cleanup html entities in filenames

This commit is contained in:
Alexander Bocken 2021-07-13 12:18:29 +02:00
parent 74c7947ae2
commit 33cd346f65
Signed by: Alexander
GPG Key ID: 1D237BE83F9B05E8

View File

@ -105,7 +105,8 @@ scan(){
files="$(paste <(paste <(echo "$files_json" | jq '.filename' | tr -d '"') <(echo "$files_json" | jq '.location' | tr -d '"')) <(echo "$files_json" | jq '.md5' | tr -d '"'))" files="$(paste <(paste <(echo "$files_json" | jq '.filename' | tr -d '"') <(echo "$files_json" | jq '.location' | tr -d '"')) <(echo "$files_json" | jq '.md5' | tr -d '"'))"
echo "$files" | while read -r file_line; do echo "$files" | while read -r file_line; do
#TODO: better cleanup like in booksplitter? #TODO: better cleanup like in booksplitter?
filename="$(echo "$file_line" | cut -f1 | tr ' ' '_')" filename="$(echo "$file_line" | cut -f1 | tr ' ' '_' |
perl -C -MHTML::Entities -pe 'decode_entities($_);')" #recode html entities as UTF-8
master_location="$(echo "$file_line" | cut -f2 | tr -d '"')" master_location="$(echo "$file_line" | cut -f2 | tr -d '"')"
filelocation="$dl_location/$filename" filelocation="$dl_location/$filename"
correct_md5="$(echo "$file_line" | cut -f3)" correct_md5="$(echo "$file_line" | cut -f3)"