My transcript does not contain the output of commands

[teto]

I had written some tests manually that are now obsolete because of internal changes. I thought of using transcript based testing instead so this is my first try but I am not sure to do it correctly:
http://cmd2.readthedocs.io/en/latest/transcript.html
After running a few commands, I did history 1:4 -t transcript.txt which saves the following transcript.txt:

iperf-mptcp-0-0.pcap> load_pcap examples\/iperf-mptcp-0-0.pcap
iperf-mptcp-0-0.pcap> lc
iperf-mptcp-0-0.pcap> ls 0
iperf-mptcp-0-0.pcap> ls 1

This lacks the output, here is the complete session:

+Ready>load_pcap examples/iperf-mptcp-0-0.pcap
>>> examples/iperf-mptcp-0-0.pcap
/home/teto/mptcpanalyzer/mptcpanalyzer/data.py:144: ParserWarning: Both a converter and dtype were specified for column tcp.flags - only the converter will be used
  "mptcp.duplicated_dsn": lambda x: list(map(int, x.split(','))) if x is not None else np.nan,
+iperf-mptcp-0-0.pcap> lc
>>> 
1 mptcp connection(s)
Description of mptcp.stream 0 
The connection has 4 subflow(s) (client/server): 
	tcp.stream 0.0 : 10.1.0.1:5001.0  <-> 10.2.0.1:5001.0 
	tcp.stream 1.0 : 10.1.1.1:43376.0  <-> 10.2.0.1:5001.0 
	tcp.stream 2.0 : 10.1.0.1:49078.0  <-> 10.2.1.1:5001.0 
	tcp.stream 3.0 : 10.1.1.1:57841.0  <-> 10.2.1.1:5001.0 
+iperf-mptcp-0-0.pcap> ls 0
>>> 0
Description of mptcp.stream 0 
The connection has 4 subflow(s) (client/server): 
	tcp.stream 0.0 : 10.1.0.1:5001.0  <-> 10.2.0.1:5001.0 
	tcp.stream 1.0 : 10.1.1.1:43376.0  <-> 10.2.0.1:5001.0 
	tcp.stream 2.0 : 10.1.0.1:49078.0  <-> 10.2.1.1:5001.0 
	tcp.stream 3.0 : 10.1.1.1:57841.0  <-> 10.2.1.1:5001.0 
+iperf-mptcp-0-0.pcap> ls 1
>>> 1
Traceback (most recent call last):
  File "/nix/store/75apj1nkg5i2sy4hhfw8bjgx2dh2ghz9-python3.6-cmd2-0.8.0/lib/python3.6/site-packages/cmd2.py", line 981, in onecmd_plus_hooks
    stop = self.onecmd(statement)
  File "/nix/store/75apj1nkg5i2sy4hhfw8bjgx2dh2ghz9-python3.6-cmd2-0.8.0/lib/python3.6/site-packages/cmd2.py", line 1179, in onecmd
    stop = func(statement)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 65, in wrapped
    return f(self, *args)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 296, in do_list_subflows
    self.list_subflows(args.mptcpstream)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 65, in wrapped
    return f(self, *args)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 300, in list_subflows
    con = MpTcpConnection.build_from_dataframe(self.data, mptcpstreamid)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/connection.py", line 215, in build_from_dataframe
    raise MpTcpException("No packet with this mptcp.stream id %r" % mptcpstreamid)
mptcpanalyzer.MpTcpException: No packet with this mptcp.stream id 1
EXCEPTION of type 'MpTcpException' occurred with message: 'No packet with this mptcp.stream id 1'
To enable full traceback, run the following command:  'set debug true'
+iperf-mptcp-0-0.pcap> history
>>> 
-------------------------[1]
load_pcap examples/iperf-mptcp-0-0.pcap
-------------------------[2]
lc
-------------------------[3]
ls 0
-------------------------[4]
ls 1
+iperf-mptcp-0-0.pcap> history 1:4 -t transcript.txt
>>> 1:4 -t transcript.txt
>>> examples/iperf-mptcp-0-0.pcap
/home/teto/mptcpanalyzer/mptcpanalyzer/data.py:144: ParserWarning: Both a converter and dtype were specified for column tcp.flags - only the converter will be used
  "mptcp.duplicated_dsn": lambda x: list(map(int, x.split(','))) if x is not None else np.nan,
>>> 
1 mptcp connection(s)
Description of mptcp.stream 0 
The connection has 4 subflow(s) (client/server): 
	tcp.stream 0.0 : 10.1.0.1:5001.0  <-> 10.2.0.1:5001.0 
	tcp.stream 1.0 : 10.1.1.1:43376.0  <-> 10.2.0.1:5001.0 
	tcp.stream 2.0 : 10.1.0.1:49078.0  <-> 10.2.1.1:5001.0 
	tcp.stream 3.0 : 10.1.1.1:57841.0  <-> 10.2.1.1:5001.0 
>>> 0
Description of mptcp.stream 0 
The connection has 4 subflow(s) (client/server): 
	tcp.stream 0.0 : 10.1.0.1:5001.0  <-> 10.2.0.1:5001.0 
	tcp.stream 1.0 : 10.1.1.1:43376.0  <-> 10.2.0.1:5001.0 
	tcp.stream 2.0 : 10.1.0.1:49078.0  <-> 10.2.1.1:5001.0 
	tcp.stream 3.0 : 10.1.1.1:57841.0  <-> 10.2.1.1:5001.0 
>>> 1
Traceback (most recent call last):
  File "/nix/store/75apj1nkg5i2sy4hhfw8bjgx2dh2ghz9-python3.6-cmd2-0.8.0/lib/python3.6/site-packages/cmd2.py", line 981, in onecmd_plus_hooks
    stop = self.onecmd(statement)
  File "/nix/store/75apj1nkg5i2sy4hhfw8bjgx2dh2ghz9-python3.6-cmd2-0.8.0/lib/python3.6/site-packages/cmd2.py", line 1179, in onecmd
    stop = func(statement)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 65, in wrapped
    return f(self, *args)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 296, in do_list_subflows
    self.list_subflows(args.mptcpstream)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 65, in wrapped
    return f(self, *args)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/cli.py", line 300, in list_subflows
    con = MpTcpConnection.build_from_dataframe(self.data, mptcpstreamid)
  File "/home/teto/mptcpanalyzer/mptcpanalyzer/connection.py", line 215, in build_from_dataframe
    raise MpTcpException("No packet with this mptcp.stream id %r" % mptcpstreamid)
mptcpanalyzer.MpTcpException: No packet with this mptcp.stream id 1
EXCEPTION of type 'MpTcpException' occurred with message: 'No packet with this mptcp.stream id 1'
To enable full traceback, run the following command:  'set debug true'
4 commands and outputs saved to transcript file 'transcript.txt'

EDIT: could that be because I use a custom prompt ?

[tleonhardt]

@teto
In your cmd2 application, how are you sending your output to the screen?

In general cmd2 makes some important assumptions regarding how you generate output for users. I'm thinking we may want to make it a little more clear in the documentation what these assumptions are. The main assumption is that all output for the user is generated using one of the following three methods:

• self.poutput() - all normal output
  • This always gets sent to self.stdout which always gets redirected when appropriate
• self.perror() - all exceptional output (errors)
  • This always gets sent to sys.stderr which almost never gets redirected
• self.pfeedback() - for printing non-essential feedback (can be silenced based on self.quiet)
  • Usually gets sent to sys.stderr, but can set self.feedback_to_output to send it to self.stdout instead

The ability to generate transcript tests from the history using history -t makes the above assumptions and it further assumes that only output written to self.stdout via self.poutput() gets written to the transcript file.

As long as you are setting your custom prompt using self.prompt, then it should be fine; transcript tests work with custom prompts. Transcript testing goes a little beyond that even - it can deal with a custom prompt which changes throughout the course of the testing and it purposely strips off any ANSI escape codes from your custom prompt to so that it only compares text and doesn't compare color.

EDIT: Given your setup, you will need to manually edit the first prompt in your transcript to match the default +Ready> since the history method of generating a transcript outputs the current prompt, which in this case wouldn't be correct.

[teto]

I had no idea, poutput doesn't appear in http://cmd2.readthedocs.io/en/latest/transcript.html.
Couldn't the function hijack the stdout temporarily though ? use its own fd and write it on the regular standout as well as in the transcript ?
I had not realized the project was so lively (seen 0.8 this morning). Thanks for your work and the new features !

[tleonhardt]

That function is already hijacking self.stdout and redirecting it to a file. I'd have to look at it - we might be able to hijack sys.stdout and also redirect it to that file. But in order to get all of the features of cmd2 to work properly, such as redirecting command output to a file and the ability to pipe command output to a shell command, it is important to use self.poutput() or self.stdout.write(). If your end users are POSIX-shell savvy, then those are really killer features. I can't tell you how many times the ability to pipe command output to grep has just rocked socks.

At the very least I will keep this issue open and try to improve the documentation to let users know they should be using poutput().

@kotfu and I have been busy lately! ;-) We managed to get a bunch of good features into the 0.8 release. I've gotten a lot of positive feedback about the new argparse decorators and I love how they open up the ability to support sub-commands. We also have some really cool stuff coming up in the 0.8.1 release - in particular the ability to have sub-menus and persistent readline history. Your thanks is appreciated and we are glad you like it.