- 1. Input API
- 1.1. A Brief Description of Input API
- Installing Input API
- 1.2. Basic Usage
- 1.3. Advanced Usage
- 1.3.1. Numbers
- 1.3.2. Strings
- 1.3.3. Booleans
- 1.3.4. Menus
- 1.3.5. RDCs
- 1.3.6. otherFunc
- 1.4. TODO
Input API is a python package made and maintained by Homer Haddock for the express purpose of making terminal I/O both easier and simpler! When using Input API you are given several options for input, including but not limited to:
- Numbers
- Both integers and floating point numbers
- Booleans
- Strings
- Menus
When receiving a copy of Input API you need to install it first, here is a short tutorial on how to install the Input API.
- Install Python (v3.7 or greater)
- Open the folder that has the .whl file for the Input API (e.g.
inputapi-2.0.0-py3-none-any.whl
) - Open the .whl file with your Python installation (Not IDLE)
- Wait for the window to close
- Input API has been globally installed on your computer
If you are using a virtual environment you just run.
For Windows:
pip install Path/To/File.whl
For Linux:
python3 -m pip install Path/To/File.whl
Now for all of the examples of use we will be importing like so:
import inputapi as inp
Note: This way of importing is easier for IDE's that don't have an "Auto-Complete" or "AI-assisted dev." feature
When importing Input API we have what is called 'The Basic Library'. The Basic Library while making usage easier it is also backwards compatible with 1.X versions of Input API without having to rewrite the code.
With numerical input you can access both floating point and integer numbers! An integer input is used as shown:
import inputapi as inp
inp.newLineInt({params})
Note: We will cover parameters in Advanced Usage
Floating point numbers can be used as shown:
import inputapi as inp
inp.newLineInt({params})
If you wish to clear the terminal by using the clearScreen function1:
import inputapi as inp
inp.clearScreen()
If you need a boolean input we have out yesNo function for that exact reason!
import inputapi as inp
inp.yesNo({params})
Fun Fact2: yesNo is the only input to be in the boolean category making it the loneliest function in the package!
This style of usage is good for if you need specific input. We will also include the parameters for the inputs with examples.
Integers serve the purpose of returning integers, integers have two ways to input:
Type of input | Best case for use |
---|---|
New Line | When there are small amounts of input |
Same Line | When there is a lot of inputs in succession |
There are two ways to use this input:
import inputapi as inp
#This is the basic way of using newLineInt
inp.newLineInt({params})
import inputapi as inp
#This is the advanced way of using newLineInt
inp.numerical.integer.newLineInt({params})
newLineInt will output the request like this:
#For this example the request string is "Number:"
Number:
>
#The ">" is where the user will input the number
The parameters for the input are as follows:
Parameter | data type | default | Purpose it serves |
---|---|---|---|
request | string | "Input an integer:" |
What the user will see when the input is requested |
allowNeg | boolean | True |
Determines if the user can give a negative number |
min | integer | No limit | What is the lowest number allowed |
max | integer | No limit | What is the highest number allowed |
clearOnLoad | boolean | False |
When the function is run, should the terminal be cleared |
clearWhenDone | boolean | False |
When the input is retrieved, should the terminal be cleared |
The sameLineInt has only one way to access the function:
import inputapi as inp
inp.numerical.integer.sameLineInt({params})
This is what sameLineInt will output:
#For this example we will have two requests, one will be "X=" and the other will be "Y="
#This is the first request
X=
#And when "X" has been given the next request is shown
Y=
#The user will enter the number at the end of the string
You can find the parameters here
When the inputs face invalid characters, we have error messages to show the user what is allowed and what isn't:
#This error message will appear when the user inputs an invalid character, in this case we will put a decimal in
Number:
>21.6
WARNING: Invalid character ['.'], to ensure the program runs correctly please enter only these characters ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '-']
#This one will be displayed when inputing a negative number when negative numbers are disabled
X=-16
WARNING: Invalid character ['-'], to ensure the program runs correctly please enter only these characters ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
#For this example we will put in a number under the minimum
Y=4
WARNING: Number [4] below minimum [5]
For going over the max there are two messages.
What is picked is depending on the amount of digits for both max and what is entered:
#A number that is greater than max (Same amount of digits as max)
Enter 433724:
>433725
WARNING: Number [433725] above maximum [433724]
#We will go above the maximum for this example (More digits than max has)
Enter a 6 digit number:
>581919921
WARNING: Too many characters, maximum amount is [6] characters, you entered [9]
Floating point inputs are great for decimal numbers (e.g. The radius of a circle, width of a box).
We have two ways for requesting an float:
Type of input | Best use case |
---|---|
newLineFloat | Small amount of inputs in succession |
sameLineFloat | Multiple inputs with each one after another, or retrieving variables |
Calling the function can be done in two ways:
import inputapi as inp
#Basic usage
inp.newLineFloat({params})
import inputapi as inp
#Advanced usage
inp.numerical.float.newLineFloat({params})
When using newLineFloat you get an output similar to newLineInt
#This time the request is "Number including decimal:"
Number including decimal:
>
The parameters for newLineFloat are very similar to integer inputs but are as follows:
Parameter | data type | default | Purpose it serves |
---|---|---|---|
request | string | "Input an integer:" |
What the user will see when the input is requested |
allowNeg | boolean | True |
Determines if the user can give a negative number |
min | integer | No limit | What is the lowest number allowed |
max | integer | No limit | What is the highest number allowed |
clearOnLoad | boolean | False |
When the function is run, should the terminal be cleared |
clearWhenDone | boolean | False |
When the input is retrieved, should the terminal be cleared |
sameLineFloat has only one way to be used:
import inputapi as inp
inp.numerical.float.sameLineFloat({params})
The input function outputs like so:
#Request string is "Float="
Float=
Parameters are found here.
For most of these can be found in the integer's error handling except for one error.
When dealing with decimal points ("."), it can't just be blocked, so we have a solution from the 1.X versions of the Input API:
#The conversion from straight string to float only works if all charecters are numbers and there are no more that one decimal point
#So this is what will happen if we try to break it
Only one decimal point:
>2.2.2
#The result is 2.22
How this works is when there is more than one point (e.g. 1.2.3
), it will only acknowledge the first decimal place and remove the others (1.2.3
-> 1.23
).
Strings works like input()
but allows for more options and helps with error handling so you don't need something like this copy and pasted 20 times:
while True:
user = input(request)
try:
convert(user)
except:
continue
else:
break
#convert() is the stand in for what you wan't the string converted to
Strings were first introduced in Input API version 1.2.0 as newLineStr, got a bug fix in 1.2.1 (Like an hour after 1.2.0 got built).
Everything3 in the Input API is dependant on the string input since it gives more viability and ease of use. There are two types of string input and they are:
Type of input | Best use case |
---|---|
newLineStr | Small amount of inputs in succession |
sameLineStr | Multiple inputs with each one after another, getting multiple strings, and replacing the character that sits left of the user input (> ) with something else |
newLineStr has (for the thousandth time) two ways to be used:
import inputapi as inp
#First way
inp.newLineStr({params})
import inputapi as inp
#Second way
inp.strings.newLineStr({params})
When using the newLineStr the user using the program will have this as the output:
#The request, "String:"
String:
>
We have the parameters for newLineStr as:
Parameter | data type | default | Purpose |
---|---|---|---|
request | string | 'Input a string:' |
Is displayed when requesting input |
minLength | integer | 0 |
Amount of characters needed for input |
maxLength | integer | None | Max amount of characters allowed in input |
allowOnly | string | None | Only allows an input if all characters are within the string, (Does nothing if default) |
clearOnLoad | boolean | False |
Clears terminal before loading input |
clearWhenDone | boolean | False |
Clears terminal after getting input |
sameLineStr has only one way to be used:
import inputapi as inp
#Only way
inp.strings.sameLineStr({params})
The output for sameLineStr is:
#The request, "String="
String=
The parameters for sameLineStr are:
Parameter | data type | default | Purpose |
---|---|---|---|
request | string | 'String=' |
Is displayed when requesting input |
minLength | integer | 0 |
Amount of characters needed for input |
maxLength | integer | None | Max amount of characters allowed in input |
allowOnly | string | None | Only allows an input if all characters are within the string, (Does nothing if default) |
clearOnLoad | boolean | False |
Clears terminal before loading input |
clearWhenDone | boolean | False |
Clears terminal after getting input |
When it comes to error handling on strings, we have three potential errors that can happen:
#This happens when the user inputs less than the minimum amount of characters
Just enter three characters:
>No
WARNING: Input requires [3] characters or more, you entered [2]
#This happens when the user inputs more than the maximum amount of characters
Three characters or less:
>Four
WARNING: Input requires [3] characters or less, you entered [4]
There are three separate messages related to allowOnly
depending on length of the amount allowed:
#When there is less than 20 characters allowed
#In this case there is only numbers allowed
Number=Boo!
WARNING: Input has invalid character [B], allowed characters are [1, 2, 3, 4, 5, 6, 7, 8, 9, 0]
#When there is more than 30 characters allowed
#In this case there is only letters and numbers allowed
Enter only what a normal person would:
>!@\#$%^&*()_+-=
WARNING: Input has invalid character [!], allowed characters are [a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t ... 5, 6, 7, 8, 9]
#This is what happens when there is more than 20 but less than 30 characters allowed
#In this case there is only lowercase letters and isn't 'a' allowed
Please don't yell:
>AAAAA
WARNING: Input has invalid character [A], allowed characters are [b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t, u ... z]
The only boolean input there is is yesNo
which is a boolean input that only allows for y
or n
as input. When input is given it will turn it into a boolean value (True
or False
).
The Booleans best case use is for when you want to get a yes or no answer from the user. It is also used in the confirm
input. (Refer to menus for more info about confirming input [Placeholder Link For Menus])
yesNo has two ways to be used:
import inputapi as inp
#The simple way
inp.yesNo({params})
import inputapi as inp
#The hard way
inp.boolean.yesNo({params})
yesNo has two different outputs:
#When numeric input is allowed (default)
#This is was added in 2.0.0
#Request is 'Do you breathe air?'
Do you breathe air?
Numeric response allowed (Y=1, n=2)
[Y/n]:
#When numeric isn't allowed
#Request is 'Is your fride running?'
Is your fridge running?
[Y/n]:
The parameters for yesNo are:
Parameter | data type | default | Purpose |
---|---|---|---|
request | string | 'Yes or no?' |
Is displayed when requesting input |
allowNumeric | boolean | True |
Allows for numeric input (Y=1, n=2) |
clearOnLoad | boolean | False |
Clears terminal before loading input |
clearWhenDone | boolean | False |
Clears terminal after getting input |
The errors in yesNo mainly happen through the sameLineStr but they are:
#When something not allowed is entered (More than 1 character)
Did you catch the fridge?
Numeric response allowed (Y=1, n=2)
[Y/n]:Still trying to
WARNING: Input requires [1] characters or less, you entered [15] characters
#Something not allowed is entered (Numeric not allowed)
Can you see me?
[Y/n]:1
WARNING: Input has invalid character [1], allowed characters are ['Y', 'N', 'y', 'n']
Menus allow the user to select a set of given options, there could be only one to select or multiple.
These menus present multiple options but only allow the user to pick one. These can be used for the 'main hub' of a program to use other functions and return after. The returned value is the value assigned to the option.
The numericSerial is the menu used in 1.X versions of Input API. All options are given a number from 1 to length of options given.
Two ways to call and use this function.
#The 1.X version of using the function
import inputapi as inp
inp.menu({params})
#The new way of using numericSerial
import inputapi as inp
inp.menus.singleSelect.numericSerial({params})
The parameters of numericSerial are as follows:
Parameter | data type | default | purpose |
---|---|---|---|
*args | Any | No default | Options that are displayed to the user |
clearOnLoad | boolean | False |
Clears terminal before displaying output |
clearWhenDone | boolean | False |
Clears terminal after user gives input |
title | string | 'Menu' |
What will be displayed as the name/purpose of the menu (---{title}--- ) |
The output of the numericSerial is:
#An example of game option selection is
#Note: Titles have a white background and black text
#but cannot be rendered properly in markdown
---Where do you go?---
1: To the blazing inferno
2: To the tree that reads books
To select an option input the number assigned to it:
>
The errors of numericSerial is the newLineInt error treatment hand handling with the following exception:
#When giving a response not within the given options it will just ignore it
---What animal do you prefer---
1: Cat
2: Dog
To select an option input the number assigned to it:
>3
To select an option input the number assigned to it:
>
The numericIndex menu is almost the same as numericSerial with the exception that instead of starting at 1 instead it starts at 0.
There is only one way to use numericIndex:
import inputapi as inp
inp.menus.singleSelect.numericIndex({params})
The parameters of numericIndex are the same as numericSerial:
Parameter | data type | default | purpose |
---|---|---|---|
*args | Any | No default | Options that are displayed to the user |
clearOnLoad | boolean | False |
Clears terminal before displaying output |
clearWhenDone | boolean | False |
Clears terminal after user gives input |
title | string | 'Menu' |
What will be displayed as the name/purpose of the menu (---{title}--- ) |
This is what the menu would output when called:
#Now lets show what a 'main hub' of a script could look like
---Cool Math Program---
0: Algebra
1: Geometry
2: Calculus
3: Quit
To select an option input the number assigned to it:
>
It's just the same as numericSerial.
The alphabetical menu is the most unique of all singleSelect menus being that instead of using numbers as the selectable options it uses the Modern Latin Alphabet (aka The English Alphabet).
Only one way to use the function:
import inputapi as inp
inp.menus.singleSelect.alphabetical({params})
The parameters are the same as numericSerial and numericIndex:
Parameter | data type | default | purpose |
---|---|---|---|
*args | Any | No default | Options that are displayed to the user |
clearOnLoad | boolean | False |
Clears terminal before displaying output |
clearWhenDone | boolean | False |
Clears terminal after user gives input |
title | string | 'Menu' |
What will be displayed as the name/purpose of the menu (---{title}--- ) |
The output of alphabetical is as follows:
#This could be used for trivia questions
#Also the correct statement is 'C'
---Guess The Correct Statement---
A: The plural word for octopus is octopi
B: Luigi`s last name is Luigi since Mario`s is Mario
C: In a State of the US it is illegal to wear a fake mustache to church if it makes someone laugh
To select an option input the letters next to it:
>
When given more options than character an additional character will be given:
#I can't bother making something creative for this
---Select One---
A: Option 1
B: Option 2
C: Option 3
...
X: Option 24
Y: Option 25
Z: Option 26
AA: Option 27
It is actually the same as numericSerial with the following exception.
#Since it uses different characters for input this is the message for invalid characters
---What is the better number?---
A: 1
B: 2
To select an option input the letters next to it:
>1
WARNING: Input has invalid character [1], allowed characters are ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O' ... 'v', 'w', 'x', 'y', 'z']
>
With multiSelect menus instead of only allowing one option to be selected you can select more than one, hence the name 'multiSelect' and 'singleSelect'.
returnSelected menus return what was selected by the user. (Since the header limit is reached the one menu will have it's documentation outside of it's normal bounds)
numericSerial has near same functionality as singleSelect numericSerial except for a few changes to allow it to be multiSelect.
The only menu there is has only one way to be used.
import inputapi as inp
inp.menus.multiSelect.returnSelected.numericSerial({params})
The parameters for the multiSelect menu are:
Parameter | data type | default | purpose |
---|---|---|---|
*args | Any | No default | Options that are displayed to the user |
confirmChoice | boolean | True |
Ensures that what the user selected was what they intended to select |
showSelectedOnRefresh | boolean | True |
Allows users to see what they selected (See Output for further detail) |
allowNoSelection | boolean | False |
Allows the user to simply select nothing |
clearOnLoad | boolean | False |
Clears terminal before displaying output |
cleanOnRefresh | boolean | False |
Clears terminal when an option is selected |
clearWhenDone | boolean | False |
Clears terminal after user selects options |
title | string | 'Menu' |
What will be displayed as the name/purpose of the menu (---{title}--- ) |
The output in a single state is almost identical to numericSerial but the difference between selections is what will be documented.
#For our first output we will have all values default
#You can replecate this with
#inp.menus.multiSelect.returnSelected.numericSerial('Option 1', 'Option 2', 'Option 3')
---Menu---
1: Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>1
---Menu---
1: *Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: *Option 1
2: *Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>3
---Menu---
1: *Option 1
2: *Option 2
3: *Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: *Option 1
2: Option 2
3: *Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>0
Option 1, Option 3
Confirm choice?
Numeric response allowed (Y=1, n=2)
[Y/n]:1
# This returned ['Option 1', 'Option 3']
But when showSelectedOnRefresh is False
you get a different output.
#This one will be recreated with
#inp.menus.multiSelect.returnSelected.numericSerial('Option 1', 'Option 2', 'Option 3', showSelectedOnRefresh=False)
---Menu---
1: Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>1
---Menu---
1: *Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: *Option 1
2: *Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>3
---Menu---
1: *Option 1
2: *Option 2
3: *Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: *Option 1
2: Option 2
3: *Option 3
0: Done
---Menu---
1: Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>1
---Menu---
1: Option 2
2: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: Option 2
0: Done
Multiple selection menu
To select an option input the number next to it:
---Menu---
1: Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>1
---Menu---
1: Option 2
2: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: Option 2
0: Done
Multiple selection menu
To select an option input the number next to it:
>1
Option 1, Option 2, Option 3
Confirm choice?
Numeric response allowed (Y=1, n=2)
[Y/n]:y
#This returned ['Option 1', 'Option 2', 'Option 3']
With showSelectedOnRefresh disabled it makes correcting errors more difficult and harder to select options but can remove the possibilities of selecting one option twice by accident. Also if all options are selected it will ask if you selected all options without you needing to select Done
.
The only possible errors are the ones from both newLineInt and yesNo since they are what the user inputs through.
returnDeselected is almost the same as returnSelected with the only difference being that instead of returning what the user selected it returns what the user didn't select.
numericSerial assigns numbers to the options from 1 to the length of *args.
There is only one way to use this menu:
import inputapi as inp
inp.menus.multiSelect.returnDeselected.numericSerial({params})
The parameters are identical to returnSelected's parameters.
Parameter | data type | default | purpose |
---|---|---|---|
*args | Any | No default | Options that are displayed to the user |
confirmChoice | boolean | True |
Ensures that what the user selected was what they intended to select |
showSelectedOnRefresh | boolean | True |
Allows users to see what they selected (See Output for further detail) |
allowNoSelection | boolean | False |
Allows the user to simply select nothing |
clearOnLoad | boolean | False |
Clears terminal before displaying output |
cleanOnRefresh | boolean | False |
Clears terminal when an option is selected |
clearWhenDone | boolean | False |
Clears terminal after user selects options |
title | string | 'Menu' |
What will be displayed as the name/purpose of the menu (---{title}--- ) |
The output is identical to returnSelected where the output is the inverse.
---Menu---
1: Option 1
2: Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>2
---Menu---
1: Option 1
2: *Option 2
3: Option 3
0: Done
Multiple selection menu
To select an option input the number next to it:
>0
Option 2
Confirm choice?
Numeric response allowed (Y=1, n=2)
[Y/n]:1
#This returned ['Option 1', 'Option 3']
Refer to returnSelected for error treatment and handling.
RDC (aka Radius Diameter Circumference) leans more on the mathematical part. RDC takes a Radius, Diameter, or Circumference of a circle and returns the radius.
RDC was implemented in the 1.X versions of Input API, this is the most basic RDC in that you need another input for it to work.
There are two ways to use RDC:
#The 1.X way to use RDC
import inputapi as inp
inp.RDC({params})
#The new way to use RDC
import inputapi as inp
inp.RDCs.RDC({params})
Parameter | data type | default | purpose |
---|---|---|---|
length | float | No default | The value to be converted to radius |
clearOnLoad | boolean | False |
Clears terminal before display |
clearWhenDone | boolean | False |
Clears terminal when converted |
The only output RDC has is:
---RDC---
1: Radius
2: Diameter
3: Circumference
To select an option input the number assigned to it:
>
All error treatment and handling is dealt by the menu RDC uses. The only error that can happen is if length is not a number.
newLineRDC doesn't require you to input length, it will simply just request the length with newLineFloat then returns radius.
Only one way to use newLineRDC:
import inputapi as inp
inp.RDCs.newLineRDC({params})
The parameters are:
Parameter | data type | default | purpose |
---|---|---|---|
clearOnLoad | boolean | False |
Clears terminal before display |
clearOnRefresh | boolean | False |
Clears terminal between inputs |
clearWhenDone | boolean | False |
Clears terminal when converted |
The output for newLineRDC isn't much:
#Let's do a radius of 5
RDC:
>5
---RDC---
1: Radius
2: Diameter
3: Circumference
To select an option input the number assigned to it:
>1
# This returned 5.0
The errors for newLineRDC are handled by newLineFloat and singleSelect's numericSerial.
sameLineRDC uses sameLineFloat as the input instead of newLineFloat, this makes the input more compact and takes up less space for input.
There is only one way to use sameLineRDC:
import inputapi as inp
inp.RDCs.sameLineRDC({params})
These are the parameters for sameLineRDC:
Parameter | data type | default | purpose |
---|---|---|---|
clearOnLoad | boolean | False |
Clears terminal before display |
clearOnRefresh | boolean | False |
Clears terminal between inputs |
clearWhenDone | boolean | False |
Clears terminal when converted |
The output for sameLineRDC is as follows:
#Lets do a diameter of 26
RDC=26
---RDC---
1: Radius
2: Diameter
3: Circumference
To select an option input the number assigned to it:
>2
#This returned 13.0
All possible errors happen through sameLineFloat and singleSelect's numericSerial.
otherFunc is what features the Input API has but isn't actually input. Just about all inputs require otherFunc to work just enough to keep afloat. You don't need to master this to actually be good at using Input API.
With pause it will stop the program until a user presses ENTER, this is really simple and is never used by any input.
standard is the first pause function and just pauses with no other noteworthy functionality.
There are two ways to use standard:
#This one actually makes sense
import inputapi as inp
inp.pause({params})
#The reimagined way
import inputapi as inp
inp.otherFunc.pause.standard({params})
The parameters for standard is:
Parameter | data type | default | purpose |
---|---|---|---|
clearOnLoad | boolean | False |
Clears terminal before pausing |
clearWhenDone | boolean | False |
Clears terminal when resuming |
The output is really lackluster:
#Anything can be printed before pausing
#If clearOnLoad is False
Press ENTER to continue:
If you can find an error that can only happen with standard... congrats at breaking the most simple function the Input API has.
This works almost identical to standard but allows you to print values while having clearOnLoad True
.
Only one way to use showValue:
import inputapi as inp
inp.otherFunc.pause.showValue({params})
The parameters for showValue are:
Parameter | data type | default | purpose |
---|---|---|---|
*values | Any | No default | The values to be displayed on pause |
clearOnLoad | boolean | False |
Clears terminal before pausing |
clearWhenDone | boolean | False |
Clears terminal when resuming |
The errors that can happen are un-treatable and will break the script.
clearScreen is the only part of Input API that is depended to the point of breaking. If clearScreen breaks then everything that wants to clear the terminal will break. clearScreen will only work with Windows, MacOS, and Linux based systems. Anything else is either lucky enough to be under the umbrella or incompatible entirely.
auto will clear the terminal while not needing info on the OS. It instead will ask Python through os
to get info on the OS to use the right command to clear the terminal. This function has no parameters or errors, so you can just call it and nothing bad will happen.
Two ways to use auto:
#The original way
import inputapi as inp
inp.clearScreen()
#The way Input API uses auto
inp.otherFunc.clearScreen.auto()
manual clearScreen gives you the ability to clear the terminal for one OS and not all. Though MacOS and Linux based systems use the same command for their clearing, it's more allow Windows to clear or allow the other two to clear.
Only one way to use manual:
import inputapi as inp
inp.otherFunc.clearScreen.manual({params})
The parameter for manual is:
Parameter | data type | default | purpose |
---|---|---|---|
OS | string | No default | What OS you want to be cleared |
When calling the function you have to tell manual the name of the OS. While it requires the spelling to be correct, manual isn't case sensitive.
#Lets go through every OS supported
import inputapi as inp
#This will make using manual easier to call
manual = inp.otherFunc.clearScreen.manual
#Windows
manual(OS='Windows')
#MacOS
manual(OS='MACOS')
#Linux based systems
manual(OS='linux')
#This will raise an error
manual(OS='WindowsOS')
#This too will raise an error
manual()
This will be some things to expect to happen as Input API gets more updates. With some coming in the next major update, others in the minor updates.
These will be added in the next major updates, some may appear in the same, others might be the only change.
NOTE: These ideas are in no order whatsoever, any combination of when these will be added is possible
Pre-set Inputs will be a way to have a series of inputs with the parameters set. A file with inputs could be read by the Input API and saved as a readable file or a compressed format to save on space. There will also be some built-in formats to make easier inputs too. This will be compatible with both older and more recent updates.
import inputapi as inp
#So instead of having something like this
inp.numeric.newLineFloat('Gimmie a number:')
inp.boolean.yesNo('Are you sure?')
#You could
fileInput = inp.presets.readFrom('/File.ext') #File extension will be given (.ext isn't it)
fileInput.getNumber()
#Or maybe
inp.presets.number.intWithConfirm('Gimmie a number', 'Are you sure?') #This is not a guarantee to be added to presets (At least not in this exact way)
This update will allow input for multiple elements of a list or array. This will allow multiple types of inputs in one.
import inputapi as inp
#This could be how the input could be used
inp.iterables.list.newLineList()
inp.iterables.tuple.sameLineTuple()
inp.iterables.set.newLineSet()
This function will allow you to determine multiple factors throughout Input API, maybe relax some restrictions or make some more strict. More power will be given to you when receiving input when this comes to existence.
import inputapi as inp
#Could be this
inp.config({params})
#Maybe this
inp.strings.config({params})
#Or this
inp.config.var = {Var}
#There is some other things tht could be how config works but this is still possible
This update will allow you to have a set of inputs within the script to be called over time and not when the line is read.
import inputapi as inp
#So instead of input being here
inp.strings.newLineStr() #This is the input *function*
#You could save it
input = inp.strings.NewLineStr({params}) #This could be a class (Name is a placeholder)
#Then call it later
if __name__ == '__main__':
input()
#Or repeat the input
while True:
user = input()
print(user)
if user == 'Quit':
break
These updates will be added between major updates or with them.
There will be new multiSelect menu types for both returnSelected and returnDeselected. These types will be from the singleSelect menus.
There will be a new singleSelect menu type that will be exclusive to singleSelect. This will have the same output as a multiSelect menu but won't let you select multiple.
There will be a changelog that will list every change from 1.0.0 to the latest update. This will be publicly available to all distributions and to the GitHub repository.
Footnotes
-
clearScreen is does not need any information about the OS to work4 It will automatically tell what OS is running the code and use the command used by the OS to clear the terminal ↩
-
Fun Fact's although being labeled as "Fun" is more fact than fun and could be a "Sad Fact" depending on the fact! ↩
-
Not everything, just what takes input, clearScreen and pause don't rely on it working to function ↩
-
Feature is only supported on Windows, MacOS, and Linux based systems ↩