# bash -e this in empty subdirectory
# with ../models/suc.cstlemma.lemmas from sparv, 100k, FBT format
# (sans \r, apparently not an issue)

mkdir models
cp ../models/suc.cstlemma.lemmas models

git clone https://github.com/kuhumcst/hashmap.git
git clone https://github.com/kuhumcst/letterfunc.git
git clone https://github.com/kuhumcst/parsesgml.git

# cstlemma0/cstlemma is the default (STREAM is 0)
# cstlemma1/cstlemma can read stdin (STREAM is 1)
# (failed to link with make all; bare make worked)

git clone https://github.com/kuhumcst/cstlemma.git cstlemma0
sed -i 's/-Wall/-Wall -Wno-reorder/' cstlemma0/src/Makefile
make -C cstlemma0/src
make -C cstlemma0/src clean

git clone https://github.com/kuhumcst/cstlemma.git cstlemma1
sed -i 's/-Wall/-Wall -Wno-reorder/' cstlemma1/src/Makefile
sed -i 's/define STREAM 0/define STREAM 1/' cstlemma1/src/defines.h
make -C cstlemma1/src
make -C cstlemma1/src clean

# empty rule set
touch empty

# dictionary compiled with both versions of cstlemma (identical results)
cstlemma0/cstlemma -D -c FBT -i models/suc.cstlemma.lemmas -o models/dict0
cstlemma1/cstlemma -D -c FBT -i models/suc.cstlemma.lemmas -o models/dict1

# These should definitely be in the dictionaries with these tags
# because they are extracted from the dictionary itself:
#
# kommer komma VB.PRS.AKT
# kommer kommer UO
#
# in tabbed: word TAB tag NL
# in slashed: word/tag NL
grep -w kommer models/suc.cstlemma.lemmas | cut -f 1,3 > tabbed
tr '\t' / < tabbed > slashed