"Can't Get Date On" Message

[QFieldBoden]

Hi, thank you for all your work on this. I've just downloaded all my Google Photos and burned the .zip files to M-Discs for long term storage. I decided to just see whether GPTH would be a good option to extract all the images should I need to in the future as I was delighted when I found it!

I've got it all installed and up and running and it's whizzed through 70,000+ photos so that was encouraging.

I don't understand the "Can't Get Date On.......file location... file name" message I'm seeing on the screen as the photos all have dates on them in Google Photos. Is it something to do with some of the images being scanned prints I've uploaded or something like that, rather then original camera data?

Any suggestions? Should I be concerned?

[N3orun]

Same is happening to me on OSX 26.1. Files pretty much all have their according supplemental-metadata.json with the correct info - for example:

{
"title": "IMG_6507.JPG",
"description": "",
"imageViews": "7",
"creationTime": {
"timestamp": "1474706463",
"formatted": "24.09.2016, 08:41:03 UTC"
},
"photoTakenTime": {
"timestamp": "1469288496",
"formatted": "23.07.2016, 15:41:36 UTC"
},
"geoData": {
"latitude": 0.0,
"longitude": 0.0,
"altitude": 0.0,
"latitudeSpan": 0.0,
"longitudeSpan": 0.0
},
"url": "https://photos.google.com/photo/0000-lj",
"googlePhotosOrigin": {
"mobileUpload": {
"deviceType": "IOS_PHONE"
}
}
}

but its still telling me it couldn't get the data.

[nazimcanislam]

TL;DR: The main reason for “Can’t get date on …” is not missing JSON, but filename mismatches: Google Takeout often exports metadata as
photo.jpg.supplemental-metadata.json / photo.jpg.metadata.json / photo.jpg.suppl.json,
while GPTH expects strictly photo.jpg.json.
Additionally, emoji / unusual Unicode in filenames prevents matching on some systems.
I fixed ~12k items by bulk-renaming those JSONs and normalizing filenames; only 2 files remained to fix manually.
(Solved together with ChatGPT—collaborative debugging + scripting.)

Root cause(s)

1. Since late 2023/2024, Takeout frequently emits sidecars like:

IMG_20210101_123456.jpg.supplemental-metadata.json
VID_20201231_235959.mp4.metadata.json
somefile.jpg.suppl.json

GPTH looks only for basename.ext.json, so it doesn’t see the JSON that’s right next to the media.

2. A tiny subset of files had emoji / non-BMP Unicode or weird Unicode forms (NFD vs NFC), causing path/regex mismatches on Windows.

What worked for me (free, CLI only)

A) Normalize sidecar names → *.json

• Windows PowerShell

Set-Location "D:\Takeout"

# Collapse common suffixes to plain ".json"
Get-ChildItem -Recurse -File -Include *.suppl.json,*.metadata.json,*.supplemental-metadata.json |
ForEach-Object {
    $dest = $_.FullName -replace '\.(suppl|metadata|supplemental-metadata)\.json$', '.json'
    if (-not (Test-Path $dest)) {
        Rename-Item -LiteralPath $_.FullName -NewName $dest
    }
}

# Optional: remove " (1)" etc. after ".json"
Get-ChildItem -Recurse -File | Where-Object { $_.Name -match '\.json \(\d+\)$' } | ForEach-Object {
    $dest = $_.FullName -replace ' \(\d+\)$',''
    if (-not (Test-Path $dest)) { Rename-Item -LiteralPath $_.FullName -NewName $dest }
}

• Bash (Linux/macOS)

find . -type f -name "*.supplemental-metadata.json" -exec bash -c 'mv "$1" "${1%.supplemental-metadata.json}.json"' _ {} \;
find . -type f -name "*.metadata.json"              -exec bash -c 'mv "$1" "${1%.metadata.json}.json"' _ {} \;
find . -type f -name "*.suppl.json"                 -exec bash -c 'mv "$1" "${1%.suppl.json}.json"' _ {} \;

# Optional: remove trailing " (n)" after .json
find . -type f -regex '.*\.json \([0-9]+\)$' -exec bash -c 'mv "$1" "${1% *}"' _ {} \;

B) Handle emoji / odd Unicode
Some filenames had emoji/non-BMP characters which broke matching. I normalized names to NFC and replaced “unsafe” characters, moving media + its sidecar together (so photo.jpg and photo.jpg.json stay in sync).
If needed I can share the full PowerShell script; key ideas were:

• Normalize([Text.NormalizationForm]::FormC)
• strip surrogate pairs: [\uD800-\uDBFF][\uDC00-\uDFFF]
• replace forbidden/odd chars with _, dedupe spaces/underscores, and avoid collisions by appending (1), (2), …

After A+B, running GPTH again processed nearly everything; only 2 emoji-heavy files needed manual attention.

---

Proposal: make GPTH tolerant to new sidecar names & Unicode

Minimal change (no heavy parsing): when searching for a JSON, also accept any basename.ext.*.json and prefer exact equality when present.

Pseudo-logic / regex idea:

• When you have a media path: .../basename.ext
• Try candidates in order:

1. Exact: basename.ext.json

2. Suffix variants: any of

   • basename.ext.supplemental-metadata.json
   • basename.ext.metadata.json
   • basename.ext.suppl.json
   • or more generally: basename\.ext\.[^.]+\.json

In Dart (conceptual sketch):

import 'package:path/path.dart' as p;

String? findSidecarFor(String mediaPath) {
  final dir = p.dirname(mediaPath);
  final base = p.basename(mediaPath);            // e.g., "photo.jpg"
  final exact = p.join(dir, '$base.json');
  if (File(exact).existsSync()) return exact;

  final stem = base; // include ext; we only allow extra token(s) before .json
  final dirList = Directory(dir).listSync().whereType<File>();
  final candidates = dirList.where((f) {
    final name = p.basename(f.path);
    // Accept "photo.jpg.<anything>.json", but avoid nested dots after .json
    return RegExp(r'^\Q$stem\E\.[^.]+\.json$', caseSensitive: false).hasMatch(name);
  }).toList();
  candidates.sort((a,b) => p.basename(a.path).length.compareTo(p.basename(b.path).length)); // shortest suffix first
  return candidates.isNotEmpty ? candidates.first.path : null;
}

Unicode/emoji matching:

• Normalize both the discovered filenames and the expected names to NFC before comparing.
• Optionally case-fold (toLowerCase() with proper locale) and use package:characters if needed.

This small expansion keeps current behavior but makes GPTH compatible with all recent Takeout variants and emoji-heavy filenames. Happy to open a PR if maintainers want this upstream.

---

Results

• Dataset: ~12,000 Google Photos items (photos + videos).
• After renaming sidecars and normalizing emoji/unicode, GPTH processed almost all without “Can’t get date…” warnings.
• Only 2 files left for manual fix (extreme emoji cases).
• Collaboration: I worked this out together with ChatGPT (prompting, scripting, and validating edge cases).

Thanks for maintaining this super helpful tool!