How can i add a custom reporter in the urlwatch?like wechat?

[ZJahon]

I want to push the msg to WeChat. The official account on WeChat provides a means, its web：http://pushplus.hxtrip.com/message, which needs to be cleared by using WeChat account.
to be clear,its Call method is very easy.just need token and content,then it submit a post or get request to the server,and my wechat account can receive the msg.

[thp]

Have a look at the custom reporters implemented here:

https://github.com/thp/urlwatch/blob/master/lib/urlwatch/reporters.py

[naquad]

How about external reporters? Sort of reporting with the given qualified class name (my.module.MyReporter) that would extend ReporterBase?

[thp]

How about external reporters? Sort of reporting with the given qualified class name (my.module.MyReporter) that would extend ReporterBase?

That would make it hard to debug if we import code from other modules, and also would mean that we need to keep the API for reporters stable (right now, because all reporters live in the source tree, we can update and improve the API between urlwatch and reporters, as all of them live in the same codebase, which makes maintaining them that much easier).

Would the shell reporter provide enough for you to implement your own reporter?

[naquad]

I'm trying to use it, although the result is pretty clumsy (parsing the job again to figure out what has changed), that's why I've asked.

[naquad]

P. S. If we could have a JSON flag for the shell...

[thp]

I'm trying to use it, although the result is pretty clumsy (parsing the job again to figure out what has changed), that's why I've asked.

What about submitting a reporter upstream here?

[naquad]

It is a DIY RSS dumper, loads the RSS, pushes the item on top (currently <pre>-formatted and title from the job), and writes to the file. That seems to be a bit ad-hoc for something in the upstream.

[thp]

I mean you could replace and close #709 by submitting one :)

[naquad]

Ok, I'll try when I have a moment. Lets see where it gets :)

[naquad]

I mean you could replace and close #709 by submitting one :)

I've made a PoC implementation. Tested it using a couple of URLs. Because of the lxml.etree quirks and API, the code is somewhat verbose even despite a try to make a multi-tool for XML operations.

Before continuing (more testing, docs, ...). Could you please give it a quick review and let me know if it has a chance of being merged?

[thp]

I mean you could replace and close #709 by submitting one :)

I've made a PoC implementation. Tested it using a couple of URLs. Because of the lxml.etree quirks and API, the code is somewhat verbose even despite a try to make a multi-tool for XML operations.

Before continuing (more testing, docs, ...). Could you please give it a quick review and let me know if it has a chance of being merged?

Sorry for the late reply :)

Yeah, I think that's already much better than #709. I can add some review comments once you open a PR. From a quick look, I'll probably have to chew on the _e() "multi-tool" a bit and in general figure out what it does, but no additional external dependencies (we already use lxml.etree, so that's fine) already is a nice thing, and it doesn't seem overly complicated and well thought out (from a surface look).

[naquad]

The point was exactly to avoid extra dependencies, and that's why there's this _e mess. Next week, I'll have some time to update the docs and create the PR.

[thp]

The point was exactly to avoid extra dependencies, and that's why there's this _e mess. Next week, I'll have some time to update the docs and create the PR.

For cross-reference, this is now here: #845