IQSS / dataverse

Open source research data repository software
http://dataverse.org
Other
878 stars 485 forks source link

Not ingested file into Dataverse #10108

Open sbondka opened 10 months ago

sbondka commented 10 months ago

I've provided a link to a binary file that was uploaded to Dataverse on December 1, 2021. You can access it here: (https://entrepot.recherche.data.gouv.fr/file.xhtml?persistentId=doi:10.15454/YUFRUW/ICGF0B&version=2.0) After downloading the file in its original .dat format and attempting to reupload it to Dataverse, the file was not successfully ingested. Please refer to the screenshot below for more details.

Capture2

qqmyers commented 10 months ago

A quick guess - Dataverse only correctly ingests Stata v 14 and 15 files correctly if they have the right version specific mimetype, e.g. application/x-stata-14 . Detecting that version-specific type may not happen for direct upload since detection requires inpecting the file bits. Fixing this is on my radar for QDR - not sure how quickly. At QDR, this particular file does get ingested now when uploaded w/o direct upload and fails due to having application/x-stata as its mimetype when direct uploaded. For now, doing a mimetype redetect and reingest via API is a work-around.