Adrian Bird created TIKA-4475:
---------------------------------
Summary: ExifTool does not work with Tika on Windows
Key: TIKA-4475
URL: https://issues.apache.org/jira/browse/TIKA-4475
Project: Tika
Issue Type: Bug
Affects Versions: 3.2.2
Environment: Windows 11 with Tika 3.2.2.
Reporter: Adrian Bird
I tried to get Tika and ExifTool to work together to process some JPEG image
files and came across a number of issues. The issues are listed here, with a
description of what I did below:
1) Tika and ExifTool doesn't work on Windows
I used the [Wiki
page|https://cwiki.apache.org/confluence/display/TIKA/EXIFToolParser] to
understand how to do the integration.
Because I wasn't getting the metadata I expected, I used the '--verbose' option
and got a Java Exception which contained this text:
"WARN [main] 07:13:34,699 org.apache.tika.parser.external.ExternalParser
problem with process exec
java.io.IOException: Cannot run program "env": CreateProcess error=2, The
system cannot find the file specified"
The exception occurs because 'env' is not a valid Windows command.
I tracked this down to the file
'org\apache\tika\parser\external\tika-external-parsers.xml' in the Tika App jar
where the command is:
'<command>env FOO=${OUTPUT} exiftool ${INPUT}</command>'
This doesn't work on Windows because 'env' does not exist.
As a test I changed the command and updated my copy of the Tika-App.jar and it
then worked with a video file. The changed command was:
'<command>exiftool ${INPUT}</command>'
2) For the same reason Tika and sox won't work on Windows
The command is:
<command>env FOO=${OUTPUT} sox --info ${INPUT}</command>
Note I didn't find any information on 'sox' on the Wiki.
3) Looking at the file
'org\apache\tika\parser\external\tika-external-parsers.xml' I noticed that it
only contains video related mime-types, meaning that I cannot use it with image
files. The Wiki page says:
'EXIFTool is a wonderful tool that reads videos, images, audio and other media
files and that extracts EXIF metadata from them.'
I took this to mean that Tika can extract metadata form all 3 file types, but
that isn't the case as it only supports video files.
Given this can I suggest that the Wiki page should be updated to make this
clear.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
