TV time account import

Ziadoa wrote 10 days ago: 1

Hi — I lost my TV Time account before I could pull the official GDPR export (the shutdown caught me out), so I rebuilt a tracking-prod-records-v2.csv from a third-party export tool (Refract's "TV Time Out") instead, matching the column names as best I could from public documentation.

First attempt gave me: "tracking-prod-records-v2.csv is missing the required TV Time columns: key, s_id, is_followed" — I added those. Second attempt now gives: "0 shows were imported, totaling 0 watched episodes and 0 votes... No importable TV series or episode rows were found in the uploaded files." No column-missing error this time, just zero rows accepted.

Could you let me know what's actually being checked at the row level (data types, a required 'type'/status field, valid ID formats, etc.), or whether there's a way to get this looked at manually? Really don't want to lose ~47,000 watched episodes just because I couldn't grab the real export in time. Happy to share a sample of my CSV if that helps debug it. Thanks for building this, by the way — it's clearly saved a lot of people already.

Jan wrote 10 days ago: 2

Hi, thanks for the detailed context. The importer is currently built around the real TV Time tracking-prod-records-v2.csv row format, so matching the column names alone is not enough.

At row level, the important checks are:

Watched episode rows must have a key starting with watch-episode- or rewatch-episode-.

Followed show rows must have a key starting with user-series- and is_followed set to exactly true.

s_id should be the TVDB show ID.

episode_id or ep_id should be the TVDB episode ID.

If the episode TVDB ID cannot be matched, we fall back to s_id + season_number/s_no + episode_number/ep_no.

Numeric ID fields need to contain numeric values, not names/slugs/URLs.

There is no separate type field being checked.

So if you got “0 shows / 0 watched episodes”, the most likely cause is that the reconstructed rows have the right headers but not the expected key values, or the IDs are not TVDB IDs that TVmaze can resolve.

Please don’t post the full file publicly, but if you can share a small anonymized sample, for example 5 watched rows and 2 show/follow rows with user_id, uuid, and any personal fields removed, we can check whether the format can be supported.

Ziadoa wrote 9 days ago: 1

@Jan wrote:
Hi, thanks for the detailed context. The importer is currently built around the real TV Time tracking-prod-records-v2.csv row format, so matching the column names alone is not enough.
At row level, the important checks are:
Watched episode rows must have a key starting with watch-episode- or rewatch-episode-.
Followed show rows must have a key starting with user-series- and is_followed set to exactly true.
s_id should be the TVDB show ID.
episode_id or ep_id should be the TVDB episode ID.
If the episode TVDB ID cannot be matched, we fall back to s_id + season_number/s_no + episode_number/ep_no.
Numeric ID fields need to contain numeric values, not names/slugs/URLs.
There is no separate type field being checked.
So if you got “0 shows / 0 watched episodes”, the most likely cause is that the reconstructed rows have the right headers but not the expected key values, or the IDs are not TVDB IDs that TVmaze can resolve.
Please don’t post the full file publicly, but if you can share a small anonymized sample, for example 5 watched rows and 2 show/follow rows with user_id, uuid, and any personal fields removed, we can check whether the format can be supported.

thank you so much that helped a lot and all worked out at the end, i'm greatful, thanks again :)