Grabcode handler

[dartar]

If time/effort allows, I'd like to see this one: it's already proved to be a "selling point". Stylesheet rules need to be improved though. --JW

 http://wikkawiki.org/GrabCodeHandler

[migrated from  http://wikkawiki.org/AdminDevelopmentDiscussions]

Updates

The code committed to the repository as of [49], [50] and [51] adds a further option to the code block syntax, allowing users to specify a filename for the code block, e.g.:

%%(php;22;myfile.php) ... %%

If a file name (like myfile.php) is specified:

• a code block header is generated with the filename;
• the filename is added to the title of the grab button;
• the filename is posted as through a form hidden field so it can be used as a name for the file generated by the grabcode handler.

A (configurable) extension (default: .txt) is appended to the filename by the grabcode handler to prevent the browser from directly running potential malicious code on download.

[DarTar]

Code committed in [49] [50] and [51]. Keep ticket open until code has been further checked for stability/security.

[DarTar]

Debugging: the regex used in the wakka formatter at line 214 for the optional filename value produces an incorrect page parsing:

source:trunk/formatters/wakka.php#L214

[DarTar]

Here's the correct reference:

source:trunk/formatters/wakka.php?rev=L214#L214

the regex pattern is:

"/^\((.+?)(;([0-9]+))?(;(.*))??\)(.*)$/s"

The tricky part is the third bracketed subpattern for the (optional) filename . The idea is that I want at this stage to match anything before the closing ")" and check later whether this is a valid filename. The problem is that using \w is too restrictive. I'll play a little bit with this. Feedback welcome.

[DarTar]

Regex replaced with the following (credits: DotMG)

/^\(([^;]+)(;(\d*?))?(;([^\)\x01-\x1f\*\?\"<>\|]*)(.*))?\)(.*)$/s

Code change committed as [56]

Now the formatter parses correctly code blocks and uses filename if valid.

[DarTar]

Mahefa, while testing the formatter I discovered that the new parser generates multiple instances of the same codeblock in a page containing at least two codeblocks. See what happens in the default FormattingRule.

(keeping ticket open)

[DarTar]

Previous issued solved by reinitializing the $output variable on line 210 (credits: DotMG - merci!).

Patch committed as [60].

Should we make the filename option configurable from wikka.config.php (much as line numbering)?

[DarTar]

Note that FormattingRules must be updated to take into account the filename option.

[DarTar]

Updated FormattingRules in [66] (both as a default page for fresh installs and as a revision for upgraders).

[DarTar]

Reopening ticket - there is a bug in the codeblock parser. When using round brackets in the code, the regex fail to correctly parse the geshi parameters. I'll try to fix this.

[DarTar]

Mahefa, why don't we just drop the $invalid match ? I don't see any use of extracting it - I think we should just extract the filename as everything before the closing bracket. If it's not valid, then filename won't be used in the codeblock header and in the grabcode form.

[NilsLindenberg]

What perhaps could be done is breaking down the regepx into smaller parts to have more oversight, like it was done in the color action.

[DarTar]

Fixed issue with code containing brackets - [101] 1.1.6.2 ready for alpha.