[
https://issues.apache.org/jira/browse/TIKA-1367?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16074175#comment-16074175
]
Gus Heck commented on TIKA-1367:
--------------------------------
So when the dust settles here, how will one build a coherent, workable one-jar
application that supports code like this that intends to make a best effort to
parse any document that might be encountered:
{code}
Tika tika = new Tika();
tika.setMaxStringLength(document.getRawData().length);
Metadata metadata = new Metadata();
try (ByteArrayInputStream bais = new ByteArrayInputStream(rawData)) {
String textContent = tika.parseToString(bais, metadata);
document.setRawData(textContent.getBytes(Charset.forName("UTF-8")));
for (String name : metadata.names()) {
document.put(sanitize(name) + plusSuffix(), metadata.get(name));
}
} catch (IOException | TikaException e) {
log.warn("Tika processing failure!", e);
// if tika can't parse it we certainly don't want random binary crap in
the index
document.setStatus(Status.DROPPED);
}
{code}
Although I notice that this is not marked as fixed yet, in 1.15, the above code
no-longer compiles... (and somehow there are no dependencies reported by
gradle...)
{code}
compile - Dependencies for source set 'main'.
+--- org.apache.tika:tika-parsers:1.15
+--- org.apache.solr:solr-solrj:5.5.0
| +--- commons-io:commons-io:2.4
{code}
vs
{code}
+--- org.apache.tika:tika-parsers:1.12
| +--- org.apache.tika:tika-core:1.12
| +--- org.gagravarr:vorbis-java-tika:0.6
| | \--- org.apache.tika:tika-core:1.5 -> 1.12
| +--- com.healthmarketscience.jackcess:jackcess:2.1.2
{code}
Which seems very much like it's totally going to break everything... if gradle
doesn't see the deps, one-jar won't package them (all I did was change a 1.12
to a 1.15 in the gradle build to cause this)
> Tika documentation should list tika-parsers parser dependencies
> ---------------------------------------------------------------
>
> Key: TIKA-1367
> URL: https://issues.apache.org/jira/browse/TIKA-1367
> Project: Tika
> Issue Type: Improvement
> Components: documentation
> Reporter: Sergey Beryozkin
> Fix For: 1.16
>
>
> tika-parsers module has many strong transitive parser dependencies. Maven
> users of tika-parsers have to exclude all the transitivie dependencies
> manually. Documenting the list of the existing transitive dependencies and
> keeping the list up to date will help developers exclude the libraries not
> needed for a given project.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
