2010-04-30 Shaun McCance

shaunm@gnome.org

This specification provides a common directory layout and URI scheme for locally installed help documents. Documents installed and referenced using this system will have better interoperability between different desktop environments and help applications.

A document is a collection of files, possibly from multiple directories in a path. Documents may be in any format, or even in multiple formats; see for details. A document has a document identifier that identifies it uniquely and specifies where it can be found on the file system. A document identifier consists of one or more characters from the following: digits (U+0030-U+0039), letters (U+0041-U+005A and U+0061-U+007A), hyphen (U+002D), underscore (U+005F), period (U+002E), and percent (U+0025). Document identifiers are not explicitly namespaced. To avoid conflicts, document identifiers should start with the name of the application or package that provides the document. In many cases, the name of the application or package alone may be used. Otherwise, the document identifier should start with the name of the package or application followed by a hyphen. Documents are installed under the help subdirectory in $XDG_DATA_HOME or in the $XDG_DATA_DIRS path. See the XDG Base Directory Specification for details on $XDG_DATA_HOME and $XDG_DATA_DIRS. Each help directory in the path contains subdirectories for languages. Each language subdirectory contains subdirectories for documents, where the name of each subdirectory matches the document identifier of a document. The document path for a given document is the list of directories of the following form: datadir/help/lang/document datadir Either $XDG_DATA_HOME or a directory in the $XDG_DATA_DIRS path. lang The language code of a language in the user's list of preferred languages. document The document identifier of the document. The document path is ordered first according to the position of datadir in the path, then by the position of lang in the user's preferred list of languages. So, for example, if the user's preferred language list is pt_BT:pt:C, then ~/.local/share/help/pt/beanstalk would take precedence over /usr/share/help/pt_BR/beanstalk.

Documents are referenced using the help: URI scheme. The help: URI scheme has the following form: help:document/page?options#anchor document The document identifier. page An identifier for a page within the document. Documents often consist of multiple logical pages, which may not be reflected in the actual files on the system. The page identifier is optional. If it is not present, the preceding / character should be omitted. options Additional options that can influence how applications present a document. Options are optional. If they are not present, the preceding ? character should be omitted. If they are present, they must conform to the application/x-www-form-urlencoded MIME type. Options may be used, for example, to override language settings or to provide keys for conditional processing. This specification makes no specific recommendations for the options. anchor An anchor point within a page. Applications should scroll to an appropriate point whenever possible. The anchor is optional. If it is not present, the preceding # character should be omitted. Page identifiers and anchors may contain any character that is valid in a document identifier. Some document types may have further restrictions on page identifiers or anchors.

Documents may be installed in any format. Not all help applications may recognize and handle the same types of documents. Whenever possible, help applications should support HTML. Documents may be installed in multiple formats. Help applications choose which format to use when multiple formats are present. For any format, a help application must map the information found in the help: URI scheme to the information in that format. This specification contains recommendations for finding and handling documents in DocBook, Mallard, HTML, and XHTML.

DocBook documents have a file named index.docbook in the document path. Any files referenced from the DocBook file, including XML fragments included through XInclude, should be looked up according to the document path. When a help application displays a DocBook document, it will often display it in multiple pages or chunks, rather than as a single long page. The page identifier specifies which chunk to display. Not all applications will split a DocBook document into chunks in the same way, so the page identifier may not map exactly to the id attribute of an element that is chunked. In this case, the applicaiton displays the nearest enclosing chunk and treats the page identifier as an anchor if no anchor was explicitly provided.

Mallard documents have a file named index.page in the document path. Other page files may be different directories in the document path. Any files referenced in any page file, including XML fragments included through XInclude, should be looked up according to the document path. The page identifier specifies the id of a Mallard page file. The anchor specifies the id of a section within a Mallard page file.

HTML documents have a file named index.html in the document path. XHTML documents have a file named index.xhtml in the document path. Other HTML or XHTML files in the document may be in different directories in the document path. Any files referenced in any HTML or XHTML page, including XML fragments included through XInclude in XHTML, should be looked up according to the document path. The page identifier specifies the base file name, without the .html or .xhtml extension, of an HTML or XHTML file in the document. The anchor specifies a named anchor within the HTML or XHTML file.