EXIFmover.py

# Name: EXIFmover
# Forked from EXIFmover.py by Brian Klug (@nerdtalker / [email protected])

# Purpose:
# Move Files into directory based on dates
# Designed to un-clusterfuck the Dropbox camera upload directory which is a mess as all files are
# uploaded to same folder.
# Moves files into /YEAR/MONTH eg /2014/03
# Creates directory if it doesn't exist, moves into that directory if it exists
# Files without "YYYY-mm-dd" pattern on its name are scanned looking for exif data
# Files without EXIF are keep at root folder
# Only files with valid extensions are moved

# This is experimental and one-way in a destructive sense, I take no responsibility
# if this absolutely destroys your directory structure for some reason

# I STRONGLY recommend making a copy of Camera Uploads, then running this on the copy, first

# Requires EXIF-PY to be installed and importable in order to check exif. If all your files are named using
# the "YYYY-mm-dd" pattern, there is no need to have EXIF-PY installed

# EXIF-PY can be obtained from https://github.com/ianare/exif-py or using pip: "sudo pip install ExifRead"

# Run simply  inside "Camera Upload" folder or whatever you have the media you want to organize
# (eg from ipython "run exifmover.py" inside "Camera Upload")
# or "python exifmover.py"

# Tested on Ubuntu
# Extremely fast: Sorted more than 1400 pictures ranging 4 years in less than a second

try: 
    import exifread
except:
    exifread = None
    print "exifread was not found, exif data won't be used"
import os
import time
import re

start_time = time.time()

current_path = os.path.abspath(os.getcwd())
# current_path = "/home/otger/Dropbox/Camera Uploads"
file_names = os.listdir(current_path)
print "Found {0} items on folder {1}".format(len(file_names), current_path)
valid_extensions = ["png", "jpg", "jpeg", "mp4"]

## expression matches: YYYY-mm-dd
re_date = re.compile(r"(\d{4})[:\-\.]?(\d{2})[:\-\.]?\d{2}")
for fname in file_names:
    file_path = os.path.join(current_path, fname)
    if os.path.isfile(file_path):

        if fname.split('.')[-1] in valid_extensions:
            year = month = None
            fname_date = re_date.search(fname)
            if fname_date:
                (year, month) = fname_date.groups()
            elif exifread:
                f = open(file_path)
                try:
                    tags = exifread.process_file(f)
                except Exception, ex:
                    print "Couldn't read tags on " + fname
                    print ex.message
                    continue
                finally:
                    f.close()
                exif_date = re_date.search(str(tags.get("Image DateTime", "NA")))
                if not  exif_date:
                    exif_date = re_date.search(str(tags.get("EXIF DateTimeOriginal", "NA")))
                    if not exif_date:
                        print "Couldn't find date on exif for " + fname
                        continue
                (year, month) = exif_date.groups()

                print file_path + "date extracted by exif: {0}-{1}".format(year,month)
            else:
                print "exifread not found, skipping " + file_path

            dst_path = os.path.join(current_path, year, month)
            dst_file_path = os.path.join(dst_path, fname)

            # print "{0} would be moved to {1}".format(fname, dst_path)
            if not os.path.exists(dst_path):
                print "Creating folder: " + dst_path
                try:
                    os.makedirs(dst_path)
                except Exception, ex:
                    print "Failed to create folder " + dst_path
                    print ex.message

            try:
                print "Moving from {0} to {1}".format(file_path, dst_file_path)
                os.rename(file_path, dst_file_path)
            except Exception, ex:
                print "Oh noes. That didn't work for some reason:"
                print ex.message
        else:
            print "Skipping " + file_path
    else:
        print "Not a file: " + file_path
print 'Done. Execution took {:0.3f} seconds'.format((time.time() - start_time))