What is the recommended technique for interactively validating content in a tkinter Entry widget?
I've read the posts about using validate=True and validatecommand=command, and it appears that these features are limited by the fact that they get cleared if the validatecommand command updates the Entry widget's value.
Given this behavior, should we bind on the KeyPress, Cut, and Paste events and monitor/update our Entry widget's value through these events? (And other related events that I might have missed?)
Or should we forget interactive validation altogether and only validate on FocusOut events?
The correct answer is, use the validatecommand attribute of the widget. Unfortunately this feature is severely under-documented in the Tkinter world, though it is quite sufficiently documented in the Tk world. Even though it's not documented well, it has everything you need to do validation without resorting to bindings or tracing variables, or modifying the widget from within the validation procedure.
The trick is to know that you can have Tkinter pass in special values to your validate command. These values give you all the information you need to know to decide on whether the data is valid or not: the value prior to the edit, the value after the edit if the edit is valid, and several other bits of information. To use these, though, you need to do a little voodoo to get this information passed to your validate command.
Note: it's important that the validation command returns either True or False. Anything else will cause the validation to be turned off for the widget.
Here's an example that only allows lowercase. It also prints the values of all of the special values for illustrative purposes. They aren't all necessary; you rarely need more than one or two.
import tkinter as tk # python 3.x
# import Tkinter as tk # python 2.x
class Example(tk.Frame):
def __init__(self, parent):
tk.Frame.__init__(self, parent)
# valid percent substitutions (from the Tk entry man page)
# note: you only have to register the ones you need; this
# example registers them all for illustrative purposes
#
# %d = Type of action (1=insert, 0=delete, -1 for others)
# %i = index of char string to be inserted/deleted, or -1
# %P = value of the entry if the edit is allowed
# %s = value of entry prior to editing
# %S = the text string being inserted or deleted, if any
# %v = the type of validation that is currently set
# %V = the type of validation that triggered the callback
# (key, focusin, focusout, forced)
# %W = the tk name of the widget
vcmd = (self.register(self.onValidate),
'%d', '%i', '%P', '%s', '%S', '%v', '%V', '%W')
self.entry = tk.Entry(self, validate="key", validatecommand=vcmd)
self.text = tk.Text(self, height=10, width=40)
self.entry.pack(side="top", fill="x")
self.text.pack(side="bottom", fill="both", expand=True)
def onValidate(self, d, i, P, s, S, v, V, W):
self.text.delete("1.0", "end")
self.text.insert("end","OnValidate:\n")
self.text.insert("end","d='%s'\n" % d)
self.text.insert("end","i='%s'\n" % i)
self.text.insert("end","P='%s'\n" % P)
self.text.insert("end","s='%s'\n" % s)
self.text.insert("end","S='%s'\n" % S)
self.text.insert("end","v='%s'\n" % v)
self.text.insert("end","V='%s'\n" % V)
self.text.insert("end","W='%s'\n" % W)
# Disallow anything but lowercase letters
if S == S.lower():
return True
else:
self.bell()
return False
if __name__ == "__main__":
root = tk.Tk()
Example(root).pack(fill="both", expand=True)
root.mainloop()
For more information about what happens under the hood when you call the register method, see Why is calling register() required for tkinter input validation?
For the canonical documentation see the Validation section of the Tcl/Tk Entry man page
After studying and experimenting with Bryan's code, I produced a minimal version of input validation. The following code will put up an Entry box and only accept numeric digits.
from tkinter import *
root = Tk()
def testVal(inStr,acttyp):
if acttyp == '1': #insert
if not inStr.isdigit():
return False
return True
entry = Entry(root, validate="key")
entry['validatecommand'] = (entry.register(testVal),'%P','%d')
entry.pack()
root.mainloop()
Perhaps I should add that I am still learning Python and I will gladly accept any and all comments/suggestions.
Use a Tkinter.StringVar to track the value of the Entry widget. You can validate the value of the StringVar by setting a trace on it.
Here's a short working program that accepts only valid floats in the Entry widget.
try:
from tkinter import *
except ImportError:
from Tkinter import * # Python 2
root = Tk()
sv = StringVar()
def validate_float(var):
new_value = var.get()
try:
new_value == '' or float(new_value)
validate_float.old_value = new_value
except:
var.set(validate_float.old_value)
validate_float.old_value = '' # Define function attribute.
# trace wants a callback with nearly useless parameters, fixing with lambda.
sv.trace('w', lambda nm, idx, mode, var=sv: validate_float(var))
ent = Entry(root, textvariable=sv)
ent.pack()
ent.focus_set()
root.mainloop()
Bryan's answer is correct, however no one mentioned the 'invalidcommand' attribute of the tkinter widget.
A good explanation is here:
http://infohost.nmt.edu/tcc/help/pubs/tkinter/web/entry-validation.html
Text copy/pasted in case of broken link
The Entry widget also supports an invalidcommand option that specifies a callback function that is called whenever the validatecommand returns False. This command may modify the text in the widget by using the .set() method on the widget's associated textvariable. Setting up this option works the same as setting up the validatecommand. You must use the .register() method to wrap your Python function; this method returns the name of the wrapped function as a string. Then you will pass as the value of the invalidcommand option either that string, or as the first element of a tuple containing substitution codes.
Note:
There is only one thing that I cannot figure out how to do: If you add validation to an entry, and the user selects a portion of the text and types a new value, there is no way to capture the original value and reset the entry. Here's an example
Entry is designed to only accept integers by implementing 'validatecommand'
User enters 1234567
User selects '345' and presses 'j'. This is registered as two actions: deletion of '345', and insertion of 'j'. Tkinter ignores the deletion and acts only on the insertion of 'j'. 'validatecommand' returns False, and the values passed to the 'invalidcommand' function are as follows: %d=1, %i=2, %P=12j67, %s=1267, %S=j
If the code does not implement an 'invalidcommand' function, the 'validatecommand' function will reject the 'j' and the result will be 1267. If the code does implement an 'invalidcommand' function, there is no way to recover the original 1234567.
Define a function returning a boolean that indicates whether the input is valid.Register it as a Tcl callback, and pass the callback name to the widget as a validatecommand.
For example:
import tkinter as tk
def validator(P):
"""Validates the input.
Args:
P (int): the value the text would have after the change.
Returns:
bool: True if the input is digit-only or empty, and False otherwise.
"""
return P.isdigit() or P == ""
root = tk.Tk()
entry = tk.Entry(root)
entry.configure(
validate="key",
validatecommand=(
root.register(validator),
"%P",
),
)
entry.grid()
root.mainloop()
Reference.
While studying Bryan Oakley's answer, something told me that a far more general solution could be developed. The following example introduces a mode enumeration, a type dictionary, and a setup function for validation purposes. See line 48 for example usage and a demonstration of its simplicity.
#! /usr/bin/env python3
# https://stackoverflow.com/questions/4140437
import enum
import inspect
import tkinter
from tkinter.constants import *
Mode = enum.Enum('Mode', 'none key focus focusin focusout all')
CAST = dict(d=int, i=int, P=str, s=str, S=str,
v=Mode.__getitem__, V=Mode.__getitem__, W=str)
def on_validate(widget, mode, validator):
# http://www.tcl.tk/man/tcl/TkCmd/ttk_entry.htm#M39
if mode not in Mode:
raise ValueError('mode not recognized')
parameters = inspect.signature(validator).parameters
if not set(parameters).issubset(CAST):
raise ValueError('validator arguments not recognized')
casts = tuple(map(CAST.__getitem__, parameters))
widget.configure(validate=mode.name, validatecommand=[widget.register(
lambda *args: bool(validator(*(cast(arg) for cast, arg in zip(
casts, args)))))]+['%' + parameter for parameter in parameters])
class Example(tkinter.Frame):
#classmethod
def main(cls):
tkinter.NoDefaultRoot()
root = tkinter.Tk()
root.title('Validation Example')
cls(root).grid(sticky=NSEW)
root.grid_rowconfigure(0, weight=1)
root.grid_columnconfigure(0, weight=1)
root.mainloop()
def __init__(self, master, **kw):
super().__init__(master, **kw)
self.entry = tkinter.Entry(self)
self.text = tkinter.Text(self, height=15, width=50,
wrap=WORD, state=DISABLED)
self.entry.grid(row=0, column=0, sticky=NSEW)
self.text.grid(row=1, column=0, sticky=NSEW)
self.grid_rowconfigure(1, weight=1)
self.grid_columnconfigure(0, weight=1)
on_validate(self.entry, Mode.key, self.validator)
def validator(self, d, i, P, s, S, v, V, W):
self.text['state'] = NORMAL
self.text.delete(1.0, END)
self.text.insert(END, 'd = {!r}\ni = {!r}\nP = {!r}\ns = {!r}\n'
'S = {!r}\nv = {!r}\nV = {!r}\nW = {!r}'
.format(d, i, P, s, S, v, V, W))
self.text['state'] = DISABLED
return not S.isupper()
if __name__ == '__main__':
Example.main()
import tkinter
tk=tkinter.Tk()
def only_numeric_input(e):
#this is allowing all numeric input
if e.isdigit():
return True
#this will allow backspace to work
elif e=="":
return True
else:
return False
#this will make the entry widget on root window
e1=tkinter.Entry(tk)
#arranging entry widget on screen
e1.grid(row=0,column=0)
c=tk.register(only_numeric_input)
e1.configure(validate="key",validatecommand=(c,'%P'))
tk.mainloop()
#very usefull for making app like calci
Here's an improved version of #Steven Rumbalski's answer of validating the Entry widgets value by tracing changes to a StringVar — which I have already debugged and improved to some degree by editing it in place.
The version below puts everything into a StringVar subclass to encapsulates what's going on better and, more importantly allow multiple independent instances of it to exist at the same time without interfering with each other — a potential problem with his implementation because it utilizes function attributes instead of instance attributes, which are essentially the same thing as global variables and can lead to problems in such a scenario.
try:
from tkinter import *
except ImportError:
from Tkinter import * # Python 2
class ValidateFloatVar(StringVar):
"""StringVar subclass that only allows valid float values to be put in it."""
def __init__(self, master=None, value=None, name=None):
StringVar.__init__(self, master, value, name)
self._old_value = self.get()
self.trace('w', self._validate)
def _validate(self, *_):
new_value = self.get()
try:
new_value == '' or float(new_value)
self._old_value = new_value
except ValueError:
StringVar.set(self, self._old_value)
root = Tk()
ent = Entry(root, textvariable=ValidateFloatVar(value=42.0))
ent.pack()
ent.focus_set()
ent.icursor(END)
root.mainloop()
This code can help if you want to set both just digits and max characters.
from tkinter import *
root = Tk()
def validate(P):
if len(P) == 0 or len(P) <= 10 and P.isdigit(): # 10 characters
return True
else:
return False
ent = Entry(root, validate="key", validatecommand=(root.register(validate), '%P'))
ent.pack()
root.mainloop()
Responding to orionrobert's problem of dealing with simple validation upon substitutions of text through selection, instead of separate deletions or insertions:
A substitution of selected text is processed as a deletion followed by an insertion. This may lead to problems, for example, when the deletion should move the cursor to the left, while a substitution should move the cursor to the right. Fortunately, these two processes are executed immediately after one another.
Hence, we can differentiate between a deletion by itself and a deletion directly followed by an insertion due to a substitution because the latter has does not change the idle flag between deletion and insertion.
This is exploited using a substitutionFlag and a Widget.after_idle().
after_idle() executes the lambda-function at the end of the event queue:
class ValidatedEntry(Entry):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.tclValidate = (self.register(self.validate), '%d', '%i', '%P', '%s', '%S', '%v', '%V', '%W')
# attach the registered validation function to this spinbox
self.config(validate = "all", validatecommand = self.tclValidate)
def validate(self, type, index, result, prior, indelText, currentValidationMode, reason, widgetName):
if typeOfAction == "0":
# set a flag that can be checked by the insertion validation for being part of the substitution
self.substitutionFlag = True
# store desired data
self.priorBeforeDeletion = prior
self.indexBeforeDeletion = index
# reset the flag after idle
self.after_idle(lambda: setattr(self, "substitutionFlag", False))
# normal deletion validation
pass
elif typeOfAction == "1":
# if this is a substitution, everything is shifted left by a deletion, so undo this by using the previous prior
if self.substitutionFlag:
# restore desired data to what it was during validation of the deletion
prior = self.priorBeforeDeletion
index = self.indexBeforeDeletion
# optional (often not required) additional behavior upon substitution
pass
else:
# normal insertion validation
pass
return True
Of course, after a substitution, while validating the deletion part, one still won’t know whether an insert will follow.
Luckily however, with:
.set(),
.icursor(),
.index(SEL_FIRST),
.index(SEL_LAST),
.index(INSERT),
we can achieve most desired behavior retrospectively (since the combination of our new substitutionFlag with an insertion is a new unique and final event.
I have a list named chosenTestHolder (imported from the my_config file) that consists of several objects each with the attribute 'sentence'.
When pressing the button 'Press' for the first time, the attribute 'sentence' of the first object in the chosenTestHolder should be displayed in the text widget. The next time the button 'Press' is pressed the attribute 'sentence' of the second object in chosenTestHolder should be displayed and so on.
I am using lambda event for binding the 'Press' button and tries to use a new sentences as its first arguments after each pressing of the 'Press' button. However, it keeps showing the first sentence.
When searching Stackoverflow I have seen in
Using lambda function to change value of an attribute that you can't use assignments in lambda expressions but by reading that I still have not figured out how to solve my problem.
Grateful for help! Code is below!
main.py
from tkinter import font
import tkinter as tk
import tkinter.ttk as ttk
import my_config
import Testlist as tl
class TestWidgetTest:
def __init__(self):
ram = tk.Frame(root)
ram.grid(in_=root,row=0, column=0)
self.myText = tk.Text(ram, height = 5)
self.myText.grid(row=0,column=1)
my_config.counter = 0
self.myButton = tk.Button(ram, text = 'Press')
self.myButton.grid(row =1, column =0, columnspan =2)
indata =[my_config.chosenTestHolder[my_config.counter] , self.myText]
self.myButton.bind('<ButtonRelease-1>',lambda event, arg=indata : self.TagConfigure(event, arg))
def TagConfigure(self, event, arg):
arg[1].delete('1.0',tk.END)
arg[1].insert('1.0',arg[0].sentence)
my_config.counter += 1
root = tk.Tk()
TestWidgetTest()
root.mainloop()
my_config.py
import Testlist as tl
testListHolder = [ ['Fabian was very tired'],
['Thomas light the fire'],
['Anna eat a red apple ']]
chosenTestHolder = []
count = 0
while count <(len(testListHolder)):
chosenTestHolder.append(tl.Testlist(testListHolder[count][0]))
count += 1
counter = 0
Testlist.py
class Testlist:
def __init__(self, sentence):
self.sentence = sentence
Your issue is the assignment of indata.
You do only assign in init.
To get your code working you need to re-configure your sentecte...
indata =[my_config.chosenTestHolder[my_config.counter] , self.myText]
self.myButton.bind('<ButtonRelease-1>',lambda event, arg=indata : self.TagConfigure(event, arg))
I would advise to keep track of the current sentence as an instance variable.
class Test_widget(tk.Frame):
def __init__(self, *args, **kwargs):
tk.Frame.__init__(self, args, kwargs)
self.sentences=["a", "b", "c", "d"] # the data
self.show = tk.StringVar() # the current displayed data
self.show.set("NULL")
self.counter=0 # the indexer
tk.Label(self, textvariable=self.show).grid(row=0)
tk.Button(self, command=self.click).grid(row=1)
def click(self, event):
self.show.set("%s"%self.sentences[self.counter]) # use the indexer to access the data
self.counter = self.counter + 1 # modify the indexer
if self.counter = len(self.sentences): # make sure you dont run in index-err
self.counter = 0
As you see, there is no need at all for the lambdas.
Edit
As to your questions:
The change in your original code was not intended.
I do not see a use case where you can use a lambda for its use inside your code.
At least none where a lambda is necessary.
Please remember to use lambda only and exclusively if there are
no ( == NULL ) other options.
Using inheritance (thats what the mechanism is called), you can inherit functions, "default" behaviour from other classes. It is a common mechanism in programming and not exclusive to python.
It is used like any normal object except you have to call the constructor of the base class (what I do using tk.Frame.__init__(self, args, kwargs) inside the init method. For more information on inheritance please refer to the uncounted manuals and tutorials available for that topic (google is your friend now that you know what the mechanism is called).
I've been desperately trying to get this section of code to work in my program. I essentially want to read in several options from a file, and create Tkinter buttons from those options. Creating the buttons is no issue; currently, I just can't make the code run the functions I want.
from Lib import StegosaurMainCode as Steg
...
class App:
def __init__(self, master, menu):
buttons = []
for counter in range(0, len(menu[0])):
text = menu[0][counter]
func = menu[1][counter]
att = menu[2][counter]
buttons.append(Button(text=text, command=lambda: Steg.func(att)))
frame = Frame(master)
for item in buttons:
item.pack()
frame.pack()
In this class, func is the function I want to call, Steg is the external code in another file, and att are the attributes for the function. I can't seem to figure out why Steg.func won't tries to call a function in Steg called "func" rather than the one described in the variable func
Have your lambda rebind its att parameter at each call.
class App:
def __init__(self, master, menu):
buttons = []
for counter in range(0, len(menu[0])):
text = menu[0][counter]
func = menu[1][counter]
att = menu[2][counter]
buttons.append(Button(text = text, command = lambda att = att: Steg.func(att)))
frame = Frame(master)
for item in buttons:
item.pack()
frame.pack()
Assuming that menu[1][counter] contains a string rather than a reference to an actual function, you need to get a reference to the function which you can then use as the value for the command attribute. You can do that with getattr:
func = getattr(steg, menu[1][counter])
Once you've done that, you can use func as if it were an actual function. However, you need to bind the variables to their current values, so you need to pass them as arguments to the lambda:
button = Button(text=text, command=lambda func=func, attr=att: func(att)))
I am working on an application that is supposed to support both running from a console and from a GUI. The application has several options to choose from, and since in both running modes the program is going to have the same options obviously, I made a generalisation:
class Option:
def __init__(self, par_name, par_desc):
self.name = par_name
self.desc = par_desc
class Mode():
def __init__(self):
self.options = []
self.options.append(Option('Option1', 'Desc1'))
self.options.append(Option('Option2', 'Desc2'))
self.options.append(Option('Option3', 'Desc3'))
self.options.append(Option('Option4', 'Desc4'))
self.options.append(Option('Option5', 'Desc5'))
#And so on
The problem is that in GUI, those options are going to be buttons, so I have to add a new field to an Option class and I'm doing it like this:
def onMouseEnter(par_event, par_option):
helpLabel.configure(text = par_option.desc)
return
def onMouseLeave(par_event):
helpLabel.configure(text = '')
return
class GUIMode(Mode):
#...
for iOption in self.options:
iOption.button = Button(wrapper, text = iOption.name, bg = '#004A7F', fg = 'white')
iOption.button.bind('<Enter>', lambda par_event: onMouseEnter(par_event, iOption))
iOption.button.bind('<Leave>', lambda par_event: onMouseLeave(par_event))
#...
There is also a "help label" showing the description of the option every time a mouse hovers over it, so there I am binding those functions.
What is happening is that while I am indeed successfully adding a new field with a button, the bind function seems to mess up and the result is this:
Help label is always showing the description of the last option added, no matter over which button I hover. The problem seems to go away if I directly modify the Option class instead, like this:
class Option:
def __init__(self, par_name, par_desc):
self.name = par_name
self.desc = par_desc
self.button = Button(wrapper, text = self.name, bg = '#004A7F', fg = 'white')
self.button.bind('<Enter>', lambda par_event: onMouseEnter(par_event, self))
self.button.bind('<Leave>', lambda par_event: onMouseLeave(par_event))
But I obviously can't keep it that way because the console mode will get those fields too which I don't really want. Isn't this the same thing, however? Why does it matter if I do it in a constructor with self or in a loop later? I therefore assume that the problem might be in a way I dynamically add the field to the class?
Here is the full minimal and runnable test code or whatever it is called, if you want to mess with it: http://pastebin.com/0PWnF2P0
Thank you for your time
The problem is that the value of iOption is evaluated after the
for iOption in self.option:
loops are complete. Since you reset iOption on each iteration, when the loop is completed iOption has the same value, namely the last element in self.options. You can demonstrate this at-event-time binding with the snippet:
def debug_late_bind(event):
print(iOption)
onMouseEnter(event, iOption)
for iOption in self.options:
iOption.button = Button(wrapper, text = iOption.name,
bg = '#004A7F', fg = 'white')
iOption.button.bind('<Enter>', debug_late_bind)
which will show that all events that iOption has the same value.
I split out the use of iOption to debug_late_bind to show that iOption comes in from the class scope and is not evaluated when the bind() call is executed. A more simple example would be
def print_i():
print(i)
for i in range(5):
pass
print_i()
which prints "4" because that is the last value that was assigned to i. This is why every call in your code to onMouseEnter(par_event, iOption) has the same value for iOption; it is evaluated at the time of the event, not the time of the bind. I suggest that you read up on model view controller and understand how you've tangled the view and the controller. The primary reason this has happened is that you've got two views (console and tk) which should be less coupled with the model.
Extracting the .widget property of the event is a decent workaround, but better still would be to not overwrite the scalar iOption, but instead use list of individual buttons. The code
for n, iOption in enumerate(self.options):
would help in creating a list. In your proposed workaround, you are encoding too much of the iOption model in the tkinter view. That's bound to bite you again at some point.
I don't know what the actual problem was with my original code, but I kind of just bypassed it. I added a dictionary with button as a key and option as a value and I just used the par_event.widget to get the option and it's description, which is working fine:
buttonOption = {}
def onMouseEnter(par_event):
helpLabel.configure(text = buttonOption[par_event.widget].desc)
return
def onMouseLeave(par_event):
helpLabel.configure(text = '')
return
class GUIMode(Mode):
def run(self):
#...
for iOption in self.options:
iOption.button = Button(wrapper, text = iOption.name, bg = '#004A7F', fg = 'white')
iOption.button.bind('<Enter>', lambda par_event: onMouseEnter(par_event))
iOption.button.bind('<Leave>', lambda par_event: onMouseLeave(par_event))
buttonOption[iOption.button] = iOption
#...