INT_MAX);

  gui_cfg.maxlev_label = guitl_tab_add_numentry(ptab,
    gettext("Max. depth of tree: "), 0, 5, USHRT_MAX);

  gui_cfg.leave_level = guitl_tab_add_numentry(ptab,
    gettext("Max. levels to leave from starting site: "), 0, 6, USHRT_MAX);

  gui_cfg.maxsize_label = guitl_tab_add_numentry(ptab,
    gettext("Max. document size: "), 2, 4, INT_MAX);

  gui_cfg.min_size = guitl_tab_add_numentry(ptab,
    gettext("Min. document size: "), 2, 5, INT_MAX);

  gui_cfg.site_level = guitl_tab_add_numentry(ptab,
    gettext("Max. site levels to leave from starting site: "), 2, 6, INT_MAX);

  frame = gtk_frame_new(NULL);
  gtk_box_pack_start(GTK_BOX(box), frame, FALSE, FALSE, 2);
  gtk_widget_show(frame);

  ptab = gtk_table_new(2, 3, FALSE);
  gtk_container_add(GTK_CONTAINER(frame), ptab);
  gtk_widget_show(ptab);

  gui_cfg.subdir_label = guitl_tab_add_path_entry(ptab,
    gettext("Working subdirectory :"), 0, 0, TRUE);

  gui_cfg.en_uexit = guitl_tab_add_entry(ptab,
    gettext("User condition script: "), 0, 1, FALSE);

  gui_cfg.follow_cmd = guitl_tab_add_entry(ptab,
    gettext("Follow command: "), 0, 2, FALSE);

}

static void limtab_patterns(GtkWidget * notebook)
{
  GtkWidget *label, *box, *col, *ptab;

  box = gtk_vbox_new(FALSE, 2);
  gtk_widget_show(box);
  label = gtk_label_new(gettext("Patterns"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), box, label);

  ptab = gtk_frame_new(gettext("Wildcard patterns"));
  gtk_box_pack_start(GTK_BOX(box), ptab, FALSE, FALSE, 2);
  gtk_widget_show(ptab);

  col = gtk_table_new(4, 2, FALSE);
  gtk_container_add(GTK_CONTAINER(ptab), col);
  gtk_widget_show(col);

  gui_cfg.pattern_label = guitl_tab_add_edit_entry(col,
    gettext("Documents matching pattern: "), NULL, 0, 0, FALSE);

  gui_cfg.skip_pattern = guitl_tab_add_edit_entry(col,
    gettext("skip: "),
    gettext("Pavuk: edit Documents matching skip pattern"), 2, 0, FALSE);

  gui_cfg.url_pattern_label = guitl_tab_add_edit_entry(col,
    gettext("URL matching pattern: "), NULL, 0, 1, FALSE);

  gui_cfg.skip_url_pattern = guitl_tab_add_edit_entry(col,
    gettext("skip: "),
    gettext("Pavuk: edit URL matching skip pattern"), 2, 1, FALSE);

#ifdef HAVE_REGEX
  ptab = gtk_frame_new(gettext("RE patterns"));
  gtk_box_pack_start(GTK_BOX(box), ptab, FALSE, FALSE, 2);
  gtk_widget_show(ptab);

  col = gtk_table_new(4, 2, FALSE);
  gtk_container_add(GTK_CONTAINER(ptab), col);
  gtk_widget_show(col);

  gui_cfg.rpattern = guitl_tab_add_edit_entry(col,
    gettext("Documents matching pattern: "), NULL, 0, 0, TRUE);

  gui_cfg.skip_rpattern = guitl_tab_add_edit_entry(col,
    gettext("skip: "),
    gettext("Pavuk: edit Documents matching skip pattern"), 2, 0, TRUE);

  gui_cfg.url_rpattern = guitl_tab_add_edit_entry(col,
    gettext("URL matching pattern: "), NULL, 0, 1, TRUE);

  gui_cfg.url_skip_rpattern = guitl_tab_add_edit_entry(col,
    gettext("skip: "),
    gettext("Pavuk: edit URL matching skip pattern"), 2, 1, TRUE);
#endif
}

static void limtab_hosts(GtkWidget * notebook)
{
  GtkWidget *brow, *label, *box, *col, *ptab;

  box = gtk_table_new(2, 3, FALSE);
  gtk_widget_show(box);
  label = gtk_label_new(gettext("Hosts"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), box, label);

  col = gtk_table_new(2, 1, FALSE);
  gtk_widget_show(col);
  gtk_table_attach(GTK_TABLE(box), col, 0, 1, 0, 1,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  gui_cfg.hosts_sw =
    gtk_check_button_new_with_label(gettext("Allow / Disallow sites"));
  gtk_widget_show(gui_cfg.hosts_sw);
  gtk_table_attach(GTK_TABLE(col), gui_cfg.hosts_sw, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 5, 5);

  brow = guitl_new_edit_list(&gui_cfg.hosts_list, &gui_cfg.hosts_entry,
    gettext("Site: "), NULL, NULL, NULL, NULL, TRUE, NULL);

  gtk_table_attach(GTK_TABLE(col), brow, 0, 1, 1, 2,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  col = gtk_table_new(2, 1, FALSE);
  gtk_widget_show(col);
  gtk_table_attach(GTK_TABLE(box), col, 1, 2, 0, 1,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  gui_cfg.domain_sw =
    gtk_check_button_new_with_label(gettext("Allow / Disallow domains"));
  gtk_widget_show(gui_cfg.domain_sw);
  gtk_table_attach(GTK_TABLE(col), gui_cfg.domain_sw, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 5, 5);

  brow = guitl_new_edit_list(&gui_cfg.domain_list, &gui_cfg.domain_entry,
    gettext("Domain: "), NULL, NULL, NULL, NULL, TRUE, NULL);

  gtk_table_attach(GTK_TABLE(col), brow, 0, 1, 1, 2,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

#ifdef HAVE_REGEX
  ptab = gtk_frame_new(gettext("IP address RE patterns"));
  gtk_table_attach(GTK_TABLE(box), ptab, 0, 2, 1, 2,
    GTK_FILL, GTK_FILL, 5, 5);
  gtk_widget_show(ptab);

  col = gtk_table_new(4, 1, FALSE);
  gtk_container_add(GTK_CONTAINER(ptab), col);
  gtk_widget_show(col);

  gui_cfg.aip = guitl_tab_add_edit_entry(col,
    gettext("Server IP address matching pattern: "), NULL, 0, 0, TRUE);

  gui_cfg.skipip = guitl_tab_add_edit_entry(col,
    gettext("skip: "),
    gettext("Pavuk: edit Server IP address matching skip pattern"),
    2, 0, TRUE);
#endif

  ptab = gtk_frame_new(gettext("Server ports"));
  gtk_table_attach(GTK_TABLE(box), ptab, 0, 2, 2, 3,
    GTK_FILL, GTK_FILL, 5, 5);
  gtk_widget_show(ptab);

  col = gtk_table_new(3, 1, FALSE);
  gtk_container_add(GTK_CONTAINER(ptab), col);
  gtk_widget_show(col);

  gui_cfg.allow_ports =
    gtk_check_button_new_with_label(gettext("Allow/deny"));
  gtk_table_attach(GTK_TABLE(col), gui_cfg.allow_ports, 0, 1, 0, 1, GTK_FILL,
    GTK_FILL, 1, 1);
  gtk_widget_show(gui_cfg.allow_ports);

  gui_cfg.ports = guitl_tab_add_edit_entry(col, gettext("Ports: "),
    NULL, 1, 0, FALSE);
}

static void limtab_docs(GtkWidget * notebook)
{
  GtkWidget *brow, *label, *box, *col;

  box = gtk_table_new(1, 2, FALSE);
  gtk_widget_show(box);
  label = gtk_label_new(gettext("Documents"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), box, label);

  col = gtk_table_new(2, 1, FALSE);
  gtk_widget_show(col);
  gtk_table_attach(GTK_TABLE(box), col, 0, 1, 0, 1,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  gui_cfg.sufix_sw =
    gtk_check_button_new_with_label(gettext("Allow / Disallow suffix"));
  gtk_widget_show(gui_cfg.sufix_sw);
  gtk_table_attach(GTK_TABLE(col), gui_cfg.sufix_sw, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 5, 5);

  brow = guitl_new_edit_list(&gui_cfg.sufixlist, &gui_cfg.sufix_label,
    gettext("Suffix: "), NULL, NULL, NULL, NULL, TRUE, NULL);

  gtk_table_attach(GTK_TABLE(col), brow, 0, 1, 1, 2,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  col = gtk_table_new(2, 1, FALSE);
  gtk_widget_show(col);
  gtk_table_attach(GTK_TABLE(box), col, 1, 2, 0, 1,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

  gui_cfg.prefix_sw =
    gtk_check_button_new_with_label(gettext("Allow / Disallow prefix"));
  gtk_widget_show(gui_cfg.prefix_sw);
  gtk_table_attach(GTK_TABLE(col), gui_cfg.prefix_sw, 0, 1, 0, 1,
    GTK_FILL, GTK_FILL, 5, 5);

  brow = guitl_new_edit_list(&gui_cfg.prefixlist, &gui_cfg.prefix_label,
    gettext("Prefix: "), NULL, NULL, NULL, NULL, TRUE, NULL);

  gtk_table_attach(GTK_TABLE(col), brow, 0, 1, 1, 2,
    GTK_FILL | GTK_EXPAND, GTK_FILL | GTK_EXPAND, 5, 5);

}

static void limtab_mime(GtkWidget * notebook)
{
  GtkWidget *col, *box, *label, *frame, *entry, *pbox;

  box = gtk_hbox_new(0, 5);
  gtk_widget_show(box);
  label = gtk_label_new(gettext("MIME types"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), box, label);

  pbox = gtk_vbox_new(FALSE, 2);
  gtk_box_pack_start(GTK_BOX(box), pbox, FALSE, FALSE, 5);
  gtk_widget_show(pbox);

  gui_cfg.mime_sw =
    gtk_check_button_new_with_label(gettext("Allow / Disallow MIME type"));
  gtk_box_pack_start(GTK_BOX(pbox), gui_cfg.mime_sw, FALSE, FALSE, 5);
  gtk_widget_show(gui_cfg.mime_sw);

  frame = gtk_frame_new(gettext("MIME types"));
  gtk_box_pack_start(GTK_BOX(pbox), frame, TRUE, TRUE, 5);
  gtk_widget_show(frame);

  col = guitl_new_edit_list(&gui_cfg.amimelist, &entry,
    gettext("MIME type: "), NULL, NULL, NULL, NULL, TRUE, mimetypes);

  gtk_container_add(GTK_CONTAINER(frame), col);
}

static void limtab_time(GtkWidget * notebook)
{
  GtkWidget *tbox, *box, *col, *label, *frame;

  tbox = gtk_vbox_new(FALSE, 1);
  gtk_widget_show(tbox);
  label = gtk_label_new(gettext("Time"));
  gtk_notebook_append_page(GTK_NOTEBOOK(notebook), tbox, label);

  box = gtk_hbox_new(FALSE, 5);
  gtk_widget_show(box);
  gtk_box_pack_start(GTK_BOX(tbox), box, FALSE, FALSE, 4);
  frame = gtk_frame_new(gettext("Lower document time limit"));
  gtk_widget_show(frame);
  gtk_box_pack_start(GTK_BOX(box), frame, FALSE, FALSE, 4);

  col = guitl_timesel_new(&gui_cfg.btime_cal, &gui_cfg.btime_h_entry,
    &gui_cfg.btime_min_entry, &gui_cfg.btime_mon, &gui_cfg.btime_yentry);
  gtk_container_add(GTK_CONTAINER(frame), col);

  gui_cfg.btime_sw =
    gtk_check_button_new_with_label(gettext
    ("Check if doc. time newer than this"));