Regarding ODF import and Export support for HistogramChart

[Regina Henschel]

Hi Devansh,

Devansh Varshney schrieb am 02.01.2025 um 09:14:
> <chart:chart svg:width="16.001cm" svg:height="8.986cm" xlink:href="..">
>    <chart:title>
>      <text:p>Histogram Chart</text:p>
>    </chart:title>
>    <chart:plot-area>

>      <!-- Series with label from A1 -->
The generated XML does not contain any comments.

>      <chart:series chart:label="Results">
A 'chart:label' attribute does not exist for the <chart:series> element. 
Instead it has the attribute 'chart:label-cell-address'.


>        <!-- Data points from A2:A101 -->
The data values are not inside <chart:data-point> elements. The data 
values are given by the attribute 'chart:values-cell-range-address' of 
the <chart:series> element.

The purpose of a <chart:data-point> element is, to specify the style of 
a data point. As the columns in the histogram do not directly reflect 
data points, a <chart:data-point> element is useless for an histogram. 
But it does not harm to keep the default, which would be
     <chart:data-point chart:repeated="100"/>
in the example. That means, that the style given in the 
'chart:style-name' attribute of the <chart:series> element is used for 
all 100 data points. You can keep it although it will actually be the 
style of the rendered bins (at least I think you will use the 
'chart:style-name' attribute of the <chart:series> element for this 
purpose).

An element <chart:data-points> does not exist. If <chart:data-point> 
elements are needed, they are simply written one after the other. The 
RELAX NG has the element <rng:zeroOrMore> for this.

>        <chart:data-points>
>          <chart:data-point chart:value="5.3"/>   <!-- A2 -->
>          <chart:data-point chart:value="12.7"/>  <!-- A3 -->
>          <chart:data-point chart:value="8.4"/>   <!-- A4 -->
>          <chart:data-point chart:value="15.2"/>  <!-- A5 -->
>          <chart:data-point chart:value="22.6"/>  <!-- A6 -->
>          <chart:data-point chart:value="30.1"/>  <!-- A7 -->
>          <chart:data-point chart:value="45.7"/>  <!-- A8 -->
>          <chart:data-point chart:value="60.3"/>  <!-- A9 -->
>          <chart:data-point chart:value="75.9"/>  <!-- A10 -->
>          <chart:data-point chart:value="95.1"/>  <!-- A11 -->
>          <!-- Add more data points up to A101 -->
>        </chart:data-points>

>        <!-- Histogram-specific configuration -->
>        <loext:histogram>

I would use element name
          loext:histogram-configuration
to be more specific. Remember, there are no comments written, thus a 
meaningful name helps.

>          <attribute name="loext:histogram-frequency-type">
>            <value>2</value>
>          </attribute>
>          <attribute name="loext:histogram-bin-count">
>            <data type="integer">10</data>
>          </attribute>
>          <optional>
>            <attribute name="loext:histogram-overflow-bin">
>              <data type="double">100</data>
>            </attribute>
>          </optional>
>          <optional>
>            <attribute name="loext:histogram-underflow-bin">
>              <data type="double">0</data>
>            </attribute>
>          </optional>
>          <optional>
>            <attribute name="loext:histogram-interval-closed">
>              <data type="boolean">true</data>
>            </attribute>
>          </optional>
>        </loext:histogram>

Here you have mixed up the content in the schema with the written out 
XML of the document. In the written out XML, it would be
       <loext:histogram-configuration loext:histogram-frequency-type="2"
          loext:histogram-bin-count="10" />

The attributes underflow-bin and overflow-bin are only written out, when 
the user sets them explicitly. Otherwise a default is used. And default 
values need not be written out.

The attribute interval-closed too should only be written out, if it is 
not the default.


>      </chart:series>
>      <!-- Define the X-axis -->
>      <chart:axis chart:dimension="x">
>        <chart:title>
>          <text:p>Values</text:p>
        It is more likely, that the user writes the unit of the data 
values into the axis title.

>        </chart:title>
>      </chart:axis>
>      <!-- Define the Y-axis -->
>      <chart:axis chart:dimension="y">
>        <chart:title>
>          <text:p>Frequency</text:p>
>        </chart:title>
>      </chart:axis>
>    </chart:plot-area>
> </chart:chart>

So you want a structure were element <loext:histogram-configuration> is 
sub-element of the <chart:series> element. That is OK.

You need to decide about the order. Where in the sequence of possible 
sub-elements of the <chart:series> element you want the 
<loext:histogram-configuration> sub-element to be written?


Now about the schema. A useful source to learn more about RELAX NG is 
https://relaxng.org/tutorial-20011203.html.

The <chart:series> element has no <rng:choice> for its sub-elements. 
Therefore a solution with combine="choice" is not possible. And as it is 
an element, a solution with combine="interleave" does neither work. You 
need to replace the existing <chart:series> element with an extended 
one. (See chapter 9.4 in the mentioned tutorial)

For replacing, you need to copy the existing <rng:define> element, that 
specifies the <chart:series> element, from the file 
schema/odf1.4/OpenDocument-v1.4-schema.rng and add it as a new 
sub-element of the <rng:inlcude> element in 
schema/libreoffice/OpenDocument-v1.4+libreoffice-schema.rng. Then you 
extend it.

You do not add the <loext:histogram-configuration> element directly, but 
add a reference to its definition. ODF uses for the name a pattern 
"prefix-elementname'. So in your case it would be a
    <rng:ref name="loext-histogram-configuration" />
Don't forget to mark it as optional.

There exist no proposal to the ODF TC to integrate a chart type 
'histogram' in ODF and if such existed, it would last several years to 
integrate it. Thus you add a comment with a pointer to that bug report, 
where the histogram chart was initially implemented in LO.

Then go to the end of the OpenDocument-v1.4+libreoffice-schema.rng file 
and add the definition for histogram-configuration after the last 
</rng:element>. Again add a comment.


In regard to loext:histogram-configuration itself:
1. <optional> is not here, but in <chart:series>
2. Your current version in Gerrit has two blocks about frequency-type. 
There must be only one. Please re-read the answer from Michael Stahl how 
to do it.

Looking forward to your next version.

Kind regards,
Regina