BUG: `to_html()` contains deprecated and poorly supported HTML

[attack68]

df = pd.DataFrame([[1,2],[3,4]], index=pd.MultiIndex.from_product([['a'], [1,2]]), columns=pd.MultiIndex.from_product([['b'], [3,4]]))
df.to_html().split('\n')

['<table border="1" class="dataframe">',
 '  <thead>',
 '    <tr>',
 '      <th></th>',
 '      <th></th>',
 '      <th colspan="2" halign="left">b</th>',
 '    </tr>',
 '    <tr>',
 '      <th></th>',
 '      <th></th>',
 '      <th>3</th>',
 '      <th>4</th>',
 '    </tr>',
 '  </thead>',
 '  <tbody>',
 '    <tr>',
 '      <th rowspan="2" valign="top">a</th>',
 '      <th>1</th>',
 '      <td>1</td>',
 '      <td>2</td>',
 '    </tr>',
 '    <tr>',
 '      <th>2</th>',
 '      <td>3</td>',
 '      <td>4</td>',
 '    </tr>',
 '  </tbody>',
 '</table>']

valign is a deprecated HTML attribute and halign is reported to be poorly supported even when it was not deprecated.
CSS solutions are recommended, but inline css like 'style="text-align:center;"' is not recommended as inline styles can break the more secure website CSPs (content security policies).

See here..
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/td

[jorisvandenbossche]

but inline css like 'style="text-align:center;"' is not recommended

Does thestyle formatter "solve" this by adding unique identifiers to the table, so that the inline css can be targetted to only that table? (no web expert though, I don't know if this is sufficient for CSPs)

[attack68]

The style formatter doesn't use inline styles. i.e. it never does this:
<tbody><tr><td style="text-align: center;">some text</td></tr></tbody>

It will either do this with identifiers as you mention:

<style> #xyz_row0_col0 {text-align: center;} </style>
...
<tbody><tr><td id="xyz_row0_col0">some text</td></tr></tbody>

or this with css classes:

<style> .my-cls {text-align: center;} </style>
...
<tbody><tr><td class="my-cls">some text</td></tr></tbody>

Both these solutions would be allowed under a no inline styles rule on a CSP

Currently anyone using df.to_html() doesn't get all the added HTML bloat that style creates to function, and it seems to have developed its own set of kwargs options.