Tools/webchecker/wcgui.py

   1 #! /usr/bin/env python
   2
   3 """GUI interface to webchecker.
   4
   5 This works as a Grail applet too!  E.g.
   6
   7   <APPLET CODE=wcgui.py NAME=CheckerWindow></APPLET>
   8
   9 Checkpoints are not (yet???  ever???) supported.
  10
  11 User interface:
  12
  13 Enter a root to check in the text entry box.  To enter more than one root,
  14 enter them one at a time and press <Return> for each one.
  15
  16 Command buttons Start, Stop and "Check one" govern the checking process in
  17 the obvious way.  Start and "Check one" also enter the root from the text
  18 entry box if one is present.  There's also a check box (enabled by default)
  19 to decide whether actually to follow external links (since this can slow
  20 the checking down considerably).  Finally there's a Quit button.
  21
  22 A series of checkbuttons determines whether the corresponding output panel
  23 is shown.  List panels are also automatically shown or hidden when their
  24 status changes between empty to non-empty.  There are six panels:
  25
  26 Log        -- raw output from the checker (-v, -q affect this)
  27 To check   -- links discovered but not yet checked
  28 Checked    -- links that have been checked
  29 Bad links  -- links that failed upon checking
  30 Errors     -- pages containing at least one bad link
  31 Details    -- details about one URL; double click on a URL in any of
  32               the above list panels (not in Log) will show details
  33               for that URL
  34
  35 Use your window manager's Close command to quit.
  36
  37 Command line options:
  38
  39 -m bytes  -- skip HTML pages larger than this size (default %(MAXPAGE)d)
  40 -q        -- quiet operation (also suppresses external links report)
  41 -v        -- verbose operation; repeating -v will increase verbosity
  42
  43 Command line arguments:
  44
  45 rooturl   -- URL to start checking
  46              (default %(DEFROOT)s)
  47
  48 XXX The command line options (-m, -q, -v) should be GUI accessible.
  49
  50 XXX The roots should be visible as a list (?).
  51
  52 XXX The multipanel user interface is clumsy.
  53
  54 """
  55
  56 # ' Emacs bait
  57
  58
  59 import sys
  60 import getopt
  61 import string
  62 from Tkinter import *
  63 import tktools
  64 import webchecker
  65 import random
  66
  67 # Override some for a weaker platform
  68 if sys.platform == 'mac':
  69     webchecker.DEFROOT = "http://grail.cnri.reston.va.us/"
  70     webchecker.MAXPAGE = 50000
  71     webchecker.verbose = 4
  72
  73 def main():
  74     try:
  75         opts, args = getopt.getopt(sys.argv[1:], 'm:qv')
  76     except getopt.error, msg:
  77         sys.stdout = sys.stderr
  78         print msg
  79         print __doc__%vars(webchecker)
  80         sys.exit(2)
  81     for o, a in opts:
  82         if o == '-m':
  83             webchecker.maxpage = string.atoi(a)
  84         if o == '-q':
  85             webchecker.verbose = 0
  86         if o == '-v':
  87             webchecker.verbose = webchecker.verbose + 1
  88     root = Tk(className='Webchecker')
  89     root.protocol("WM_DELETE_WINDOW", root.quit)
  90     c = CheckerWindow(root)
  91     if args:
  92         for arg in args[:-1]:
  93             c.addroot(arg)
  94         c.suggestroot(args[-1])
  95     root.mainloop()
  96
  97
  98 class CheckerWindow(webchecker.Checker):
  99
 100     def __init__(self, parent, root=webchecker.DEFROOT):
 101         self.__parent = parent
 102
 103         self.__topcontrols = Frame(parent)
 104         self.__topcontrols.pack(side=TOP, fill=X)
 105         self.__label = Label(self.__topcontrols, text="Root URL:")
 106         self.__label.pack(side=LEFT)
 107         self.__rootentry = Entry(self.__topcontrols, width=60)
 108         self.__rootentry.pack(side=LEFT)
 109         self.__rootentry.bind('<Return>', self.enterroot)
 110         self.__rootentry.focus_set()
 111
 112         self.__controls = Frame(parent)
 113         self.__controls.pack(side=TOP, fill=X)
 114         self.__running = 0
 115         self.__start = Button(self.__controls, text="Run", command=self.start)
 116         self.__start.pack(side=LEFT)
 117         self.__stop = Button(self.__controls, text="Stop", command=self.stop,
 118                              state=DISABLED)
 119         self.__stop.pack(side=LEFT)
 120         self.__step = Button(self.__controls, text="Check one",
 121                              command=self.step)
 122         self.__step.pack(side=LEFT)
 123         self.__cv = BooleanVar(parent)
 124         self.__cv.set(self.checkext)
 125         self.__checkext = Checkbutton(self.__controls, variable=self.__cv,
 126                                       command=self.update_checkext,
 127                                       text="Check nonlocal links",)
 128         self.__checkext.pack(side=LEFT)
 129         self.__reset = Button(self.__controls, text="Start over", command=self.reset)
 130         self.__reset.pack(side=LEFT)
 131         if __name__ == '__main__': # No Quit button under Grail!
 132             self.__quit = Button(self.__controls, text="Quit",
 133                                  command=self.__parent.quit)
 134             self.__quit.pack(side=RIGHT)
 135
 136         self.__status = Label(parent, text="Status: initial", anchor=W)
 137         self.__status.pack(side=TOP, fill=X)
 138         self.__checking = Label(parent, text="Idle", anchor=W)
 139         self.__checking.pack(side=TOP, fill=X)
 140         self.__mp = mp = MultiPanel(parent)
 141         sys.stdout = self.__log = LogPanel(mp, "Log")
 142         self.__todo = ListPanel(mp, "To check", self.showinfo)
 143         self.__done = ListPanel(mp, "Checked", self.showinfo)
 144         self.__bad = ListPanel(mp, "Bad links", self.showinfo)
 145         self.__errors = ListPanel(mp, "Pages w/ bad links", self.showinfo)
 146         self.__details = LogPanel(mp, "Details")
 147         webchecker.Checker.__init__(self)
 148         if root:
 149             root = string.strip(str(root))
 150             if root:
 151                 self.suggestroot(root)
 152         self.newstatus()
 153
 154     def reset(self):
 155         webchecker.Checker.reset(self)
 156         for p in self.__todo, self.__done, self.__bad, self.__errors:
 157             p.clear()
 158
 159     def suggestroot(self, root):
 160         self.__rootentry.delete(0, END)
 161         self.__rootentry.insert(END, root)
 162         self.__rootentry.select_range(0, END)
 163
 164     def enterroot(self, event=None):
 165         root = self.__rootentry.get()
 166         root = string.strip(root)
 167         if root:
 168             self.__checking.config(text="Adding root "+root)
 169             self.__checking.update_idletasks()
 170             self.addroot(root)
 171             self.__checking.config(text="Idle")
 172             try:
 173                 i = self.__todo.items.index(root)
 174             except (ValueError, IndexError):
 175                 pass
 176             else:
 177                 self.__todo.list.select_clear(0, END)
 178                 self.__todo.list.select_set(i)
 179                 self.__todo.list.yview(i)
 180         self.__rootentry.delete(0, END)
 181
 182     def start(self):
 183         self.__start.config(state=DISABLED, relief=SUNKEN)
 184         self.__stop.config(state=NORMAL)
 185         self.__step.config(state=DISABLED)
 186         self.enterroot()
 187         self.__running = 1
 188         self.go()
 189
 190     def stop(self):
 191         self.__stop.config(state=DISABLED, relief=SUNKEN)
 192         self.__running = 0
 193
 194     def step(self):
 195         self.__start.config(state=DISABLED)
 196         self.__step.config(state=DISABLED, relief=SUNKEN)
 197         self.enterroot()
 198         self.__running = 0
 199         self.dosomething()
 200
 201     def go(self):
 202         if self.__running:
 203             self.__parent.after_idle(self.dosomething)
 204         else:
 205             self.__checking.config(text="Idle")
 206             self.__start.config(state=NORMAL, relief=RAISED)
 207             self.__stop.config(state=DISABLED, relief=RAISED)
 208             self.__step.config(state=NORMAL, relief=RAISED)
 209
 210     __busy = 0
 211
 212     def dosomething(self):
 213         if self.__busy: return
 214         self.__busy = 1
 215         if self.todo:
 216             l = self.__todo.selectedindices()
 217             if l:
 218                 i = l[0]
 219             else:
 220                 i = 0
 221                 self.__todo.list.select_set(i)
 222             self.__todo.list.yview(i)
 223             url = self.__todo.items[i]
 224             self.__checking.config(text="Checking "+url)
 225             self.__parent.update()
 226             self.dopage(url)
 227         else:
 228             self.stop()
 229         self.__busy = 0
 230         self.go()
 231
 232     def showinfo(self, url):
 233         d = self.__details
 234         d.clear()
 235         d.put("URL:    %s\n" % url)
 236         if self.bad.has_key(url):
 237             d.put("Error:  %s\n" % str(self.bad[url]))
 238         if url in self.roots:
 239             d.put("Note:   This is a root URL\n")
 240         if self.done.has_key(url):
 241             d.put("Status: checked\n")
 242             o = self.done[url]
 243         elif self.todo.has_key(url):
 244             d.put("Status: to check\n")
 245             o = self.todo[url]
 246         else:
 247             d.put("Status: unknown (!)\n")
 248             o = []
 249         if self.errors.has_key(url):
 250             d.put("Bad links from this page:\n")
 251             for triple in self.errors[url]:
 252                 link, rawlink, msg = triple
 253                 d.put("  HREF  %s" % link)
 254                 if link != rawlink: d.put(" (%s)" %rawlink)
 255                 d.put("\n")
 256                 d.put("  error %s\n" % str(msg))
 257         self.__mp.showpanel("Details")
 258         for source, rawlink in o:
 259             d.put("Origin: %s" % source)
 260             if rawlink != url:
 261                 d.put(" (%s)" % rawlink)
 262             d.put("\n")
 263         d.text.yview("1.0")
 264
 265     def setbad(self, url, msg):
 266         webchecker.Checker.setbad(self, url, msg)
 267         self.__bad.insert(url)
 268         self.newstatus()
 269
 270     def setgood(self, url):
 271         webchecker.Checker.setgood(self, url)
 272         self.__bad.remove(url)
 273         self.newstatus()
 274
 275     def newlink(self, url, origin):
 276         webchecker.Checker.newlink(self, url, origin)
 277         if self.done.has_key(url):
 278             self.__done.insert(url)
 279         elif self.todo.has_key(url):
 280             self.__todo.insert(url)
 281         self.newstatus()
 282
 283     def markdone(self, url):
 284         webchecker.Checker.markdone(self, url)
 285         self.__done.insert(url)
 286         self.__todo.remove(url)
 287         self.newstatus()
 288
 289     def seterror(self, url, triple):
 290         webchecker.Checker.seterror(self, url, triple)
 291         self.__errors.insert(url)
 292         self.newstatus()
 293
 294     def newstatus(self):
 295         self.__status.config(text="Status: "+self.status())
 296         self.__parent.update()
 297
 298     def update_checkext(self):
 299         self.checkext = self.__cv.get()
 300
 301
 302 class ListPanel:
 303
 304     def __init__(self, mp, name, showinfo=None):
 305         self.mp = mp
 306         self.name = name
 307         self.showinfo = showinfo
 308         self.panel = mp.addpanel(name)
 309         self.list, self.frame = tktools.make_list_box(
 310             self.panel, width=60, height=5)
 311         self.list.config(exportselection=0)
 312         if showinfo:
 313             self.list.bind('<Double-Button-1>', self.doubleclick)
 314         self.items = []
 315
 316     def clear(self):
 317         self.items = []
 318         self.list.delete(0, END)
 319         self.mp.hidepanel(self.name)
 320
 321     def doubleclick(self, event):
 322         l = self.selectedindices()
 323         if l:
 324             self.showinfo(self.list.get(l[0]))
 325
 326     def selectedindices(self):
 327         l = self.list.curselection()
 328         if not l: return []
 329         return map(string.atoi, l)
 330
 331     def insert(self, url):
 332         if url not in self.items:
 333             if not self.items:
 334                 self.mp.showpanel(self.name)
 335             # (I tried sorting alphabetically, but the display is too jumpy)
 336             i = len(self.items)
 337             self.list.insert(i, url)
 338             self.list.yview(i)
 339             self.items.insert(i, url)
 340
 341     def remove(self, url):
 342         try:
 343             i = self.items.index(url)
 344         except (ValueError, IndexError):
 345             pass
 346         else:
 347             was_selected = i in self.selectedindices()
 348             self.list.delete(i)
 349             del self.items[i]
 350             if not self.items:
 351                 self.mp.hidepanel(self.name)
 352             elif was_selected:
 353                 if i >= len(self.items):
 354                     i = len(self.items) - 1
 355                 self.list.select_set(i)
 356
 357
 358 class LogPanel:
 359
 360     def __init__(self, mp, name):
 361         self.mp = mp
 362         self.name = name
 363         self.panel = mp.addpanel(name)
 364         self.text, self.frame = tktools.make_text_box(self.panel, height=10)
 365         self.text.config(wrap=NONE)
 366
 367     def clear(self):
 368         self.text.delete("1.0", END)
 369         self.text.yview("1.0")
 370
 371     def put(self, s):
 372         self.text.insert(END, s)
 373         if '\n' in s:
 374             self.text.yview(END)
 375
 376     def write(self, s):
 377         self.text.insert(END, s)
 378         if '\n' in s:
 379             self.text.yview(END)
 380             self.panel.update()
 381
 382
 383 class MultiPanel:
 384
 385     def __init__(self, parent):
 386         self.parent = parent
 387         self.frame = Frame(self.parent)
 388         self.frame.pack(expand=1, fill=BOTH)
 389         self.topframe = Frame(self.frame, borderwidth=2, relief=RAISED)
 390         self.topframe.pack(fill=X)
 391         self.botframe = Frame(self.frame)
 392         self.botframe.pack(expand=1, fill=BOTH)
 393         self.panelnames = []
 394         self.panels = {}
 395
 396     def addpanel(self, name, on=0):
 397         v = StringVar(self.parent)
 398         if on:
 399             v.set(name)
 400         else:
 401             v.set("")
 402         check = Checkbutton(self.topframe, text=name,
 403                             offvalue="", onvalue=name, variable=v,
 404                             command=self.checkpanel)
 405         check.pack(side=LEFT)
 406         panel = Frame(self.botframe)
 407         label = Label(panel, text=name, borderwidth=2, relief=RAISED, anchor=W)
 408         label.pack(side=TOP, fill=X)
 409         t = v, check, panel
 410         self.panelnames.append(name)
 411         self.panels[name] = t
 412         if on:
 413             panel.pack(expand=1, fill=BOTH)
 414         return panel
 415
 416     def showpanel(self, name):
 417         v, check, panel = self.panels[name]
 418         v.set(name)
 419         panel.pack(expand=1, fill=BOTH)
 420
 421     def hidepanel(self, name):
 422         v, check, panel = self.panels[name]
 423         v.set("")
 424         panel.pack_forget()
 425
 426     def checkpanel(self):
 427         for name in self.panelnames:
 428             v, check, panel = self.panels[name]
 429             panel.pack_forget()
 430         for name in self.panelnames:
 431             v, check, panel = self.panels[name]
 432             if v.get():
 433                 panel.pack(expand=1, fill=BOTH)
 434
 435
 436 if __name__ == '__main__':
 437     main()