**UNDT Processing Functions**
Button	Description
	Retrieves UNDT English and French metadata for all years and builds the UNDT table entries
	Retrieves UNDT English and French metadata for this year and builds the UNDT table entries
	Retrieves UNDT English and French metadata for all years
	UNDT Process Management Tools
	Stop UNDT build process

**UNAT Processing Functions**
Button	Description
	Retrieves UNAT English and French metadata for all years and builds the UNAT table entries
	Retrieves UNAT English and French metadata for this year and builds the UNAT table entries
	Processes UNAT data and Downloads/OCR's as needed(
	UNAT Process Management Tools
	Stop UNAT build process

**ILOAT Processing Functions**
Button	Description
	Processes ILOAT data and Downloads/OCR's as needed
	Finds sessions ILO has published that aren't harvested yet, and pulls them (rate-limited) into the scratch area - never touches production
	Same as above, capped at 5 documents - a quick sanity check before running a full session's backlog
	Reattach to the background harvest (running/complete/failed - runs on the server regardless of this page being open)
	Copies sessions/documents already harvested into scratch into the real production database and corpus - only after you've tested them
	Read-only cross-check of our ILOSessions ranges against what ILO's site currently lists - HTML pages only, no PDF/document hits
	Same as above, capped at 5 sessions - a quick sanity check before checking every session on ILO's site
	Reattach to the background consistency check and see any discrepancies found so far
	ILOAT Process Management Tools
	Stop ILOAT build process
Results:	Ready

**System Testing Functions**
Button	Description
	Test collection info update end editing
	Display connection pool info
	See which files need downloading and ocring
	Fetches one ILOAT judgment (EN) into a scratch temp file - confirms scrape.do + ilo.org's URL pattern still work
	Fetches one UNAT judgment into a scratch temp file
	Fetches one UNDT judgment into a scratch temp file
Re-OCR - corrects documents OCR'd with the wrong tesseract language model before this fix (French documents were OCR'd as English). Overwrites existing text output in place; does not re-download anything.
	Runs at low priority in the background - can take hours for the full corpus. Check status any time, even after navigating away.
Test Completeness of Collection - read-only gap report: missing database rows (ILOAT only, via its session case-range), missing PDFs, missing OCR text.
	Runs synchronously - just filesystem/database checks, no network calls or OCR
Language Tag Check - flags documents whose actual OCR'd text doesn't match their tagged language (e.g. a French document mistakenly tagged English).
	"Flag" resets both language versions of each mismatched case for redownload/re-OCR and clears their current (wrong) search index entries - run "Check Only" first to see the scope

Incremental Update

TBD

Testing

Simple function testing