feat(unitproc/bkt-replace_contractions):Use backreferences
[BK-2020-03.git] / user / bkmpv2
1 #!/usr/bin/env bash
2 # Desc: Wrapper for mpv that accepts directory paths via posargs or stdin lines
3 # Usage: bkmpv2 [DIR]
4 # Version: 0.0.1
5 # Depends: GNU Parallel, GNU Bash v5.1.16, mpv v0.34.1, bc v1.07.1
6 # Ref/Attrib: [1] Tange, Ole. GNU Parallel with Bash Array. 2019-03-24. https://unix.stackexchange.com/a/508365/411854
7 # Example: find $HOME/Music -type d | bkmpv2
8 # Example: bkmpv2 $HOME/Music/
9 # Note: Does not follow symlinks
10
11 # Find settings
12 firegex=".+\(aac\|aif\|aiff\|flac\|m4a\|mp3\|mp4\|ogg\|opus\|wav\)$"; # update according to `find . -type f | grep -Eo "\.([[:alnum:]])+$" | sort -u`
13 fsize="10k"; # default: minimum "10k"
14 fdepth_posarg="10"; # find depth for positional arguments
15 fdepth_stdin="1"; # find depth for stdin
16 fc_falloff="1000"; # characteristic file count falloff in the output
17 fc_redbase="10"; # logarithm base to reduce output file count
18
19 export firegex fsize ; # export for parallel
20
21 #===Declare local functions===
22 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
23 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
24 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
25 checkapp() {
26 # Desc: If arg is a command, save result in assoc array 'appRollCall'
27 # Usage: checkapp arg1 arg2 arg3 ...
28 # Version: 0.1.1
29 # Input: global assoc. array 'appRollCall'
30 # Output: adds/updates key(value) to global assoc array 'appRollCall'
31 # Depends: bash 5.0.3
32 local returnState
33
34 #===Process Args===
35 for arg in "$@"; do
36 if command -v "$arg" 1>/dev/null 2>&1; then # Check if arg is a valid command
37 appRollCall[$arg]="true";
38 if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
39 else
40 appRollCall[$arg]="false"; returnState="false";
41 fi;
42 done;
43
44 #===Determine function return code===
45 if [ "$returnState" = "true" ]; then
46 return 0;
47 else
48 return 1;
49 fi;
50 } # Check that app exists
51 checkfile() {
52 # Desc: If arg is a file path, save result in assoc array 'fileRollCall'
53 # Usage: checkfile arg1 arg2 arg3 ...
54 # Version: 0.1.1
55 # Input: global assoc. array 'fileRollCall'
56 # Output: adds/updates key(value) to global assoc array 'fileRollCall';
57 # Output: returns 0 if app found, 1 otherwise
58 # Depends: bash 5.0.3
59 local returnState
60
61 #===Process Args===
62 for arg in "$@"; do
63 if [ -f "$arg" ]; then
64 fileRollCall["$arg"]="true";
65 if ! [ "$returnState" = "false" ]; then returnState="true"; fi;
66 else
67 fileRollCall["$arg"]="false"; returnState="false";
68 fi;
69 done;
70
71 #===Determine function return code===
72 if [ "$returnState" = "true" ]; then
73 return 0;
74 else
75 return 1;
76 fi;
77 } # Check that file exists
78 checkdir() {
79 # Desc: If arg is a dir path, save result in assoc array 'dirRollCall'
80 # Usage: checkdir arg1 arg2 arg3 ...
81 # Version 0.1.1
82 # Input: global assoc. array 'dirRollCall'
83 # Output: adds/updates key(value) to global assoc array 'dirRollCall';
84 # Output: returns 0 if app found, 1 otherwise
85 # Depends: Bash 5.0.3
86 local returnState
87
88 #===Process Args===
89 for arg in "$@"; do
90 if [ -d "$arg" ]; then
91 dirRollCall["$arg"]="true";
92 if ! [ "$returnState" = "false" ]; then returnState="true"; fi
93 else
94 dirRollCall["$arg"]="false"; returnState="false";
95 fi
96 done
97
98 #===Determine function return code===
99 if [ "$returnState" = "true" ]; then
100 return 0;
101 else
102 return 1;
103 fi
104 } # Check that dir exists
105 displayMissing() {
106 # Desc: Displays missing apps, files, and dirs
107 # Usage: displayMissing
108 # Version 0.1.1
109 # Input: associative arrays: appRollCall, fileRollCall, dirRollCall
110 # Output: stderr: messages indicating missing apps, file, or dirs
111 # Depends: bash 5, checkAppFileDir()
112 local missingApps value appMissing missingFiles fileMissing
113 local missingDirs dirMissing
114
115 #==BEGIN Display errors==
116 #===BEGIN Display Missing Apps===
117 missingApps="Missing apps :";
118 #for key in "${!appRollCall[@]}"; do echo "DEBUG:$key => ${appRollCall[$key]}"; done
119 for key in "${!appRollCall[@]}"; do
120 value="${appRollCall[$key]}";
121 if [ "$value" = "false" ]; then
122 #echo "DEBUG:Missing apps: $key => $value";
123 missingApps="$missingApps""$key ";
124 appMissing="true";
125 fi;
126 done;
127 if [ "$appMissing" = "true" ]; then # Only indicate if an app is missing.
128 echo "$missingApps" 1>&2;
129 fi;
130 unset value;
131 #===END Display Missing Apps===
132
133 #===BEGIN Display Missing Files===
134 missingFiles="Missing files:";
135 #for key in "${!fileRollCall[@]}"; do echo "DEBUG:$key => ${fileRollCall[$key]}"; done
136 for key in "${!fileRollCall[@]}"; do
137 value="${fileRollCall[$key]}";
138 if [ "$value" = "false" ]; then
139 #echo "DEBUG:Missing files: $key => $value";
140 missingFiles="$missingFiles""$key ";
141 fileMissing="true";
142 fi;
143 done;
144 if [ "$fileMissing" = "true" ]; then # Only indicate if an app is missing.
145 echo "$missingFiles" 1>&2;
146 fi;
147 unset value;
148 #===END Display Missing Files===
149
150 #===BEGIN Display Missing Directories===
151 missingDirs="Missing dirs:";
152 #for key in "${!dirRollCall[@]}"; do echo "DEBUG:$key => ${dirRollCall[$key]}"; done
153 for key in "${!dirRollCall[@]}"; do
154 value="${dirRollCall[$key]}";
155 if [ "$value" = "false" ]; then
156 #echo "DEBUG:Missing dirs: $key => $value";
157 missingDirs="$missingDirs""$key ";
158 dirMissing="true";
159 fi;
160 done;
161 if [ "$dirMissing" = "true" ]; then # Only indicate if an dir is missing.
162 echo "$missingDirs" 1>&2;
163 fi;
164 unset value;
165 #===END Display Missing Directories===
166
167 #==END Display errors==
168 } # Display missing apps, files, dirs
169 check_depends() {
170 if ! checkapp mpv parallel bkshuf bc b2sum; then
171 displayMissing;
172 die "FATAL:Missing apps.";
173 fi;
174 return 1;
175 }; # check dependencies
176 checkInt() {
177 # Desc: Checks if arg is integer
178 # Usage: checkInt arg
179 # Input: arg: integer
180 # Output: - return code 0 (if arg is integer)
181 # - return code 1 (if arg is not integer)
182 # Example: if ! checkInt $arg; then echo "not int"; fi;
183 # Version: 0.0.1
184 local returnState
185
186 #===Process Arg===
187 if [[ $# -ne 1 ]]; then
188 die "ERROR:Invalid number of arguments:$#";
189 fi;
190
191 RETEST1='^[0-9]+$'; # Regular Expression to test
192 if [[ ! $1 =~ $RETEST1 ]] ; then
193 returnState="false";
194 else
195 returnState="true";
196 fi;
197
198 #===Determine function return code===
199 if [ "$returnState" = "true" ]; then
200 return 0;
201 else
202 return 1;
203 fi;
204 } # Checks if arg is integer
205 read_stdin() {
206 # Desc: Consumes stdin; outputs as stdout lines
207 # Input: stdin (consumes)
208 # Output: stdout (newline delimited)
209 # Example: printf "foo\nbar\n" | read_stdin
210 # Depends: GNU bash (version 5.1.16)
211 # Version: 0.0.1
212 local input_stdin output;
213
214 # Store stdin
215 if [[ -p /dev/stdin ]]; then
216 input_stdin="$(cat -)";
217 fi;
218
219 # Store as output array elements
220 ## Read in stdin
221 if [[ -n $input_stdin ]]; then
222 while read -r line; do
223 output+=("$line");
224 done < <(printf "%s\n" "$input_stdin");
225 fi;
226
227 # Print to stdout
228 printf "%s\n" "${output[@]}";
229 }; # read stdin to stdout lines
230 read_psarg() {
231 # Desc: Reads arguments; outputs as stdout lines
232 # Input: args
233 # Output: stdout (newline delimited)
234 # Example: read_psarg "$@"
235 # Depends: GNU bash (version 5.1.16)
236 # Version: 0.0.1
237 local input_psarg output;
238
239 # Store arguments
240 if [[ $# -gt 0 ]]; then
241 input_psarg="$*";
242 fi;
243
244 # Store as output array elements
245 ## Read in positional arguments
246 if [[ -n $input_psarg ]]; then
247 for arg in "$@"; do
248 output+=("$arg");
249 done;
250 fi;
251
252 # Print to stdout
253 printf "%s\n" "${output[@]}";
254 }; # read positional argument to stdout lines
255 find_flist() {
256 # Desc: print file list to stdout via `find` using script parameters
257 # Input: arg1: path to dir
258 # var: find_depth
259 # var: pattern_find_iregex
260 # var: find_size
261 if [[ ! -d "$1" ]]; then return 1; fi;
262 must find "$1" -maxdepth "$fdepth" -type f -iregex "$firegex" -size +"$fsize";
263 }; # print file list to stdout from dir with script parameters
264 main() {
265 # Input: var: firegex find iregex file name pattern
266 # var: fsize find minimum file siz
267 # var: fc_falloff characteristic file count falloff in the output
268 # var: fc_redbase logarithm base to reduce output file count
269
270 local re_dotfile;
271 declare -a dirs_stdin dirs_psarg;
272 declare -a paths_files;
273 declare list_paths_files;
274 declare -a cmd_args;
275 check_depends;
276
277 yell "STATUS:$SECONDS:Started.";
278 #Populate dirs_stdin and dirs_psarg arrays
279 ## Read stdin as lines
280 re_dotfile="^\."; # first char is a dot
281 while read -r line; do
282 line="$(readlink -e "$line")";
283 # Check if dir
284 if [[ ! -d "$line" ]]; then
285 echo "ERROR:Not a dir:$line" 1>&2;
286 continue;
287 fi;
288 dir_name="$(basename "$line")";
289 # Exclude dotdirs
290 if [[ "$dir_name" =~ $re_dotfile ]]; then
291 echo "ERROR:Is a dotdir:$line" 1>&2;
292 continue
293 fi;
294 dirs_stdin+=("$line");
295 done < <(read_stdin);
296 yell "STATUS:$SECONDS:Read stdin.";
297 ## Read positional arguments as lines
298 re_dotfile="^\."; # first char is a dot
299 while read -r line; do
300 line="$(readlink -e "$line")";
301 # Check if dir
302 if [[ ! -d "$line" ]]; then
303 echo "ERROR:Not a dir:$line" 1>&2;
304 continue;
305 fi;
306 dir_name="$(basename "$line")";
307 # Exclude dotdirs
308 if [[ "$dir_name" =~ $re_dotfile ]]; then
309 echo "ERROR:Is a dotdir:$line" 1>&2;
310 continue
311 fi;
312 dirs_psarg+=("$line");
313 done < <(read_psarg "$@");
314 yell "STATUS:$SECONDS:Read posargs.";
315
316 # Catch all arrays empty
317 if [[ "${#dirs_stdin[@]}" -le 0 ]] && [[ "${#dirs_psarg[@]}" -le 0 ]]; then
318 die "FATAL:No valid directories provided.";
319 fi;
320
321 # Generate file list
322 ## Call find_filelist() in parallel for positional argument input
323 if [[ "${#dirs_psarg[@]}" -gt 0 ]]; then
324 fdepth="$fdepth_posarg"; export fdepth; # for dirs from positional arguments
325 paths_files+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_psarg[@]}" )"); # See [1]
326 fi;
327 ## Call find_filelist() in parallel for stdin input
328 if [[ "${#dirs_stdin[@]}" -gt 0 ]]; then
329 fdepth="$fdepth_stdin"; export fdepth; # 1 for dirs from stdin
330 paths_files+=("$( parallel find_flist {} "$fdepth" "$firegex" "$fsize" ::: "${dirs_stdin[@]}" )"); # See [1]
331 fi;
332
333 # Convert paths_files array into file list
334 for i in "${!paths_files[@]}"; do
335 list_paths_files="$(printf "%s\n%s" "${paths_files[$i]}" "$list_paths_files")";
336 done;
337 # Get stats
338 fc="$(wc -l < <(echo -n "$list_paths_files"))";
339 echo "STATUS:$SECONDS:file count:fc:$fc"
340 yell "STATUS:$SECONDS:Generated file list.";
341
342 # Sort, remove duplicate paths
343 list_paths_files="$(echo "$list_paths_files" | sort -u | tr -s '\n')";
344 yell "STATUS:$SECONDS:Sorted and deduplicated list.";
345
346 # Remove paths with dotfiles
347 list_paths_files="$(echo "$list_paths_files" | grep -viE "/\." )";
348 yell "STATUS:$SECONDS:Removed dotfiles.";
349
350 # Write playlist
351 list_paths_files_tmp="/dev/shm/$(date +%Y%m%dT%H%M%S.%N%z)"..mpv_paths.txt;
352 ## Reduce output file count if greater than falloff
353 if [[ $fc -gt $fc_falloff ]]; then
354 bc_exp="$fc_falloff * (1 + l( $fc / $fc_falloff )/l($fc_redbase))";
355 fc_out="$( echo "$bc_exp" | bc -l )";
356 else
357 fc_out="$fc";
358 fi;
359 ## Reduce output file count by fixed fraction (bkshuf optimization)
360 fc_out="$(echo "$fc_out * 0.75" | bc -l)";
361 ## Select subset via bkshuf
362 ### Specify bkshuf-specific environment variables
363 export BKSHUF_PARAM_LINC=1000000; # for these numbers of lines of input..
364 export BKSHUF_PARAM_GSIZE=25; # target this group size
365 ### Round file count down
366 fc_out="$(printf "%.0f" "$fc_out")"; # round float down to int
367 ### Get neighbor-preserving shuffled subset (size $fc_out)
368 yell "STATUS:$SECONDS:Selecting $fc_out files via bkshuf...";
369 must \
370 echo -n "$list_paths_files" | \
371 bkshuf "$fc_out" > "$list_paths_files_tmp";
372 yell "STATUS:$SECONDS:Wrote playlist.";
373
374 # Print stats
375 yell "STATUS:$SECONDS:Built file list in $SECONDS seconds.";
376
377 # Run mpv with filelist
378 ## Form command
379 cmd_args+=("mpv");
380 cmd_args+=("--audio-display=no"); # disable video for audio files
381 cmd_args+=("--vid=no"); # donʼt show video track
382 cmd_args+=("--image-display-duration=0"); # don't show album art
383 cmd_args+=("--af=scaletempo=stride=15:overlap=1:search=15"); # improve scrubbing
384 cmd_args+=("--playlist=$list_paths_files_tmp"); # read playlist
385 declare -p cmd_args; # debug
386 ## Execute command
387 must "${cmd_args[@]}";
388 must rm "$list_paths_files_tmp";
389 };
390 export -f yell die must read_stdin read_psarg find_flist;
391
392 main "$@";
393
394 # Author: Steven Baltakatei Sandoval
395 # License: GPLv3+