Improve library to use region and not full screen for speed and accuracy

[rajeshtaneja]

While using this library over jython based approach, there are few hurdles:

1. Current implementation use full screen, it would be nice to use region for fast response.
2. No way to pass similarity to image, hence end up Setting minSimilarity
3. Limited exposed Sikuli api's.
4. Still using 1.1.2, would be nice to upgrade to 1.1.4
5. in pom.xml python3 is specified, whereas default python install create python.exe and not python3.exe

[rainmanwy]

@rajeshtaneja , thanks for your suggestion. Now i have updated sikulixapi version to 1.1.4.

For 1: So you mean user could define region, right? For 2, Yes, currectly user could setting minSimilarity, it is a global setting. For 3, This library do not expose sikuli api directly. Could you provid a keyword list, and i could check how to implement. For 4, it is upated to 1.1.4 For 5, I use linux to develop this library, and i have python2 and python3 in my env. You may change it to python when you package this library youself.

[rajeshtaneja]

Thanks for the quick reply @rainmanwy . Really appreciate your work :)

1. That is correct, rather than using screen, we should use region.
2. In attached patch, user can send image.png = 0.85 that will find a pattern with image.png having a similarity of 0.85
3. I have attached a patch which has set of keywords which might be useful. Some of them are:

• click on nth element
• Get image coordinates
• Allow ROI
• Search in ROI
• Allow highlight of ROI
• Each match, should return coordinates, so they can be used if needed. I am using it to save image region on screen, and update them automatically showing user the difference from what has changed.
• Paste text has incorrect condition check.
• Click on region without a pattern/image passed.
• Application open and close (Using python is not really a good thing for heavy applications)

4. 👍
5. 👍

In attached pull request it exposes lot of Sikuli api's and return appropriate values for ROI or match with optional similarity. IMHO it will be helpful for users.

[rainmanwy]

Wow, Thank you @rajeshtaneja . I have checked your PR, looks very good. I will have a try in my env later.

Thank you again!

[rainmanwy]

Thank you @rajeshtaneja !

[rajeshtaneja]

Pleasure is all mine :)

On Wed, Nov 14, 2018 at 8:51 AM Rainmanwy [email protected] wrote:

Thank you @rajeshtaneja https://github.com/rajeshtaneja !

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/rainmanwy/robotframework-SikuliLibrary/issues/79#issuecomment-438495824, or mute the thread https://github.com/notifications/unsubscribe-auth/AAxGh1FOlIxY4Z0z6mulBrExKh0Gb8nmks5uu2j-gaJpZM4X6oUm .