#!/usr/bin/python
#
# Removes duplicate values for multi-value EXIF tags, like IPTC keywords, from images.
#
# Usage: remove-exif-duplicates ExifTag1,ExifTag2,...,ExifTagN image1.jpg image2.jpg ... imagen.jpg
#
# Copyright: Miguel Angel Vilela <miguev@gmail.com>

import pyexiv2
import sys

if len(sys.argv) < 3:
  print "Usage: %s ExifTag1,ExifTag2,...,ExifTagN" \
      "image1.jpg image2.jpg ... imageN.jpg" % sys.argv[0]
  sys.exit(1)

tags = sys.argv[1].split(',')
images = sys.argv[2:]
while "*.jpg" in images: images.remove("*.jpg")
while "*.JPG" in images: images.remove("*.JPG")

for image in images:
  im = pyexiv2.Image(image)
  im.readMetadata()
  for tag in tags:
    if tag not in im.iptcKeys():
      continue
    tag_values = im[tag]
    if type(tag_values) == type(""):
      continue
    im[tag] = list(set(tag_values))
  im.writeMetadata()
  print "Done with %s" % image