Description

When you do an anchor macro where the name of the anchor contains an underscore, the resulting anchor has the name that has the underscore removed. Thus, [[#anchor_name]] with the underscore will fail.

Steps to reproduce

1. edit wiki page and put in an anchor w/the name containing an underscore <<Anchor(anchor_name)>>

2. also put an anchor link [[#anchor_name]].

3. Save page.

4. the anchor link won't work because the anchor is now named "anchorname". [[#anchorname]] will work however.

Example

underscore

#foo_bar

...

colon

#foo:bar

...

(see discussion below)

Component selection

• Anchor macro, wikiutil.anchor_name_from_text < does someone understand why it strips the _ ?

Details

this wiki

Workaround

don't use anchor names with underscore, or remove underscore when creating the link.

Discussion

Relevant Specs:

• http://www.w3.org/TR/html4/types.html#h-6.2 - must begin with a letter ([A-Za-z]) and may be followed by any number of letters, digits ([0-9]), hyphens ("-"), underscores ("_"), colons (":"), and periods (".").

Other Specs:

• http://dev.w3.org/html5/spec/Overview.html#the-id-attribute - at least one char, no spaces (no whitespace)

• http://www.w3.org/TR/xhtml1/#guidelines - strings matching the pattern [A-Za-z][A-Za-z0-9:_.-]* should be used

Conclusion:

• [A-Za-z][A-Za-z0-9:_.-]* matches all specs and recommendations

• http://hg.moinmo.in/moin/1.8/rev/a7dc3cc36362 was too restrictive

(22:56) <  TheSheep> as long as _ are not used in default themes for stylable element IDs, I think it's fine
(23:04) < ThomasWal> we can avoid that by convention, I think
(23:05) <  TheSheep> exactly

Proposed code

def anchor_name_from_text(text):
    '''
    Generate an anchor name from the given text
    This function generates valid HTML IDs matching: [A-Za-z][A-Za-z0-9:_.-]*
    '''
    quoted = urllib.quote_plus(text.encode('utf-7'))
    res = quoted.replace('%', '.').replace('+', '_')
    if not res[:1].isalpha():
        return 'A%s' % res
    return res

This code is a revert of changeset http://hg.moinmo.in/moin/1.8/rev/a7dc3cc36362.

Further ideas

If we run anchor IDs (link targets) as well as the fragment part of links through the same transformation, it will ever match.

If the transformation is an identity transform for valid IDs, there should be no problem.

The current code is almost an identity for valid IDs, except for ":" which is transformed to ".3A":

>       assert u'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789:_.-' ==
               u'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789.3A_.-'
(fails)

We could modify the function to keep ":" in IDs, then stuff like this could work:

[[#Übler:Dübel]]
...
<<Anchor(Übler:Dübel)>>

Done by http://hg.moinmo.in/moin/1.8/rev/1f5b1e0423fb.

Plan

• Priority:
• Assigned to:

• Status: fixed by http://hg.moinmo.in/moin/1.8/rev/30ffb215bf6e

CategoryMoinMoinBugFixed