That’s a very general question. Surely the answer will depend on exactly what you want to automate. If you could provide some more details of what you are trying to achieve, I’m sure someone here will give you an answer.
I have to take specific data from a source that is produced in a CSV file.
I will remove the first line as it is headers. Then I have to remove all colons, semi-colons, single quotes as well as double quotes. Re-save the file.
Next is I have to load a specialized 3rd party software that will convert that data into a proprietary format.
Next I have to use WinSCP login to a specific server. Transfer the data file to a certain folder. Then I have to start a Putty session to my server and run a proprietary server script.
That’s it in a nutshell.
I have used other automation platforms before but they are focused on web processing (botting) and they do not work well for automating Windows-based applications.
Today I did more research and found a few others; AutoHotKey, Sikuli, Do It Again, Macro Express and Keyboard Express along with your suggestion of AutoIt.
I think I have that marked in my top two.
WinAutomation looks great but the price is also waaaaaay out there and work won’t pay for it.
I used to work on game bots a lot as a hobby in college, if that counts.
AutoIt was a pretty good macro language, probably the most mature out there. I’ve seen people do some pretty amazing things with it and I took it to levels it probably never should have gone to. It was partly BASIC partly PHP inspired.
I would probably do what you’re wanting to do in some other scripting language, it doesn’t sound like you need to simulate mouse clicks? I could be wrong. If not it would be a good reason to learn Python.
The 3rd party software which converts the data does not have any keyboard shortcuts so I do have a few mouse-clicks. I will probably use AutoIt.
Python is definitely on my radar to learn. I just need more spare time. LOL
I am considered to be an xpert in “You bot” (if you catch my drift there). lol I really pushed that to the limits…I once created a script to automate MS Paint. It was pretty cool. I mainly just tinker and teach it…no BH stuff.
If you need to read the screen and can’t get a hang of the color matching in AutoIt, try AutoHotkey. It’s a fork of AutoIt but it has image matching instead of Color Matching… which in reality is just a lot of color matching. It’s much slower. Just try to find enough points that don’t change and put in fuzzy logic. (Like match 3 out of 5 points)
Python is definitely on my radar to learn. I just need more spare time. LOL
That’s exactly where I am with Python and Node. I wrote a couple small Python scripts and liked it.
I needed a way to process lots of DOS command in addition to some 3rd party DOS apps so I have been disappoint so far in that AutoIt does not have a way to process within a visible window. Apparently it can with a hidden window but that will work against me.
So I will work with part of the project that is GUI based.
Interesting to hear people’s different experiences. I think Sikuli works extremely well. (Note that there’s a stable version (1.0.1) and a beta version (1.1.0). I’ve only ever used the stable version.)
Sikuli is screenshot based, so it can automate anything on the screen, doesn’t matter what kind of application. Its documentation is excellent. It uses Python as its scripting language, so you have the whole language and libraries at your disposal. It’s X-OS. I can personally confirm that it works well in both Windows and Mac, and it’s supposed to work in Linux too. And even if you need to automate some other OS, that’s still doable because Sikuli can interact with a VM or a VNC viewer. They’re just pictures on the screen, after all.
Interesting. When I was trying to use it daily (about 4 years ago), it just wasn’t reliable for me (on Windows). It struggled finding the images of the application I was trying to automate. I ultimately gave up with any Graphical Tester and developed an API for the presentation layer to use. I could write tests against the API and expect it to behave consistently giving the Presentation layer a much needed separation (which ultimately made it easier to test).
Then over the past 2 years, I’ve been playing with automation of graphical interfaces again (just off and on), and although I gave Sikuli a second chance, it just never really outlived the problems I remembered 4 years ago. Ultimately I eventually dropped all automation for GUIs as they all were too cumbersome/unreliable for what I needed – if I had to still watch over them, then they weren’t going to do the job I needed.
I still think Sikuli was a great project, just didn’t fit into my path. I’m thrilled someone else has had better experience with it than I. That’s good to hear.
It also looks like Sikuli is back under active development. For a while I was fairly certain it stalled and wasn’t really be developed, but just maintained. I may have to give it another shot sometime (should I ever have a GUI that warrants automated testing).
That’s the brilliant side of testing by being screenshot based… The bad side is that it uses test server’s screen resolution to find pixel perfect picture. So, if you wrote the script on a different resolution screen then you’re out of luck.
Hasn’t been a problem for me so far. That was something I made a point to check. On the machine running the Sikuli automation, I once deliberately lowered the screen resolution just to see what it would do. It went on like everything was normal. The only exception is if the resolution is so low that windows end up off the edge of the screen. For obvious reasons, things still need to be visible.
But it’s entirely possible the resolution thing used to be more of an issue. I just started using Sikuli about a year ago. Sounds like cpradio and others tried using it several years earlier. It probably wasn’t as robust back then.
I wrote a couple small Python scripts and liked it.