Resume attachment downloads #8940
Conversation
services/src/main/java/org/fao/geonet/api/records/attachments/AttachmentsApi.java
Outdated
Show resolved
Hide resolved
| String originalFilename, int size) throws IOException { | ||
| // Set headers before processing image | ||
| String filename = originalFilename.substring(0, originalFilename.lastIndexOf('.')) + ".png"; | ||
| response.setHeader(HttpHeaders.CONTENT_DISPOSITION, "inline; filename=\"" + filename + "\""); |
There was a problem hiding this comment.
Not sure if inline is the best option here, probably it should use attachment?
There was a problem hiding this comment.
Do you also think all of the resource downloads should use attachment? I don't see why only the resized images would use attachment while all other resource downloads use inline.
The use of inline was pre-existent and I figured the reasoning was so that users could view images or other attachments that are viewable in the browser without needing to download it and open it manually in the browser or another viewer. If the browser does not support displaying the file in the browser it seems to revert to downloading it as an attachment so this seems to give the best of both worlds.
If others agree that all attachment downloads should be downloaded directly instead of displaying in the browser when possible I will update to attachment.
There was a problem hiding this comment.
it seems fine to me @tylerjmchugh
services/src/main/java/org/fao/geonet/api/records/attachments/AttachmentsApi.java
Show resolved
Hide resolved
Co-authored-by: Juan Luis Rodríguez Ponce <[email protected]>
There was a problem hiding this comment.
Tested to upload a big file and download with curl interrupting the download and continuing, and the file is
curl --limit-rate 1000k -O http://localhost:8080/geonetwork/srv/api/records/78f93047-74f8-4419-ac3d-fc62e4b0477b/attachments/bigfile.zip
CTRL+C
curl --limit-rate 1000k -C - -O http://localhost:8080/geonetwork/srv/api/records/78f93047-74f8-4419-ac3d-fc62e4b0477b/attachments/bigfile.zip
The download is resumed and the file is correct also.
Please check the small comment added.
| long skipped = 0; | ||
| while (skipped < start) { | ||
| long toSkip = start - skipped; | ||
| long actuallySkipped = bufferedStream.skip(toSkip); | ||
| if (actuallySkipped <= 0) break; // EOF | ||
| skipped += actuallySkipped; | ||
| } |
Currently attachment downloads cannot be resumed. This means that if an attachment download is almost complete and an error is encountered, the entire download needs to be restarted. Depending on the size of the file this may be problematic and time consuming.
This PR aims to fix this issue by adding support for the
Rangeheader. This enables the browser to resume the download by specifying which bytes it still needs to download.Checklist
mainbranch, backports managed with labelREADME.mdfilespom.xmldependency management. Update build documentation with intended library use and library tutorials or documentation