Closed orangewolf closed 1 year ago
Dev Notes: NNP QA - Diem includes an example for how she tested this: newman-numismatic-portal#303 So far when testing in dev, I verified that OCR is generated. We are able to search for text in the home page and find the related work. https://share.getcloudapp.com/4guZZnlW However, I've also noticed that we don't have a universal viewer and created a ticket for that. I believe we will need to have the ability of searching in the UV for the exact word.
No code changes have been needed, since pulling in hyku update. Tested on STAGING:
NOTE: The UV is tiny for some reason. However, I was able to OCR functionality by expanding the UV to full screen.
[ ] We are able to search for OCR text from the home page. It returns relevant works. Demo: https://share.getcloudapp.com/RBuymY4L
[ ] We are able to expand the UV to full screen and search for OCR text in the search bar. If there are results, it will return how many were found within the document and highlight them as well.
Search for OCR txt: "Messrs"
Test with manual upload for a work.
I am able to pass this on staging:
I made a work here- https://lv-hyku-staging.notch8.cloud/concern/generic_works/024c40a1-5355-460c-89a5-cc92e2152fc0
uploaded a panera bread menu pdf.
Searched for "Avocado", a string that is contained in the pdf and it returned the work successfully.
The Leader test in Shana's screenshot doesn't appear to be on this sandbox, so I was unable to test it.
The Cardinal example from 2/18/2022 is great! I only provided a PDF to you for this (although we also have image files for each page). I take it this means that a PDF will suffice? Still trying to determine the best, most efficient approach for upload, since filling in page-level source identifiers and file names for a collection of this size will be arduous if JPEGs are preferred.
@rachel.howard are you ok with closing this? the install does appear to have happened correctly, but we are aware of the uv search issue with ocr and it's being tracked on #90@rachel.howard are you ok with closing this? the install does appear to have happened correctly, but we are aware of the uv search issue with ocr and it's being tracked on #90 (closed)
Testing