[Patch] Unicode field names in SQLTable macro

[theYT <dev@…>]

Error occured when using non-ASCII text as field name (as table header). Values are ok.

Sample:

SELECT 'test' as '한글'

Patch file attached. (tested with r13683)

[Jun Omae]

No. to_unicode() is not needed. We should use ​get_column_names() instead of cursor.description directly.

[Ryan J Ollos]

Thanks for the feedback. I'll apply a patch that uses get_column_names.

[Ryan J Ollos]

In 14524:

0.3dev: Use Trac 1.0 DB API.

• Macro now requires Trac 1.0 and later.
• Genshi is used to construct the table.
• Queries can use non-ASCII table names. Fixes #11708.
• Read-only database connection is used.

Please report any issues with the latest changes.

[theYT <dev@…>]

Checked with [14524]:

• Trac 1.1.4 database API changes: OK
• non-ASCII Table names: OK
• SQLTable macro error (MySQL DB): ticket #12269
• Now Wiki text field displayed as one-line (format_to_html → format_to_oneliner); I don't know which one would be better...

[Ryan J Ollos]

Replying to theYT <dev@…>:

• Now Wiki text field displayed as one-line (format_to_html → format_to_oneliner); I don't know which one would be better...

Thank you for reviewing and testing. It seems either format_to_html or format_to_oneliner might be preferred depending on the data being returned.

I made the change to oneliner to avoid the margins on the paragraph, however format_to_html could be used with some CSS to remove the margins (similar to [14529]). Another option would be to support an argument for the WikiProcessor to specify paragraph or oneliner.

[Ryan J Ollos]

In 14734:

0.3dev: Partial revert of [14524]. Refs #11708.

Reverts to format_to_html but adds CSS to shrink the cell margins.