bib

#!/bin/bash

#This script extract text from trs and save to file

#Global variables input from user
file=bak
final=$2
field=$3
path=$4

#Global variables for this program help nothing more
space=" "
st="<s>"
end="</s>"
br1="("
br2=")"
#NO nedd for this up just variable"char" no +

sed '1,6d' $1 > 1
sed '$d' 1 > 2
sed '$d' 2 > 1
sed '$d' 1 > 2
cp 2 bak
rm 1 2

#Name of file for sphinx tran
name=$1
#Remove .trs
echo $1 | sed 's/.trs//g' > name
name=`cat name`
#Append sync
name=$name"_Sync0"
#Name for number more than 9
name2=$name
name=$name"0"
rm name

#This is for get number lines where is <*/*>
grep -n \<*/*\> $file | sed 's/:.*//g' >> temp

var=0
lines=`cat temp | wc -l`
#Fix because it dont read last number
lines=`expr $lines + 1`

#Read temp and save number of lines to line array
while [ "$var" -lt "$lines" ]
do
    orig=`cat temp | head -n $var | tail -n 1 `
    tx="$orig"
    line[$var]=$tx
    var=`expr $var + 1`          
done

#Save input to temp and then append to new file called tran
#+ also create files field
i=1
#Need  to correct vaule because of starting lines at 0
while [ "$i" -lt "$var" ]
do
    #Asagin vaules of lines
    x=`echo ${line[$i]}`
    y=`echo  `${line[`expr $i+1`]}``

    #Get differ of lines
    a="$((${y}-${x}))"

    #Get lines we Need
    head -n $y $file > temp3
    tail -n $a temp3 > temp2
    
    #remove what is left from trash
    sed '/<.*>/d' temp2  > temp

    #remove empty lines
    sed '/^$/d' temp > temp2

    #Make all upper case and clean with java program
    text=`cat temp2`
    echo $text | tr '[:lower:]' '[:upper:]' > temp3
    text=`cat temp3`
    tmp=$(java CleanText $text)
    
    #Change name from number higer than 9
    if [[ $i -eq 10 ]]; then
        name=$name2
    fi;

    #Make input for sphinx traning append <s>*<s/> (filename) 
    tmp=$st$space$tmp$space$end$space$br1$name$i$br2 

    #Save to final file 
    echo $tmp >>  $final


    ##############
    #Create fileds
    ##############
    #Variable that will be writen in field
    f=$path$name$i
    echo $f >> $field 
    
    i=`expr $i + 1`          
done

#Hack because we get line more than expected
sed '$d' $final > temp
#Return bak to final
cp temp  $final

#Hack because we get line more than expected
sed '$d' $field > temp
#Return bak to final
cp temp  $field

#Last cleanLw
rm temp temp2 temp3  bak
exit