user/gthumb/gthumb-analyze.sh

   1 #!/bin/bash
   2 # Desc: Analyzes photo files selected in gThumb; shows size/time/GPS
   3 #   statistics in a zenity dialog. Intended for evaluating burst-mode
   4 #   sequences where larger JPEG size correlates with sharpness.
   5 # Usage: gthumb-analyze.sh FILE…
   6 # Example: $HOME/.local/bin/gthumb-analyze.sh %F
   7 # Depends: bash, GNU coreutils, awk, bc, column (util-linux 2.39.3)
   8 #   exiftool (12.x), GNU sed 4.9, zenity
   9 # Version: 0.0.9
  10
  11 fdebug=/tmp/gtAnalyzeLog.txt;
  12
  13 yell() { echo "$0: $*" >&2; } # print script path and all args to stderr
  14 die() { yell "$*"; exit 111; } # same as yell() but non-zero exit status
  15 must() { "$@" || die "cannot $*"; } # runs args as command, reports args if command fails
  16 main() {
  17     # Validate inputs
  18     local fina fin cmd;
  19     fina=("$@");
  20     if [[ $# -le 0 ]]; then die "FATAL:No arguments:$#"; fi;
  21     for fin in "${fina[@]}"; do
  22         if [[ ! -f "$fin" ]]; then die "FATAL:Not a file:${fin}"; fi;
  23     done;
  24     for cmd in awk bc column cut exiftool sed sort zenity; do
  25         command -v "$cmd" >/dev/null 2>&1 || die "FATAL:Missing dependency:${cmd}";
  26     done;
  27
  28     # Single batched exiftool pass. Columns (tab-separated):
  29     #   col 1: FileName
  30     #   col 2: FileSize [bytes]
  31     #   col 3: SubSecCreateDate [float POSIX seconds]
  32     #   col 4: GPSPosition [±Lat, ±Lon]
  33     #   col 5: FNumber [float]
  34     #   col 6: ExposureTime [str seconds]
  35     #   col 7: ISO [int]
  36     #   col 8: FocalLength [str %.01f mm]
  37     # Note: Missing tags emit '-' in -T mode.
  38     local tsv;
  39     tsv="$(exiftool -q -T -FileName '-FileSize#' \
  40         -SubSecCreateDate -d '%s%f' \
  41         -GPSPosition -c '%+0.4f' \
  42         -FNumber -ExposureTime -ISO -FocalLength \
  43         -- "${fina[@]}")" \
  44         || die "FATAL:exiftool failed";
  45     yell "DEBUG:tsv:"; printf '%s\n' "$tsv" 1>&2;
  46
  47     local n fMax fMaxArr fMin fMinArr mean sdev median span gps;
  48     n="$(printf '%s\n' "$tsv" | wc -l)";
  49
  50     # First and last by file name
  51     local fFirst fLast;
  52     fFirst="$(printf '%s\n' "$tsv" | sort -t$'\t' -k1,1 | head -n1 | cut -f1)";
  53     fLast="$(printf '%s\n' "$tsv" | sort -t$'\t' -k1,1 | tail -n1 | cut -f1)";
  54
  55     # Largest and smallest files by size (numeric sort on column 2)
  56     fMax="$(printf '%s\n' "$tsv" | sort -t$'\t' -k2,2n | tail -n1)";
  57     fMin="$(printf '%s\n' "$tsv" | sort -t$'\t' -k2,2n | head -n1)";
  58     fMaxArr[0]="$(cut -f1 <<< "$fMax")";
  59     fMaxArr[1]="$(cut -f2 <<< "$fMax")";
  60     fMinArr[0]="$(cut -f1 <<< "$fMin")";
  61     fMinArr[1]="$(cut -f2 <<< "$fMin")";
  62     ## Avoid & or < being parsed as markup by zenity
  63     fMaxArr[0]="$(sed -e 's/&/\&amp;/g' -e 's/</\&lt;/g' <<< "${fMaxArr[0]}")";
  64     fMinArr[0]="$(sed -e 's/&/\&amp;/g' -e 's/</\&lt;/g' <<< "${fMinArr[0]}")";
  65
  66     # Mean and population standard deviation (÷N, not N−1: the
  67     #   selection is the entire population of interest).
  68     read -r mean sdev < <(printf '%s\n' "$tsv" | cut -f2 \
  69         | awk '{s+=$1; ss+=$1*$1;} END {m=s/NR; printf "%.0f %.0f\n", m, sqrt(ss/NR - m*m);}');
  70
  71     # Median file size
  72     median="$(printf '%s\n' "$tsv" | cut -f2 | sort -n \
  73     | awk '{a[i++]=$1;} END {if (i%2==0) {printf "%.0f\n", (a[i/2-1]+a[i/2])/2;} else {print a[int(i/2)];};}')";
  74
  75     # Time span in seconds via bc; ignore files lacking dates.
  76     local tMin tMax;
  77     tMin="$(printf '%s\n' "$tsv" | cut -f3 | grep -v '^-$' | sort -n | head -n1)";
  78     tMax="$(printf '%s\n' "$tsv" | cut -f3 | grep -v '^-$' | sort -n | tail -n1)";
  79     if [[ -n "$tMin" && -n "$tMax" ]]; then
  80         span="$(printf "%'.3f s" "$(bc -l <<< "${tMax} - ${tMin}")")"; # seconds
  81     else
  82         span="n/a (missing SubSecCreateDate)";
  83     fi;
  84
  85     # GPS: single unique value, none, or multiple.
  86     local gpsUniq gpsUniqSamp gpsCount;
  87     gpsUniq="$(printf '%s\n' "$tsv" | cut -f4 | sort -u | grep -v '^-$')";
  88     gpsCount="$(printf '%s\n' "$gpsUniq" | grep -c .)";
  89     case "$gpsCount" in
  90         0) gps="none";;
  91         *) gps="(${gpsCount} distinct)";
  92            mapfile -t gpsUniqSamp < <(shuf <<< "$gpsUniq" | head -n8 | sort);;
  93     esac;
  94
  95     # Exposure combos: unique (FNumber, ExposureTime, ISO, FocalLength) tuples.
  96     # Missing tags appear as '-' and still compare consistently.
  97     local expUniq expUniqSamp expCount exp;
  98     expUniq="$(printf '%s\n' "$tsv" | cut -f5-8 | sort -u \
  99         | awk -F'\t' '{printf "f/%s\t%s s\tISO %s\t%s\n", $1, $2, $3, $4;}')";
 100     mapfile -t expUniqSamp < <(shuf <<< "$expUniq" | head -n8 | sort -t $'\t' -k4,4n);
 101
 102     expCount="$(printf '%s\n' "$expUniq" | grep -c .)";
 103     case "$expCount" in
 104         1) exp="uniform";;
 105         *) exp="${expCount} distinct combinations";;
 106     esac;
 107
 108     # Report
 109     local msgSec1Arr msgSec1 msgSec2Arr msgSec2 msgSec3Arr msgSec3 msg;
 110     ## Assemble message section 1 (file size, time span)
 111     msgSec1Arr+=("$(printf "%d files analyzed (About %d s)" "$n" "$SECONDS")");
 112     msgSec1Arr+=("$(printf "First:\t\t%s" "$fFirst")");
 113     msgSec1Arr+=("$(printf "Last:\t\t%s" "$fLast")");
 114     msgSec1Arr+=("$(printf "%s" "----" )");
 115     msgSec1Arr+=("$(printf "Largest:\t%'d B\t%s" "${fMaxArr[1]}" "${fMaxArr[0]}")");
 116     msgSec1Arr+=("$(printf "Smallest:\t%'d B\t%s" "${fMinArr[1]}" "${fMinArr[0]}")");
 117     msgSec1Arr+=("$(printf "Mean size:\t%'d B" "$mean")");
 118     msgSec1Arr+=("$(printf "Median size:\t%'d B" "$median")");
 119     msgSec1Arr+=("$(printf "Std. dev.:\t%'d B" "$sdev")");
 120     msgSec1Arr+=("$(printf "Time span:\t%s" "$span")");
 121     msgSec1="$(printf '%s\n' "${msgSec1Arr[@]}" | column -s $'\t' -t --table-right 2)";
 122     ## Assemble message section 2 (location)
 123     msgSec2Arr+=("$(printf "GPS:\t%s" "$gps")");
 124     msgSec2Arr+=("$(printf '%s\n' "${gpsUniqSamp[@]}")");
 125     msgSec2="$(printf '%s\n' "${msgSec2Arr[0]}"; printf '%s\n' "${msgSec2Arr[@]:1}" | column -s $'\t' -t | sed -e 's/^/\t/')";
 126     ## Assemble message section 3 (exposure)
 127     msgSec3Arr+=("$(printf "Exposure:\t%s" "$exp")");
 128     msgSec3Arr+=("$(printf '%s\n' "${expUniqSamp[@]}")");
 129     msgSec3="$(printf '%s\n' "${msgSec3Arr[0]}"; printf '%s\n' "${msgSec3Arr[@]:1}" | column -s $'\t' -t | sed -e 's/^/\t/')";
 130     ## Assemble combined message
 131     msg="$(printf '%s\n--------\n' "$msgSec1" "$msgSec2" "$msgSec3")";
 132     ## Display message
 133     zenity --info --title="gThumb-analyze.sh stats" --no-wrap --text="<span font='monospace'>${msg}</span>";
 134 };
 135
 136 main "$@" 1>>"$fdebug" 2>&1;
 137
 138 # Author: Steven Baltakatei Sandoval
 139 # License: GPLv3+
 140