Starting to fill in bits for detecting duplicate uploads

flickrscripts.py

@@ -1,4 +1,5 @@
 import json
+from hashlib import sha1
 from os.path import expanduser
 
 import flickrapi
@@ -15,6 +16,30 @@ def setup_flickr(api_key, api_secret):
     return flickr
 
 
+def get_photo_signature(flickr, photo_id):
+    info = flickr.photos_getInfo(photo_id=photo_id).find('photo')
+    exif = flickr.photos_getExif(photo_id=photo_id).find('photo')
+
+    if not info or not exif:
+        raise InvalidPhotoIdError(photo_id)
+
+    return [
+        info.get('originalformat'), info.find('owner').get('location'),
+        info.find('dates').get('taken')
+    ] + sorted(
+        (e.get('label'), e.get('tag'), e.get('tagspace'), e.get('tagspaceid'),
+         e.find('raw').text) for e in exif.findall('exif')
+    )
+
+
+def get_photo_sha1sum(flickr, photo_id):
+    return sha1(str(get_photo_signature(flickr, photo_id)))
+
+
 def get_flickr_from_rc_file(rc_file=expanduser('~/.flickrscripts.json')):
     rc_conf = json.load(open(rc_file))
     return setup_flickr(rc_conf['APIKEY'], rc_conf['APISECRET'])
+
+
+class InvalidPhotoIdError(ValueError):
+    pass