Update of @cliss's update of Dr. Drang's photo management scripts. Replaced shutil calls with subprocess calls. Added a log file that gets written to the destination directory. Source can now be a directory tree (using os.walk). Replaced comments/docstrings with slightly better code organization and function names. Replaced path string concatena…

organize-photos.py

#!/usr/bin/env python

"""
organize-photos.py - Organize an unstructured folder tree of photos and movie
    files into a month-and-year folder structure.

Note: This is a minor rewrite of @cliss's extension [1,2] of @drdrang's
photo management scripts [3], and includes a tweak from @jamiepinkham [4,5].
The lists of raw [6] and video [7] file extensions were found elsewhere.

[1] https://gist.github.com/cliss/6854904
[2] http://tumblr.caseyliss.com/day/2013/10/06
[3] http://www.leancrew.com/all-this/2013/10/photo-management-via-the-finder/
[4] https://gist.github.com/jamiepinkham/6984369
[5] https://twitter.com/drdrang/status/389952079763996672
[6] http://www.file-extensions.org/filetype/extension/name/digital-camera-raw-files
[7] http://www.fileinfo.com/filetypes/video
"""

import sys
import os, os.path
import subprocess
from datetime import datetime

## Edit these paths (relative to home) to set input and output locations
sourceDir = "Pictures/Photo Imports"
destDir = "Pictures/Photos"
## No more editing (unless you're fixing/improving the script)

JPG_EXTENSIONS = (  '.jpg', '.jpeg', '.jpe')
RAW_EXTENSIONS = (  '.3fr','.3pr','.arw','.ce1','.ce2','.cib','.cmt','.cr2','.craw','.crw',
                    '.dc2','.dcr','.dng','.erf','.exf','.fff','.fpx','.gray','.grey','.gry',
                    '.iiq','.kc2','.kdc','.mdc','.mef','.mfw','.mos','.mrw','.ndd','.nef','.nop',
                    '.nrw','.nwb','.orf','.pcd','.pef','.ptx','.ra2','.raf','.raw','rw2','.rwl',
                    '.rwz','.sd0','.sd1','.sr2','.srf','.srw','.st4','.st5','.st6','.st7','.st8',
                    '.stx','.x3f','.ycbcra')
PHOTO_EXTENSIONS = JPG_EXTENSIONS + RAW_EXTENSIONS
MOVIE_EXTENSIONS = ('.3g2','.3gp','.asf','.asx','.avi','.flv','.m4v','.mov','.mp4','.mpg',
                    '.rm','.srt','.swf','.vob','.wmv','.aepx','.ale','.avp','.avs','.bdm',
                    '.bik','.bin','.bsf','.camproj','.cpi','.dash','.divx','.dmsm','.dream',
                    '.dvdmedia','.dvr-ms','.dzm','.dzp','.edl','.f4v','.fbr','.fcproject',
                    '.hdmov','.imovieproj','.ism','.ismv','.m2p','.mkv','.mod','.moi',
                    '.mpeg','.mts','.mxf','.ogv','.otrkey','.pds','.prproj','.psh','.r3d',
                    '.rcproject','.rmvb','.scm','.smil','.snagproj','.sqz','.stx','.swi','.tix',
                    '.trp','.ts','.veg','.vf','.vro','.webm','.wlmp','.wtv','.xvid','.yuv')
SIDECAR_EXTENSIONS = ('.aae',)
VALID_EXTENSIONS = PHOTO_EXTENSIONS + MOVIE_EXTENSIONS

def get_source_date_time(f):
    try:
        if os.path.splitext(f)[1].lower() in MOVIE_EXTENSIONS:
            raise TypeError
        cDate = subprocess.check_output(['sips', '-g', 'creation', f])
        cDate = cDate.split('\n')[1].lstrip().split(': ')[1]
        return datetime.strptime(cDate, "%Y:%m:%d %H:%M:%S")
    except:
        return datetime.fromtimestamp(os.path.getmtime(f))

def get_source_filenames(d):
    src = []
    is_valid = lambda f: os.path.splitext(f)[1].lower() in VALID_EXTENSIONS
    for dirpath, dirnames, filenames in os.walk(d):
        path = os.path.join(d, dirpath)
        src.extend(map(lambda f: os.path.join(path, f), filter(is_valid, filenames)))
    return src

home = os.environ['HOME']
if not sourceDir.startswith(os.path.sep):
    sourceDir = os.path.join(home, sourceDir)
if not destDir.startswith(os.path.sep):
    destDir = os.path.join(home, destDir)
errorDir = os.path.join(destDir, 'Unsorted')
print 'Moving from %s to %s.' % (sourceDir, destDir)

sources = get_source_filenames(sourceDir)
print 'Found %d photos and videos to process.' % len(sources)

if not os.path.exists(destDir):
    os.makedirs(destDir)
if not os.path.exists(errorDir):
    os.makedirs(errorDir)

lastMonth = 0
lastYear = 0
fmt = "%Y-%m-%d %H-%M-%S"
problems = []

# Open a log file to record copy operations and errors
logfd = file(os.path.join(destDir, 'organize-photos.log'), 'w')

for original in sources:
    suffix = 'a'
    orig_base, orig_ext = os.path.splitext(original)
    ext = orig_ext.lower()
    if ext in JPG_EXTENSIONS:
        ext = '.jpg'

    sidecar = None
    sidecar_ext = None
    for sc_ext in SIDECAR_EXTENSIONS:
        for f in (str.upper, str.lower):
            sc = orig_base + f(sc_ext)
            if os.path.exists(sc):
                sidecar = sc
                sidecar_ext = f(sc_ext)
                break

    try:
        pDate = get_source_date_time(original)
        yr = pDate.year
        mo = pDate.month

        if (mo, yr) != (lastMonth, lastYear):
            sys.stdout.write('\nProcessing %04d-%02d...' % (yr, mo))
            lastMonth = mo
            lastYear = yr
        elif ext in MOVIE_EXTENSIONS:
            sys.stdout.write(':')
        else:
            sys.stdout.write('.')

        newname = pDate.strftime(fmt)
        thisDestDir = os.path.join(destDir, '%04d' % yr, '%02d' % mo)
        if ext in MOVIE_EXTENSIONS:
            thisDestDir = os.path.join(thisDestDir, 'movies')
        if not os.path.exists(thisDestDir):
            os.makedirs(thisDestDir)

        duplicate = os.path.join(thisDestDir, newname + ext)
        while os.path.exists(duplicate):
            duplicate = os.path.join(thisDestDir, newname + suffix + ext)
            suffix = chr(ord(suffix) + 1)

        if subprocess.call(['cp', '-p', original, duplicate]) != 0:
            raise Exception
        print >>logfd, 'Copied: %s -> %s' % (original, duplicate)

        if sidecar:
            sidecar_copy = os.path.splitext(duplicate)[0] + sidecar_ext.lower()
            if subprocess.call(['cp', '-p', sidecar, sidecar_copy]) != 0:
                print >>logfd, 'Failed to copy sidecar: %s' % sidecar
            else:
                print >>logfd, 'Copied: %s -> %s' % (sidecar, sidecar_copy)

        sys.stdout.flush()

    except Exception:
        unsorted_file = os.path.join(errorDir, os.path.basename(original))
        subprocess.call(['cp', '-p', original, unsorted_file])
        problems.append(original[len(home):])
        print >>logfd, 'Error: unable to organize %s' % original

    except:
        sys.exit("Execution stopped.")

if len(problems) > 0:
    print "\nProblem files:"
    print "\n\t".join(problems)
    print "These can be found in: %s" % errorDir
elif len(sources):
    sys.stdout.write('\n')

logfd.close()
sys.exit(0)

Good fork!

Can it be modified to take into account to manage also the movies captured by the mobile phone?

Absolutely! That was actually the next thing I planned to do, but I hadn't decided how I wanted to deal with movies yet. Basically the extension check on Line 36 should be generalized to movies, and then the move file extensions need to persist until the destination (duplicate) filenames are created on Lines 84 and 86. Right now everything is clobbered into a standard '.jpg', which obviously won't work for movie files.

(Hmm, it would help to get notifications on comments on gists, sorry about that delay!)

The latest revision now organizes RAW camera files in addition to JPGs. Movie files are now handled as well, they are copied into the same year/month-year directory tree but collected under "movies/" subfolders.

Hi, would it be feasable to modify the script to move instead of copy the files? (or delete successfully copied files the moment they are copied)?

Also, great work!!

Sure, I just think copying is generally safer! (Also, it's not clear what people would want to do with the leftover directory tree, so it's better for the script to leave as is.) You could change Line 122 so that the call goes to the shell move command:

        if subprocess.call(['mv', original, duplicate]) != 0:
            raise Exception

        print >>logfd, 'Moved: %s -> %s' % (original, duplicate)

And thanks!