[
https://issues.apache.org/jira/browse/FOP-3165?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17932243#comment-17932243
]
Greg K. commented on FOP-3165:
------------------------------
| # If PDF 1.5+ is generated (which includes PDF/UA-1 mode on my point of view)
there is no reason why the table row groups are not tagged.
With version set to 1.5 (or 1.7 for that matter) in fop.xconf, the PDF/UA-1
identifier is only written if pdf-ua-mode is set to PDF/UA-1.
| # If the table row groups {*}are not tagged{*}, the order of the rows should
be {*}in the correct order{*}.
This could possibly be true, but if they're not tagged, what use are they to
screen readers, which is one of the main reasons for generating UA output? It
just makes no sense to map the FO table row group elements to the right tags
_for non-UA output_, but then _not to do so for UA-mode output_.
Indeed, your examples bear out this analysis in testing. PDF 1.7 output, with
UA-1 mode disabled, get the table row group tags. The same with UA-1 mode
enabled get no such tags.
> [PATCH] FOP tags Tables in PDF 1.4 style if PDF/UA-1 mode active
> ----------------------------------------------------------------
>
> Key: FOP-3165
> URL: https://issues.apache.org/jira/browse/FOP-3165
> Project: FOP
> Issue Type: Bug
> Affects Versions: 2.9
> Environment: Windows 10, PAC 2021 for tagging analyzes.
> Reporter: Nico Kutscherauer
> Priority: Major
> Labels: Accessibility, Table, Tagging
> Attachments: FOP-3165.diff, table-14.jpg, table-ua.jpg, table.jpg
>
>
> Hi,
> this small GitHub project shows the problem:
> [https://github.com/nkutsche/fop-pdf-ua-table-tagging-issue]
>
> A
> [table|https://github.com/nkutsche/fop-pdf-ua-table-tagging-issue/blob/main/table.fo]
> is rendered to PDF using FOP 2.9 with three different configs. The
> differences are:
> [Config 1: PDF 1.5 is
> requested|https://github.com/nkutsche/fop-pdf-ua-table-tagging-issue/blob/main/config/fop-config.xml].
> [Config 2: PDF 1.5 is requested and PDF/UA-1 mode is
> active.|https://github.com/nkutsche/fop-pdf-ua-table-tagging-issue/blob/main/config/fop-config-ua.xml]
> [Config 3: PDF 1.4 is
> requested.|https://github.com/nkutsche/fop-pdf-ua-table-tagging-issue/blob/main/config/fop-config14.xml]
>
> I analyzed the result PDFs with PAC 2021 - this are the screenshots focosing
> the table footer:
>
> ||Config 1: PDF 1.5||Config 2: PDF/UA-1||Config 3: PDF 1.4||
> |!table.jpg|width=481,height=200!|!table-ua.jpg|width=386,height=200!|!table-14.jpg|width=468,height=200!|
>
>
> You see that if the {{PDF/UA-1}} mode is *active* the tagging of
> {{{}TBody{}}}, {{THead}} and {{TFoot}} are lost and the {{TFoot}} row is
> missplaced (at least on my point of view).
> The reason for adding the PDF 1.4 version was, that PDF 1.4 knows only the
> tags {{{}Table{}}}, {{{}TR{}}}, {{TH}} and {{TD}} (See [PDF 1.4
> specification|https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/pdfreference1.4.pdf]
> Table 9.22). With PDF 1.5 {{{}TBody{}}}, {{THead}} and {{TFoot}} was
> introduced (see [PDF 1.5
> specification|https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/pdfreference1.5_v6.pdf]
> Table 10.22).
> *Conclusion:* FOP tags tables in PDF 1.4 style if PDF/UA-1 mode is *active*
> even if PDF 1.5 is generated. If PDF/UA-1 mode is *inactive* it tags tables
> in PDF 1.5 style even if PDF 1.4 is generated.
> Am I correct that this is not the desired behavior? I think we could live
> with the PDF 1.4 style if the table footer wouldn't be on top of the body. On
> my point of view this is also in the PDF 1.4 style wrong.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
