May the names of celebrity and professional dancers to full names. Note that most entries are currently just the first name, but some entries hint at the surname (e.g. Emma B.).

Series 7, 8, 9, and 10 are missing their initial week or two, where the dances are broken up into multiple tables.

The number of judges has varied from 3 to 5. Hence, the absolute score does not reflect fairly the quality of each dance. Detect the number of judges and create a new column for average score.

Some table cells for couples include a footnote marker (e.g. <td>Anastacia &amp; Gorka<sup>1</sup>\n</td>. This results in 14,5,Anastacia,Gorka,,1 - no dance and a score of 1.

Using lol_html this gives three text nodes, each of which has last_in_text_node() returning true. This causes the td selector to update the state before the end of the <td> element. It is possible to suppress the inner tag text (see cloudflare/lol-html#93) but there are still the two nodes Anastacia & Gorka and \n.

• Can we identify a way to merge these using the existing code?
• Should we move to using nth-of-type selector, remembering to deal with rowspans (e.g. 1st child is usually couple, but may be score).

8,4,Michelle Williams,Ian*,Tango,27,4,6.75,
9,7,Holly Valance,Brendan*,Rumba,34,4,8.5,
9,9,Anita Dobson,Brendan*,Cha-Cha-Cha,30,4,7.5,
The asterisk in the couple name gets through to the CSV data.

Series 7 week 11 has a group dance with all the results in one row, separated by linebreaks:

<tr>
<td>Ali &amp; Brian<br />Chris &amp; Ola<br />Laila &amp; Anton<br />Natalie &amp; Vincent<br />Ricky &amp; Natalie
</td>
<td>10<br />7<br />8<br />7<br />9
</td>
<td>Viennese Waltz
</td>

Currently this is ignored, but we could handle this.

The serde serialization adds a header row before the rows of each series.

• Can we suppress the header for all but the first week?
• Can we merge all rows into a single structure to serialize at once?