Settings and management¶
The interpreter can do a number of other useful things. They are outlined here.
You should be able to store most of the objects you create in memory using the
> store result as 'good_result' > show store > fetch 'good_result' as result
A more permanent solution is to use save and load:
> save result as 'good_result' > ls saved_interrogations > load 'good_result' as result
An alternative approach is to create variables using the
> search corpus for words matching any > call result anyword > calculate anyword as percentage of self
A variable can also be a simple string, which you can then add into searches:
> call '/NN.?/ >># NP' headnoun > search corpus for trees matching headnoun
To forget a variable, just do remove <name>.
Toggles and settings¶
toggle interactive, You can switch between interactive mode, where results and concordances are shown in a way that you can manipulate directly, and non-interactive mode, where results and concordances are simply printed to the console.
toggle conc, you can tell corpkit not to produce concordances. This can be much faster, especially when there are a lot of results.
toggle commawill display thousands separators in results
toggle annotationis used to switch from dry-run to actual modification of corpus files when annotating
- You can set the number of decimals displayed when viewing results with
set decimal to <n>
set max_rows to <n>and
set max_cols to <n>limit the amount of data loaded into results lists. This can speed up interactive viewing.
Switching to IPython¶
When the interpreter constrains you, you can switch to IPython with
ipython. Your objects are available there under the same name. When you’re done there, do
quit to return to the corpkit interpreter.
You can also write and run scripts. If you make a file,
#!/usr/bin/env corpkit set mydata-parsed as corpus search corpus for function matching roles.participant showing lemma export result as csv to part.csv
You can run it from the terminal with:
corpkit participants.cki # or, directly, if there's a shebang and chmod +x: ./participants.cki
which will leave you with a CSV file at
exported/part.csv. This approach can be handy if you need to pipe
stderr, or if you want to call corpkit within a shell script.
Finally, just like Python, you can use the
-c flag to pass code in on the command line:
corpkit -c "set 2 ; search corpus for features ; export result as csv to feat.csv"
When running a script, interactivity will automatically be switched off, and concordancing disabled if the script does not appear to need it.