commit 084d62225cd39ab4733216d6c201c4aeec2e77a7
Author: Tobias Langner <tlangner@bitvalve.org>
Date:   Tue Apr 2 16:09:34 2024 +0200

    Initial commit of bash script.

diff --git a/photos_bestof.sh b/photos_bestof.sh
new file mode 100755
index 0000000..49192e6
--- /dev/null
+++ b/photos_bestof.sh
@@ -0,0 +1,201 @@
+#!/bin/bash
+
+source "$( dirname $( readlink -f "$0" ) )/shflags.sh"
+
+# Filter options
+DEFINE_integer min_rating 3 "The min rating for files to be processed (0-5)."
+DEFINE_integer max_age_days -1 "The maximum age (since the creation date) of an image to be processed."
+DEFINE_string exiftool_filter "" "Only process images that satisfy the condition, which must be a string that can be evaluated by exiftool's -if statement."
+
+# Path options
+DEFINE_string photos_dir "" "The directory that contains the photos to be processed."
+DEFINE_string bestof_dir "" "The directory where the photos above --min_rating should be copied to. Path will be created if it does not exist."
+
+# Process options
+DEFINE_boolean resize false "Whether to resize the photos before copying to --resolution."
+DEFINE_string resize_to 1920 "The resolution to which to resize the photos if --resize is set. Can be anything that \\\`convert -resize\\\` accepts."
+DEFINE_boolean flatten false "Whether to flatten the directory structure to have all image files in 'best_of' directory."
+DEFINE_boolean force "false" "Whether to force reprocessing of all files instead of just new files."
+DEFINE_boolean prune "false" "Whether to remove files already existing in the bestof folder that do not meet the filter criteria."
+DEFINE_boolean use_hardlinks "false" "Whether to use hardlinks instead of copying. Cannot be combined with --resize."
+
+# Debug options
+DEFINE_boolean verbose false "Whether to print all actions performed."
+DEFINE_boolean dry_run false "Only print what would be done but do not do anything."
+DEFINE_boolean debug false "Whether to print bash debug output."
+
+set -euo pipefail
+
+EXCLUDE_MARKER=".bestof-exclude"
+PROCESSED_FILES_STORE=".bestof-processed"
+
+function error() {
+  echo "$@" >&2
+}
+
+function verbose() {
+  if [[ ${FLAGS_verbose} -eq ${FLAGS_TRUE} ]]; then
+    error "$@"
+  fi
+}
+
+function paths_in_same_filesystem() {
+  fs_id1=$( stat -c "%d" "$1" )
+  fs_id2=$( stat -c "%d" "$2" )
+  [[ "$fs_id1" = "$fs_id2" ]]
+}
+
+function validate_flags() {
+  [[ ! -d "${FLAGS_photos_dir}" ]] && error "--photos_dir must be set and point to a valid directory." && return 1
+  [[ -z "${FLAGS_bestof_dir}" ]] && error "--bestof_dir must be set." && return 1
+  if [[ ${FLAGS_resize} -eq ${FLAGS_TRUE} ]]; then
+    [[ -z "${FLAGS_resize_to}" ]] && error "--resize_to cannot be empty when --resize is given." &&  return 1
+  fi
+
+  if [[ ${FLAGS_use_hardlinks} -eq ${FLAGS_TRUE} ]]; then
+    if [[ ${FLAGS_resize} -eq ${FLAGS_TRUE} ]]; then
+      error "--use_hardlinks and --resize are incompatible."
+      return 1
+    fi
+
+    if ! paths_in_same_filesystem "${FLAGS_photos_dir}" "${FLAGS_bestof_dir}"; then
+      error "--use_hardlinks can only be used when --photos_dir and --bestof_dir are in the same filesystem."
+      return 1
+    fi
+  fi
+  return 0
+}
+
+function process_flags() {
+  if [[ ${FLAGS_dry_run} -eq ${FLAGS_TRUE} ]]; then
+    echo "Dry-run: Not doing anything..."
+  fi
+
+  if [[ ${FLAGS_debug} -eq ${FLAGS_TRUE} ]]; then
+    set -x
+  fi
+  mkdir -p "${FLAGS_bestof_dir}"
+
+  return 0
+}
+
+function is_already_processed() {
+  if [[ ${FLAGS_force} -eq ${FLAGS_FALSE} ]] && \
+     [[ -f "$PROCESSED_FILES_STORE_PATH" ]] && \
+     look -b "$photo_path" "$PROCESSED_FILES_STORE_PATH" &> /dev/null; then
+    # Return that file does not need to be processed again.
+    return 0
+  fi
+
+  # Mark file as processed for next time.
+  echo "$photo_path" >> "$PROCESSED_FILES_STORE_PATH"
+
+    # Return that file should be processed.
+  return 1
+}
+
+function maybe_run() {
+  if [[ ${FLAGS_dry_run} -eq ${FLAGS_FALSE} ]]; then
+    # Run the command
+    "$@"
+  fi
+}
+
+function file_passes_filters() {
+  local file_path
+  file_path="$1"
+
+  condition='$Rating >= '${FLAGS_min_rating}
+  if [[ ${FLAGS_max_age_days} -gt -1 ]]; then
+    condition="$condition and ( not ShiftTime(\$now,\"-${FLAGS_max_age_days} 0\") and \$CreateDate gt \$now )"
+  fi
+  if [[ -n "${FLAGS_exiftool_filter}" ]]; then
+    condition="$condition and ${FLAGS_exiftool_filter}"
+  fi
+  # Use -TagDoesNotExist to suppress any output by exiftool as we only care about the return value.
+  verbose "exiftool -if \"$condition\" -TagDoesNotExist -q -q -m \"$photo_path\""
+  exiftool -if "$condition" -TagDoesNotExist -q -q -m "$photo_path"
+}
+
+function main() {
+  validate_flags
+  process_flags
+
+  PROCESSED_FILES_STORE_PATH="${FLAGS_bestof_dir}/$PROCESSED_FILES_STORE"
+  [[ -f "$PROCESSED_FILES_STORE_PATH" ]] && LC_COLLATE=C sort -u -d -o "$PROCESSED_FILES_STORE_PATH" "$PROCESSED_FILES_STORE_PATH"
+
+  # Exclude directories containing the .bestof-exclude marker. See https://unix.stackexchange.com/questions/468734/find-exclude-path-containing-specific-files for details.
+  # Also exclude the bestof dir if it is part of the photos_dir
+  total=$( find "${FLAGS_photos_dir}" \( -exec [ -f {}/.bestof-exclude ] \; -prune \) \
+    -o -path "${FLAGS_bestof_dir}" -prune \
+    -o -iname "*.jpg" -print | wc -l )
+  
+  count=0
+  processed_cnt=0
+  while read photo_path; do
+    count=$(( $count + 1 ));
+    if [[ $(( $count % 10 )) -eq 0 ]] || [[ $count -eq $total ]]; then
+      echo "$( date +%T ) Visited $count/$total image files. Images copied to best-of folder: $processed_cnt..."
+    fi
+
+    verbose "Processing file $photo_path."
+
+    # Only record processed state when not doing a dry run.
+    if [[ ${FLAGS_dry_run} -eq ${FLAGS_FALSE} ]] && is_already_processed; then
+      verbose "  Skipping file as it was already processed."
+      continue
+    fi
+  
+    rel_path="${photo_path#${FLAGS_photos_dir}}"
+    if [[ ${FLAGS_flatten} -eq ${FLAGS_TRUE} ]]; then
+      # Drop the complete path if we are flattening.
+      rel_path=$( basename "$rel_path" )
+    fi
+    bestof_path="${FLAGS_bestof_dir}/${rel_path}"
+
+    # Delete file with the same name if it already exists.
+    [[ -f "${bestof_path}" ]] && rm "${bestof_path}"
+  
+    if ! file_passes_filters "$photo_path"; then
+      verbose "  File does not pass filters."
+      continue
+    fi
+
+    # Create all parent directories.
+    mkdir -p "${bestof_path%/*}"
+    if [[ ${FLAGS_resize} -eq ${FLAGS_TRUE} ]]; then
+      verbose "  Resizing file into target directory."
+      maybe_run convert "${photo_path}" -resize "${FLAGS_resize_to}" "${bestof_path}"
+    else
+      if [[ ${FLAGS_use_hardlinks} -eq ${FLAGS_TRUE} ]]; then
+        verbose "  Hardlinking file into target directory."
+        maybe_run ln "${photo_path}" "${bestof_path}"
+      else
+        verbose "  Copying file into target directory."
+        maybe_run cp "${photo_path}" "${bestof_path}"
+      fi
+    fi
+    processed_cnt=$(( $processed_cnt + 1 ))
+  done < <( find "${FLAGS_photos_dir}" \( -exec [ -f "{}/$EXCLUDE_MARKER" ] \; -prune \) \
+              -o -path "${FLAGS_bestof_dir}" -prune \
+              -o -iname "*.jpg" -print )
+
+  pruned_cnt=0
+  if [[ ${FLAGS_prune} -eq ${FLAGS_TRUE} ]]; then
+    echo "Pruning filtered files in ${FLAGS_bestof_dir}..."
+    while read photo_path; do
+      echo "Processing file $photo_path."
+      if ! file_passes_filters "$photo_path"; then
+        verbose "  Pruning file as it does not pass the filters."
+        maybe_run rm "$photo_path"
+        pruned_cnt=$(( $pruned_cnt + 1 ))
+      fi
+    done < <( find "${FLAGS_bestof_dir}" -iname "*.jpg" )
+    [[ $pruned_cnt -gt 0 ]] && echo "Pruned $pruned_cnt files from best-of directory."
+  fi
+}
+
+# Parse the command-line
+FLAGS "$@" || exit $?
+eval set -- "${FLAGS_ARGV}"
+main "$@"