#!/usr/bin/python
# -*- encoding: utf-8 -*-
#
# Removes IPTC keywords from images.
#
# Usage: remove-iptc-keywords [encoding] keyword1,keyword2,...,keywordn image1.jpg image2.jpg ... imagen.jpg
#
# Copyright: Miguel Angel Vilela <miguev@gmail.com>

import pyexiv2
import sys

IptcTag = 'Iptc.Application2.Keywords'

if len(sys.argv) < 3:
  print "Usage: %s [enc] keyword1,keyword2,...,keywordN " \
      "image1.jpg image2.jpg ... imageN.jpg" % sys.argv[0]
  sys.exit(1)

encoding = sys.argv[1]
keywords = sys.argv[2].split(',')
images = sys.argv[3:]
if encoding not in ('latin-1', 'utf-8'): keywords = encoding + keywords
while "*.jpg" in images: images.remove("*.jpg")
while "*.JPG" in images: images.remove("*.JPG")

for image in images:
  im = pyexiv2.Image(image)
  im.readMetadata()
  if IptcTag not in im.iptcKeys():
    continue
  image_keywords = im[IptcTag]
  if type(image_keywords) == type(""):
    image_keywords = [image_keywords]
  else:
    image_keywords = list(image_keywords)
  if encoding == 'latin-1':
    # Convert ISO-8859-1 text in old (Spanish) tags into UTF-8
    image_keywords = [k.decode('latin-1').encode('utf-8') for k in image_keywords]
  elif encoding == 'utf-8':
    image_keywords = [k.decode('utf-8').encode('latin-1') for k in image_keywords]
  else:
    pass
  print "==", image_keywords
  for keyword in keywords:
    while keyword in image_keywords:
      print "-", keyword
      image_keywords.remove(keyword)
  print ":=", image_keywords
  im[IptcTag] = image_keywords 
  im.writeMetadata()
  print "Done with %s" % image