Add Base64 encoded screenshot and additional screenshot creation from XPATH element

[HackXIt]

This PR adds another logging function, to directly embed screenshots as base64 encoded strings in the output log.

It also expands the screenshot functionality to allow for screenshot creation based on a provided element XPATH.
It will create a screenshot based on that element, so a screenshot of a window will contain the whole window, a screenshot of a button, will only contain that button.

Fixes #192 and fixes #193

[Nepitwin]

Please update Changelog file to upcoming unreleased new keywords.

[Nepitwin]

Is it possible to extend some Screenshots tests under ATests to see if log are resolved?

(Optional would be nice)

[Nepitwin]

Probably a switch would be nice to decide default behavior from screenshots to create files or base64 like

Set Screenshot Log Mode (Base64 | File)

[HackXIt]

I'll get to address the changes right away, please keep in mind that I noticed that my commits made were using the wrong user configuration by mistake.
I have rebased the commits since then to overwrite the author with my correct information.

It should be visible in a moment.

[HackXIt]

This will always create a screenshot, even if the element could not be found. If the element is not found, the whole screen is captured like in default.
Might be better if the screenshot will not get created if the element was not found?

[HackXIt]

I'm done from my point of view, please have another look @Nepitwin

[HackXIt]

It triggers me hard that they are not working.

I should be better at this ;D

[Nepitwin]

@HackXIt

Feel free this is the challenge of testing to execute them correctly on a testing system 📦

[HackXIt]

@Nepitwin I do need some help on this last test case. I do not know what runs differently on appveyor.

When I run the test locally, it creates the screenshot file and finds it correctly, but on appveyor it doesn't. I changed the format to use ${TEST_NAME} instead of hardcoding, but that didn't change results.

By the way, if you also have some suggestions on further necessary test cases, I can add those as well, but I believe with the feature being so small, the ones created should suffice.
Here's a list of the new ones:

• Take Screenshot As Base64
• Take Screenshot Of Window (this is the failing one)
• Take Screenshot Of Window As Base64

Attached is my local log file of the successful execution of that test case. I did not run it via the keen.bat, since it executes quite a lot of test cases and I do not wish to run the full suite to verify a single one. It would be nice to have a mechanism in keen.bat to at least run suites seperately, but that's a change for another day.
result_take_screenshot_of_window.zip

Execution log

C:; cd 'C:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui\atests'; & 'C:\Users\RINI\AppData\Local\Temp\rf-ls-run\run_env_00_38e9sp6_.bat' '-u' 'c:\Users\RINI\.vscode\extensions\robocorp.robotframework-lsp-1.12.0\src\robotframework_debug_adapter\run_robot__main__.py' '--port' '7111' '--no-debug' '--listener=robotframework_debug_adapter.events_listener.EventsListenerV2' '--listener=robotframework_debug_adapter.events_listener.EventsListenerV3' '--outputdir' 'C:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui/result/tmp' '--pythonpath' 'c:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui/.venv/lib/python3.8/site-packages' '--pythonpath' 'c:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui/atests' '--variable' 'UIA:UIA3' '--prerunmodifier=robotframework_debug_adapter.prerun_modifiers.FilteringTestsSuiteVisitor' 'c:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui\atests\Screenshot.robot' s\x5cRINI\x5c#Entwicklung_TechNICK\x5c0_Git_Stash\x5crobotframework-flaui\x5catests\x5cScreenshot.robot' ============================================================================== Screenshot :: Test suite for screenshot keywords. XPath not found error han... ============================================================================== Take Screenshot Of Window | PASS | ------------------------------------------------------------------------------ Screenshot :: Test suite for screenshot keywords. XPath not found ... | PASS | 1 test, 1 passed, 0 failed ============================================================================== Output: C:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui\result\tmp\output.xml Log: C:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui\result\tmp\log.html Report: C:\Users\RINI\#Entwicklung_TechNICK\0_Git_Stash\robotframework-flaui\result\tmp\report.html

[HackXIt]

It would be better to always have it in the teardown, as currently it will not be executed on failure for the test that has to also stop the application. I'll add a keyword I think to combine the two.

[ghost]

@Nepitwin I do need some help on this last test case. I do not know what runs differently on appveyor.

When I run the test locally, it creates the screenshot file and finds it correctly, but on appveyor it doesn't. I changed the format to use ${TEST_NAME} instead of hardcoding, but that didn't change results.

By the way, if you also have some suggestions on further necessary test cases, I can add those as well, but I believe with the feature being so small, the ones created should suffice. Here's a list of the new ones:

• Take Screenshot As Base64
• Take Screenshot Of Window (this is the failing one)
• Take Screenshot Of Window As Base64

Attached is my local log file of the successful execution of that test case. I did not run it via the keen.bat, since it executes quite a lot of test cases and I do not wish to run the full suite to verify a single one. It would be nice to have a mechanism in keen.bat to at least run suites seperately, but that's a change for another day. result_take_screenshot_of_window.zip

Execution log

@HackXIt

Run a case in a robotfile, you can execute command like

.\keen.bat install

cd .\atests\

python -m robot --name "UIA3" --variable UIA:UIA3 --outputdir ../result/uia3 --test "Take Screenshot Of Window" .\Screenshot.robot

--test is the casename you edit in robotfile

[HackXIt]

There seems to be an unrelated problem on AppVeyor. The last commit had all tests successful, but the script had a failure somewhere. See console log for details on AppVeyor.

...
5 File(s) copied
Command exited with code 28
$wc = New-Object 'System.Net.WebClient'
$wc.UploadFile("https://ci.appveyor.com/api/testresults/junit/$($env:APPVEYOR_JOB_ID)", (Resolve-Path .\result\xunit.xml))

[ghost]

There seems to be an unrelated problem on AppVeyor. The last commit had all tests successful, but the script had a failure somewhere. See console log for details on AppVeyor.

...
5 File(s) copied
Command exited with code 28
$wc = New-Object 'System.Net.WebClient'
$wc.UploadFile("https://ci.appveyor.com/api/testresults/junit/$($env:APPVEYOR_JOB_ID)", (Resolve-Path .\result\xunit.xml))

@HackXIt

I debuged that, job "call:pylint" in keen.bat maybe retrurn the %ERRORLEVEL%=28, so the %result%=28 and it donnot change in left jobs, so keen.bat ended with EXIT /B %result%=28. EXIT code not be 0, it is marked as a Failure.

I add the attachment debug-flaui.zip with debugcode and EXIT code with 0
tip: use keen.bat in attachment and command ".\keen.bat test" will run only ./Screenshot.robot, no other robotfile

EXIT code with 28 in log, job "call:pylint" and job "tidy" print like

EXIT code with 0, should print like

[HackXIt]

@JimRevolutionist

Alright, applied the changes to keen.bat. (only the changes relating to errorlevel)

I also adjusted all screenshot test cases with a proper teardown keyword for resetting to default state, which should also run on error to not influence other test cases.

I also ran the screenshot suite beforehand, should all be green.

Fingers crossed, we should be good to go for merge after this.

Do you want me to create issues for the keen.bat so we can reference it in the PR?

I should also update Changelog to reflect the minor adjustments made.

[ghost]

@JimRevolutionist

Alright, applied the changes to keen.bat. (only the changes relating to errorlevel)

I also adjusted all screenshot test cases with a proper teardown keyword for resetting to default state, which should also run on error to not influence other test cases.

I also ran the screenshot suite beforehand, should all be green.

Fingers crossed, we should be good to go for merge after this.

Do you want me to create issues for the keen.bat so we can reference it in the PR?

I should also update Changelog to reflect the minor adjustments made.

@HackXIt

optmize keen.bat should be an Independent issue,pr separate from screenshot issue,pr. We should not put keenbat commits into screenshot pr.

[HackXIt]

Oh. I'll revert the change then.

[Nepitwin]

@HackXIt

Oh yes i forgot to update all related verifiction from pylint to see in error logs from appveyor. Sorry for this mistake. If you got already an solution you can create a new pr from it.

Because all issues can be seen afterwards in artifact files to guarantee python code style guides and robotframework has to be fixxed.

@JimRevolutionist

Sometimes api call failed from appveyor --> Command exited with code 28

It's not nice if a run was failed but for this case i would rerun the testcase from my perspective.

Sorry for late response.

[Nepitwin]

@JimRevolutionist

By actuall solution error code is handled but appveyor does not upload zip file :(

ERRORLEVEL after call xcopy uia3 is: 0
ResultCode after call xcopy uia3: 1
ERRORLEVEL after call parsly is: 0
ResultCode after call parsly is: 1
Command exited with code 1
$wc = New-Object 'System.Net.WebClient'

In zip file from artifact the reasons from pylint is described what to fix.

[Nepitwin]

But it can be seen in logs

************* Module src.FlaUILibrary.flaui.module.screenshot
src\FlaUILibrary\flaui\module\screenshot.py:115:8: R1705: Unnecessary "elif" after "return", remove the leading "el" from "elif" (no-else-return)
src\FlaUILibrary\flaui\module\screenshot.py:111:4: R1710: Either all return statements in a function should return an expression, or none of them should. (inconsistent-return-statements)
************* Module src.FlaUILibrary.keywords.screenshot
src\FlaUILibrary\keywords\screenshot.py:20:0: C0303: Trailing whitespace (trailing-whitespace)
src\FlaUILibrary\keywords\screenshot.py:47:0: C0301: Line too long (150/120) (line-too-long)
src\FlaUILibrary\keywords\screenshot.py:4:0: W0611: Unused robotlog imported from FlaUILibrary.robotframework (unused-import)
************* Module src.FlaUILibrary.robotframework.robotlog
src\FlaUILibrary\robotframework\robotlog.py:48:8: C0209: Formatting a regular string which could be an f-string (consider-using-f-string)

[Nepitwin]

Verification locally can be done with

keen.bat pylint

But you should first install new package with

keen.bat install
keen.bat pylint

[HackXIt]

Alright, after all that… I think we finally have a valid one.

I fixed all the pylint errors, so appveyor should also give green light if my luck hasn't forsaken me.

Unfortunately, due to my bad behavior of pushing too much, it will take a bit until it gets to the last commit.

[HackXIt]

Yay.

Thank you @Nepitwin

I might do another PR soon regarding #199