Customizable html output structure

[memetb]

Following up on #310, I've made a basic modification to the html structure and how it is generated.

I also added a new hook called pytest_html_report_final_dom, the semantics of which are fairly straightforward.

Commit e8bd795 in itself would be sufficient for addressesing issue #310. Commit 8c97045 is an added bonus to make life easier.

I would also recommend that perhaps in the future, a migration away from py.xml.html might be beneficial (in favor of core python libraries that support native tree walking etc). The py package is noted as being in maintenance mode and new projects are recommended not to be started using that library.

Given that the changes required to upgrade pytest-html are actually somewhat limited, it might be a good migration target for the future.

Please let me know if or how I should amend this PR to meet project guidelines.

[memetb]

I might need some help with the npm/grunt tests.

Also, I'm seeing 45 tests failing on master.

Comments?

[BeyondEvil]

This is really nice work @memetb ! 💪

I would really like @ssbarnea input on this as well. 😊

I would also recommend that perhaps in the future, a migration away from py.xml.html might be beneficial (in favor of core python libraries that support native tree walking etc). The py package is noted as being in maintenance mode and new projects are recommended not to be started using that library.

Yes, I agree 100% and it's been on my ToDo for a long time - I just have not had the time. :(

Are there core python libraries that support HTML natively?

[memetb]

Thanks. Much appreciated.

With regards to html libraries: I really like the whole html.div( blah blah ) verbosity as a library consumer, i.e. when I just want to quickly spit out some html. However, as a library provider, I think the most important thing should be walkability/visitability.

So very high up on my list of requirements for that document variable that's passed to the hook is that I could use an xpath query on it.

To that effect, the xml module that comes stock with python is pretty powerful.

I can take that document object as it currently is and simply do:

import xml.etree.ElementTree as ET 
doc = ET.fromstring(document.unicode())
doc.findall(".//div") # xpath query

To output from an element tree, one would have to do an io.ByteBuffer to get a string, or can simply call:

ET.ElementTree(doc).write( filename )

The weakness of ElementTree comes from the building of the tree (it's not as neat and easy as py.xml), but I think that's a bullet we can bite as a library provider.

[BeyondEvil]

Thanks. Much appreciated.

With regards to html libraries: I really like the whole html.div( blah blah ) verbosity as a library consumer, i.e. when I just want to quickly spit out some html. However, as a library provider, I think the most important thing should be walkability/visitability.

So very high up on my list of requirements for that document variable that's passed to the hook is that I could use an xpath query on it.

To that effect, the xml module that comes stock with python is pretty powerful.

I can take that document object as it currently is and simply do:

import xml.etree.ElementTree as ET 
doc = ET.fromstring(document.unicode())
doc.findall(".//div") # xpath query

To output from an element tree, one would have to do an io.ByteBuffer to get a string, or can simply call:

ET.ElementTree(doc).write( filename )

The weakness of ElementTree comes from the building of the tree (it's not as neat and easy as py.xml), but I think that's a bullet we can bite as a library provider.

Interesting! 🤔

[memetb]

@BeyondEvil: I've been dogfooding this addition, and I think the hook for the dom finalizer should include a session parameter to avoid ugly globals. I will add a final commit to this affect.

[memetb]

@BeyondEvil , @ssbarnea: any notes or comments on this PR?

[ssbarnea]

@memetb Please resolve conflicts.

[memetb]

@ssbarnea: Merge is done, however I'll have to check what has changed since I wrote this. evidently something no longer passes the node test.

[memetb]

Fellas, (@ssbarnea, @BeyondEvil) I'm at a loss for what's going wrong here.

The show_hide_extras test is opaque to me in what it's trying to do, and it isn't my code. I actually don't see how it's even related to the code in this branch given that it finds the element based on class name ('.passed'). (I guess I understand that it's looking at the collapsed status and toggling it, however I don't understand the design intent of using .firstElementChild.firstElementChild given the DOM, and this seems to be what is failing.)

Likewise, the sort_column_test error I'm seeing is seemingly unrelated to this revision.

A second pair of eyes would be appreciated here.

[BeyondEvil]

As the plugin has seen major rework in v4, I'm not sure how relevant this is anymore.

If you the reader want to contribute with something here, please feel free to reopen.