txt detect as application/octet-stream

[MrSwed]

Example txt txt.txt detect as application/octet-stream

[gabriel-vasile]

I tested your file using the latest release of mimetype and it is correctly detected as text/plain.

[MrSwed]

the problem occurs with a large buffer, which is necessary to determine the office files #230

Details, click to open

[gabriel-vasile]

Oh, I see. The problem appears when you call mimetype.SetLimit(1024*1024) before detection. Let me see what's the problem and how to fix it.

[gabriel-vasile]

The problem is the test file contains the 0x0B (vertical tab) character at index 4478.

Computer files can be either binary files or text files.
mimetype uses the MIME sniffing standard to detect text files.
This standard says: If the resource header contains no binary data bytes, the computed MIME type is "text/plain".

0x0B is considered binary data byte in this standard so the file is not detected as text/plain.

If you want these types of files to be detected as text/plain you can extend mimetype to detect other formats.

[gabriel-vasile]

There are different approaches to what is considered a text file.
One is the MIME sniffing standard I linked above.
The Linux file utility for example, considers text files any file that does not contain the null 0x00 byte. For this reason, file detects your file as text/plain.

[zdiff]

Do you think it would benefit this project to allow users to choose the binary file detection method: MIME sniffing standard or the looser Linux file standard?