Bounding boxes are missing for many feeds with good stops.txt files

[dancory-urbanfootprint]

Describe the bug

Many feeds have good stops.txt files do not have bounding boxes in the API response or the spreadsheet. They did have bounding boxes in the old spreadsheet.

A few examples:
mdb-195
mdb-631
mdb-210
mdb-125
mdb-2122
mdb-2064
mdb-37
mdb-233
mdb-234
mdb-2299

Steps/Code to Reproduce

Navigate to https://mobilitydatabase.org/feeds/gtfs/mdb-195

Expected Results

There should be a bounding box for this feed

Actual Results

Unable to generate bounding box from the feed's stops.txt file.

Screenshots

No response

Files used

No response

Additional notes

No response

[welcome]

Thanks for opening your first issue in this project! If you haven't already, you can join our slack and join the #mobility-database channel to meet our awesome community. Come say hi 👋!

Welcome to the community and thank you for your engagement in open source! 🎉

[dancory-urbanfootprint]

My guess is this related to #918 but I don't know for sure. Certainly it's apparent with the switch to the new spreadsheet.

[koshy1123]

Looks like mdb-195 and mdb-631 now do have bounding boxes when they were updated over the weekend. The others are still missing bounding boxes.

[emmambd]

Hi @koshy1123 and @dancory-urbanfootprint — thanks for looking into this! Our team will investigate. The old spreadsheet only extracted the bounding box once, at time of pull request creation. Now we update it every time a new dataset is created, so the issue is likely in our extraction process or a problem with the stops.txt file. We'll keep you updated as we look further.

[qcdyx]

Followup issue #1230 created.

[emmambd]

Next steps for this issue:

Add task in cloud task function to identify all feeds containing a stops.txt file but missing a bounding box. we need to create a task executor that checks all the feeds that are have stops.txt and no bounding box.

Tasks:

• Create a task as part of the tasks executor with dryrun parameter.
• The tasks should get all feeds without a bounding box and verify the existence of the stop.txt(GTFS)
• For the feeds without a bounding box and a stop.txt present, create a task to extract locations
• If dryrun is true, don't create a task only response with how many feeds miss the bounding box
• Execute the task in PROD

Post release:

• Trigger it
• Check if all feeds without a location are updated or if any remain on Metabase

[emmambd]

Outstanding feeds that return no bounding box (some of these have seemed to resolve themselves):

mdb-125
mdb-2064
mdb-233
mdb-234