Sandro Lackner created TIKA-4394:
------------------------------------
Summary: DXF files with a comment in the first two lines are
sometimes detected as text/plain
Key: TIKA-4394
URL: https://issues.apache.org/jira/browse/TIKA-4394
Project: Tika
Issue Type: Bug
Affects Versions: 3.1.0, 2.9.3
Reporter: Sandro Lackner
Attachments: testDXF_long_top_comment-1.dxf, testDXF_no_comment.dxf,
testDXF_original_comment.dxf, testDXF_short_top_comment.dxf,
testDXF_shorter_top_comment.dxf
When working with DXF files i realized that Tika fails to detect the correct
MIME type for some of them. Instead of "image/vnd.dxf;format=ascii" Tika
detects them as "text/plain".
After some analysis I found out that longer comments in the beginning of the
DXF files lead to this behaviour. Attached you'll find my test files from which
only the "shorter" and "no_comment" ones are detected correctly.
The "original" DXF file contains the comment which lead to this finding in the
first place. We got it from a customer who uses the GEONIS framework for DXF
file creation, which is a ESRI ArcGIS-Desktop extension. It seems to use the
netDXF framework for creation, which writes this comment in the first line.
{noformat}
Dxf file generated by netDxf https://netdxf.codeplex.com, Copyright(C)
2009-2016 Daniel Carvajal, Licensed under LGPL
{noformat}
Looking through the Tika GitHub I realized that the tika-mimetypes.xml contains
a regular expression in the magic match for the mime type
"image/vnd.dxf;format=ascii" that limits the length of a comment at the
beginning of the file to 64 characters.
This regex was added for TIKA-3550 with the following
[commit|https://github.com/apache/tika/commit/e9f36cb7425f25a768eb58936505277ccbecfedf].
In my opinion the limit of 64 characters should be removed and replaced by a
more fitting regular expression for ascii dxf file detection.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
