Context Navigation

#368 closed defect (fixed)

Calling HTML("<? ?>") causes a ValueError?

genshi_HTML_PI.diff (627 bytes) - added by Neil Muller 11 years ago.: Accept empty data in processing instructions in HTMLParser
genshi_HTML_PI.2.diff (627 bytes) - added by Neil Muller 11 years ago.: less buggy version of the path
genshi_HTML_PI_with_tests.diff (1.8 KB) - added by Neil Muller 11 years ago.: Patch with extra tests
genshi_HTML_PI_with_tests.2.diff (1.8 KB) - added by Neil Muller 11 years ago.: tests should have distinct names

Download all attachments as: .zip

I think that is correct. <? ?> is not a valid PI.

<?foo ?> will yield a correct object.

Accept empty data in processing instructions in HTMLParser

'<?foo ?>' will also raise a KeyError? on current genshi trunk. Genshi assumes that all processing instructions have data attached - i.e. <?foo x > or <?page break>. This is incorrect - consider this example from the HTML 4.01 specifications (http://www.w3.org/TR/1999/REC-html401-19991224/appendix/notes.html#h-B.3.6):

'<?experiment> ... <?/experiment>'

The genshi_HTML_PI.diff accepts empty data on PI - it doesn't break the test suite, but probably needs further testing

less buggy version of the path

Patch with extra tests

tests should have distinct names

Patch committed to trunk in r1219. Merged to 0.7.x in r1229 and to 0.6.x in r1222. Thanks!

Note: See TracTickets for help on using tickets.