Refactor how timestamps are stored in AWS adapter

[KenEucker]

At one point, with Imgur, the foundTime and mysteryTime were taken from the image itself, which were stored in epoch time. Then, after needing to migrate images from one album to another, we changed to storing the time as a readable string in the description of the image. This broke several things, including the autopost function which checks if a tag has been posted for longer than an amount of minutes.

Now, with the move to Digital Ocean Spaces, and using an object store with metadata fields, we kept the storing of metadata in a title and description field just like Imgur. Unfortunately, it was not possible to send these human readable strings across the wire from the browser so we were forced to base64 encoded these strings. Keep the same title and description fields with the image was a bad decision, and now we are faced, nearly immediately, with the consequences of that bad decision.

Instead, what we need to do, is just store the tag data in the metadata as json. It will still be base64 encoded (as it is now), but stored as json will be faster and simpler than using the regular expressions to pull out data from a human readable string.

[KenEucker]

STEP 1

Update the file src/aws/helper.ts to add new helper functions for encoding and decoding structured base64-encoded JSON metadata for DigitalOcean Spaces tag uploads.

These helpers will replace the legacy human-readable string metadata previously stored in the x-amz-meta-description field. The new format should be stored in the x-amz-meta-data field.

Implement the following:
getMysteryMetadata(tag: Tag, metadataOverrides?: Record<string, any>): string

Extract only these fields from the Tag:

tagnumber, playerId, mysteryPlayer, mysteryTime, hint, discussionUrl, shareUrl, mentionUrl

Shallow merge any overrides from metadataOverrides before encoding

Return a base64-encoded JSON string

getFoundMetadata(tag: Tag, metadataOverrides?: Record<string, any>): string

Extract only:

tagnumber, playerId, foundPlayer, foundTime, foundLocation, gps, confirmedBoundary

Merge any values from metadataOverrides

Return as base64-encoded JSON

mergeMetadata(mysteryMeta: string | undefined, foundMeta: string | undefined, metadataOverrides?: Record<string, any>): Tag

Decode both metadata strings (if present) and merge them together

Shallow-merge metadataOverrides into the combined result

Use the existing createTagObject() function from src/common to construct and return a full Tag

Throw an error if tagnumber is missing or invalid

Notes:
All metadata keys must use camelCase to match the Tag interface

Do not include mysteryImage, mysteryImageUrl, foundImage, or foundImageUrl in the metadata (they are inferred from the image key)

If JSON parsing fails, return an empty object (we may add legacy regex fallback later)

Place these helpers near the existing metadata logic in src/aws/helper.ts

STEP 2

Refactor Tag Methods to Use JSON Metadata Helpers in biketag-api
Update the following functions in src/aws/helper.ts to use the new JSON metadata helpers:

uploadTagImage

updateTag

loadIndexFromImages

Changes:
uploadTagImage:
If uploading a mystery image, call getMysteryMetadata(tag, metadataOverrides) and assign the result to the x-amz-meta-data field.

If uploading a found image, call getFoundMetadata(tag, metadataOverrides) instead.

Remove logic that generates the legacy human-readable description string.

updateTag:
When preparing S3 CopyObjectCommand metadata for the updated tag, use getMysteryMetadata or getFoundMetadata based on the target image being updated.

Ensure you preserve tagnumber and playerId in all updates.

loadIndexFromImages:
Replace logic that builds Tag objects from raw S3 metadata by calling mergeMetadata(mysteryMeta, foundMeta) where mysteryMeta and foundMeta are pulled from the x-amz-meta-data field (base64 JSON).

Pass any necessary overrides (e.g., inferred slug, URLs, or game) into the third parameter of mergeMetadata.

Notes:
Ensure all putObjectCommand and copyObjectCommand calls include the updated Metadata object with the base64 JSON under x-amz-meta-data

Continue to support legacy images that only include x-amz-meta-description by keeping fallback logic in mergeMetadata

[KenEucker]

This has been implemented.

We still need to go back and update all previous tags to use the new way of storing metadata, so I will leave this ticket open until that work is completed.