Immich WhatsApp Fix

README.md

This is an script that searches all photos and videos from WhatsApp (name contains -WA), on your Immich library, and replaces their date with the one provided in the file name.

Usage:

API_KEY=XXXX BASE_URL=http://localhost:2283 ./immich_fix_whatsapp.sh

It adds a tag to the assets (WHATSAPP_FIXED) to identify those modified resources, and not update them every time the script is executed.

immich_fix_whatsapp.sh

#!/bin/env bash
# Requires: bash >= 4.0, jq >= 1.5, curl >= 7.58.0
#
# This script uses the immich API to find images that have WhatsApp-like filenames,
# then extracts dates from filenames and timestamps, compares them, and updates the API
# if mismatches are found. It assumes that the filename is correct enough for your needs,
# and that the timestamp is not.
#
# --- Usage ---
# To run this script from the command line, ensure that it is executable and that you
# have the required environment variables set.
# API_KEY: The key used for authenticating API requests.
# BASE_URL: The base URL of the API.
# > API_KEY=XXXX BASE_URL=http://localhost:2283 ./whatsapp.sh

set -euo pipefail

# Ensure API_KEY and BASE_URL are set
if [[ -z "${API_KEY:-}" ]]; then
  echo "Error: API_KEY is not set." >&2
  exit 1
fi

if [[ -z "${BASE_URL:-}" ]]; then
  echo "Error: BASE_URL is not set." >&2
  exit 1
fi

# --- Functions ---
# icurl: A wrapper around curl for making API requests with JSON headers and
# authentication.
icurl() {
  local path="$1"
  shift
  local response
  response=$(/usr/bin/curl "$@" -s -w "\n%{http_code}" --header 'Content-Type: application/json' --header 'Accept: application/json' --header "X-Api-Key: ${API_KEY}" "${BASE_URL}/api${path}")
  local http_code
  http_code=$(echo "$response" | tail -n1)
  local body
  body=$(echo "$response" | head -n-1)

  if [[ $http_code -ge 200 && $http_code -le 299 ]]; then
    echo "$body" | /usr/bin/jq
  else
    echo -e "Error: API request failed with status code $http_code\n$body" >&2
    exit 1
  fi
}

# fetch_metadata: Fetches metadata from the API for a given page.
fetch_metadata() {
  local page="$1"
  local json_payload
  json_payload=$(jq -n --argjson page "$page" '{page: $page, withExif: true, isVisible: true, language: "en-GB", originalFileName: "-WA"}')
  icurl /search/metadata -d "$json_payload"
}

# Tagging constants
TAG_NAME="WHATSAPP_FIXED"
TAG_ID=""
ASSETS_TO_TAG=()

# get_or_create_tag: Retrieves the tag ID or creates it if it doesn't exist.
get_or_create_tag() {
  local tags_response
  tags_response=$(icurl /tags -X GET)
  
  # Try to find existing tag
  TAG_ID=$(echo "$tags_response" | jq -r --arg name "$TAG_NAME" '.[] | select(.name == $name) | .id')
  
  if [[ -z "$TAG_ID" || "$TAG_ID" == "null" ]]; then
    echo "Tag '$TAG_NAME' not found. Creating..."
    local create_response
    local json_payload
    json_payload=$(jq -n --arg name "$TAG_NAME" '{"type":"CUSTOM","name":$name}')
    create_response=$(icurl /tags -X POST -d "$json_payload")
    TAG_ID=$(echo "$create_response" | jq -r '.id')
    echo "Created tag '$TAG_NAME' with ID: $TAG_ID"
  else
    echo "Found tag '$TAG_NAME' with ID: $TAG_ID"
  fi
}

# bulk_tag_assets: Adds the fixed tag to all collected assets.
bulk_tag_assets() {
  if [[ ${#ASSETS_TO_TAG[@]} -eq 0 ]]; then
    echo "No assets to tag."
    return
  fi

  echo "Tagging ${#ASSETS_TO_TAG[@]} assets..."

  # Construct JSON payload using jq to ensure correct formatting
  local json_payload
  # Generate a JSON array of strings from the bash array
  local ids_json
  ids_json=$(printf '%s\n' "${ASSETS_TO_TAG[@]}" | jq -R . | jq -s .)
  
  # Construct the final object
  json_payload=$(jq -n --argjson ids "$ids_json" '{"ids": $ids}')

  icurl /tags/"${TAG_ID}"/assets -X PUT -d "$json_payload" > /dev/null
  echo "Successfully tagged all assets."
}

# Initialize counters
inspected_count=0
changed_count=0
skipped_count=0

# process_asset: Processes an asset by comparing and updating date information if
# necessary. Checks for existing tag to skip.
process_asset() {
  local asset_id="$1"
  local file_name="$2"
  local timestamp="$3"
  local tags_str="$4"
  local exif_date="$5"

  # Check if already tagged
  if [[ "$tags_str" == *"$TAG_ID"* ]]; then
    skipped_count=$((skipped_count + 1))
    return
  fi

  # Increment inspected count
  inspected_count=$((inspected_count + 1))

  # Extract date from filename
  local date_from_filename=""
  if [[ "$file_name" =~ ([0-9]{8})-WA ]]; then
    date_from_filename="${BASH_REMATCH[1]}"
  fi

  if [[ -z "$date_from_filename" ]]; then
    # echo "Skipping $file_name (no date in filename)"
    return
  fi

  # Check against EXIF date first (source of truth)
  local date_from_exif=""
  if [[ -n "$exif_date" && "$exif_date" != "null" ]]; then
    # EXIF format is usually YYYY:MM:DD HH:mm:ss or YYYY-MM-DDTHH:mm:ss
    # Remove separators to get YYYYMMDD
    date_from_exif=$(echo "$exif_date" | cut -c 1-10 | tr -d ':-')
  fi

  if [[ "$date_from_exif" == "$date_from_filename" ]]; then
    # Date is already correct in EXIF, skip
    return
  fi

  # Fallback check against localDateTime if EXIF was missing (or just mismatched)
  # But technically if EXIF was missing, date_from_exif is empty != date_from_filename, so we proceed.
  # If EXIF was present but mismatched, we modify.

  local time_part
  time_part=$(echo "$timestamp" | cut -d'T' -f2-)
  local year=${date_from_filename:0:4}
  local month=${date_from_filename:4:2}
  local day=${date_from_filename:6:2}
  local new_date_part="${year}-${month}-${day}"
  local new_timestamp="${new_date_part}T${time_part}"

  local json_payload
  json_payload=$(jq -n --arg dt "$new_timestamp" '{"dateTimeOriginal": $dt}')
  
  local new_date_time_original
  # We only care if the command succeeds, capturing output for logging
  new_date_time_original=$(icurl /assets/"${asset_id}" -X PUT -d "$json_payload" | jq -r '.exifInfo.dateTimeOriginal')

  echo "Asset: $asset_id, $file_name (EXIF: $exif_date) -> $new_date_time_original"

  # Queue for tagging
  ASSETS_TO_TAG+=("$asset_id")

  # Increment changed count
  changed_count=$((changed_count + 1))
}

# --- Main Execution ---

# 1. Get or Create Tag
get_or_create_tag
if [[ -z "$TAG_ID" || "$TAG_ID" == "null" ]]; then
  echo "Error: Failed to obtain TAG_ID for '$TAG_NAME'. Exiting." >&2
  exit 1
fi

# --- Pagination ---
# PAGE: Tracks the current page of results.
# CONTINUE: Controls whether to continue fetching pages.
PAGE=1
CONTINUE=true
while [[ "$CONTINUE" == true ]]; do
  # Fetch metadata from the API.
  RESULT=$(fetch_metadata "$PAGE")
  if [[ $(echo "$RESULT" | jq '.assets.items | length') -eq 0 ]]; then
    CONTINUE=false
  else
    # Parse items, including tags and dateTimeOriginal.
    while IFS=$'\t' read -r asset_id file_name timestamp tags exif_date; do
      process_asset "$asset_id" "$file_name" "$timestamp" "$tags" "$exif_date"
    done <<< "$(echo "$RESULT" | jq -r '.assets.items[] | [.id, .originalFileName, .localDateTime, ((.tags // []) | map(.id) | join(",")), .exifInfo.dateTimeOriginal] | @tsv')"
    PAGE=$((PAGE + 1))
  fi
done

# Bulk tag assets
bulk_tag_assets

# Print summary
echo "Summary: $inspected_count assets inspected, $changed_count dates changed, $skipped_count skipped."

This worked great. But it ends with an error and the tag is not there. Looks like it is too many pictures. But everything has a date.

I put line 210 "bulk_tag_assets" inside of the while-loop line 205 and get the tag after every 250 items (every page). Therefore it doesn't get too many items and confuses some library.