lifo-dijkstraserv/scripts/geojson/get_rendered_meta.sh

50 lines
1.2 KiB
Bash
Raw Permalink Normal View History

#!/bin/bash
BASE="https://wiki.linux-forks.de"
mode="find_description"
thumbnail=""
categories=""
data=`curl -s "$1"`
temp=`echo "$data" | sed -n 's/<p>\(.\+\).*/\1/p' | head -n1`
if [ "$temp" != "" ]; then
temp=`echo "$temp" | sed "s#href=\"#href=\"$BASE#g" | sed 's/"/\\\\"/g' | sed 's/\t//g'`
description="$temp"
mode="find_infobox";
fi
IFS=$'>';
for line in $data; do
if [ "$mode" == "find_infobox" ]; then
if [ "`echo \"$line\" | grep 'infobox'`" != "" ]; then
mode="image";
fi
elif [ "$mode" == "image" ]; then
temp=`echo "$line" | sed -n 's/.*img.*src="\([^"]\+\).*/\1/p'`;
if [ "$temp" != "" ]; then
thumbnail="$BASE$temp"
mode="find_cat"
fi
elif [ "$mode" == "find_cat" ]; then
if [ "`echo \"$line\" | grep 'mw-normal-catlinks'`" != "" ]; then
mode="cat";
fi
elif [ "$mode" == "cat" ]; then
temp=`echo "$line" | sed -n 's/.*title="Category:\([^"]\+\).*/\1/pg' | grep -v 'page does not exist'`
if [ "$temp" != "" ]; then
if [ "$categories" != "" ]; then
categories="$categories,"
fi
categories="$categories\"$temp\""
fi
fi
done
IFS=" ";
echo "\"categories\": [$categories],"
echo "\"image\": \"$thumbnail\","
echo "\"description\": \"$description\""