#!/bin/sh

# Input format: word tag <lemma>
# lemma is optional, tag must be in the second position

gawk 'BEGIN{OFS=FS="\t";getline;word=$1;tag=$2;lemma=$3}\
((tag~"V.FIN" || tag~"V.INF")\
&& $2~"^[$][.,]"\
&& (word~"[erlu]n$" && word!~"[^aeiou]e*ten$") && word!~".zu.....") \
{\
  if (flag||zu)\
  {\
    if (tag=="VVFIN") tag="VVINF";\
    else if (tag=="VAFIN") tag="VAINF";\
    else if (tag=="VMFIN") tag="VMINF";\
  }\
  else\
  {\
    if (tag=="VVINF") tag="VVFIN";\
    else if (tag=="VAINF") tag="VAFIN";\
    else if (tag=="VMINF") tag="VMFIN";\
  }\
}\
tag~"^V[VAM]FIN$"\
{\
   flag=1;\
}\
tag~"^[$][.,]"\
{\
   flag=0;\
}\
\
{if (tag=="PTKZU") zu=1; else zu=0;}\
{print word,tag,lemma; word=$1;tag=$2;lemma=$3;}
END\
{print word,tag,lemma}' $* |
sed 's/[ 	]*$//'