[TIKA-403] Refactor log library usage in tika-parsers

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 0.7
• Fix Version/s: 0.8
• Assignee: Jukka Zitting
• Reporter: Attila Király
• Created: 11/Apr/10 14:13
• Updated: 29/Mar/17 15:41
• Resolved: 14/Apr/10 13:27

Description

Currently tika has the following direct depencdencies declared in its pom.xml-s:

tika-core:

• none

tika-parent:

• none

tika-parsers:

• commons-logging 1.1.1
• log4j 1.2.14

In tika-parsers pom.xml it is stated that commons-logging is only a transitive dependency and it is not used in tika only in poi and it is only there so tika pulls the 1.1.1 version of commons-logging instead of the 1.1 version which is pulled by poi. For this usecase there is a better option in maven: move this depencency declaration from the <depencencies> element to the <dependencyManagement> element. Like so:
<dependencyManagement>
<dependencies>
<dependency>
<groupId>commons-logging</groupId>
<artifactId>commons-logging</artifactId>
<version>1.1.1</version>
</dependency>
</dependencies>
</dependencyManagement>
The result will be the same but tika-parsers will no more say that it directly needs commons-logging as dependency.

I did a search for "log" in the java files of tika-core and tike-parsers and it seems that only one class uses logging and its org.apache.tika.parser.mbox.MboxParser referencing log4j. I think it should be trivial to modify this class to not depend on log4j and so the log4j depencency could also be removed from the direct dependencies of tika-parsers.