feat(unitproc/bkt-get_media_bitrate):Add function to measure bitrate

[BK-2020-03.git] / user / bk-copy-rand-music

diff --git a/user/bk-copy-rand-music b/user/bk-copy-rand-music
old mode 100644 (file)
new mode 100755 (executable)
index bc21e5b..f3ddb55
--- a/user/bk-copy-rand-music
+++ b/user/bk-copy-rand-music
@@ -1,7 +1,7 @@
 #!/usr/bin/env bash
 # Desc: Copies random audio files
 # Usage: bk-copy-rand-music [dir SOURCE] [dir DEST] [int DURATION] ([int BYTES])
 #!/usr/bin/env bash
 # Desc: Copies random audio files
 # Usage: bk-copy-rand-music [dir SOURCE] [dir DEST] [int DURATION] ([int BYTES])

-# Version: 0.1.1
+# Version: 0.5.0

 # Depends: BK-2020-03: bkshuf v0.1.0
 
 declare -Ag appRollCall # Associative array for storing app status
 # Depends: BK-2020-03: bkshuf v0.1.0
 
 declare -Ag appRollCall # Associative array for storing app status


@@ -10,14 +10,16 @@ declare -Ag dirRollCall # Associative array for storing dir status
 declare -a music_codecs # Array for storing valid codec names (e.g. "aac" "mp3")
 
 # Adjustable parameters
 declare -a music_codecs # Array for storing valid codec names (e.g. "aac" "mp3")
 
 # Adjustable parameters

-music_codecs=("vorbis" "aac" "mp3" "flac" "opus"); # whitelist of valid codec_names ffprobe might return
+music_codecs=("vorbis" "aac" "mp3" "flac" "opus" "eac3"); # whitelist of valid codec_names ffprobe might return
+ext_ignore=".ots\$|.mid\$|.json\$|.gz\$|.jpg\$|.png\$|.asc\$|.pdf\$|.txt\$|.vtt\$|\.SUM|.zip\$|.xz\$|.org\$|.txt\$"; # blacklist of file extensions for 'grep -Evi'

 max_filename_length="255"; # max output filename length
 max_filename_length="255"; # max output filename length

-min_file_duration="10"; # minimum duration per music file
+min_file_duration="30"; # minimum duration per music file

 max_file_duration="3600"; # maximum duration per music file
 min_file_size="100000"; # minimum size per music file (bytes)
 max_file_size="100000000"; # maximum size per music file (bytes)
 siz_dest="600000000"; # default destination size limit: 600 MB
 max_find_depth="10"; # max find depth
 max_file_duration="3600"; # maximum duration per music file
 min_file_size="100000"; # minimum size per music file (bytes)
 max_file_size="100000000"; # maximum size per music file (bytes)
 siz_dest="600000000"; # default destination size limit: 600 MB
 max_find_depth="10"; # max find depth

+limit_bitrate="320000"; # maximum bitrate (bps) for output audio files

 
 # Load env vars (bkshuf defaults for typical music albums)
 if [[ ! -v BKSHUF_PARAM_LINEC ]]; then export BKSHUF_PARAM_LINEC=1000000; fi;
 
 # Load env vars (bkshuf defaults for typical music albums)
 if [[ ! -v BKSHUF_PARAM_LINEC ]]; then export BKSHUF_PARAM_LINEC=1000000; fi;


@@ -292,6 +294,7 @@ get_media_length() {
     # Output: stdout: seconds (float)
     # Depends: ffprobe 4.1.8
     # Ref/Attrib: [1] How to get video duration in seconds? https://superuser.com/a/945604
     # Output: stdout: seconds (float)
     # Depends: ffprobe 4.1.8
     # Ref/Attrib: [1] How to get video duration in seconds? https://superuser.com/a/945604

+    

     local file_in
     file_in="$1";
     if [[ ! -f $file_in ]]; then
     local file_in
     file_in="$1";
     if [[ ! -f $file_in ]]; then


@@ -367,6 +370,35 @@ checkIsInArray() {
        return 1;
     fi;
 } # Check if string is element in array
        return 1;
     fi;
 } # Check if string is element in array

+get_media_bitrate() {
+    # Use ffprobe to get audio/visual media container bitrate (bits per second integer)
+    # Usage: get_media_bitrate arg1
+    # Input:  arg1:    str  path to file
+    # Output: stdout:  int  bitrate (bits per second)
+    # Version: 0.0.1
+    # Depends: ffprobe 4.4.2
+    # Ref/Attrib: [1] How to get video duration in seconds? https://superuser.com/a/945604
+    #             [2] Determine video bitrate using ffmpeg https://superuser.com/questions/1106343/determine-video-bitrate-using-ffmpeg
+    local file_in
+    file_in="$1";
+    if [[ ! -f $file_in ]]; then
+       die "ERROR:Not a file:$file_in";
+    fi;
+    ffprobe -v error -show_entries format=bit_rate -of default=noprint_wrappers=1:nokey=1 "$file_in";
+
+    yell "DEBUG:Finished get_media_bitrate with $? on:$1";
+} # Get media container length in seconds via stdout
+transcode_copy() {
+    # Desc: Transcode high bitrate file into smaller opus file
+    # Note: Meant for downsizing large lossless FLAC
+    # Input: arg1                str  path to input audio file
+    #        arg2                str  output file path
+    #        var  limit_bitrate  int  max output transcode bitrate (bps)
+    # Output: file
+
+    ffmpeg -nostdin -i "$1" -c:a libopus -b:a "${limit_bitrate}" "${2}.opus";    
+}; # transcode to lower bitrate audio file
+

 main() {
     # Desc: Main program
     # Input: arg1: path to source tree
 main() {
     # Desc: Main program
     # Input: arg1: path to source tree


@@ -397,15 +429,19 @@ main() {
 
     # Check env vars
     if ! checkInt "$BKSHUF_PARAM_LINEC"; then
 
     # Check env vars
     if ! checkInt "$BKSHUF_PARAM_LINEC"; then

-        die "FATAL:Not an int:BKSHUF_PARAM_LINEC:$BKSHUF_PARAM_LINEC"; fi;
+        die "FATAL:Not an int:BKSHUF_PARAM_LINEC:${BKSHUF_PARAM_LINEC}"; fi;

     if ! checkInt "$BKSHUF_PARAM_GSIZE"; then
     if ! checkInt "$BKSHUF_PARAM_GSIZE"; then

-        die "FATAL:Not an int:BKSHUF_PARAM_LINEC:$BKSHUF_PARAM_GSIZE"; fi;
+        die "FATAL:Not an int:BKSHUF_PARAM_LINEC:${BKSHUF_PARAM_GSIZE}"; fi;
+
+    # Check adjustable parameters
+    if ! checkInt "$limit_bitrate"; then
+        die "FATAL:Not an int:limit_bitrate:${limit_bitrate}"; fi;

     
     ## Check duration
     if checkInt "$arg3"; then
        dur_dest="$arg3";
     else
     
     ## Check duration
     if checkInt "$arg3"; then
        dur_dest="$arg3";
     else

-       die "FATAL:Duration (seconds) not an int:$arg3"
+       die "FATAL:Duration (seconds) not an int:${arg3}"

     fi;
 
     ## Check size
     fi;
 
     ## Check size


@@ -413,7 +449,7 @@ main() {
         if checkInt "$arg4"; then
             siz_dest="$arg4";
         else
         if checkInt "$arg4"; then
             siz_dest="$arg4";
         else

-            die "FATAL:Size (bytes) not an int:$arg4";
+            die "FATAL:Size (bytes) not an int:${arg4}";

         fi;
     fi;
     
         fi;
     fi;
     


@@ -438,7 +474,9 @@ main() {
     # Populate list_files array
     while read -r line; do
         list_files+=("$line");
     # Populate list_files array
     while read -r line; do
         list_files+=("$line");

-    done < <(find -L "$dir_source" -maxdepth "$max_find_depth" -type f | sort);
+    done < <(find -L "$dir_source" -maxdepth "$max_find_depth" -type f | \
+                 grep -Ev "$ext_ignore" | \
+                 sort);

 
     # Test and add random elements of list_files to list_copy
     dur=0; # Initialize duration
 
     # Test and add random elements of list_files to list_copy
     dur=0; # Initialize duration


@@ -449,12 +487,23 @@ main() {
     ## Get element count of list_files array
     file_count="${#list_files[@]}";
     while read -r line && \
     ## Get element count of list_files array
     file_count="${#list_files[@]}";
     while read -r line && \

-            [[ $dur -le $dur_dest ]] && \
-            [[ $siz -le $siz_dest ]] && \
+            [[ $dur -le $((dur_dest * 95 / 100)) ]] && \
+            [[ $siz -le $((siz_dest * 95 / 100)) ]] && \

             [[ $n -le $file_count ]]; do
             [[ $n -le $file_count ]]; do

-        #yell "DEBUG:list_copy building loop:$n";
+        ((n++));
+
+        yell "DEBUG:list_copy building loop:$n/$file_count"; # debug
+        printf "DEBUG:%8d,%8d,%8d/%8d,%8d/%8d\n" "$dur_cand" "$siz_cand" "$dur" "$dur_dest" "$siz" "$siz_dest"; # debug
+

         path_candfile="$line"; # path of candidate file
 
         path_candfile="$line"; # path of candidate file
 

+        ### Check size
+        siz_cand="$(du -Lb "$path_candfile" | awk '{ print $1 }')"; # size in bytes
+       if ! checkInt "$siz_cand"; then continue; fi; # reject
+        if [[ "$((siz + siz_cand))" -gt "$siz_dest" ]]; then continue; fi; # reject
+       if [[ "$siz_cand" -lt "$min_file_size" ]]; then continue; fi; # reject
+        if [[ "$siz_cand" -gt "$max_file_size" ]]; then continue; fi; # reject
+        

        ### Check if has valid codec
        if ! check_parsable_audio_ffprobe "$path_candfile"; then continue; fi; # reject
        
        ### Check if has valid codec
        if ! check_parsable_audio_ffprobe "$path_candfile"; then continue; fi; # reject
        


@@ -462,31 +511,28 @@ main() {
        file_format="$(get_audio_format "$path_candfile")";
        if ! checkIsInArray "$file_format" "${music_codecs[@]}"; then continue; fi; # reject
 
        file_format="$(get_audio_format "$path_candfile")";
        if ! checkIsInArray "$file_format" "${music_codecs[@]}"; then continue; fi; # reject
 

-       ### Check and save duration
-       dur_cand="$(get_media_length "$path_candfile")";
+        ### Check duration
+       dur_cand="$(get_media_length "$path_candfile")"; # length in seconds (float)

        dur_cand="${dur_cand%%.*}"; # convert float to int
        dur_cand="${dur_cand%%.*}"; # convert float to int

-        if [[ "$((dur + dur_cand))" -gt "$dur_dest" ]]; then break; fi; # no more
-        dur_cand_wnow="$(printf "%s" "$dur_cand" | wc -m)"; # duration width count
-        if [[ $dur_cand_wnow -gt $dur_cand_w ]]; then
-            dur_cand_w="$dur_cand_wnow"; fi;
-       if ! checkInt "$dur_cand"; then continue; fi; # reject
+        if ! checkInt "$dur_cand"; then continue; fi; # reject
+        if [[ "$((dur + dur_cand))" -gt "$dur_dest" ]]; then continue; fi; # reject

        if [[ "$dur_cand" -lt "$min_file_duration" ]]; then continue; fi; # reject
         if [[ "$dur_cand" -gt "$max_file_duration" ]]; then continue; fi; # reject
 
        if [[ "$dur_cand" -lt "$min_file_duration" ]]; then continue; fi; # reject
         if [[ "$dur_cand" -gt "$max_file_duration" ]]; then continue; fi; # reject
 

-        ### Check and save size
-        siz_cand="$(du -b "$path_candfile" | awk '{ print $1 }')"; # size in bytes
-        if [[ "$((siz + siz_cand))" -gt "$siz_dest" ]]; then break; fi; # no more
+        ### Update stats digits widths
+        #### duration
+        dur_cand_wnow="$(printf "%s" "$dur_cand" | wc -m)"; # duration width count
+        if [[ $dur_cand_wnow -gt $dur_cand_w ]]; then
+            dur_cand_w="$dur_cand_wnow"; fi;
+        #### size

         siz_cand_wnow="$(printf "%s" "$siz_cand" | wc -m)"; # size  width count
         if [[ $siz_cand_wnow -gt $siz_cand_w ]]; then
             siz_cand_w="$siz_cand_wnow"; fi;
         siz_cand_wnow="$(printf "%s" "$siz_cand" | wc -m)"; # size  width count
         if [[ $siz_cand_wnow -gt $siz_cand_w ]]; then
             siz_cand_w="$siz_cand_wnow"; fi;

-       if ! checkInt "$siz_cand"; then continue; fi; # reject
-       if [[ "$siz_cand" -lt "$min_file_size" ]]; then continue; fi; # reject
-        if [[ "$siz_cand" -gt "$max_file_size" ]]; then continue; fi; # reject
-
+        

        ### Add/update candfile to array:
         ###   list_copy (array with "duration, size, path")
        #yell "DEBUG:Adding $path_candfile";
        ### Add/update candfile to array:
         ###   list_copy (array with "duration, size, path")
        #yell "DEBUG:Adding $path_candfile";

-        #printf "DEBUG:%8d,%8d,%s\n" "$dur_cand" "$siz_cand" "$path_candfile" 1>&2;
+        printf "DEBUG:%8d,%8d,%s\n" "$dur_cand" "$siz_cand" "$path_candfile" 1>&2;

         #printf "DEBUG:dur:%s\n" "$dur" 1>&2;
         #printf "DEBUG:siz:%s\n" "$siz" 1>&2;
         list_copy+=("$dur_cand,$siz_cand,$path_candfile"); # for copying with order
         #printf "DEBUG:dur:%s\n" "$dur" 1>&2;
         #printf "DEBUG:siz:%s\n" "$siz" 1>&2;
         list_copy+=("$dur_cand,$siz_cand,$path_candfile"); # for copying with order


@@ -494,10 +540,8 @@ main() {
        ### Update total duration $dur and total size $siz
         dur="$((dur + dur_cand))";
         siz="$((siz + siz_cand))";
        ### Update total duration $dur and total size $siz
         dur="$((dur + dur_cand))";
         siz="$((siz + siz_cand))";

-       #yell "DEBUG:dur:$dur";
-       #yell "DEBUG:siz:$siz";
-
-       ((n++));
+       yell "DEBUG:dur:$dur";
+       yell "DEBUG:siz:$siz";

     done < <(printf "%s\n" "${list_files[@]}" | bkshuf);
 
     #yell "DEBUG:BKSHUF_PARAM_LINEC:$BKSHUF_PARAM_LINEC";
     done < <(printf "%s\n" "${list_files[@]}" | bkshuf);
 
     #yell "DEBUG:BKSHUF_PARAM_LINEC:$BKSHUF_PARAM_LINEC";


@@ -511,26 +555,37 @@ main() {
     # Copy files in list_copy to dir_dest;
     while read -r line; do
         #yell "DEBUG:line:$line"; # debug
     # Copy files in list_copy to dir_dest;
     while read -r line; do
         #yell "DEBUG:line:$line"; # debug

-        fdur="$(printf "%s" "$line" | cut -d',' -f1)";
-        fsize="$(printf "%s" "$line" | cut -d',' -f2)";
+        fdur="$(printf "%s" "$line" | cut -d',' -f1)";  # duration in seconds (integer)
+        fsize="$(printf "%s" "$line" | cut -d',' -f2)"; # size in bytes

         fpath="$(printf "%s" "$line" | cut -d',' -f3-)";
         fpath="$(printf "%s" "$line" | cut -d',' -f3-)";

+        fbitrate="$(get_media_bitrate "$fpath")"; # bitrate in bps
+        if ! checkInt "$fbitrate"; then die "FATAL:Invalid bitrate. $(declare -p fpath fbitrate)"; fi;

         ## Get basename of path
         file_basename="$(basename "$fpath")";
         ## Get basename of path
         file_basename="$(basename "$fpath")";

+        ### Get basename without unprintable non-ASCII characters
+        file_basename_compat="$(printf "%s" "$file_basename" | tr -dc '[:graph:][:space:]' )";

 
         ## Get 16-character b2sum fingerprint (for different files that share basename)
        fingerprint="$(b2sum -l32 "$fpath" | awk '{print $1}' )";
 
         ## Form output filename
         num="$(printf "$num_fmt" "$n")";
 
         ## Get 16-character b2sum fingerprint (for different files that share basename)
        fingerprint="$(b2sum -l32 "$fpath" | awk '{print $1}' )";
 
         ## Form output filename
         num="$(printf "$num_fmt" "$n")";

-       file_name="$num"_"$fingerprint".."$file_basename";
+       file_name="$num"_"$fingerprint".."$file_basename_compat";

        file_name="${file_name:0:$max_filename_length}"; # Limit filename length (e.g. Windows has max of 255 characters)
 
        ## Form output path
        path_output="$dir_dest"/"$file_name";
        
        ## Copy
        file_name="${file_name:0:$max_filename_length}"; # Limit filename length (e.g. Windows has max of 255 characters)
 
        ## Form output path
        path_output="$dir_dest"/"$file_name";
        
        ## Copy

-       must cp "$fpath" "$path_output" && yell "NOTICE:Copied ($fdur seconds): $fpath ";
-       #yell "DEBUG:Copied $file_basename to $dur_dest.";
+        yell "DEBUG:$(declare -p fdur fsize fbitrate fpath file_basename file_basename_compat fingerprint num file_name path_output)";
+        if [[ "$fbitrate" -lt "$limit_bitrate" ]]; then
+           must cp "$fpath" "$path_output" && yell "NOTICE:Copied ($(printf "%""$dur_cand_w"d "$fdur") seconds): $fpath ";
+            yell "DEBUG:Copied ${file_basename} to ${dir_dest}.";
+        elif [[ "$fbitrate" -ge "$limit_bitrate" ]]; then
+            must transcode_copy "$fpath" "$path_output";
+            yell "DEBUG:Wrote ${limit_bitrate}kbps transcoded ${file_basename} to ${dir_dest}.";
+        fi;
+               

 
        ## Append log
         fpath_can="$(readlink -f "$fpath")"; # resolve symlinks to canonical path
 
        ## Append log
         fpath_can="$(readlink -f "$fpath")"; # resolve symlinks to canonical path


@@ -538,7 +593,7 @@ main() {
         printf "$log_fmt" "$num" "$fingerprint" "$fdur" "$fsize" "$fpath_can" >> "$path_log_output";
 
        ((n++));
         printf "$log_fmt" "$num" "$fingerprint" "$fdur" "$fsize" "$fpath_can" >> "$path_log_output";
 
        ((n++));

-       unset file_basename path_output
+       unset file_basename file_basename_compat path_output;

     done < <(printf "%s\n" "${list_copy[@]}");
 
     # Report total duration and size
     done < <(printf "%s\n" "${list_copy[@]}");
 
     # Report total duration and size