PDF tagging: Some bad semantics: footnote, P tags, etc

[ronaldtse]

Various other bad semantics here and there e.g. footnote indicator marked as “Note”, various P tags covering >1 para, invalid reading order in various places seemingly associated with cross-document links, etc.

From: metanorma/mn2pdf#201

[Intelligent2013]

Example P in P:

This occurs due the block element note inside the block element p:

<p id="_24f2eb3c-d04a-ae7f-b7f6-d7508208e30d">The mass fraction of extraneous matter and defective kernels in husked and milled rice, whether or not parboiled, determined in accordance with <xref target="AnnexA">
		<span class="citeapp">Annex A</span>
	</xref>, shall not be greater than the values specified in <xref target="table1">
		<span class="citetbl">Table 1</span>
	</xref>.<note id="_37eebd65-fcfd-1ff3-167d-42106bc02dc2">
		<name>NOTE</name>
		<p id="_cffbd604-0bed-6331-229b-1db2259caaa8">Lower mass fractions of moisture are sometimes needed for certain destinations depending on the climate, duration of transport and storage. For further details, see <xref type="inline" target="ISO6322-1">ISO 6322-1</xref>, <xref type="inline" target="ISO6322-2">ISO 6322-2</xref> and <xref type="inline" target="ISO6322-3">ISO 6322-3</xref>.</p>
	</note>
</p>

The fo:block for note should be marked with role="Note".

[Intelligent2013]

The fo:block for note should be marked with role="Note".

No, it's wrong.

From Tagged PDF Best Practice Guide::

I.e. note should be moved after p (outside of p). I have to fully refactored the XSLT (with more testing, visual checking, etc.)

[Intelligent2013]

May be there is more simple solution - post-processing PDF tags tree editing via PDFBox after Apache FOP PDF generation. I'll study it.

[Intelligent2013]

Div...Div:

occurs due the helper fo:block-container for alignment:

<fo:block-container margin-left="0mm">
	<fo:block-container margin-left="0mm" margin-right="0mm">

To do:

• add role="SKIP" for such fo:block-containers

[Intelligent2013]

• empty tags should be removed
  For instance:
  

These empty tags occurs due empty XSL-FO elements fo:block, by two reason:

• fo:block intended for PDF layout
• fo:block doesn't contain the text by some condition

There are two approach to remove empty tags:

• don't add such tags, or pre-process tags tree - need deep dive into Apache FOP sources
• add role="SKIP" for fo:block if it doesn't contain text - need big XSL-FO refactoring
  1st approach is more preferable, because in XSL-FO there are a lot of workaround solutions for Apache FOP issues.

[Intelligent2013]

empty tags should be removed

In Apache FOP there is a special attribute:

To remove empty fo:blocks from the structure tree, you can set the keep-empty-tags to false.
<accessibility keep-empty-tags="false">true</accessibility>

Added.

[Intelligent2013]

Current footnote tags:

Expected structure:

<P> {
	<Reference> {
		<Lbl>
		}
	<Note> {
		<Lbl>
		<P>
	}
}

[Intelligent2013]

Tags tree for footnotes almost done:

By some reason, Apache FOP adds the tag Div for fo:block-container (marked with role="SKIP"):

<fo:footnote-body role="Note">
	<fo:block-container text-indent="0" start-indent="0" role="SKIP">
		<fo:block font-weight="normal" font-style="normal" text-indent="0" start-indent="0" font-size="10pt" margin-bottom="12pt" role="SKIP">
			<fo:inline keep-with-next.within-line="always" padding-right="3mm" role="Lbl" id="footnote_en_1_2">2)</fo:inline>
			<fo:inline role="P">Formerly denoted as 15 % (<fo:inline role="SKIP" keep-together.within-line="always">m/m</fo:inline>).</fo:inline>
		</fo:block>
	</fo:block-container>
</fo:footnote-body>

[Intelligent2013]

• the inner cover page tags tree contains nested 5 Divs:
  

[Intelligent2013]

• wrong tags structure for Figure:
  

Expected:

[Intelligent2013]

• each Bibliography item represents as a standalone list (L) with one list item (LI):
  

[Intelligent2013]

• outer Div for list L:
  

[Intelligent2013]

• P inside H1:
  

[Intelligent2013]

• empty Span between section number and title
  

[Intelligent2013]

• wrong tags structure for term title
  

[Intelligent2013]

• wrong Span tag instead of P for inline paragraph:
  

[Intelligent2013]

Note moved under the paragraph: