cherrypy/cherrypy

CherryPy is a pythonic, object-oriented HTTP framework. | post your issues to: github.com/cherrypy/cherrypy/issues | FAQ: https://stackoverflow.com/questions/tagged/cheroot+or+cherrypy

Sviatoslav Sydorenko

@webknjaz

It's also recommended to create feature-branches in your fork and avoid sending PRs from master because you can't submit separate PRs while updating only one branch

MikHulk

@MikHulk

Hi. Is there a mean to execute specific code when the server get shutdown in cherrypy ?

Sviatoslav Sydorenko

@webknjaz

try/finally?

MikHulk

@MikHulk

ah, ok

thx

bmxp

@bmxp

I got a little problem again, it's with a response and a strange behaviour of a dict. As soon as I give the HeaderReader an existing dict it turns out to have strange keys and values:

Consider the following code snippet running on debian buster with Python 3.7.3 with cherrypy 18.4.0:

import tempfile
import cherrypy
from cheroot import server

response = tempfile.SpooledTemporaryFile(max_size = 1000, mode = "w+b")

response.write(b'Status: 302 Found\r\nLocation: index.php?\r\nContent-type: text/html; charset=UTF-8\r\n\r\n')

emptydict ={} 
response.seek(0)
hr = server.HeaderReader()
headers = hr(response,cherrypy.serving.response.headers)
print(headers)

print(80*'-')

response.seek(0)
headers = hr(response,emptydict)
print(headers)
print(80*'-')

otherdict ={ 'Content-Type':'text/html'} 
response.seek(0)
headers = hr(response,otherdict)
print(headers)

It will give

{'Content-Type': 'text/html', 'Server': 'CherryPy/18.4.0', 'Date': 'Sat, 23 Nov 2019 20:10:51 GMT', "B'Status'": b'302 Found', "B'Location'": b'index.php?', "B'Content-Type'": b'text/html; charset=UTF-8'}
--------------------------------------------------------------------------------
{b'Status': b'302 Found', b'Location': b'index.php?', b'Content-Type': b'text/html; charset=UTF-8'}
--------------------------------------------------------------------------------
{'Content-Type': 'text/html', b'Status': b'302 Found', b'Location': b'index.php?', b'Content-Type': b'text/html; charset=UTF-8'}

The "B'Status'": b'302 Found' does not look right ...

Sviatoslav Sydorenko

@webknjaz

Interesting. Perhaps there's something like str(b'status').upper() inside...

@bmxp do you want to submit a PR with these test cases to Cheroot?

Also, a full bug report in the Cheroot repo would be very helpful. We need to keep the records of such things in one place and have it well-googlable for history.

bmxp

@bmxp

I opened up an issue on github cherrypy/cheroot#250. Unfortunately I can't help with a PR to fix this since I do not have enough understanding of what's under the hood of cheroot/cherrypy and any coding idea might break other things. Thus I can not help with a test case besides from the code provided above to reproduce. Sorry I can be of no further help ...

bmxp

@bmxp

Just a small followup:

The dict used by cherrypy.serving.response.headers is from cherrypy.lib.httputil.HeaderMap()
There is a member function:
```
@staticmethod
  def transform_key(key):
      return str(key).title()
```
This will cause a bytes object like b'status' to result in "B'Status'" which I would think is wrong. At this place I think it must be decided to first transform the bytes into a string using encoding and not using simply str()

Sviatoslav Sydorenko

@webknjaz

@bmxp thanks for opening an issue! It seems like you feel unqualified or somehow afraid to contribute PRs but you're still very welcome to try :) Even if your contribution isn't accepted as is, it's still very useful for everyone because it's a good demo.

What I meant earlier is adding a failing test to the project and marking it as "expected to fail" (@pytest.mark.xfail). This way, we'd have some tests that wouldn't crash our CI but also would be a great starting point for somebody else to fix the issue in the future. I'll show you how it'd look in a bit.

Sviatoslav Sydorenko

@webknjaz

oh...

@bmxp so it looks that this is happening because CherryPy is already Python-3-only and Cheroot is not

What is your use-case anyway?

bmxp

@bmxp

I wand to use the cgi for an external program to handle non static and non python script files.

Sviatoslav Sydorenko

@webknjaz

But why did you want to mix headers like this?

bmxp

@bmxp

Well I don't want to mix headers.

Sviatoslav Sydorenko

@webknjaz

Let's rephrase: how did you hit this error? It seems to me like you're only using Cheroot and for some reason importing one helper from CherryPy. Right?

bmxp

@bmxp

Yes.

Sviatoslav Sydorenko

@webknjaz

I'm pretty sure that Cheroot just operates on bytes (it's usually what we want when working with HTTP)

And CherryPy should have some adapter for this

bmxp

@bmxp

Wait a minute, I need to log in at home to get the code.

Sviatoslav Sydorenko

@webknjaz

You should probably just feed it with unicode:

-headers = hr(response, cherrypy.serving.response.headers)
+headers = hr(response.decode(), cherrypy.serving.response.headers)

bmxp

@bmxp

I needed a function to read the headers returned from the cgi program and found that cheroot.HeaderReader() would do the job.

Sviatoslav Sydorenko

@webknjaz

I see

well, decoding bytes should do the job

Oh, wait. I was looking at the HeaderMap, not HeaderReader. Let me check it closer

And why do you need cherrypy.serving.response.headers?

Sviatoslav Sydorenko

@webknjaz

So I think you just need bytes in that initial headers dict. Why do you use it in the first place? That arg is optional.

-otherdict = {'Content-Type': 'text/html'}
+otherdict = {b'Content-Type': b'text/html'}

bmxp

@bmxp

And why do you need cherrypy.serving.response.headers?

I thought it was a good idea to use the standard headers as a start....

Sviatoslav Sydorenko

@webknjaz

define "standard"

bmxp

@bmxp

Like info on version of cherrypy e.g.

Sviatoslav Sydorenko

@webknjaz

Also, cherrypy.serving.response.headers is something that is populated on each request that CherryPy processes. It may give weird results if used outside.

But if you don't use CherryPy in the first place, why do you need to hit this?

bmxp

@bmxp

Well I need CherryPy in other sections of the software anyway. It serves other things like dynamic data.

Sviatoslav Sydorenko

@webknjaz

Are you sure you don't need to separate projects?

bmxp

@bmxp

Jep :-)

Sviatoslav Sydorenko

@webknjaz

Or maybe just use CherryPy's capabilities instead of directly using its dependency

bmxp

@bmxp

I am working on a plugin to extend the bigger project

Sviatoslav Sydorenko

@webknjaz

If you tell the details, maybe I could point you to some better way

bmxp

@bmxp

Jep. I could also just change the encoding of the source for HeaderReader.

I will try some things during the week and let you know about the forthcoming. Got to proceed with my normal job now ;-)

Sviatoslav Sydorenko

@webknjaz

:+1:

bmxp

@bmxp

I decided to Clone the functionality of HeaderReader together with direkt conversion to a utf-8 str.

Where communities thrive

People

Repo info

Activity