MimeTypeDetector fails for local URIs

[henini]

When a URL encoded local URI is passed to the function

MimeTypeDetector.isPdfContentType(String url)

the return value turns out to be False if url is a local URI.
If one likes to implement FullTextFinder using a temporary copy of some fetched PDF this needs to be changed.

[stefan-kolb]

You are thinking of URIs like file:/C:/Users/Stefan/Desktop/Github/jabref/build/resources/test/net/sf/jabref/logic/io/empty.pdf?

[henini]

Exactly, the current implementation seems to work for http URLs only.

I also tried to write a test, but trying to get absolute paths in Java annoyed me a bit.

[stefan-kolb]

The question is what URLs/Paths should MimeTypeDetector be able to categorize.
Is C:/Users/Stefan/Desktop/Github/jabref/build/resources/test/net/sf/jabref/logic/io/empty.pdf also a valid input here? Do we really need to check local URLs and paths? At the moment there is no use case (FullTextFinder) for this.

[henini]

URLConnection works for local URIs as well as http URLs.
Some institutional APIs return a PDF file instead of a link to a PDF. Generalising MimeTypeDetector eases writing custom fetchers a bit.

[simonharrer]

👍 for adding a test :)

[tobiasdiez]

Mhhh... the code checks the HTTP header Content-Type and thus fails for local URIs since these don't have a HTTP header.

To be honest, I don't see the use-case to allow local URIs. Are you trying to check that a file with .pdf ending is really a pdf? Can you please elaborate a bit more on what you are trying to achieve.

[henini]

java.net.URLConnection has getContentType(), which returns the Content-Type of a file:// URI as well as the Content-Type of a http:// URL, replacing the code in MimeTypeDetector does not break the current tests

[stefan-kolb]

I already created a test, but will only push it later so it doesn't break the build.