Improvements to keypad library

[regicidalplutophage]

Some feature I find missing from keypad library:

• There's no way to use normally closed switches without external pullups.
  In keypad.Keys if pull=True, the pin is always pulled to the opposite of value_when_pressed which makes it so that the only way to use normally closed switch is with keypad.Keys and external pullups.
  Allow the use of normally closed switches in all the classes;
• Add support for SPI bus to be used for keypad.ShiftRegisterKeys;
• Add support for more configurations of shift registers to be used, mainly a shift register matrix with GPIO outputs and shift register inputs;
• Add support for Seesaw.

That's all I can think of for now.
Thanks!

[s-light]

@regicidalplutophage to 1.: so we would need to have pull=UP and pull=DOWN as option?!
and a (small) change to line 54 & 55 in Keys.c

in this regard i would like to add a wish:
allow for arrays for pull and value_when_pressed - so we can use the same object for all three buttons on the [Adafruit ESP32-S* Reverse TFT Feather](https://www.adafruit.com/search?q=tft+reverse)
(Button D0 is opposite to D1 & D2 for the sake of deep-sleep wake-up..)
maybe there are other boards with similar setups...

is it good to extract each request in its own issue?

[regicidalplutophage]

@s-light

so we would need to have pull=UP and pull=DOWN as option?!

That would solve the usage of normally closed switches with keypad.Keys.
For matrices pressed and released would have to be inverted and same for shift register.

is it good to extract each request in its own issue?

I don't have an opinion on that.

[dhalbert]

Please do split issues up, since they can be fixed one at a time.

so we would need to have pull=UP and pull=DOWN as option?!

Instead of that, I think I would add a normally_open=True optional argument, to handle normally closed switches.

I'm not sure we want to add lots more features to keypad, because it makes it larger and it might stop fitting on smaller builds. But adding some kind of access to the Event queues would make it possible to do some of this in Python instead of in C.

[regicidalplutophage]

Before I start opening new issues, let's think if one change can solve a bunch of problems at once.
Then I think keypad should at least take digitalio.DigitalInOut, adafruit_tca8418.DigitalInOut, adafruit_74hc595.DigitalInOut and the like.
This way we can remove shift register scanner from keypad library altogether without losing shift register support and still gain support for shift register or mixed matrices.

[kolkmvd]

Add support for more configurations of shift registers to be used, mainly a shift register matrix with GPIO outputs and shift register inputs;

How do you mean this? I have a board where I have 4 HC165 in parallel, sharing the same CLOCK and LATCH, but each on a different DATA. To make this supported I was thinking about adding a function ShiftRegister.addDataPin(data: microcontroller.Pin, keycount: int) that would resize scanarrays and increase the key_count after the initial set-up. the data and key_count arguments would be extended to allow a list as well.

Would this interfere with you suggestion?

[kolkmvd]

Created #8143, which allows the data and key_count arguments to be a either a single pin as currently, but also a list as well.

[s-light]

i created two feature-requests out of these comments here..
i think - as @regicidalplutophage already formulated - especially #8316 would solve a bunch of things..
also half of #8315