= What is the BotGames?
The BotGames was a series of RPA Challenges promoted by Automation Anywhere (AA). This event had a considerable scope. It went from Windows to API automation, which helps any RPA Developer to improve.
= What about this repository?
This repository holds all the solutions I created to solve the challenges. To run it, you will need to configure your environment.
= Ok, what do I need to configure?
To install all the dependencies, please follow these steps:
== 1. Python configuration
You will need the following Python packages:
Also, you will need tesseract installed in this path: C:/Program Files/Tesseract-OCR/tesseract.exe. You can find this program https://github.com/UB-Mannheim/tesseract/wiki[here].
== 2. TagUI configuration
Regarding the RPA tool, TagUI, you can follow the steps described https://tagui.readthedocs.io/en/latest/setup.html[here].
After you install it, you can improve the performance by following https://github.com/tebelorg/RPA-Python/issues/120#issuecomment-610532082[this] issue on GitHub, so you should reach similar results.
Also, I modified the tagui_config.txt file on the \tagui\src folder to make sure the run command for TagUI does not timeout with VBScript execution.
== 3. Excel configuration
No configuration is needed. You only need Excel installed on your machine.
= Great. What about the challenges?
You can find the description of the challenges on the AA website. Until the day 26/Aug/2021, the websites are:
I have some thoughts and considerations about each one of the challenges.
== Week 1 and 2: Understanding your weapons
TagUI is not the tool I was most familiar with. I decided to pick it to know how it performs in practice, and the two first weeks served it well for this.
This tool behaves like a human, so you need some modifications to run at faster speeds. From my experience, most of the websites can handle unhuman interaction speed, besides some fields and loading screens. So I changed the TagUI files to achieve better performance.
Another point about the tool is the documentation. Until this point, part of it exists as GitHub Issues, but not on the main one. It is an ongoing task.
Besides this, TagUI was able to complete, by itself, the Week 1 challenge. The Week 2 problem required a little help from VBScript since this RPA tool can only deal with desktop apps by computer vision.
Another point is: I refreshed the page after setting up the bot for the challenge. I did this because interacting with different pages is only possible for TagUI if the second page is accessible from the first one, considerably increasing the number of seconds to complete the challenge.
This discussion leads me to further decisions for the ongoing weeks: https://github.com/tebelorg/RPA-Python[use the Python version].
== Week 3: Saving Private Ryan, with Python
I believe you do not need introductions for this. The swiss knife of programming, AKA Python, was used to automate on Windows. The piwinauto package is more optimal for this type of task.
It is good to point: Python alone can be cumbersome to check the field names, so if you follow a similar approach in your projects, I recommend you to use https://accessibilityinsights.io/docs/en/windows/overview/[Accessibility Insights for Windows]. This program brings the information from the Windows apps fields.
=== What about TagUI?
On this challenge, TagUI performed a secondary but needed role. This situation happened because I could not do the same speed improvement configurations as I did for Week 1 and Week 2. In this case, I chose to use the input filling with JavaScript directly, accessing DOM using TagUI.
== Week 4: The Good, the Bad and the Ugly
I have mixed feelings about this challenge. From one point, I was able to use OCR with Tesseract, this is The Good. t the same time, I did not reach this technology's full potential due to my lack of knowledge in this field, and here we have The Bad.
To fill this knowledge gap, I used different queries in REGEX o parse the content of each invoice, and this, my friends, is The Ugly.
Some of you can think: Why? It solves the issue, right?
[quote,The programmer who quit]
If it ain't broke, don't fix it
Yes, it is solved, but this is one aspect of programming. As a programmer, I need to think in other aspects, such as:
Following these aspects, I do not see REGEX as the core of a solution for most occasions. In real life, I would only use this on one or more of those occasions:
The reason is simple. If you are a programmer who used REGEX at some point, I bet you can remember what the query means after a day. REGEX is hard to read and maintain if you do not work with this daily.
To mitigate this issue, I separated the queries per function. Also, I put each REGEX query in a variable, so the variable’s name describes what the query does.
If you use REGEX, I strongly recommend you do similar things. REGEX without documentation can become a nightmare fast.
= Conclusion
First of all, thank you for your attention and your time reading this. I will use the results from these challenges to guide my next steps. If you want to test your skills as an RPA Developer, I recommend trying this out!