From: Alexander Schmidt <alex@treefish.org>
Date: Fri, 16 Oct 2020 13:59:48 +0000 (+0200)
Subject: extract meta time for video files
X-Git-Url: http://git.treefish.org/~alex/photosort.git/commitdiff_plain/727a9ebdde342105d608e7f657ad24462967891e

extract meta time for video files
---

diff --git a/src/misc.py b/src/misc.py
index d47a9cf..cd31858 100644
--- a/src/misc.py
+++ b/src/misc.py
@@ -2,7 +2,9 @@ import datetime
 import mimetypes
 import os
 import PIL.Image
+import re
 import shutil
+import subprocess
 
 def walk_media_files(dir_path):
     for root, dirs, files in os.walk(dir_path):
@@ -11,15 +13,17 @@ def walk_media_files(dir_path):
             if _is_media_file(file_path):
                 yield (f, file_path)
 
-def extract_timestamp(file_path, use_exif=False):
-    if use_exif and _is_media_file(file_path, types=['image']):
-        with PIL.Image.open(file_path) as image:
-            exif = image._getexif()
-            if exif and 36867 in exif:
-                return int( datetime.datetime
-                            .strptime(exif[36867], '%Y:%m:%d %H:%M:%S')
-                            .timestamp() )
-    return os.path.getmtime(file_path)
+def extract_timestamp(file_path, use_meta=False):
+    time = None
+    if use_meta:
+        if _is_media_file(file_path, types=['image']):
+            time = _extract_image_timestamp(file_path)
+        elif _is_media_file(file_path, types=['video']):
+            time = _extract_video_timestamp(file_path)
+    if time:
+        return time
+    else:
+        return os.path.getmtime(file_path)
 
 def find_file(dir_path, file_name, file_size, exclude_dir):
     for root, dirs, files in os.walk(dir_path):
@@ -55,3 +59,23 @@ def _is_media_file(file_path, types=['image', 'video']):
     if not mime_type.split('/')[0] in types:
         return False
     return True
+
+def _extract_image_timestamp(file_path):
+    with PIL.Image.open(file_path) as image:
+        exif = image._getexif()
+        if exif and 36867 in exif:
+            return datetime.datetime\
+                           .strptime(exif[36867], '%Y:%m:%d %H:%M:%S')\
+                           .timestamp()
+    return None
+
+def _extract_video_timestamp(file_path):
+    p = subprocess.run(['ffmpeg', '-i', file_path],
+                       capture_output=True, encoding='UTF-8')
+    for line in p.stderr.splitlines():
+        m = re.search('^.*creation_time.*: ([^ ]+)$', line)
+        if m:
+            return datetime.datetime\
+                           .strptime(m.group(1), '%Y-%m-%dT%H:%M:%S.%fZ')\
+                           .timestamp()
+    return None
diff --git a/src/photosort.py b/src/photosort.py
index a4b319e..efe1790 100755
--- a/src/photosort.py
+++ b/src/photosort.py
@@ -24,10 +24,10 @@ logging.basicConfig(format='[%(asctime)s] %(levelname)s: %(message)s',
 for src_file_name, src_file_path in misc.walk_media_files(args.SOURCE_DIR):
     logging.info('Processing %s...', src_file_name)
 
-    exif_time = misc.extract_timestamp(src_file_path, use_exif=True)
+    meta_time = misc.extract_timestamp(src_file_path, use_meta=True)
 
     dst_dir = os.path.join(args.DEST_DIR,
-                           datetime.datetime.fromtimestamp(exif_time).strftime("%Y/%m"))
+                           datetime.datetime.fromtimestamp(meta_time).strftime("%Y/%m"))
     dst_file_path = os.path.join(dst_dir, src_file_name)
 
     if not os.path.exists(dst_file_path):