Subversion Repositories wimsdev

#! /bin/sh

datadir=factbook/print

list=`cd $datadir; ls *.html | sed 's/.html//'`

rm -f area* population* capital*
for c in $list
do
 awk 'BEGIN {a=0};
        a==0 && /Area:/ {a=1};
        a==1 && /total:/ {print "'$c': " $0; exit}' $datadir/$c.html \
    | sed 's!:.*</i>!:!;s/million/000/g;s/sq km/@/' | cut -d@ -f 1 \
    | tr -d ' .,' >>area.1
 awk 'BEGIN {a=0};
        a==0 && /Population:/ {a=1};
        a==2 && /hab/ {exit}
        a==2 && /[0-9]/ {print "'$c': " $0; exit}
        a==1 && /<td/ {a=2};' $datadir/$c.html \
    | sed 's/:[:space:]*/: /;s/million/000/g' | cut -d '(' -f 1 \
    | tr -d '   .,' >>population.1
 awk 'BEGIN {a=0};
        a==0 && /Capital:/ {a=1};
        a==2 && /[a-z]/ {print "'$c': " $0; exit}
        a==1 && /<td/ {a=2};' $datadir/$c.html \
    | sed 's/:[:space:]*/: /;s/million/000/g' | cut -d '(' -f 1 \
    | tr -d '   .,' >>capital.1
done

for f in area population
do
 sort -nr -k2 -t: <$f.1 >$f
 rm -f $f.1
done

sort <capital.1 | cut -d ';' -f 1 | grep -v -E 'capital| none' | tr -s ' ' >capital
rm -f capital.1