Compile a parser with two gradle tasks:
gradle buildStemsgradle fstThe buildStems task reads in simple delimited text files with your data for stem dictionaries and for inflectional rules, and translates them into SFST-PL, the language of the Stuttgart FST toolkit. Optionally, use the datadir to specify a directory where you store your data tables, e.g., gradle -Pdatadir. The default value is the sampledata directory in the distribution. Within datadir, you should have one or more subdirectories with a complete set of tables for a given corpus. You may choose which one to use with the dataset property. gradle -Pdataset=smyth buildStems will translate the tables in the smyth directory within the default source directory (sampledata).
The fst task compiles a binary parser. The output is written in the build directory, in a subdirectory named for the current set. You can easily compile multiple distinct parsers. The following two lines illustrate this:
gradle -Pdataset=smyth buildStems && gradle -Pdataset=smyth fst
gradle -Pdataset=attic buildStems && gradle -Pdataset=attic fst
Both lines use the default source directory (since no datadir property was specified). The first line compiles a parser for the smyth corpus in build/smyth, while the second line compiles a parser for the attic corpus in build/attic. The binary parser in each case will be named greek.a
The compiled parser can be used in multiple ways. You can work directly with the utility programs of the SFST toolkit, such as fst-mor. (See the SFST toolkit’s documentation for information.) For example, run fst-mor build/smyth/greek.a to enter forms interactively for analysis. Note, however, that the FST works on unaccented forms.
You can also parse fully accented forms in a simple text file with one word per line using the included greek.sh shell script. It writes to standard output, so redirect it if you want to save to a file, e.g., greek.sh WORDSTOPARSE.txt > analyses.txt to save analyses of WORDSTOPARSE.txt in analyses.txt.