Standard input and input options on command line interface
Closed this issue · 16 comments
jdeck88 commented
Great if we can have standard in and standard out using the command line interface. Or, having the ability to specify in and out using command line arguments (supporting both would be optimal). This would make it easier to use the pre-reasoner with other tools....
stuckyb commented
Okay, @jdeck88, I've implemented a basic inferencing pipeline mode for OntoPilot. You can run
$ ontopilot.py inference_pipeline -i INPUT -o OUTPUT
or
$ ontopilot.py ipl -i INPUT -o OUTPUT
where INPUT is the input data file and OUTPUT is the destination path for the reasoned data.
Note that OntoPilot still needs a config file to specify how to do the reasoning. This can either be in the same directory from where you launch OntoPilot, in which case OntoPilot will find it automatically, or you can specify its location via the -c/--config_file option.
The feature's not quite done -- I will next work on adding support for stdin/stdout. Thus, this feature's not yet been merged into the elk_pipeline branch, but you can start experimenting with it in master.
jdeck88 commented
Ok, cool! it worked.
It complained about not having the [Ontology] section of the configuration
file and that it couldn't find the ontology_file designation, so i put one
in there to get it to be quiet, like:
[Ontology]
ontology_file = Outgoing/dummy_file_that_is_never_written.owl
If we specify an output file on the command line we should not require that
section.
Will the elk_pipeline branch merge into master at some point?
…On Thu, Apr 20, 2017 at 12:53 PM, stuckyb ***@***.***> wrote:
Okay, @jdeck88 <https://github.com/jdeck88>, I've implemented a basic
inferencing pipeline mode for OntoPilot. You can run
$ ontopilot.py inference_pipeline -i INPUT -o OUTPUT
or
$ ontopilot.py ipl -i INPUT -o OUTPUT
where INPUT is the input data file and OUTPUT is the destination path for
the reasoned data.
Note that OntoPilot still needs a config file to specify how to do the
reasoning. This can either be in the same directory from where you launch
OntoPilot, in which case OntoPilot will find it automatically, or you can
specify its location via the -c/--config_file option.
The feature's not quite done -- I will next work on adding support for
stdin/stdout. Thus, this feature's not yet been merged into the
elk_pipeline branch, but you can start experimenting with it in master.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABGdxX3oZRk31YIQeqSz8v83GKY7tvIgks5rx7e5gaJpZM4MtHdC>
.
--
John Deck
(541) 914-4739
stuckyb commented
Good point, John. I'll work on improving configuration file checking.
Yes, I do want to merge the elk_pipeline branch at some point. The
tricky thing is that it has the custom (i.e., hacked) version of ELK,
and I don't want to merge that into the master branch. I'll get that
sorted out as soon as I have some time. What I'll probably do is merge
all of the new OntoPilot features (in particular, the DL over EL
reasoning implementation) and leave the custom ELK in its own branch for
now.
…On 04/20/2017 06:58 PM, John Deck wrote:
Ok, cool! it worked.
It complained about not having the [Ontology] section of the configuration
file and that it couldn't find the ontology_file designation, so i put one
in there to get it to be quiet, like:
[Ontology]
ontology_file = Outgoing/dummy_file_that_is_never_written.owl
If we specify an output file on the command line we should not require
that
section.
Will the elk_pipeline branch merge into master at some point?
On Thu, Apr 20, 2017 at 12:53 PM, stuckyb ***@***.***>
wrote:
> Okay, @jdeck88 <https://github.com/jdeck88>, I've implemented a basic
> inferencing pipeline mode for OntoPilot. You can run
>
> $ ontopilot.py inference_pipeline -i INPUT -o OUTPUT
>
> or
>
> $ ontopilot.py ipl -i INPUT -o OUTPUT
>
> where INPUT is the input data file and OUTPUT is the destination
path for
> the reasoned data.
>
> Note that OntoPilot still needs a config file to specify how to do the
> reasoning. This can either be in the same directory from where you
launch
> OntoPilot, in which case OntoPilot will find it automatically, or
you can
> specify its location via the -c/--config_file option.
>
> The feature's not quite done -- I will next work on adding support for
> stdin/stdout. Thus, this feature's not yet been merged into the
> elk_pipeline branch, but you can start experimenting with it in master.
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#62 (comment)>,
> or mute the thread
>
<https://github.com/notifications/unsubscribe-auth/ABGdxX3oZRk31YIQeqSz8v83GKY7tvIgks5rx7e5gaJpZM4MtHdC>
> .
>
--
John Deck
(541) 914-4739
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ADKD7_epy3PhrzAoXFtFOEm_woDNBc5Oks5rx-McgaJpZM4MtHdC>.
stuckyb commented
@jdeck88 -- Inference pipeline mode now supports reading from stdin (happens automatically if no input file is provided) and writing to stdout (if no output path is provided), so it should work with shell pipes and i/o redirects.
jdeck88 commented
awesome... i'll try this out.
…On Fri, Apr 21, 2017 at 9:05 AM, stuckyb ***@***.***> wrote:
@jdeck88 <https://github.com/jdeck88> -- Inference pipeline mode now
supports reading from stdin (happens automatically if no input file is
provided) and writing to stdout (if no output path is provided), so it
should work with shell pipes and i/o redirects.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABGdxQEisLaO0Sly2WmljMi1UAkm2fN7ks5ryNPOgaJpZM4MtHdC>
.
--
John Deck
(541) 914-4739
stuckyb commented
@jdeck88, the issue with checking configuration files should now be fixed. In fact, I overhauled configuration file handling in general, so that you can even run the inference pipeline mode without any configuration file at all. If you do that, though, you're stuck with the default settings!
stuckyb commented
@jdeck88 -- One other thing. When you have a chance to test this, could you let me know (or comment on the issue thread) if this feature is working and is good enough to do you what you need? If so, then I'll close the issue. Thanks!
jdeck88 commented
This looks good Brian. The features we discussed are working. One last
thing that i noticed when playing around with this which i didn't think
about earlier...
i've added an import statement to the unreasoned data files. Seeing that
the import statement is in the data itself it seems redundant to import the
ontology again in the configuration file. if i remove the import related
statements from the configuration file, it goes back to the default
location and complains with:
*The top-level imports source file could not be found:
/Users/jdeck/IdeaProjects/pheno_paper/bin/src/imports/imported_ontologies.csv. *
However, i'd rather that it simply look instead in the input datafile for
the imported ontology and not have to specify the import file at all.
Maybe, a "no import" flag??
John
…On Sat, Apr 22, 2017 at 12:30 PM, stuckyb ***@***.***> wrote:
@jdeck88 <https://github.com/jdeck88> -- One other thing. When you have a
chance to test this, could you let me know (or comment on the issue thread)
if this feature is working and is good enough to do you what you need? If
so, then I'll close the issue. Thanks!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABGdxWl719FhMbtyEBwmujXWi69ik0f0ks5rylVigaJpZM4MtHdC>
.
--
John Deck
(541) 914-4739
stuckyb commented
@jdeck88, OntoPilot should only be trying to add imports if you are running one of the make commands (e.g., make ontology). If you run it in "inference pipeline" mode, it shouldn't be doing anything with the imports at all. Can you check how you are running OntoPilot?
jdeck88 commented
Ah--- ok, yes. i had it pointed to the master branch for checking the
command-syntax... is this now all implemented in the elk_pipeline branch?
…On Sat, Apr 22, 2017 at 5:47 PM, stuckyb ***@***.***> wrote:
@jdeck88 <https://github.com/jdeck88>, OntoPilot should only be trying to
add imports if you are running one of the make commands (e.g., make
ontology). If you run it in "inference pipeline" mode, it shouldn't be
doing anything with the imports at all. Can you check how you are running
OntoPilot?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABGdxb37iIbuOer1Iw5AQywrb6jkVDyyks5ryp-RgaJpZM4MtHdC>
.
--
John Deck
(541) 914-4739
jdeck88 commented
ah, and i do see you mentioned this was running in the inference pipeline
earlier... i'll test against that...
…On Sat, Apr 22, 2017 at 5:57 PM, John Deck ***@***.***> wrote:
Ah--- ok, yes. i had it pointed to the master branch for checking the
command-syntax... is this now all implemented in the elk_pipeline branch?
On Sat, Apr 22, 2017 at 5:47 PM, stuckyb ***@***.***> wrote:
> @jdeck88 <https://github.com/jdeck88>, OntoPilot should only be trying
> to add imports if you are running one of the make commands (e.g., make
> ontology). If you run it in "inference pipeline" mode, it shouldn't be
> doing anything with the imports at all. Can you check how you are running
> OntoPilot?
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#62 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/ABGdxb37iIbuOer1Iw5AQywrb6jkVDyyks5ryp-RgaJpZM4MtHdC>
> .
>
--
John Deck
(541) 914-4739
--
John Deck
(541) 914-4739
jdeck88 commented
so, i don't see those arguments available in the elk_pipeline branch so
looks like you haven't updated those features there yet.
and, i did have it in inference_pipeline mode using master.
did you try testing on your side by removing "
*bin/src/imports/imported_ontologies.csv"?*
*John*
…On Sat, Apr 22, 2017 at 5:58 PM, John Deck ***@***.***> wrote:
ah, and i do see you mentioned this was running in the inference pipeline
earlier... i'll test against that...
On Sat, Apr 22, 2017 at 5:57 PM, John Deck ***@***.***> wrote:
> Ah--- ok, yes. i had it pointed to the master branch for checking the
> command-syntax... is this now all implemented in the elk_pipeline branch?
>
> On Sat, Apr 22, 2017 at 5:47 PM, stuckyb ***@***.***>
> wrote:
>
>> @jdeck88 <https://github.com/jdeck88>, OntoPilot should only be trying
>> to add imports if you are running one of the make commands (e.g., make
>> ontology). If you run it in "inference pipeline" mode, it shouldn't be
>> doing anything with the imports at all. Can you check how you are running
>> OntoPilot?
>>
>> —
>> You are receiving this because you were mentioned.
>> Reply to this email directly, view it on GitHub
>> <#62 (comment)>,
>> or mute the thread
>> <https://github.com/notifications/unsubscribe-auth/ABGdxb37iIbuOer1Iw5AQywrb6jkVDyyks5ryp-RgaJpZM4MtHdC>
>> .
>>
>
>
>
> --
> John Deck
> (541) 914-4739
>
--
John Deck
(541) 914-4739
--
John Deck
(541) 914-4739
stuckyb commented
Just tested and it works fine for me if I remove imported_ontologies.csv. E.g.:
$ ontopilot.sh -q inference_pipeline -i Incoming/unreasoned_data.ttl
runs without complaint if there is no imports file.
Note that Annie and I just finished the Java-only branch and merged that into master, and the Jython launch script is now in bin/ontopilot.sh.
stuckyb commented
I just merged master into elk_pipeline, so you should be good to go with that branch now, too.
jdeck88 commented
ok, i tested it all out again this morning and working for me now... thanks!
…On Sat, Apr 22, 2017 at 7:37 PM, stuckyb ***@***.***> wrote:
I just merged master into elk_pipeline, so you should be good to go with
that branch now, too.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#62 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABGdxVVXCCGougkZfFnwZTIPHAIdL6NJks5ryrl7gaJpZM4MtHdC>
.
--
John Deck
(541) 914-4739
stuckyb commented
Excellent! Considering this feature done, then.