Acorn Browse HTML Support

Document Status

Distribution:   General Release

Title:   Acorn Browse HTML support

Drawing Number:   1216,211/T

Issue:   1

Author(s):   Andrew Hodgkinson

Date:   07/09/1998

Revision:   1.00

Change Number:   N/A

Last Issue:   N/A

Document Status
Issue / Revision History
Supported Elements
Supported Entites
Supported Colours
Broken HTML
Further reading
1. HTML 4.0 Specification
2. Acorn Browse Plug-In Handling Description

Issue / Revision History

Issue 1 (General release)

1.00 07/09/1998 Created, using JavaScript in places since this allows easy maintenance base on application source code (ADH)

This document uses JavaScript. A non-JavaScript version is available, but note that the copy you are reading now is the master - compare dates to see how up to date the non-JavaScript version is.

Supported Elements

The browser's parser recognises the following set of elements and the browser itself will act on the following element attributes:

Element	Attributes
`A`	`coords href name shape target`
`ACRONYM`	None
`ADDRESS`	n/a
`APPLET`	`align` [1] `alt code codebase height hspace object vspace width`
`AREA`	`alt coords href shape target`
`B`	n/a
`BASE`	`href target`
`BASEFONT`	`color` [4] `size`
`BIG`	n/a
`BLINK`	[2]
`BLOCKQUOTE`	n/a
`BODY`	`alink background` [6] `bgcolor` [4] `link text vlink`
`BR`	n/a
`CENTER`	[4]
`CITE`	n/a
`CODE`	n/a
`DD`	n/a
`DEL`	n/a
`DFN`	n/a
`DIR`	[2]
`DIV`	`align` [4]
`DL`	None
`DT`	n/a
`EM`	n/a
`EMBED`	`align` [1] `height src width`
`FONT`	n/a
`FORM`	`action method target`
`FRAME`	`bordercolor` [4] `name noresize scrolling src`
`FRAMESET`	`border bordercolor` [4] `cols frameborder framespacing rows`
`H1`	`align`
`H2`	`align`
`H3`	`align`
`H4`	`align`
`H5`	`align`
`H6`	`align`
`HEAD`	n/a
`HR`	`align noshade size width`
`I`	n/a
`IMG`	`align` [1] `alt border height hspace ismap src` [6] `usemap vspace width`
`INPUT`	`align` [5] `alt checked maxlength name size src` [6] `type usemap value`
`INS`	n/a
`ISINDEX`	[2]
`KBD`	n/a
`LI`	None
`LINK`	[2]
`MAP`	`name`
`MENU`	None
`META`	`content http-equiv` [3]
`NOBR`	[2]
`NOFRAMES`	n/a
`NOSCRIPT`	n/a
`OBJECT`	`align` [1] `border classid codebase codetype data height hspace name shapes standby type usemap vspace width`
`OL`	`start type`
`OPTION`	`selected value`
`P`	`align` [4]
`PARAM`	`id name type value valuetype`
`PRE`	n/a
`Q`	n/a
`S`	n/a
`SAMP`	n/a
`SCRIPT`	None
`SELECT`	`multiple size`
`SMALL`	n/a
`STRIKE`	n/a
`STRONG`	n/a
`STYLE`	[2]
`SUB`	n/a
`SUP`	n/a
`TABLE`	`align` [1] `bgcolor` [4] `border cellpadding cellspacing height width`
`TD / TH`	`align bgcolor` [4] `colspan height rowspan valign` [4] `width`
`TEXTAREA`	`cols rows`
`TITLE`	n/a
`TR`	`align bgcolor` [4] `colspan height rowspan valign` [4] `width`
`TT`	n/a
`U`	n/a
`UL`	n/a
`VAR`	n/a
`XMP`	[5]

Footnotes from the above table:

Alignment - Browse does not support floating objects, so text cannot flow around images and navigation tables cannot float next to content tables for example. It will attempt to at least align objects centred or to the left or right margin where possible. See [4] also.
Unsupported - The browser does nothing with this element.
META - only used for client pull presently.
UK English - for some common items Browse will recognise UK and 'American English' spellings, e.g. "color" and "colour", "center" and "centre".
Partial support - Some values may not work as expected (typically alignment specification).
Internal URLs - Navigator (but not MSIE) knows certain "magic" URLs such as 'internal-gopher-image'. Web proxies have been known to pass these out to people using FTP through them. Browse does not support these magic values.

Supported Entities

The full range of HTML 4 entities are supported.

The following table is generated with JavaScript using a fragment of a source file from the HTML parser. The Unicode values are listed in hex. If the browser you are using does not support JavaScript, this isn't really a problem as the list of supported colours is self-evident from the script source.

Supported Colours

The following table of colours is generated with JavaScript using a fragment of a source file from the HTML parser. The red, green and blue values are listed in hex. If the browser you are using does not support JavaScript, this isn't really a problem, as the list of supported colours is self-evident from the script source.

Broken HTML

The browser will cope with a variety of types of broken HTML, not least of which includes incorrect structure (documents with TITLE inside BODY sections and so-forth). The measures detailed below have been arrived at over a lengthy development cycle where behaviour of Navigator, Internet Explorer, and other browsers is compared on various sites which failed in Browse. Over time, the parser and browser have become progressively more robust to broken HTML, though there is always work to do to improve this.

Steps taken include:

Entities - progressive matching is used so that when semi-colons are missed off the end of entity names, a good chance of finding the right match is made (if no match is found the entire entity including the '&' is displayed).
Comments - '<!' is used to start and to '>' end a comment sequence, with matching angle bracket pairs counted in between. Comment handling is usually in a state of flux as the parser cannot currently do a relaxed match, waiting until the end of the document, then track back and do a strict match to '-->' as in, say, Navigator. This means that not all cases of broken comments that would manage to work in Navigator will work in Browse.
Mismatched quotes - Single and double quote marks are allowed for attribute values, and missing quotes are acceptable. A missing opening quote is also survived, as are multiple consecutive quotes to close attributes values. In a similar manner to comment handling, a missing closing quote is fatal because the parser cannot backtrack and clean up afterwards.
Colour names - All spaces inside colour names are collapsed, so whether people write "dark red" or "d A r K R e D", the correct result is obtained. If a named colour is not found, the value is checked to see if it is legal hex - if so, this is used instead (so the '#' may be missed off). In addition, hex values specified explicitly with '#' may contain the letter O instead of zero. To avoid collapsing all colours to the same value (e.g. particularly foreground or background colours), a random colour choice is used if the colour name is not recognised despite all of the above mechanisms. Note that the greyscale extended colour names (e.g. "grey20", "grey40") are not supported at present.
Frame specifiers - Rubbish in "cols" or "rows" attributes values is stripped, so values such as "30 pixels, *, 100 %, 50px" is handled gracefully.
Common unusual constructs - Certain cases are specifically hard-coded for, e.g. <LI><P>... means put a paragraph before both the bullet point and associated line of text, rather than between the bullet point and text.

This is not a complete list, but it covers the most important things.