I downloaded the html page suggested by Jayesh Shende on TIKA-2382, and I've dumped the proposed output in the RecursiveParserWrapper format.

There are 10 metadata objects. The first contains the main page, and then there are 9 scripts.

I'm not sure what we should do with the src= info, when a script relies on an external resource rather than inlining the code.

Dumb question: what other types besides js can we have? Should we have a mapping from type= to mimetype that we can pass in to the child's metadata?

For now, we're still ignoring <style> elements.

I'd want to require users to turn this behavior on via an HTMLParserConfig.

Big question, what do you think? Other areas for improvements?

draft of patch

For now, this requires that there be content in the <script><script> for there to be an embedded docuement. If there's a src attribute, as in <script src="something.js"></script>, this is represented in the xhtml as it was before, but no MACRO embedded document/attachment is created.

FAILURE: Integrated in Jenkins build Tika-trunk #1293 (See https://builds.apache.org/job/Tika-trunk/1293/)
TIKA-2391 – extract <script> elements as embedded documents (tallison: https://github.com/apache/tika/commit/132d3e7ff0387234c65cfc2983dd2b6ba59ff38b)

• (add) tika-parsers/src/test/resources/org/apache/tika/parser/html/tika-config.xml
• (edit) tika-parsers/src/main/java/org/apache/tika/parser/html/HtmlHandler.java
• (edit) tika-parsers/src/test/java/org/apache/tika/parser/html/HtmlParserTest.java
• (edit) tika-parsers/src/main/java/org/apache/tika/parser/html/HtmlParser.java
• (edit) CHANGES.txt
• (add) tika-core/src/main/java/org/apache/tika/metadata/HTML.java