iNaturalistReviewer

Automatic iNaturalist reviewer for Commons.

Usage

inrbot

One-shot:

• inrbot.py --file File:Example.png Runs the bot on File:Example.png only.

Bot mode:

• inrbot.py --auto Runs the bot continuously.
• inrbot.py --auto --total 5 Runs the bot on 5 files.

Optional args:

• --simulate Prints the wikitext output instead of writing to the wiki.
• --ignore-runpage Skips the on-wiki runpage check for use in development. Sets --simulate unless overridden with --no-simulate.
• -h --help Prints help information
• --version Prints version information

inrcli

inrcli is a modified interface to inrbot that makes manual reviewing of files easier. You will be prompted to manually input the correct photo URL or to re-attempt automatic detection. Confirmation of the license status is usually handled automatically, unless there is an archived copy in the Wayback Machine that needs to be manually reviewed. To skip a file, press Ctrl+C.

Reviewing specific files:

• inrcli.py File:Example.png Reviews File:Example.png only.
• inrcli.py File:Example.png --url https://www.inaturalist.org/photos/12345 Reviews File:Example.png with https://www.inaturalist.org/photos/12345 as the iNaturalist source URL.
• inrcli.py ask Repeatedly prompts for files to review.

Reviewing previously-reviewed files:

• inrcli.py auto Review files from https://commons.wikimedia.org/wiki/Category:INaturalist_images_needing_human_review
• inrcli.py errors Review files from https://commons.wikimedia.org/wiki/User:INaturalistReviewBot/untagged_error_log

Deployment

This bot runs on Toolforge as inaturalistreviewer with the python3.11 Kubernetes container. It uses the toolforge jobs configuration in jobs.yaml and assumes that there is a python 3.11 virtualenv in /data/project/inaturalistreviewer/iNaturalistReviewer/venv/.

To stop the bot: toolforge jobs delete inrbot

To start the bot in automatic mode: toolforge jobs load ~/iNaturalistReviewer/jobs.yaml --job inrbot

To get the pod status: toolforge jobs list

Updating the bot

With only code changes:

$ cd iNaturalistReviewer
$ git pull
$ toolforge jobs restart inrbot

With code changes and/or dependency updates:

$ cd iNaturalistReviewer
$ git pull
$ toolforge jobs load jobs.yaml

Commons intergration

The bot is controlled by a runpage on Commons. If the runpage does not end with True, the bot will stop cleanly. Blocking the bot will also stop it from running. Using the runpage is preferred as it is faster and easier for everyone involved.

The bot looks for images in Category:iNaturalist review needed that transclude {{iNaturalistreview}}. It will also attempt to review files that are likely sourced to iNaturalist that have not yet been tagged with the template.

If the bot can not automatically determine the license status, it sets |status=error with the review date and the reviewer parameters. Files that were not previously tagged will instead be listed at User:INaturalistReviewBot/untagged error log to avoid spurious tags.

If the bot determines that the image is freely licensed on iNaturalist, it sets |status=pass with the author, source URL, review date, reviewer, and review license parameters.

If the bot determines that the image is freely licensed on iNaturalist but the Commons license is wrong, it will change the license template on Commons and set |status=pass-change with the author, source URL, review date, reviewer, review license, and upload license parameters.

If the bot determines that the image is under a non-free license on iNaturalist, it will add {{copyvio|Bot license review NOT PASSED: iNaturalist author is using <review_license> to the top of the file and set |status=fail with the author, source url, review date, reviewer, and review license parameters. Older files will instead be tagged with {{No permission since}} to allow the uploader a chance to correct the license.

|reason= parameter

The bot will add a |reason= parameter when it tags files. This parameter is not read by the template, and exists mostly for debugging.

• nourl: No iNaturalist /photos/ or /observations/ URL was found on the file page.
• apierr: No data was recieved from the iNaturalist API. This usually indicates a problem with how the API is called, not the API itself.
• notmatching: No photo on iNaturalist was found that matched the Commons file. That usually means the Commons file has been modified or the URL is wrong.
• sha1: The Commons file was matched to an iNaturalist photo using a SHA-1 hash. Errors with this reason typically indicate a problem comparing the licenses.
• phash: The Commons file was matched to an iNaturalist photo based on a perceptual hashing algorithm.
• crop_resistant_hash The Commons file was matched to an iNaturalist photo using a crop-resistant hashing algorithm.
• manual: The Commons file was manually matched to an iNaturalist photo using inrcli.
• Other Python exceptions when reviewing fails in an interesting way.