feat(user/transcribe_whisper.sh):Apply to various media files

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

diff --git a/user/bkfeh b/user/bkfeh
index 7dbf9fb067916155befd05c2d64fc050d5e719ec..fba697cdc22b1db875c46a5a20dfa8d360191691 100755 (executable)
--- a/user/bkfeh
+++ b/user/bkfeh
@@ -1,7 +1,8 @@
 #!/usr/bin/env bash
 #!/usr/bin/env bash

-# Version: 0.0.7
+# Desc: Wrapper for feh that accepts directory paths via posargs or stdin lines.
+# Version: 0.3.0

 # Ref/Attrib: [1] Tange, Ole. GNU Parallel with Bash Array. 2019-03-24. https://unix.stackexchange.com/a/508365/411854
 # Ref/Attrib: [1] Tange, Ole. GNU Parallel with Bash Array. 2019-03-24. https://unix.stackexchange.com/a/508365/411854

-# Depends: GNU Parallel, GNU Bash v5.1.16, feh 3.6.3
+# Depends: GNU Parallel, GNU Bash v5.1.16, feh 3.6.3, GNU Coreutils 8.32 (b2sum)

 
 #===Declare local functions===
 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
 
 #===Declare local functions===
 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr


@@ -152,9 +153,41 @@ displayMissing() {
     #==END Display errors==
 } # Display missing apps, files, dirs
 check_depends() {
     #==END Display errors==
 } # Display missing apps, files, dirs
 check_depends() {

-    if ! checkapp feh parallel; then displayMissing; die "FATAL:Missing apps."; fi;
+    if ! checkapp feh parallel bkshuf b2sum; then
+        displayMissing;
+        die "FATAL:Missing apps.";
+    fi;

     return 1;
 }; # check dependencies
     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)
 read_stdin() {
     # Desc: Consumes stdin; outputs as stdout lines
     # Input: stdin (consumes)


@@ -214,6 +247,81 @@ find_flist() {
     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
     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

+save_sample() {
+    # Usage: save_sample arg1
+    # Input: arg1   list_paths        (list of files to take samples from)
+    #        envvar BKFEH_SAMPLE_DIR  (environment variable set outside of this script)
+    #        envvar BKFEH_SAMPLE_SIZE (space limit for sample dir files)
+    # Depends: GNU Parallel, GNU find, GNU Coreutils 8.32 (cut, find, du)
+    #   BK-2020-03: bkshuf (0.0.1), yell()
+    local list_paths
+    sample_count="100"; # max number of images to put in sample dir
+    sample_max_space="10000000"; # max bytes to put in sample dir
+
+    # Load environment variables if set
+    if [[ ! -v BKFEH_SAMPLE_DIR ]]; then return 0; fi; # return early if environment var not set.
+    if [[ -v BKFEH_SAMPLE_SIZE ]] && checkInt "$BKFEH_SAMPLE_SIZE"; then
+        sample_max_space="$BKFEH_SAMPLE_SIZE";
+    fi;
+    if [[ -v BKFEH_SAMPLE_COUNT ]] && checkInt "$BKFEH_SAMPLE_COUNT"; then
+        sample_count="$BKFEH_SAMPLE_COUNT";
+    fi;
+    
+    if [[ -n "$1" ]]; then
+        list_paths="$1"; # newline-delimited list of file paths to sample from
+    else
+        yell "ERROR:NO paths available to sample.";
+    fi;
+    
+    if [[ -d "$BKFEH_SAMPLE_DIR" ]]; then
+        #sample_dir="$BKFEH_SAMPLE_DIR";
+        yell "STATUS:Environment variable BKFEH_SAMPLE_DIR set. Clearing and saving samples...";
+        
+        ## clear previous sample
+        count_prev_samples="$(find "$BKFEH_SAMPLE_DIR" -maxdepth 1 -type f | wc -l)";
+        yell "STATUS:Deleting $count_prev_samples previous samples...";
+        find "$BKFEH_SAMPLE_DIR" -maxdepth 1 -type f -exec rm '{}' \; ;
+        
+        ## save random sample
+        yell "STATUS:Saving random sample of size $sample_count to $BKFEH_SAMPLE_DIR...";
+        list_paths_sample="$(echo "$list_paths" | bkshuf "$sample_count" | head -n"$sample_count")";
+        n_samp=0; # init sample file counter
+        sample_log="$BKFEH_SAMPLE_DIR"/paths.txt;
+        printf "%s,%s,%s\n" "n_samp" "file_hash" "file_path" >> "$sample_log";
+        while read -r line; do
+            if [[ -z "$line" ]]; then continue; fi;
+            ### check size limit
+            sample_act_space="$(du -bd1 "$BKFEH_SAMPLE_DIR" | cut -f1 )"; # actual used space
+            cand_space="$(du -bd1 "$line" | cut -f1 )"; # size of candidate file to add
+            sample_req_space="$((sample_act_space + cand_space))";
+
+            ### Customize file names
+            n_samp_w="$(printf "%s" "$sample_count" | wc -c)";
+            n_samp_fmt="%0""$n_samp_w""d";
+            n_samp_dd="$(printf "$n_samp_fmt" "$n_samp")"; # sample number fixed-width
+            file_path="$line";
+            #file_dir="$(dirname "$line")";
+            file_name="$(basename "$line")";
+            file_hash="$(b2sum -l32 "$line" | awk '{print $1}')"; # use file hash to avoid clobbering
+            file_ext="${file_name##*.}";
+            file_name="${file_name%.*}";
+            file_shortname="${file_name:0:32}";
+            file_name_new="$n_samp_dd"_"$file_hash".."$file_shortname"."$file_ext";
+            file_path_new="$BKFEH_SAMPLE_DIR"/"$file_name_new"            
+            if [[ "$sample_req_space" -lt "$sample_max_space" ]]; then
+                #### add file to sample dir
+                must cp -n "$file_path" "$file_path_new";
+                #### note path in sample dir log
+                printf "%s,%s,%s\n" "$n_samp_dd" "$file_hash" "$file_path" \
+                       >> "$sample_log";
+            fi;
+            ((n_samp++));
+        done < <( echo "$list_paths_sample" );
+    else
+        yell "ERROR:Does not exist: $BKFEH_SAMPLE_DIR";
+    fi;    
+}; # save sample of files
+

 main() {
     # Depends: read_stdin_psarg() v0.0.1, check_depends()
     local re_dotfile;
 main() {
     # Depends: read_stdin_psarg() v0.0.1, check_depends()
     local re_dotfile;


@@ -226,6 +334,7 @@ main() {
     ## Read stdin as lines
     re_dotfile="^\."; # first char is a dot
     while read -r line; do
     ## 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;
         # Check if dir
         if [[ ! -d "$line" ]]; then
             echo "ERROR:Not a dir:$line" 1>&2;


@@ -242,6 +351,7 @@ main() {
     ## Read positional arguments as lines
     re_dotfile="^\."; # first char is a dot
     while read -r line; do
     ## 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;
         # Check if dir
         if [[ ! -d "$line" ]]; then
             echo "ERROR:Not a dir:$line" 1>&2;


@@ -273,7 +383,7 @@ main() {
     fi;
     ## Call find_filelist() in parallel for stdin input
     if [[ "${#dirs_stdin[@]}" -gt 0 ]]; then
     fi;
     ## Call find_filelist() in parallel for stdin input
     if [[ "${#dirs_stdin[@]}" -gt 0 ]]; then

-        fdepth=1; export fdepth; # 1 ofr dirs from stdin
+        fdepth=1; export fdepth; # 1 for dirs from stdin

         paths_images+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_stdin[@]}" )"); # See [1]
     fi;
     
         paths_images+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_stdin[@]}" )"); # See [1]
     fi;
     


@@ -300,15 +410,18 @@ main() {
     yell "STATUS:Built file list in $SECONDS seconds.";
 
     # Run feh with filelist
     yell "STATUS:Built file list in $SECONDS seconds.";
 
     # Run feh with filelist

-    feh --full-screen --auto-zoom --draw-filename --filelist "$list_paths_images_tmp";
+    feh --full-screen --auto-zoom --draw-filename --filelist "$list_paths_images_tmp" && \
+        must rm "$list_paths_images_tmp" &

 
 

-    # Cleanup
-    must rm "$list_paths_images_tmp";
+    # Save sample to path in env. var. BKFEH_SAMPLE_DIR if set
+    save_sample "$list_paths_images";
+    

 };
 export -f yell die must read_stdin read_psarg find_flist;
 #==END Define local functions==
 
 main "$@";
 
 };
 export -f yell die must read_stdin read_psarg find_flist;
 #==END Define local functions==
 
 main "$@";
 

+

 # Author: Steven Baltakatei Sandoval
 # License: GPLv3+
 # Author: Steven Baltakatei Sandoval
 # License: GPLv3+