[
https://issues.apache.org/jira/browse/TIKA-4692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18070914#comment-18070914
]
Tim Allison commented on TIKA-4692:
-----------------------------------
SAX is now a fair amount better than DOM. It is able to handle ooxml-strict,
uses less memory/fewer dependencies and is faster (7 min vs 12 min) on a
freshly pulled commoncrawl set of 92k docx/pptx.
> Move PPTX and DOCX SAX parsers closer to parity with DOM based parsers
> ----------------------------------------------------------------------
>
> Key: TIKA-4692
> URL: https://issues.apache.org/jira/browse/TIKA-4692
> Project: Tika
> Issue Type: Task
> Reporter: Tim Allison
> Priority: Major
> Attachments: eval-reports-sax-v-dom-pptx-docx.zip
>
>
> SAX processing of pptx and docx is more robust in cases where elements are
> embedded more than usual. We should work to promote the sax parsers so that
> we can have parity with features extracted by the DOM based parsers.
> Ideally, we'd move to using sax based as the default in 4.x on a separate
> ticket), but we should back port the updates to 3.x as well.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
