What: Added AUR Package

Why: It can be installed from the AUR directly.

How: Creating the PKGBUILD

Checklist:

• [x] Documentation added
• [ ] Tests added
• [ ] No linting errors / broken tests
• [x] Merge ready

Summary

This PR adds support for Wikipedia!!

I added an option, -W, which will select a random article from Wikipedia and then make you type the summary of the article.

I tried to add all the logic into its own file, to avoid cluttering the main.rs file.

Changes

• Cargo.toml → Added dependency reqwest to make requests to Wikipedia's API
• main.rs → Added the module wiki, imported the appropriate function and added code to have a flag and check if the flag is present. The call to random_article() is a bit bloated because wikipedia can sometimes return an empty article (rarely), so I had to do a match (although most errors are caught inside the wiki module).
• wiki.rs → Contains all the structs and the two functions to make the requests. It's honestly very simple, but seems complicated because all the structs and all the parsing...

Why

I think this is a cool addition to the program! Typing real-world text is better than randomly generated chains of words.

In the future, an option could be added to change the article's language, or for the user to be able to provide an article title and select that one.

How

With reqwest! Wikipedia has an awesome API (here is the documentation), so I just tell it to give me a random article and then make a summary of that article. That summary is what will be displayed.

Really all the complexity is because all the error checking and parsing.

Checklist:

• [x] Allow edits from maintainers option checked
• [x] Branch name is prefixed with [your_username]/ (ex. coloradocolby/featureX)
• [ ] Documentation added
• [ ] Tests added
• [x] No failing actions
• [x] Merge ready

I tried to comment all the relevant code. I would love feedback on how you would like to handle the errors when parsing or making a request.

Hope you like this addition! 😄

Problem

When I press Ctrl+Backspace, I'd expect the last word to be deleted – this is the behavior of most text input boxes, and Monkeytype. Instead, it adds a new character.

Solution

I don't know exactly how to implement this feature. It could be an optional argument in the CLI.

Additional context

Video demonstrating the behavior. I mistype the word "possible" and attempt to delete the whole word.

https://user-images.githubusercontent.com/62364808/166745675-61dd4d1f-3972-4353-b151-3eec35ea3da9.mp4

… to indicate that the user wishes to type-test full, grammatical sentences.

What: Added -f as a flag to allow for the generation of random sentences instead of just words.

Why: Personally, I felt that thokr's random words didn't really help certain typists as much (such as myself) due to the lack of structure. While nonsensical, the sentences generated by thokr with this new flag are grammatical, and have the necessary full stops and spaces.

How: Through use of cgisf_lib. I shamelessly reused one of my own side-projects to expedite the process of sentence generation.

Checklist:

• [x] Allow edits from maintainers option checked
• [x] Documentation added
• [ ] Tests added N/A
• [x] No failing actions
• [x] Merge ready

• Do note that get_random_sentence() has to return two results because there would be no other way to get the necessary word count for analysis later on.
• I have also manually tested it myself and everything seems to be working.

What: Fix process always panicking on exit on Windows.

Why: This bug isn't critical as it mainly occurs when the process is already exiting, but I think it would be better for the process to exit nominally instead of panicking so as not to confuse the user.

How: The problem is caused by resize event arriving when the main thread and, therefore, the receiver end of the channel is already dead. All I did was replacing unwrap with a conditional break. Even though the problem presents itself only with the resize event at the moment, I figured it would be better and more consistent to change all three cases to a conditional break. Another option would be to return the join handles main and join them in the end, but since the code inside both threads is just an infinite cycle, breaking seems like a more straightforward and cleaner choice.

Checklist:

• [ ] Documentation added
• [ ] Tests added
• [x] No linting errors / broken tests
• [x] Merge ready

What: Logging historical data to $XDG_DATA_HOME/thokr.log

Why: So nerds like me can plot progress over time

How: Results are logged whenever results are calculated. It may be smart to add a configuration to disable this, but that could be another comit

Checklist: N/A

• [x] Documentation added
• [ ] Tests added
• [x] No linting errors / broken tests
• [x] Merge ready

Description

After running a typing test and pressing t, nothing happens.

To Reproduce

1. Run in Docker (docker run -it coloradocolby/thokr)
2. Complete a typing test
3. Press t

Expected behavior

I would expect some sort of Twitter-friendly text to copy paste or Twitter opens up with the tweet pre-filled.

Environment

• OS: macOS
• Terminal Emulator: xterm
• Crate version: 0.1.0

Description

When a multiple-byte codepoint is rendered in the prompt, thokr panics and dies.

To Reproduce

Run thokr -p "ä" Observe the panic.

Expected behavior

Thokr should render the character properly, without panicking.

Environment

• OS: Devuan GNU/Linux Ceres on amd64
• Terminal Emulator: gnome-terminal
• Font: JetBrains Mono Nerd Font
• Crate version: 0.3.0

Additional context

I suspect the source of the panic can be found on line 64 in thok.rs. https://github.com/thatvegandev/thokr/blob/9453468032cae753570b660146103abd55609df2/src/thok.rs#L64

Description

When I start a fresh terminal and run thokr for the first time it doesn't load any graphic until I run it again

To Reproduce

Run thokr in a fresh warp terminal.

Expected behavior

It's expected that it shows some graphics, isn't it?

Screenshots

first run in a fresh terminal: after retry:

video:

https://user-images.githubusercontent.com/39351332/175514836-d22ee8f4-8641-4ca4-83cb-bdaa07ecdcca.mov

Environment

• OS: macOS Monterey 12.4 (21F79)
• Terminal Emulator: warp
• Font: hack (warp default)
• Thokr Version: 0.4.0

Additional context

I'm not an English speaker so sorry for any language mistake...

In iTerm it runs normally...

Description

I have a test file with around 50 words that have this structure : (word in each line with no space at the end)

dependencies
visualization
description
parse
duplicates
possible
nootropics
scissors

now when i run thokr -p "$(cat Notes/typing/words)" the app show like this : ( wiht no spaces between words)

dependenciesvisualizationdescriptionparseduplicatespossible
nootropicsscissors

To Reproduce

I tried to add space after each word in the text file and it worded almost : the problem i had is that i need to type twice for each

Expected behavior

To show the words with spaces

Environment

• OS: Linux Fedora 36 KDE
• Terminal Emulator: kitty
• Font: JetBrainsMono Nerd Font
• Thokr Version: 0.3.0

What: Remove tweet functionality if no browser is available.

Why: Makes it so users running thokr in a docker container aren't shown a tweet option that won't work.

This How: Uses the newly updated 0.7.1 version of webbrowser-rs which exposes Browser::is_available().

Checklist:

• [x] Documentation added
• [ ] Tests added
• [x] No linting errors / broken tests
• [x] Merge ready

This PR attempts to support multiple languages by implementing grapheme cluster inputs and comparison using the unicode-segmentation crate. I've tested it with Finnish prompts (which can contain letters with diacritical marks, such as ä and ö), and they seem to work fine (under fish-shell on tmux, anyways).

Hi ! Thanks for creating thokr, i'm having fun playing it in random terminals :)

My only issue is that I struggle noticing where my caret is sometimes (when i'm trying to see whether the key i hit registered or not); the underline is not always visible enough to me.

It would be cool to see an option to use different caret types (block, beam), perhaps even following current terminal settings ! (That could perhaps expand into cool fluff like pace carets like the other issue describes as well).

Problem

There's no way to visualize historical test data, such as WPM and accuracy.

Solution

I would love to help write some code (in Python, if you're alright with that) to generate graphs that show average WPM and accuracy over time, for example. The logs would have to record the game mode for this to work. Let me know what you think.

Alternatives

A simple statistics summary page could show one's PB for each game mode, total time spent typing, etc.

Additional context

Statistics graph in Monkeytype.

Problem

One of my favorite features of Monkeytype is called the "pace caret":

Displays a second caret that moves at constant speed. The 'average' option averages the speed of last 10 results.

This gives me a good idea of how fast I'm typing in each trial and helps me keep pace.

Solution

A pace caret feature in thokr could include "PB", "average", and "custom" WPM options, and would work just like in Monkeytype.

Alternatives

A live WPM counter could serve a similar purpose but is hard to pay attention to when focusing on typing. Also, it only indicates how fast I'm typing in the moment, and doesn't give me any idea how well I've been doing throughout the entire test (how likely I am to get a PR).

Additional context

Pace caret demo in Monkeytype. The underline caret moves at a consistent 130 WPM, but jumps when I miss a word.

https://user-images.githubusercontent.com/62364808/166748602-748a802f-a758-4075-8918-af2dcf3a9568.mp4

What

This PR aims to address #6. It adds a scrolling functionality to the text field, so that the user isn't overwhelmed by having all the words displayed at once.

Why

Besides being a better user experience, rendering fewer words at a single time makes the application faster since there aren't as many characters to render/keep track of.

How

The difficult part was figuring out when the user is at the end of a line. I don't think there's a way to do this using rust-tui so I've had to do the math myself. This logic is present in the get_skip_count function. Once we have calculated the amount of characters to skip, the implementation is relatively straightforward.

Checklist

• [ ] Documentation added
• [ ] Tests added
• [ ] No linting errors / broken tests
• [ ] Merge ready

Once this is merged, I think it would be nice to increase the HORIZONTAL_MARGIN (dynamically) to center the text. It would result in a better UI in my opinion.

Problem

Presently, when you pass -s to set a time limit, the test doesn't run for that amount of time. Instead, it runs until the you've typed all the words(which is 15 by default)

Solution

I really like 60 seconds typing tests and on something like monkeytype, the number of words isn't static - if you're typing fast, it increases the number of words automatically. There are 3 lines and as soon as you complete the 2nd line, a new line of words is added. So you never reach the last line.

This makes more sense to me.

Alternatives

However, I have no clue if doing that is feasible in a terminal. If it isn't, I think doing something similar to →(but without resetting the timer) would be nice.

Of course, I can do something like -s 60 -w 2000 but having so many words displayed at once makes it much more difficult for me to focus.