c6c3e1e9c2
This commit removes the old 'scrap_pokemon.py' script in favor of the new 'scrap_data.sh' script. Now we don't need python anymore!
54 lines
2.1 KiB
Bash
Executable File
54 lines
2.1 KiB
Bash
Executable File
#!/bin/sh
|
|
|
|
#
|
|
# This script scraps some pokémon pictures from Bulbapedia.
|
|
#
|
|
|
|
bulbapedia_page_url="http://bulbapedia.bulbagarden.net/wiki/List_of_Pok%C3%A9mon_by_Kanto_Pok%C3%A9dex_number"
|
|
bulbapedia_page_name="bulbapedia.html"
|
|
scrap_folder="`pwd`/scrapped-data"
|
|
|
|
# Make sure the directory for the scrapped data is there.
|
|
mkdir -p "$scrap_folder"
|
|
|
|
# Download the bulbapedia page if it doesn't already.
|
|
if [ ! -e "scrapped-data/bulbapedia.html" ]; then
|
|
echo " > Downloading '$bulbapedia_page_url' to '$scrap_folder/$bulbapedia_page_name' ..."
|
|
wget "$bulbapedia_page_url" -O "$scrap_folder/$bulbapedia_page_name" -q
|
|
echo " > Downloaded."
|
|
fi
|
|
|
|
# Dear furure me,
|
|
#
|
|
# If you are in need to maintain this part of the code... I am
|
|
# realy sorry for you (T.T). This was the best I could do... But
|
|
# I will try to explain things here a little bit.
|
|
# 'cat' will read the file and pipe its output to 'sed'. 'sed'
|
|
# will filter the html searching for the Pokémon name and its
|
|
# image url. 'sed' will output the Pokémons in this format:
|
|
# "<POKEMON_NAME>=<POKEMON_URL>".
|
|
# Then, the output of 'sed' goes into the while loop, which will
|
|
# read the output one line at a time. Within the while loop, I
|
|
# extract the pokemon name and the url from the read line. And
|
|
# then, it just downloads the url to a file.
|
|
# Again... I'm sorry for all the trouble. But I hope you will
|
|
# grow stronger and may be able to turn this code into something
|
|
# more readable.
|
|
#
|
|
# Kind regards,
|
|
# Yourself from the past.
|
|
|
|
cat "$scrap_folder/$bulbapedia_page_name" | \
|
|
sed -nr 's;^.*<img alt="(.*)" src="(http://cdn.bulbagarden.net/upload/.*\.png)" width="40" height="40" />.*$;\1=\2;p' | \
|
|
while read line
|
|
do
|
|
pokemon_name="${line%=*}"
|
|
pokemon_url="${line#*=}"
|
|
|
|
# Unescape HTML characters... Damn "Farfetch'd".
|
|
pokemon_name=$(echo "$pokemon_name" | sed "s/'/'/")
|
|
|
|
echo " > Downloading '$pokemon_name' from '$pokemon_url' to '$scrap_folder/$pokemon_name.png' ..."
|
|
wget "$pokemon_url" -O "$scrap_folder/$pokemon_name.png" -q
|
|
done
|