[PPTX] view html does not preserve several drawing, text, and equation properties

[ueno-koki]

Environment

• OfficeCLI: 1.0.144
• Command: officecli view sample.pptx html

Summary

Several PPTX properties are rendered incorrectly in the raw HTML generated directly by OfficeCLI.

1. A bent connector is partially missing

A bentConnector4 connector is only partially visible.

The generated SVG uses:

viewBox="0 0 100 100"

However, its polyline contains an X coordinate as large as 2136.27, so part of the connector falls
outside the SVG view box.

Example source object:

• Path: /slide[5]/connector[@id=16]
• Size: approximately 2.42pt × 103.82pt
• Rotation: 180

Expected: the complete connector is visible.

Actual: part of the connector is clipped or missing.

2. Alignment inherited from a slide layout is ignored

A slide-number placeholder is right-aligned in ppt/slideLayouts/slideLayout1.xml using:

<a:lvl1pPr algn="r">

The generated HTML paragraph does not contain:

text-align: right;

As a result, the slide number appears on the left side of its placeholder.

Direct right alignment works correctly. A normal text box containing:

<a:pPr algn="r"/>

is rendered with:

text-align: right;

Expected: the effective alignment inherited from the slide layout is applied.

Actual: the inherited right alignment is omitted.

3. noFill text outlines are rendered as black strokes

Some text runs contain:

<a:ln>
  <a:noFill/>
</a:ln>

OfficeCLI converts this to:

-webkit-text-stroke: 1px #000000;
paint-order: stroke fill;

Expected: no visible text stroke.

Actual: a 1px black stroke is added, making the text appear blurred or darker.

4. Equation font size and inline/display layout are not preserved

The font size associated with an OfficeMath equation is not applied to the generated KaTeX element.

For example, an equation associated with:

<a:rPr sz="1200"/>

should be rendered at 12pt, but the generated HTML uses:

<div class="para" style="font-size:18pt;line-height:1.2">
  <span class="katex-formula" data-formula="..."></span>
</div>

The equation span has no font size of its own and therefore uses the default paragraph size instead
of the equation's run size.

The generated script also unconditionally calls:

katex.render(el.dataset.formula, el, {
  throwOnError: false,
  displayMode: true
});

This treats inline <m:oMath> equations as display equations, changing their vertical spacing and
placement relative to surrounding text.

Expected:

• Preserve the effective font size of each equation.
• Render <m:oMath> as inline math.
• Render <m:oMathPara> as display math.

Actual:

• The equation uses a default or inherited font size.
• Every equation is rendered in display mode.

Related issue: #228

That issue fixed missing equations and extraction of multiple equations, but the proposed equation-
size and inline/display-mode changes were not included in the accepted fix.

5. Simplified Chinese fonts are preferred for Japanese fallback

When the source font is unavailable, the generated font family resembles:

font-family:
  '<source font>',
  'PingFang SC',
  'Microsoft YaHei',
  'Noto Sans CJK SC',
  'Hiragino Sans GB',
  sans-serif;

For Japanese text, this may select Simplified Chinese glyphs before an appropriate Japanese font.

Expected: for Japanese content, Japanese fallback fonts such as the following should be preferred
before Simplified Chinese fonts:

Noto Sans JP
Noto Sans CJK JP
Yu Gothic
Meiryo
Hiragino Sans

All of these problems are present in the unmodified HTML produced directly by OfficeCLI.

[goworm]

Thanks — these are concrete and useful. To fix them faithfully I need the source sample.pptx, because the details are geometry/inheritance-specific and I want to compare our render against real PowerPoint (officeshot) rather than guess:

1. bentConnector4 clipped — our bent-connector polyline is authored in a 0..100 viewBox; a point at X≈2136 means the connector's adj guides place an elbow far outside that box. Whether PowerPoint draws it out there or clamps it depends on the exact adj1/2/3 values and the connector's start/end anchors, which I can only see in your file. A blind "expand the viewBox" or "clamp" risks distorting normal connectors, so I do not want to change it without your connector[@id=16].
2. layout-inherited alignment ignored and the equation/text items likewise need the specific slide + layout to reproduce.

Could you attach the sample.pptx (even a single-slide cut containing connector[@id=16] and the right-aligned slide-number placeholder)? With it I can officeshot-compare and fix each item precisely.

[ueno-koki]

Thank you for the careful response.

I have attached a minimal reproduction: sample.pptx.
sample.pptx

I also need to correct my previous explanation. The polyline contains an X coordinate of 2136.27, outside:

viewBox="0 0 100 100"

However, it is not clipped because the SVG uses:

overflow: visible;

The actual cause appears to be non-uniform scaling:

SVG size: 3.52px × 151.40px
scaleX: approximately 0.0352
scaleY: approximately 1.514
vertical stroke width: approximately 0.07px
horizontal stroke width: approximately 3.03px

This makes the vertical segments and arrowhead nearly invisible.
I apologize for incorrectly describing this as viewBox clipping.

The sample covers:

1. The original bentConnector4 geometry
2. Alignment inherited from the slide layout
3. An <a:noFill/> text outline rendered as a black stroke
4. Equation size and inline/display behavior

The equation examples are:

Display: a + b = c
Inline source: The value is [x + y].

Only x + y is inside <m:oMath>, but OfficeCLI moves it before the surrounding text:

<span class="katex-formula" data-formula="x + y"></span>
<span>The value is </span>
<span>.</span>

It also renders every equation with:

displayMode: true

Therefore, the inline equation is displayed separately instead of remaining inside the sentence.

===== Preview =====
OfficeCLI Preview:

Powerpoint Preview: