bbcode (unfinished)

C implementation of the BBCode markup used on many web forums. It reads BBCode markup code and produces HTML. Unlike many other BBCode implementations, this one uses a different approach in that it does not use Regexp string replacing functions, and maintains a hierarchy of the written HTML tags even if it is supplied with non-hierarchical BBCode.

Installation

Requires (at least) autoconf, automake and libtool.

git clone https://github.com/ivartj/bbcode
cd bbcode
autoreconf --install
./configure
make all install clean

Usage

bbcode [ -o <output-HTML-file> ] [ <input-BBCode-file> ]

If no output file is given, it writes to standard output.

If no input file is given, it reads from standard input.

For use of the [align=], [left], [center] and [right] tags, bbcode expects that CSS style rules are defined for the bbcode-left, bbcode-center and bbcode-right HTML classes.

TODO

Relativley complicated tasks include:

• Lists
• Ensure that XSS is not possible with stuff like [url=javascript:alert('hello')] from untrusted users

Spec

(not all of this is implemented yet)

BBCodes have the following anatomies:

• A start tag:

  [tag]
  

• A start tag with a parameter:

  [tag=value]
  

• A stop tag:

  [/tag]
  

• A list item tag

  [*]
  

Registered tag names are divided into the following types:

• Inline-level tags ([url], [b], [i])
• Block-level tags ([quote], [list])
• Content tags ([code], [img], [youtube])
• Delimiter tags ([*])

Unmatched start and stop tags are printed verbatim. Content tags alter how the source text between the tags is treated. The [*] tag only has effect if the first block-level ancestor is a [list]; it is otherwise printed verbatim.

Inline-level tags can overlap with each other

[b] test [i] test [/b] test [/i]

<b> test <i> test </i></b><i> test </i>

Inline-level tags can overlap with block-level tags

[b] test [quote] test [/b] test [/quote]

<b> test </b><blockquote><b> test </b> test </blockquote>

Block-level tags can not overlap with each other

[list] test [quote] test [/list] test [/quote]

<ul> test [quote] test </ul> test [/quote]

They can still be nested

[quote] test [list] test [/list] test [/quote]

<blockquote> test <ul> test </ul> test </blockquote>

Operation

The parser maintains a list of parsed tags and text content.

Any start tag is stored in the list immediately. When encountering stop tags the parser will attempt to find a previous matching start tag, and if found, it will set the match member of the two tags to each other. Start tags with match set can no longer be matched to any stop tags. Block-level stop tags can only be matched to block-level start tags in the same nesting (ignoring unviable unmatched block-level tags). When a match for content type tags are found, all the tags and text content between them are removed from the list and their matches set to unmatched.

The printer maintains a doubly-linked lists of printed inline-level start tags and 'unwinds' and 'rewinds' it as needed to emulate overlapping formatting. It prints the source of unmatched tags verbatim.

Why?

• Nostalgia
• It's a programming exercise
• Markdown isn't any good for poetry (without adjustments)

I am using it to write my homepage at ivartj.org.