[
https://issues.apache.org/jira/browse/PDFBOX-6249?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18110015#comment-18110015
]
Tilman Hausherr edited comment on PDFBOX-6249 at 8/31/26 6:43 PM:
------------------------------------------------------------------
I am beginning to understand what happens. The "filter by CIDSystemInfo, if
given" part doesn't match. PDFBox ends up using DroidSans, which is a mess.
(Had there be no fonts at all on the system, it would have used the Liberation
Sans fallback and it would have worked). The Droid Sans font does not have a
proper "A". 0xFF21 is "Fullwidth Latin Capital Letter A". The real "A" would be
0x0041. I think that font should be avoided like we already do for barcode
fonts.
!image-2026-08-31-20-35-16-505.png!
was (Author: tilman):
I am beginning to understand what happens. The "filter by CIDSystemInfo, if
given" part doesn't match (the CJK fonts are OTF fonts, not TTF). PDFBox ends
up using DroidSans, which is a mess. (Had there be no fonts at all on the
system, it would have used the Liberation Sans fallback and it would have
worked). The Droid Sans font does not have a proper "A". 0xFF21 is "Fullwidth
Latin Capital Letter A". The real "A" would be 0x0041. I think that font should
be avoided like we already do for barcode fonts.
!image-2026-08-31-20-35-16-505.png!
> Non-embedded Adobe-GB1/CNS1/Japan1 fonts render as .notdef even when Noto CJK
> / Source Han fonts are installed
> --------------------------------------------------------------------------------------------------------------
>
> Key: PDFBOX-6249
> URL: https://issues.apache.org/jira/browse/PDFBOX-6249
> Project: PDFBox
> Issue Type: Task
> Components: Rendering
> Reporter: Tim Allison
> Priority: Minor
> Attachments: cjk-cid-repro.pdf, image-2026-08-31-18-22-38-997.png,
> image-2026-08-31-20-35-16-505.png
>
>
> I ran a rendering bake-off with pdftoppm vs PDFBox on the 1k pdfs in [digital
> corpora
> 0000.zip|https://digitalcorpora.s3.amazonaws.com/corpora/files/CC-MAIN-2021-31-PDF-UNTRUNCATED/zipfiles/0000-0999/0000.zip]
>
> There was a small handful of files where pdftoppm and pdfium were handling
> missing cjk fonts a bit better than PDFBox.
>
> Specifically:
> ||file||non-embedded CID font||collection||
> |0000000.pdf|MSungStd-Light-Acro|Adobe-CNS1|
> |0000063.pdf|STSongStd-Light-Acro|Adobe-GB1|
> |0000750.pdf|MSungStd-Light-Acro|Adobe-CNS1|
> |0000877.pdf|KozGoPro-Medium-Acro|Adobe-Japan1|
> |0000908.pdf|MSungStd-Light-Acro|Adobe-CNS1|
>
> I had Claude generate a small repro file (attached).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
