BBCode

Reader\BBCode\BBCodeReader reads BBCode into a UJO document, which phil/ujo renders as HTML. It reads the tags of forums, and extensions written in the same style that give BBCode every feature of the Markdown reader: footnotes, definition lists, a table of contents, attributes, boxes, columns, page breaks, stat blocks and the other blocks of printed pages. Formats puts each tag beside the syntax of the other formats.

$reading = new BBCodeReader()->read("[h2]Rules[/h2]\n[quote=Ada]Roll [b]two[/b] dice.[/quote]");

$reading->document->blocks[0];   // a `header` block, level 2, its anchor `rules`
$reading->document->blocks[1];   // a `quote` block, its caption `Ada`

new BBCodeReader(true) keeps raw HTML; DefaultCompiler asks for it at the level all only. A second argument replaces the phil/ujo normalizer the document is checked by.

Tags

A tag is written [name], [name=option], [name="option" key=value flag] or [name key=value flag], its name in any case. An option without quotes runs up to the first key=. An attribute value holding spaces is quoted.

Text

Colors, spoilers and icons reach the page as far as the level purifies them: basic and forum keep their text.

Blocks

Attributes

Every block tag takes id — the anchor tune —, align — the alignment tune —, width, height and wrap=left or wrap=right — the layout tune —, and class, a list of the look's classes that wraps the block in a container where nesting allows it, as the attributes of the Markdown reader do:

[h2 id=intro class="lead wide" align=center]Introduction[/h2]
[img=Map width=300px wrap=right]https://example.com/map.png[/img]

Pages and game books

A source that is not well formed

Any text is a document: the reader never throws.

• A tag the reader does not know, or written over two lines, stays as written.
• A tag left open is closed at the end of the source. A closing tag closes the nearest open tag of its name and the tags opened inside it; a closing tag nothing opened stays as written.
• An item, a row, a cell, a column, a term or a definition closes the one of its kind left open in the same parent: [list][*]a[*]b[/list]. Outside its parent, it stays as written.
• Text and tags between the items of a list, the rows of a table, the cells of a row or the columns of [columns] make an item, a row, a cell or a column of their own; blank text between them is left out.
• An inline tag holding a block is left out, its content kept; a [spoiler] holding blocks becomes a toggle.
• A heading, a quote, a list item, a cell, a term, a definition and a footnote fold the blocks they hold into their inline content; a block with no inline content — a box, columns — is left out there, its paragraphs kept.
• Tags nested deeper than 32 stay as written. Boxes and other containers of blocks nest as deep as UJO allows; one deeper is left out, its blocks kept.
• The content of [code], [icode], [img], [html], [var], [metadata] and [style] runs up to its closing tag, in any case, or to the end of the source.

Linear time

A member who writes a post can write a hostile source. The reader takes time in proportion to the length of the source, whatever its shape:

• the cursor of the analyser reads each byte once; a bracket that opens no tag is text, and the scan for its ] stops at the next [;
• the reader keeps, for each tag name, the positions its tags are open at: a closing tag finds its opening tag, or learns there is none, without searching the stack, and each tag is opened and closed once;
• the search for the closing tag of [code] and the others consumes what it passes;
• headings of the same text, footnotes and images take constant time each.

BBCodeReaderTest reads sources of 100 KB written to be slow, and BBCodeBench reads them at 25, 50 and 100 KB: the time doubles with the size.