#!/bin/zsh

file="../ga.wp.txt.bz2"
bz="../c/bzipreader -f $file"
pfile="../gawiki-latest-pages-articles.xml.bz2.processed"

pushd ../c; make clean; make; popd

function fail() {
 echo "FAIL:" $* >/dev/stdout
 exit -1
}

if ! (eval "$bz -x -n 1 2>/dev/null" | grep -q Daoine); then
  fail decompression 
fi

if ! (eval "$bz -x -b 0" | grep -q Esperanto); then
  fail decompression with 0 offset
fi

last=`eval "$bz -l | tail -1 | awk '{print \\$2;}'"`
if ! (eval "$bz -x -b $last" | grep -q Nikon); then
  fail decompression with $last offset
fi

mv $pfile ../ga.tmp
rm ../gawiki-*
mv ../ga.tmp $pfile
../sh/process `echo "$pfile" | sed -E 's/\.proc.+//'`