Set file modification time for exporting images

piwigo_export.py

@@ -24,6 +24,7 @@ python3 piwigo_export.py ... --metadata exif iptc xmp
 """
 
 import argparse
+import datetime
 import json
 import math
 import os
@@ -198,6 +199,46 @@ def _xmp_datetime(s) -> str:
     return f'{s[:10]}T{t}'
 
 
+def _parse_datetime(s) -> datetime.datetime | None:
+    """Parse a DB or EXIF date string into a datetime, or return None."""
+    s = str(s).strip()
+    for fmt in ('%Y-%m-%d %H:%M:%S', '%Y-%m-%d',
+                '%Y:%m:%d %H:%M:%S', '%Y:%m:%d'):
+        try:
+            return datetime.datetime.strptime(s[:len(fmt)], fmt)
+        except ValueError:
+            continue
+    return None
+
+
+# EXIF/IPTC/XMP tags that carry a capture or creation date, in preference order.
+_DATE_TAGS_IN_IMAGE = (
+    'EXIF:DateTimeOriginal',
+    'EXIF:CreateDate',
+    'XMP-xmp:CreateDate',
+    'IPTC:DateCreated',
+    'EXIF:ModifyDate',
+)
+
+
+def _earliest_image_date(image_path: pathlib.Path) -> datetime.datetime | None:
+    """Return the earliest datetime found in the image's embedded metadata.
+
+    Returns None if exiftool is not available or no date tags are found.
+    """
+    if not shutil.which('exiftool'):
+        return None
+    existing = _read_existing_metadata(image_path)
+    dates = []
+    for tag in _DATE_TAGS_IN_IMAGE:
+        val = existing.get(tag)
+        if val:
+            dt = _parse_datetime(str(val))
+            if dt:
+                dates.append(dt)
+    return min(dates) if dates else None
+
+
 def _build_metadata_tags(metadata: dict, fmt: str) -> dict:
     """
     Build a dict of  { 'GROUP:TagName': value }  for everything we want to write.
@@ -469,11 +510,15 @@ def export_image(
     tags     = tags_by_image.get(image_id, [])
     cat_ids  = cats_by_image.get(image_id, [])
 
+    # Collect both date sources before building the metadata dict.
+    date_embedded_dt = _earliest_image_date(src_file)
+
     metadata = {
         'title':         image_row.get('name'),
         'author':        image_row.get('author'),
         'date_created':  str(image_row['date_creation'])  if image_row.get('date_creation')  else None,
         'date_added':    str(image_row['date_available']) if image_row.get('date_available') else None,
+        'date_embedded': str(date_embedded_dt)            if date_embedded_dt                else None,
         'description':   image_row.get('comment'),
         'tags':          tags,
         'albums':        [category_display_path(cid, categories) for cid in cat_ids],
@@ -486,6 +531,12 @@ def export_image(
         'original_path': image_row['path'],
     }
 
+    # Print both date sources so the user can see any discrepancy.
+    piwigo_str   = metadata['date_created'] or '—'
+    embedded_str = metadata['date_embedded'] or '—'
+    filename_str = pathlib.Path(image_row['path']).name
+    print(f'  {filename_str}  piwigo: {piwigo_str}   embedded: {embedded_str}')
+
     dest_dirs = (
         [output_dir / category_fs_path(cid, categories) for cid in cat_ids]
         if cat_ids
@@ -539,53 +590,8 @@ def export_image(
 # Entry point
 # ---------------------------------------------------------------------------
 
-def main():
+def cmd_export(args):
-    parser = argparse.ArgumentParser(
+    """Run the export subcommand (database → filesystem)."""
-        description='Export Piwigo photos with JSON metadata sidecars.',
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-        epilog=__doc__,
-    )
-
-    db = parser.add_argument_group('database')
-    db.add_argument('--dbhost',     metavar='HOST')
-    db.add_argument('--dbuser',     required=True, metavar='USER')
-    db.add_argument('--dbpassword', metavar='PASS')
-    db.add_argument('--dbname',     required=True, metavar='NAME')
-    db.add_argument(
-        '--db-prefix', default='piwigo_', metavar='PREFIX',
-        help='Piwigo table prefix (default: %(default)s)',
-    )
-
-    io = parser.add_argument_group('paths')
-    io.add_argument(
-        '--src-path', required=True, metavar='DIR',
-        help='Root of the Piwigo installation; piwigo_images.path is relative to this.',
-    )
-    io.add_argument(
-        '--output-dir', required=True, metavar='DIR',
-        help='Directory to write exported files into (created if absent).',
-    )
-
-    behaviour = parser.add_argument_group('behaviour')
-    behaviour.add_argument(
-        '--metadata', choices=['exif', 'iptc', 'xmp'], nargs='+', metavar='FORMAT',
-        help='Also embed metadata into the exported image copy using exiftool. '
-             'One or more of: exif, iptc, xmp. '
-             'Example: --metadata exif iptc xmp',
-    )
-    behaviour.add_argument(
-        '--overwrite', action='store_true',
-        help='Re-export image files that already exist in the output directory. '
-             'JSON sidecars are always refreshed.',
-    )
-    behaviour.add_argument(
-        '--no-overwrite-metadata', action='store_true',
-        help='When embedding metadata, never overwrite a tag that already has a '
-             'value in the file — skip it silently instead of prompting.',
-    )
-
-    args = parser.parse_args()
-
     if args.metadata:
         check_exiftool()
 
@@ -646,5 +652,173 @@ def main():
     )
 
 
+def cmd_set_dates(args):
+    """Walk an export directory and set each image's mtime from its sidecar dates."""
+    output_dir = pathlib.Path(args.output_dir)
+    if not output_dir.is_dir():
+        sys.exit(f'ERROR: output directory not found: {output_dir}')
+
+    sidecars = sorted(output_dir.rglob('*.json'))
+    if not sidecars:
+        print('No JSON sidecars found.')
+        return
+
+    applied = skipped = noted = 0
+
+    for sidecar in sidecars:
+        # Load the sidecar.
+        try:
+            data = json.loads(sidecar.read_text(encoding='utf-8'))
+        except (json.JSONDecodeError, OSError) as exc:
+            print(f'WARNING: could not read {sidecar}: {exc}', file=sys.stderr)
+            continue
+
+        # Find the corresponding image file (same stem, any non-.json extension).
+        image_path = None
+        for candidate in sidecar.parent.iterdir():
+            if candidate.stem == sidecar.stem and candidate.suffix.lower() != '.json':
+                image_path = candidate
+                break
+        if image_path is None:
+            print(f'WARNING: no image file found for {sidecar.name}', file=sys.stderr)
+            continue
+
+        # Parse the two candidate dates from the sidecar.
+        dt_piwigo   = _parse_datetime(data['date_created'])  if data.get('date_created')  else None
+        dt_embedded = _parse_datetime(data['date_embedded']) if data.get('date_embedded') else None
+
+        # Determine which date to apply.
+        if args.use == 'piwigo':
+            chosen = dt_piwigo
+            if chosen is None:
+                print(f'  NOTE: {image_path.name}: no piwigo date in sidecar; skipping.')
+                noted += 1
+                continue
+
+        elif args.use == 'embedded':
+            chosen = dt_embedded
+            if chosen is None:
+                print(f'  NOTE: {image_path.name}: no embedded date in sidecar; skipping.')
+                noted += 1
+                continue
+
+        else:
+            # Interactive mode.
+            if dt_piwigo is None and dt_embedded is None:
+                print(f'  NOTE: {image_path.name}: no dates available; skipping.')
+                noted += 1
+                continue
+
+            if dt_piwigo == dt_embedded or (dt_piwigo and dt_embedded is None):
+                chosen = dt_piwigo
+            elif dt_embedded and dt_piwigo is None:
+                chosen = dt_embedded
+            else:
+                # Both present and different — ask the user.
+                print(f'\n{image_path.name}')
+                print(f'  [1] piwigo   : {dt_piwigo}')
+                print(f'  [2] embedded : {dt_embedded}')
+                print(f'  [s] skip')
+                while True:
+                    choice = input('Choice [1/2/s]: ').strip().lower()
+                    if choice in ('s', 'skip', ''):
+                        chosen = None
+                        break
+                    if choice == '1':
+                        chosen = dt_piwigo
+                        break
+                    if choice == '2':
+                        chosen = dt_embedded
+                        break
+                    print('  Please enter 1, 2, or s.')
+
+                if chosen is None:
+                    skipped += 1
+                    continue
+
+        ts = chosen.timestamp()
+        os.utime(image_path, (ts, ts))
+        applied += 1
+
+    print(
+        f'\nDone.  {applied} mtime(s) set, {skipped} skipped, {noted} with no date.'
+    )
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description='Piwigo photo export and date-management utilities.',
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog=__doc__,
+    )
+    sub = parser.add_subparsers(dest='command', required=True)
+
+    # ------------------------------------------------------------------ export
+    ep = sub.add_parser(
+        'export',
+        help='Export photos from Piwigo to a directory tree with JSON sidecars.',
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+
+    db = ep.add_argument_group('database')
+    db.add_argument('--dbhost',     metavar='HOST')
+    db.add_argument('--dbuser',     required=True, metavar='USER')
+    db.add_argument('--dbpassword', metavar='PASS')
+    db.add_argument('--dbname',     required=True, metavar='NAME')
+    db.add_argument(
+        '--db-prefix', default='piwigo_', metavar='PREFIX',
+        help='Piwigo table prefix (default: %(default)s)',
+    )
+
+    io = ep.add_argument_group('paths')
+    io.add_argument(
+        '--src-path', required=True, metavar='DIR',
+        help='Root of the Piwigo installation; piwigo_images.path is relative to this.',
+    )
+    io.add_argument(
+        '--output-dir', required=True, metavar='DIR',
+        help='Directory to write exported files into (created if absent).',
+    )
+
+    behaviour = ep.add_argument_group('behaviour')
+    behaviour.add_argument(
+        '--metadata', choices=['exif', 'iptc', 'xmp'], nargs='+', metavar='FORMAT',
+        help='Also embed metadata into the exported image copy using exiftool. '
+             'One or more of: exif, iptc, xmp. '
+             'Example: --metadata exif iptc xmp',
+    )
+    behaviour.add_argument(
+        '--overwrite', action='store_true',
+        help='Re-export image files that already exist in the output directory. '
+             'JSON sidecars are always refreshed.',
+    )
+    behaviour.add_argument(
+        '--no-overwrite-metadata', action='store_true',
+        help='When embedding metadata, never overwrite a tag that already has a '
+             'value in the file — skip it silently instead of prompting.',
+    )
+
+    # --------------------------------------------------------------- set-dates
+    dp = sub.add_parser(
+        'set-dates',
+        help='Set each exported image\'s mtime from dates recorded in its JSON sidecar.',
+    )
+    dp.add_argument(
+        '--output-dir', required=True, metavar='DIR',
+        help='Directory containing the exported files and JSON sidecars.',
+    )
+    dp.add_argument(
+        '--use', choices=['piwigo', 'embedded'], metavar='SOURCE',
+        help='Auto-select a date source (piwigo or embedded) instead of '
+             'prompting for each image.',
+    )
+
+    args = parser.parse_args()
+    if args.command == 'export':
+        cmd_export(args)
+    else:
+        cmd_set_dates(args)
+
+
 if __name__ == '__main__':
     main()