Small photo archiving script for personal use

Photo_Archiver.py

import Tkinter
import tkFileDialog
import os
import hashlib
import datetime
import pytz
import shutil
from PIL import Image
from PIL.ExifTags import TAGS

UK_TIME_ZONE = pytz.timezone("Europe/London")
SEATTLE_TIME_ZONE = pytz.timezone("US/Pacific")
IMAGE_EXTENSIONS = ["png","bmp","raw","jpg","jpeg","tiff"]

def get_md5(file_path):
    return hashlib.md5(open(file_path, 'rb').read()).hexdigest()

def get_directory(prompt, initial_directory="/"):
    tk_root = Tkinter.Tk()
    directory = tkFileDialog.askdirectory(parent=tk_root, initialdir=initial_directory, title=prompt)
    if len(directory) > 0:
        return directory
    else:
        return None

def get_photos(directory):
    for dir_name, subdir_list, file_list in os.walk(directory):
        for filename in file_list:
            ext = filename.split(".")[-1]
            if ext.lower() in IMAGE_EXTENSIONS:
                yield os.path.join(dir_name, filename)

def parse_exif_datetime(value):
    return datetime.datetime.strptime(value, '%Y:%m:%d %H:%M:%S')

def get_exif(photo_path):
    photo = Image.open(str(photo_path))
    exif_data = {}

    # Populate Exif Data
    try:
        exif_raw_data = photo._getexif()
        for tag, value in exif_raw_data.items():
            decoded = TAGS.get(tag, tag)
            exif_data[decoded] = value
    except:
        pass

    if len(exif_data.keys()) == 0:
        return {}

    # Normalise Exif Data
    for key, value in exif_data.items():

        if type(key) != str:
            exif_data[str(key)] = exif_data[key]
            del exif_data[key]
            key = str(key)

        if type(value) == str or type(value) == unicode:
            exif_data[key] = value.strip()

        if key.startswith("DateTime"):
            exif_data[key] = parse_exif_datetime(value)

    if "MakerNote" in exif_data.keys():
        del exif_data["MakerNote"]

    return exif_data

def set_date_info(photo_info):

    new_time = None

    for key in ["DateTimeOriginal", "DateTimeDigitized", "DateTime"]:
        try:
            new_time = photo_info["exif"][key]
            break
        except:
            pass

    if new_time == None:
        try:
            new_time = datetime.datetime.strptime(photo_info["raw_name"], '%Y-%m-%d %H.%M.%S')
        except:
            return False

    photo_info["new_name"] = new_time.strftime('%Y-%m-%d_%H.%M.%S')
    photo_info["year"] = "%02d" % (new_time.year,)
    photo_info["month"] = "%02d" % (new_time.month,)
    photo_info["day"] = "%02d" % (new_time.day,)
    photo_info["hour"] = "%02d" % (new_time.hour,)
    photo_info["minute"] = "%02d" % (new_time.minute,)
    photo_info["second"] = "%02d" % (new_time.second,)

    return True

def get_name_and_extension(photo_info):
    photo_info["extension"] = photo_info["path"].split(".")[-1]
    tempname = photo_info["path"]
    tempname = tempname.replace("\\","/")
    tempname = tempname.split("/")[-1]
    tempname = tempname[:-len("." + photo_info["extension"])]
    photo_info["raw_name"] = tempname

def get_photo_info(photos, directory):
    for photo in get_photos(directory):
        photo_info = {"path":photo}
        get_name_and_extension(photo_info)
        photo_info["exif"] = get_exif(photo_info["path"])
        if not set_date_info(photo_info):
            print "Unable to find date: " + photo_info["path"]
            continue
        photo_info["md5"] = get_md5(photo)
        #print photo_info
        photos.append(photo_info)

def get_name_count(file_path, name, file_hash):
    test_name = os.path.splitext(name)[0]

    matches = 0

    for existing_file in os.listdir(file_path):
        if existing_file.startswith(name):
            if get_md5(os.path.join(file_path,existing_file)) == file_hash:
                print "Existing: " + os.path.join(file_path,existing_file)
                return -1
            matches += 1

    return matches + 1

def add_extension(filename, extension):
    temp_name = filename
    temp_ext = extension

    if temp_name[-1] == "/" or temp_name[-1] == "\\":
        temp_name = temp_name[:-1]

    if extension[0] == ".":
        temp_ext = extension[1:]

    return temp_name + "." + temp_ext.lower()

def move(photo_info, photos_dir):

    # Check if photos_dir exists
    if not os.path.isdir(photos_dir):
        os.makedirs(photos_dir)

    # Check if full dir with year and month exists
    full_dir = os.path.join(photos_dir, photo_info["year"], photo_info["month"])
    if not os.path.isdir(full_dir):
        os.makedirs(full_dir)

    # Check if file with same name exists
    count = get_name_count(full_dir, photo_info["new_name"], photo_info["md5"])

    if count > 1:
        photo_info["new_name"] += "_" + str(count)

    if count == -1:
        # There is an identical file so don't bother moving, and just delete
        print "Deleting file due to already existing: ",
        os.remove(photo_info["path"])
        return

    full_path = os.path.join(full_dir, photo_info["new_name"])

    # Add extension back in
    full_path = add_extension(full_path, photo_info["extension"])

    # Copy the file
    shutil.move(photo_info["path"], full_path)

def main():
    photos_dir = get_directory('Where should the photos be moved to?')
    new_dir = get_directory("Where are the photos to be moved?")
    photos = []
    get_photo_info(photos, new_dir)

    print "Found all info. Beginning move..."

    for photo in photos:
        move(photo, photos_dir)

    print photos

if __name__ == "__main__":
    main()