unitproc/find_erotica.sh

   1 #!/bin/bash
   2 # Desc: Searches for erotic phrases in text files in specified working directory
   3 # Usage: find_erotica.sh [dir]
   4 # Depends: GNU Find utils, GNU Parallel 20210822, GNU Grep, GNU Coreutils 8.32, bc
   5 # Version: 0.3.0
   6
   7
   8 FIN_MAX=400;
   9 export SHOW_FILENAME=false;
  10 export CONTEXT_NOUN=128;
  11 export CONTEXT_ADJ=64;
  12 export CONTEXT_VERB=32;
  13 export CONTEXT_MIX_1=32
  14 export CONTEXT_MIX_2=32
  15 export CONTEXT_PHRASE=16;
  16
  17
  18 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
  19 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  20 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  21 search_thread() {
  22     # Desc: Apply filters to a file
  23     # Depends: find_nouns();
  24     # Input:  arg1    file path
  25     # Output: stdout  search results
  26
  27     if [[ ! -f "$1" ]]; then die "FATAL:Not a file:$1"; fi;
  28     local fin="$1";
  29     # yell "STATUS:Starting search_thread() on ${fin}"  # debug;
  30     # yell "Processing: ${fin}..."; # debug
  31
  32     mapfile -t output < <(
  33         find_nouns "$1" | \
  34             find_adjectives | \
  35             find_verbs | \
  36 #           find_mix_1 | \
  37 #           find_mix_2 | \
  38             find_phrase | \
  39             grep -v -- '-$'; # ADJUST ME
  40     );
  41
  42     if [[ "${#output[@]}" -gt 1 ]]; then
  43         printf "%s\n" "$(basename "${fin}" )";
  44         printf "Result: %d lines.\n" "${#output[@]}";
  45         printf "%s\n" "${output[@]}";
  46         printf "\n";
  47     fi;
  48
  49     # yell "STATUS:Finished search_thread() on ${fin}"  # debug;
  50 };
  51 find_nouns() {
  52     # Desc: Find nouns
  53     # Usage: find_nouns [path]
  54     # Input:  arg1          text file path
  55     #         CONTEXT_NOUN  grep parameter
  56     # Output: stdout  lines matching grep patterns
  57
  58     if [[ ! -f "$1" ]]; then die "FATAL:Not a file:$1"; fi;
  59     fin="$1";
  60
  61     #yell "STATUS:Starting find_nouns() on $fin"  # debug;
  62
  63     local cmd+=('grep');
  64     cmd+=('--ignore-case');
  65     cmd+=('--extended-regexp');
  66     if [[ "$SHOW_FILENAME" == "true" ]]; then cmd+=('--with-filename'); fi;
  67     cmd+=('--line-number');
  68     cmd+=('--context' "$CONTEXT_NOUN");
  69     cmd+=('-e' 'nipple');
  70     cmd+=('-e' 'areola');
  71     cmd+=('-e' 'breast');
  72     cmd+=('-e' 'vulva');
  73     cmd+=('-e' 'womanhood');
  74     cmd+=('-e' 'manhood');
  75     cmd+=('-e' 'his member');
  76     cmd+=('-e' 'slit');
  77     cmd+=('-e' 'pussy');
  78     cmd+=('-e' 'vagina');
  79     cmd+=('-e' ' cock');
  80     cmd+=('-e' 'dildo');
  81     cmd+=('-e' 'shaft');
  82     cmd+=('-e' 'blood');
  83     cmd+=('-e' ' lust');
  84     cmd+=('-e' 'tongue');
  85     cmd+=('-e' 'neck');
  86     cmd+=('-e' 'voyeur');
  87     cmd+=('-e' 'ecstacy');
  88     cmd+=('-e' 'pleasure');
  89     cmd+=('-e' 'tingle');
  90     cmd+=('-e' 'orgasm');
  91     cmd+=('-e' 'lips');
  92     cmd+=('-e' 'lovers');
  93     cmd+=('-e' 'hard-on');
  94     cmd+=('-e' 'sex');
  95     cmd+=('-e' 'throat');
  96     cmd+=('-e' 'cleavage');
  97     cmd+=('-e' 'kiss');
  98     cmd+=('--');
  99     cmd+=("$fin");
 100     local output; mapfile -t output < <( "${cmd[@]}"; );
 101     printf "%s\n" "${output[@]}";
 102     #printf "Found %d matching lines for nouns.\n" "${#output[@]}" 1>&2; # debug
 103     # yell "STATUS:Finished find_nouns() on $fin"  # debug;
 104 }; # find nouns
 105 find_adjectives() {
 106     # Desc: Find adjectives
 107     # Usage: cat *.txt | find_adjectives
 108     # Input:  stdin        input lines
 109     #         CONTEXT_ADJ  grep parameter
 110     # Output: stdout  lines matching grep patterns
 111
 112     local cmd+=('grep');
 113     cmd+=('--ignore-case');
 114     cmd+=('--extended-regexp');
 115     cmd+=('--context' "$CONTEXT_ADJ");
 116     cmd+=('-e' 'handsome');
 117     cmd+=('-e' 'gorgeous');
 118     cmd+=('-e' 'beautiful');
 119     cmd+=('-e' 'bewitching');
 120     cmd+=('-e' 'seductive');
 121     cmd+=('-e' 'lusty');
 122     cmd+=('-e' 'comely');
 123     cmd+=('-e' 'stunning');
 124     cmd+=('-e' 'elegant');
 125     cmd+=('-e' 'exquisite');
 126     cmd+=('-e' 'luxurious');
 127     cmd+=('-e' 'ravishing');
 128     cmd+=('-e' 'magnificent');
 129     cmd+=('-e' 'glistening');
 130     cmd+=('-e' 'lovely');
 131     cmd+=('-e' 'loving');
 132     cmd+=('-e' 'sexy|sexual|sexiest');
 133     cmd+=('-e' 'sweet');
 134     cmd+=('-e' 'sensual');
 135     cmd+=('-e' 'arousing');
 136     cmd+=('-e' 'warm');
 137     cmd+=('-e' 'moist');
 138     cmd+=('-e' 'soaked');
 139     cmd+=('-e' 'hot');
 140     cmd+=('-e' 'gentle');
 141     cmd+=('-e' 'supple');
 142     cmd+=('-e' 'tender');
 143     cmd+=('-e' 'pleasurabl');
 144     cmd+=('-e' 'buxom|busty|curvy');
 145     cmd+=('-e' 'naked|nude');
 146     cmd+=('--');
 147     cmd+=('-');
 148
 149     local output; mapfile -t output < <( "${cmd[@]}"; );
 150     printf "%s\n" "${output[@]}";
 151     #printf "Found %d matching lines for adjectives.\n" "${#output[@]}" 1>&2; # debug
 152 }; # find adjectives
 153 find_verbs() {
 154     # Desc: Find verbs
 155     # Usage: cat *.txt | find_verbs
 156     # Input:  stdin         input lines
 157     #         CONTEXT_VERB  grep parameter
 158     # Output: stdout  lines matching grep patterns
 159
 160     local cmd+=('grep');
 161     cmd+=('--ignore-case');
 162     cmd+=('--extended-regexp');
 163     cmd+=('--context' "$CONTEXT_VERB");
 164     cmd+=('-e' 'love');
 165     cmd+=('-e' 'nuzzle');
 166     cmd+=('-e' 'pleasure');
 167     cmd+=('-e' 'make love');
 168     cmd+=('-e' 'fuck');
 169     cmd+=('-e' 'arouse');
 170     cmd+=('-e' 'quiver');
 171     cmd+=('-e' 'tremble');
 172     cmd+=('-e' 'caress');
 173     cmd+=('-e' 'knead');
 174     cmd+=('-e' 'kiss');
 175     cmd+=('-e' 'moan');
 176     cmd+=('-e' 'suck|suckle');
 177     cmd+=('-e' 'fondle');
 178     cmd+=('-e' 'grope');
 179     cmd+=('-e' 'feel up|felt up|felt (me|her|him) up');
 180     cmd+=('--');
 181     cmd+=('-');
 182     local output; mapfile -t output < <( "${cmd[@]}"; );
 183     printf "%s\n" "${output[@]}";
 184     #printf "Found %d matching lines for verbs.\n" "${#output[@]}" 1>&2; # debug
 185 }; # find verbs
 186 find_mix_1() {
 187     # Desc: Filter mix 1
 188     # Usage: cat *.txt | find_mix_1
 189     # Input:  stdin         input lines
 190     #         CONTEXT_VERB  grep parameter
 191     # Output: stdout  lines matching grep patterns
 192     local cmd+=('grep');
 193     cmd+=('--ignore-case');
 194     cmd+=('--context' "$CONTEXT_MIX_1");
 195     cmd+=('-e' 'kiss');
 196     cmd+=('-e' 'breast');
 197     cmd+=('-e' 'tits');
 198     cmd+=('-e' 'manhood');
 199     cmd+=('-e' 'vagina');
 200     cmd+=('-e' 'sex');
 201     cmd+=('--');
 202     cmd+=('-');
 203     local output; mapfile -t output < <( "${cmd[@]}"; );
 204     printf "%s\n" "${output[@]}";
 205 };
 206 find_mix_2() {
 207     # Desc: Filter mix 2
 208     # Usage: cat *.txt | find_mix_2
 209     # Input:  stdin         input lines
 210     #         CONTEXT_VERB  grep parameter
 211     # Output: stdout  lines matching grep patterns
 212
 213     local cmd+=('grep');
 214     cmd+=('--ignore-case');
 215     cmd+=('--context' "$CONTEXT_MIX_2");
 216     cmd+=('-e' 'nipple');
 217     cmd+=('-e' 'womanhood');
 218     cmd+=('-e' 'manhood');
 219     cmd+=('-e' 'slit');
 220     cmd+=('-e' 'pussy');
 221     cmd+=('-e' 'vagina');
 222     cmd+=('-e' 'cock');
 223     cmd+=('-e' 'shaft');
 224     cmd+=('-e' '\ lust');
 225     cmd+=('-e' 'voyeur');
 226     cmd+=('-e' 'ecstacy');
 227     cmd+=('-e' 'tingle');
 228     cmd+=('-e' 'orgasm');
 229     cmd+=('-e' 'gorgeous');
 230     cmd+=('-e' 'bewitching');
 231     cmd+=('-e' 'seductive');
 232     cmd+=('-e' 'lusty');
 233     cmd+=('-e' 'stunning');
 234     cmd+=('-e' 'ravishing');
 235     cmd+=('-e' 'lovely');
 236     cmd+=('-e' 'sexy');
 237     cmd+=('-e' 'sweet');
 238     cmd+=('-e' 'sexiest');
 239     cmd+=('-e' 'sensual');
 240     cmd+=('-e' 'arousing');
 241     cmd+=('-e' 'moist');
 242     cmd+=('-e' 'love');
 243     cmd+=('-e' 'nuzzle');
 244     cmd+=('-e' 'pleasure');
 245     cmd+=('-e' 'make\ love');
 246     cmd+=('-e' 'sex');
 247     cmd+=('-e' 'fuck');
 248     cmd+=('-e' 'arouse');
 249     cmd+=('-e' 'kiss');
 250     cmd+=('-e' 'breast');
 251     cmd+=('-e' 'tits');
 252     cmd+=('-e' 'manhood');
 253     cmd+=('-e' 'vagina');
 254     cmd+=('-e' 'lips');
 255     cmd+=('-e' 'lovers');
 256     cmd+=('--');
 257     cmd+=('-');
 258     local output; mapfile -t output < <( "${cmd[@]}"; );
 259     printf "%s\n" "${output[@]}";
 260 };
 261 find_phrase () {
 262     # Desc: Filter phrase
 263     # Usage: cat *.txt | find_phrase
 264     # Input:  stdin         input lines
 265     #         CONTEXT_VERB  grep parameter
 266     # Output: stdout  lines matching grep patterns
 267
 268     local cmd+=('grep');
 269     cmd+=('--ignore-case');
 270     cmd+=('--context' "$CONTEXT_PHRASE");
 271     cmd+=('--extended-regexp');
 272     cmd+=('--color=always');
 273     cmd+=('-e' '( his| her| my) [^\.|;]{,48}( his| her| my) (breast|pussy|tits|nipple|womanhood|cock( |\.|,)|manhood|dick|cunt)');
 274     cmd+=('-e' '( his| her| my) [^\.|;]{,24}nipple.{,24}harden');
 275     cmd+=('-e' '(suck|lick)[^\.|;]{,12} (nipple|cock( |\.|,)|dick)');
 276     cmd+=('-e' '(slap)[^\.|;]{,12}( ass)');
 277     cmd+=('-e' '( his| her| my) [^\.|;]{,12}(cunt|pussy( |\.|,)|cock( |\.|,)|dick|shaft( |\.|,|;))');
 278     cmd+=('--');
 279     cmd+=('-');
 280     local output; mapfile -t output < <( "${cmd[@]}"; );
 281     printf "%s\n" "${output[@]}";
 282     #printf "Found %d matching lines for phrases in %s.\n" "${#output[@]}" "$fin" 1>&2; # debug
 283 }; # find phrase
 284 main () {
 285     din="$1";
 286     if [[ ! -d "$din" ]]; then die "FATAL:Not a dir:${din}"; fi;
 287
 288     find "$din" -type f -name "*.txt" | shuf | head -n "$FIN_MAX" | \
 289         #parallel search_thread '{}' | less --RAW-CONTROL-CHARS -S;
 290         parallel search_thread '{}';
 291 };
 292 export -f yell die must search_thread find_nouns find_adjectives find_verbs find_mix_1 find_mix_2 find_phrase;
 293
 294 main "$@";