[
https://issues.apache.org/jira/browse/PDFBOX-2711?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14535850#comment-14535850
]
John Hewson commented on PDFBOX-2711:
-------------------------------------
This is related to the problem from PDFBOX-2509, where the PDF spec defines
some extra rules for Type 0 fonts which use Identity-H/V encodings:
{quote}
if the font is composite and uses a predefined cmap (excluding Identity-H/V)
or if its descendant font uses Adobe-GB1/CNS1/Japan1/Korea1 then ...
{quote}
So there's special handling where we need to check if the dependent font
identifies itself as CJK in its CIDSystemInfo. If so, then we know which UCS-2
CMap to use. I was actually been working on a patch for this already.
> Japanese text not extracted
> ---------------------------
>
> Key: PDFBOX-2711
> URL: https://issues.apache.org/jira/browse/PDFBOX-2711
> Project: PDFBox
> Issue Type: Bug
> Affects Versions: 1.8.8, 2.0.0
> Reporter: Daniel Bonniot de Ruisselet
> Assignee: John Hewson
> Attachments: 150218-pdfbox-1.8.8.txt, 150218-pdfbox-2.0.0.txt,
> 150218-pdftotext.txt, 150218.pdf
>
>
> ExtractText does not return the text content of this PDF. There are just a
> few real characters when running 1.8.8, and none with today's 2.0.0 snapshot.
> I also attach the output from pdftotext 0.26.5 (from poppler-utils), which
> seems to get it mostly right.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
