feat(user/bkmpv2):Add bkshuf wrapper for mpv

[BK-2020-03.git] / user / bkmpv2

#!/usr/bin/env bash
# Desc: Wrapper for mpv that accepts directory paths via posargs or stdin lines
# Usage: bkmpv2 [DIR]
# Version: 0.0.1
# Depends: GNU Parallel, GNU Bash v5.1.16, mpv v0.34.1, bc v1.07.1
# Ref/Attrib: [1] Tange, Ole. GNU Parallel with Bash Array. 2019-03-24. https://unix.stackexchange.com/a/508365/411854
# Example: find $HOME/Music -type d | bkmpv2
# Example: bkmpv2 $HOME/Music/
# Note: Does not follow symlinks

# Find settings
firegex=".+\(aac\|aif\|aiff\|flac\|m4a\|mp3\|mp4\|ogg\|opus\|wav\)$"; # update according to `find . -type f | grep -Eo "\.([[:alnum:]])+$" | sort -u`
fsize="10k"; # default: minimum "10k"
fdepth_posarg="10"; # find depth for positional arguments
fdepth_stdin="1";  # find depth for stdin
fc_falloff="1000"; # characteristic file count falloff in the output
fc_redbase="10"; # logarithm base to reduce output file count

export firegex fsize ; # export for parallel

#===Declare local functions===
yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
checkapp() {
    # Desc: If arg is a command, save result in assoc array 'appRollCall'
    # Usage: checkapp arg1 arg2 arg3 ...
    # Version: 0.1.1
    # Input: global assoc. array 'appRollCall'
    # Output: adds/updates key(value) to global assoc array 'appRollCall'
    # Depends: bash 5.0.3
    local returnState    

    #===Process Args===
    for arg in "$@"; do
	if command -v "$arg" 1>/dev/null 2>&1; then # Check if arg is a valid command
	    appRollCall[$arg]="true";
	    if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
	else
	    appRollCall[$arg]="false"; returnState="false";
	fi;
    done;

    #===Determine function return code===
    if [ "$returnState" = "true" ]; then
	return 0;
    else
	return 1;
    fi;
} # Check that app exists
checkfile() {
    # Desc: If arg is a file path, save result in assoc array 'fileRollCall'
    # Usage: checkfile arg1 arg2 arg3 ...
    # Version: 0.1.1
    # Input: global assoc. array 'fileRollCall'
    # Output: adds/updates key(value) to global assoc array 'fileRollCall';
    # Output: returns 0 if app found, 1 otherwise
    # Depends: bash 5.0.3
    local returnState

    #===Process Args===
    for arg in "$@"; do
	if [ -f "$arg" ]; then
	    fileRollCall["$arg"]="true";
	    if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
	else
	    fileRollCall["$arg"]="false"; returnState="false";
	fi;
    done;
    
    #===Determine function return code===
    if [ "$returnState" = "true" ]; then
	return 0;
    else
	return 1;
    fi;
} # Check that file exists
checkdir() {
    # Desc: If arg is a dir path, save result in assoc array 'dirRollCall'
    # Usage: checkdir arg1 arg2 arg3 ...
    # Version 0.1.1
    # Input: global assoc. array 'dirRollCall'
    # Output: adds/updates key(value) to global assoc array 'dirRollCall';
    # Output: returns 0 if app found, 1 otherwise
    # Depends: Bash 5.0.3
    local returnState

    #===Process Args===
    for arg in "$@"; do
	if [ -d "$arg" ]; then
	    dirRollCall["$arg"]="true";
	    if ! [ "$returnState" = "false" ]; then returnState="true"; fi
	else
	    dirRollCall["$arg"]="false"; returnState="false";
	fi
    done
    
    #===Determine function return code===
    if [ "$returnState" = "true" ]; then
	return 0;
    else
	return 1;
    fi
} # Check that dir exists
displayMissing() {
    # Desc: Displays missing apps, files, and dirs
    # Usage: displayMissing
    # Version 0.1.1
    # Input: associative arrays: appRollCall, fileRollCall, dirRollCall
    # Output: stderr: messages indicating missing apps, file, or dirs
    # Depends: bash 5, checkAppFileDir()
    local missingApps value appMissing missingFiles fileMissing
    local missingDirs dirMissing
    
    #==BEGIN Display errors==
    #===BEGIN Display Missing Apps===
    missingApps="Missing apps  :";
    #for key in "${!appRollCall[@]}"; do echo "DEBUG:$key => ${appRollCall[$key]}"; done
    for key in "${!appRollCall[@]}"; do
	value="${appRollCall[$key]}";
	if [ "$value" = "false" ]; then
	    #echo "DEBUG:Missing apps: $key => $value";
	    missingApps="$missingApps""$key ";
	    appMissing="true";
	fi;
    done;
    if [ "$appMissing" = "true" ]; then  # Only indicate if an app is missing.
	echo "$missingApps" 1>&2;
    fi;
    unset value;
    #===END Display Missing Apps===

    #===BEGIN Display Missing Files===
    missingFiles="Missing files:";
    #for key in "${!fileRollCall[@]}"; do echo "DEBUG:$key => ${fileRollCall[$key]}"; done
    for key in "${!fileRollCall[@]}"; do
	value="${fileRollCall[$key]}";
	if [ "$value" = "false" ]; then
	    #echo "DEBUG:Missing files: $key => $value";
	    missingFiles="$missingFiles""$key ";
	    fileMissing="true";
	fi;
    done;
    if [ "$fileMissing" = "true" ]; then  # Only indicate if an app is missing.
	echo "$missingFiles" 1>&2;
    fi;
    unset value;
    #===END Display Missing Files===

    #===BEGIN Display Missing Directories===
    missingDirs="Missing dirs:";
    #for key in "${!dirRollCall[@]}"; do echo "DEBUG:$key => ${dirRollCall[$key]}"; done
    for key in "${!dirRollCall[@]}"; do
	value="${dirRollCall[$key]}";
	if [ "$value" = "false" ]; then
	    #echo "DEBUG:Missing dirs: $key => $value";
	    missingDirs="$missingDirs""$key ";
	    dirMissing="true";
	fi;
    done;
    if [ "$dirMissing" = "true" ]; then  # Only indicate if an dir is missing.
	echo "$missingDirs" 1>&2;
    fi;
    unset value;
    #===END Display Missing Directories===

    #==END Display errors==
} # Display missing apps, files, dirs
check_depends() {
    if ! checkapp mpv parallel bkshuf bc b2sum; then
        displayMissing;
        die "FATAL:Missing apps.";
    fi;
    return 1;
}; # check dependencies
checkInt() {
    # Desc: Checks if arg is integer
    # Usage: checkInt arg
    # Input: arg: integer
    # Output: - return code 0 (if arg is integer)
    #         - return code 1 (if arg is not integer)
    # Example: if ! checkInt $arg; then echo "not int"; fi;
    # Version: 0.0.1
    local returnState

    #===Process Arg===
    if [[ $# -ne 1 ]]; then
	die "ERROR:Invalid number of arguments:$#";
    fi;
    
    RETEST1='^[0-9]+$'; # Regular Expression to test
    if [[ ! $1 =~ $RETEST1 ]] ; then
	returnState="false";
    else
	returnState="true";
    fi;

    #===Determine function return code===
    if [ "$returnState" = "true" ]; then
	return 0;
    else
	return 1;
    fi;
} # Checks if arg is integer
read_stdin() {
    # Desc: Consumes stdin; outputs as stdout lines
    # Input: stdin (consumes)
    # Output: stdout (newline delimited)
    # Example: printf "foo\nbar\n" | read_stdin
    # Depends: GNU bash (version 5.1.16)
    # Version: 0.0.1
    local input_stdin output;

    # Store stdin
    if [[ -p /dev/stdin ]]; then
        input_stdin="$(cat -)";
    fi; 
    
    # Store as output array elements
    ## Read in stdin
    if [[ -n $input_stdin ]]; then
        while read -r line; do
            output+=("$line");
        done < <(printf "%s\n" "$input_stdin");
    fi;

    # Print to stdout
    printf "%s\n" "${output[@]}";
}; # read stdin to stdout lines
read_psarg() {
    # Desc: Reads arguments; outputs as stdout lines
    # Input: args
    # Output: stdout (newline delimited)
    # Example: read_psarg "$@"
    # Depends: GNU bash (version 5.1.16)
    # Version: 0.0.1
    local input_psarg output;
    
    # Store arguments
    if [[ $# -gt 0 ]]; then
        input_psarg="$*";
    fi;
    
    # Store as output array elements
    ## Read in positional arguments
    if [[ -n $input_psarg ]]; then
        for arg in "$@"; do
            output+=("$arg");
        done;
    fi;

    # Print to stdout
    printf "%s\n" "${output[@]}";
}; # read positional argument to stdout lines
find_flist() {
    # Desc: print file list to stdout via `find` using script parameters
    # Input: arg1: path to dir
    #        var: find_depth
    #        var: pattern_find_iregex
    #        var: find_size
    if [[ ! -d "$1" ]]; then return 1; fi;
    must find "$1" -maxdepth "$fdepth" -type f -iregex "$firegex" -size +"$fsize";
}; # print file list to stdout from dir with script parameters
main() {
    # Input: var: firegex     find iregex file name pattern
    #        var: fsize       find minimum file siz
    #        var: fc_falloff  characteristic file count falloff in the output
    #        var: fc_redbase  logarithm base to reduce output file count

    local re_dotfile;
    declare -a dirs_stdin dirs_psarg;
    declare -a paths_files;
    declare list_paths_files;
    declare -a cmd_args;
    check_depends;

    yell "STATUS:$SECONDS:Started.";
    #Populate dirs_stdin and dirs_psarg arrays
    ## Read stdin as lines
    re_dotfile="^\."; # first char is a dot
    while read -r line; do
        line="$(readlink -e "$line")";
        # Check if dir
        if [[ ! -d "$line" ]]; then
            echo "ERROR:Not a dir:$line" 1>&2;
            continue;
        fi;
        dir_name="$(basename "$line")";
        # Exclude dotdirs
        if [[ "$dir_name" =~ $re_dotfile ]]; then
            echo "ERROR:Is a dotdir:$line" 1>&2;
            continue
        fi;
        dirs_stdin+=("$line");
    done < <(read_stdin);
    yell "STATUS:$SECONDS:Read stdin.";
    ## Read positional arguments as lines
    re_dotfile="^\."; # first char is a dot
    while read -r line; do
        line="$(readlink -e "$line")";
        # Check if dir
        if [[ ! -d "$line" ]]; then
            echo "ERROR:Not a dir:$line" 1>&2;
            continue;
        fi;
        dir_name="$(basename "$line")";
        # Exclude dotdirs
        if [[ "$dir_name" =~ $re_dotfile ]]; then
            echo "ERROR:Is a dotdir:$line" 1>&2;
            continue
        fi;
        dirs_psarg+=("$line");
    done < <(read_psarg "$@");
    yell "STATUS:$SECONDS:Read posargs.";
    
    # Catch all arrays empty
    if [[ "${#dirs_stdin[@]}" -le 0 ]] && [[ "${#dirs_psarg[@]}" -le 0 ]]; then
        die "FATAL:No valid directories provided.";
    fi;

    # Generate file list
    ## Call find_filelist() in parallel for positional argument input
    if [[ "${#dirs_psarg[@]}" -gt 0 ]]; then
        fdepth="$fdepth_posarg"; export fdepth; # for dirs from positional arguments
        paths_files+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_psarg[@]}" )"); # See [1]
    fi;
    ## Call find_filelist() in parallel for stdin input
    if [[ "${#dirs_stdin[@]}" -gt 0 ]]; then
        fdepth="$fdepth_stdin"; export fdepth; # 1 for dirs from stdin
        paths_files+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_stdin[@]}" )"); # See [1]
    fi;

    # Convert paths_files array into file list
    for i in "${!paths_files[@]}"; do
        list_paths_files="$(printf "%s\n%s" "${paths_files[$i]}" "$list_paths_files")";
    done;
    # Get stats
    fc="$(wc -l < <(echo -n "$list_paths_files"))";
    echo "STATUS:$SECONDS:file count:fc:$fc"
    yell "STATUS:$SECONDS:Generated file list.";

    # Sort, remove duplicate paths
    list_paths_files="$(echo "$list_paths_files" | sort -u | tr -s '\n')";
    yell "STATUS:$SECONDS:Sorted and deduplicated list.";

    # Remove paths with dotfiles
    list_paths_files="$(echo "$list_paths_files" | grep -viE "/\." )";
    yell "STATUS:$SECONDS:Removed dotfiles.";

    # Write playlist
    list_paths_files_tmp="/dev/shm/$(date +%Y%m%dT%H%M%S.%N%z)"..mpv_paths.txt;
    ## Reduce output file count if greater than falloff
    if [[ $fc -gt $fc_falloff ]]; then
            bc_exp="$fc_falloff * (1 + l( $fc / $fc_falloff )/l($fc_redbase))";
            fc_out="$( echo "$bc_exp" | bc -l )";
    else
        fc_out="$fc";        
    fi;
    ## Reduce output file count by fixed fraction (bkshuf optimization)
    fc_out="$(echo "$fc_out * 0.75" | bc -l)";
    ## Select subset via bkshuf
    ### Specify bkshuf-specific environment variables
    export BKSHUF_PARAM_LINC=1000000; # for these numbers of lines of input..
    export BKSHUF_PARAM_GSIZE=25;     # target this group size
    ### Round file count down
    fc_out="$(printf "%.0f" "$fc_out")"; # round float down to int
    ### Get neighbor-preserving shuffled subset (size $fc_out)
    yell "STATUS:$SECONDS:Selecting $fc_out files via bkshuf...";
    must \
        echo -n "$list_paths_files" | \
        bkshuf "$fc_out" > "$list_paths_files_tmp";
    yell "STATUS:$SECONDS:Wrote playlist.";

    # Print stats
    yell "STATUS:$SECONDS:Built file list in $SECONDS seconds.";

    # Run mpv with filelist
    ## Form command
    cmd_args+=("mpv");
    cmd_args+=("--audio-display=no"); # disable video for audio files
    cmd_args+=("--vid=no"); # donʼt show video track
    cmd_args+=("--image-display-duration=0"); # don't show album art
    cmd_args+=("--af=scaletempo=stride=15:overlap=1:search=15"); # improve scrubbing
    cmd_args+=("--playlist=$list_paths_files_tmp"); # read playlist
    declare -p cmd_args; # debug
    ## Execute command
    must "${cmd_args[@]}";
    must rm "$list_paths_files_tmp";
};
export -f yell die must read_stdin read_psarg find_flist;

main "$@";

# Author: Steven Baltakatei Sandoval
# License: GPLv3+