remove-ads

#!/bin/sh
# cleans srt formatted subtitles of common blocks that may be considered unwanted, works well as a post-process script for software such as Bazarr or Sub-Zero
# please consider leaving or modifying this regex to properly credit the hard work that is put into providing these subtitles

### usage:
## Download this file from the command line to your current directory:
# curl https://raw.githubusercontent.com/brianspilner01/media-server-scripts/master/sub-clean.sh > sub-clean.sh && chmod +x sub-clean.sh
## Run this script across your whole media library:
# find /path/to/library -name '*.srt' -exec /path/to/sub-clean.sh "{}" \;
## Add to Bazarr (Settings > Subtitles > Use Custom Post-Processing > Post-processing command):
# /path/to/sub-clean.sh "{{subtitles}}" --
## Add to Sub-Zero (in Plex > Settings > under Manage > Plugins > Sub-Zero Subtitles > Call this executable upon successful subtitle download (near the bottom):
# /path/to/sub-clean.sh %(subtitle_path)s
## Test out what lines this script would remove:
# REGEX_TO_REMOVE='(br|dvd|web).?(rip|scr)|english (- )?us|sdh|srt|(yahoo|mail|book|fb|4m|hd)\. ?com|(sub(title)?(bed)?(s)?(fix)?|encode(d)?|correct(ed|ion(s)?)|caption(s|ed)|sync(ed|hroniz(ation|ed))?|english)(.pr(esented|oduced))?.?(by|&)|[^a-z]www\.|http|\. ?(co|pl|link|org|net|mp4|mkv|avi|pdf)([^a-z]|$)|©|™'
# REGEX_TO_REMOVE2='opensubtitles|sub(scene|rip)|podnapisi|addic7ed|titlovi|bozxphd|sazu489|psagmeno|normita|anoxmous|isubdb|americascardroom'
# awk 'tolower($0) ~ '"/$REGEX_TO_REMOVE/" RS='' ORS='\n\n' "/path/to/sub.srt"
# awk 'tolower($0) ~ '"/$REGEX_TO_REMOVE2/" RS='' ORS='\n\n' "/path/to/sub.srt"

# specify file ownership
CHMOD=666

SUB_FILEPATH="$1"

# check usage
[ ! -f "$SUB_FILEPATH" ] && { echo "usage: sub-clean.sh [FILE]" ; echo "Warning: subtitle file does not exist" ; exit 1 ; }

# lowercase list of regex (gore/magic?) that will be removed from srt
REGEX_TO_REMOVE='(br|dvd|web).?(rip|scr)|english (- )?us|sdh|srt|(yahoo|mail|facebook|fb|4m|hd)\. ?com|(sub(title)?(bed)?(s)?(fix)?|encode(d)?|correct(ed|ion(s)?)|caption(s|ed)|sync(ed|hroniz(ation|ed))?|english)(.pr(esented|oduced))?.?(by|&)|[^a-z]http|\. ?(co|pl|link|mp4|mkv|avi|pdf)([^a-z]|$)|©|™'
# regex lists seperated for compatibility with old implementations of awk that require <400 characters
REGEX_TO_REMOVE2='opensubtitles|sub(scene|rip)|podnapisi|addic7ed|titlovi|bozxphd|sazu489|psagmeno|normita|anoxmous|isubdb|americascardroom'

if [ "$(echo "$SUB_FILEPATH" | grep '\.srt$')" ] # only operate on srt files
then
    cp "$SUB_FILEPATH" "$SUB_FILEPATH.original"
    # convert any DOS formatted files to UNIX (remove carriage return line endings)
    awk '{ sub("\r$", ""); print }' "$SUB_FILEPATH" > "${SUB_FILEPATH}.bak" && mv "${SUB_FILEPATH}.bak" "$SUB_FILEPATH"

    ### each record (in awk) is defined as a block of srt formatted subs (record seperator RS is essentially \n\n+, see docs), with each line of the block a seperate field .i.e.:
    # LINE NUMBER
    # TIMESTAMP --> TIMESTAMP
    # SUB LINE 1
    # SUB LINE 2
    # ...
    #

    awk 'tolower($0) !~ /'"$REGEX_TO_REMOVE"'/ { $1 = VAR++ ; print ; next } { print >> TRASH }' RS='' FS='\n' OFS='\n' ORS='\n\n' VAR=1 TRASH="$SUB_FILEPATH.trash.tmp" "$SUB_FILEPATH" > "$SUB_FILEPATH.tmp" && \
    mv "$SUB_FILEPATH.tmp" "$SUB_FILEPATH" && \
    awk 'tolower($0) !~ /'"$REGEX_TO_REMOVE2"'/ { $1 = VAR++ ; print ; next } { print >> TRASH }' RS='' FS='\n' OFS='\n' ORS='\n\n' VAR=1 TRASH="$SUB_FILEPATH.trash.tmp" "$SUB_FILEPATH" > "$SUB_FILEPATH.tmp" && \
    mv "$SUB_FILEPATH.tmp" "$SUB_FILEPATH" && \
    chmod $CHMOD "$SUB_FILEPATH" && \
    echo "sub-clean.sh succesfully processed $SUB_FILEPATH"

    if [ -f "$SUB_FILEPATH.trash.tmp" ]
    then
        REMOVED_LINES=$(cat "$SUB_FILEPATH.trash.tmp")
        # rm "$SUB_FILEPATH.trash.tmp"

        if [ "$REMOVED_LINES" ]
        then
            echo "The following lines were removed:"
            echo "$REMOVED_LINES"
        fi
    else
        rm "$SUB_FILEPATH.original"
    fi

else
    echo "Provided file must be .srt"
    exit 1
fi