Instructions on how to generate docsets (for Dash) are found below. There is no method that is best for every case, so you'll have to consider all and decide which is the best for you. When your docset is ready, please contribute it to Dash.
AppleDoc can be used to generate docsets from Objective-C source files. Please note that if you are trying to generate a docset for an Objective-C framework, it's very likely that a docset for that framework is already available in Preferences > Downloads > Cocoa Docsets.
When you're done generating your docset, check out some tips on how to improve it.
Use http://pypi.python.org/pypi/doc2dash to generate docsets from Python, Sphinx or PyDoctor-generated documentation. When you're done generating your docset, check out some tips on how to improve it.
You can also install docsets for Java libraries that are on Maven.org from Preferences > Downloads > Java Docsets.
You can install docsets for any Ruby Gem from Preferences > Downloads > Ruby Docsets.
Use sbt-dash or mkscaladocset to generate docsets from Scaladoc-generated API documentation. Use scala-dash to generate docsets for Scala manuals. When you're done generating your docset, check out some tips on how to improve it.
You can also install docsets for Scala libraries that are on Maven.org from Preferences > Downloads > Scala Docsets.
You can install docsets for any Go Package from Preferences > Downloads > Go Docsets.
You can install docsets for any Hex Package from Preferences > Downloads > Hex Docsets.
You can install docsets for any Dart Package from Preferences > Downloads > Dart Docsets.
You can install docsets for any Haskell Package from Preferences > Downloads > Haskell Docsets.
Doxygen can generate docsets from source files of C, C++, C#, PHP, Objective-C, Java, Python (and some others).
These are the entries you need to add into your Doxygen config file to make it generate a docset (note: the last 3 entries are optional):
GENERATE_DOCSET = YES DISABLE_INDEX = YES SEARCHENGINE = NO GENERATE_TREEVIEW = NO
When Doxygen is done generating the documentation, run
make inside the generated folder. Afterwards, check out some tips on how to improve your docset.
Docsets are essentially just a folder containing the HTML documentation and a SQLite database that indexes the files.
If the HTML docs you want to index are well-formatted, it might be easier to generate the docset using Dashing instead of writing your own generation script.
The docset folder structure can be created using this Terminal command:
mkdir -p <docset name>.docset/Contents/Resources/Documents/
You can also manually create the docset structure if you want, they're just folders.
Copy the HTML documentation you already have to this folder:
Download and edit this sample Info.plist and place it in the
<docset name>.docset/Contents/ folder. Editing should be straightforward, just set the values to whatever name you want for your docset.
Create a SQLite database in the file
<docset name>.docset/Contents/Resources/docSet.dsidx with the following query:
CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);
Recommended: you can easily prevent adding duplicate entries to the index by also using this query:
CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);
You need to create a script (or application or whatever) that will go through your HTML documentation and add appropriate rows into the SQLite database. Rows can be added using this query:
INSERT OR IGNORE INTO searchIndex(name, type, path) VALUES ('name', 'type', 'path');
The values are:
nameis the name of the entry. For example, if you are adding a class, it would be the name of the class. This is the column that Dash searches.
typeis the type of the entry. For example, if you are adding a class, it would be "Class". For a list of types that Dash recognises, see below.
pathis the relative path towards the documentation file you want Dash to display for this entry. It can contain an anchor (#). Alternatively, Dash also supports
You can find a few generation script examples here.
Please contact me if none of the currently supported types are suitable for what you're trying to index.
To allow easy navigation inside a large page, Dash is able to show a table of contents at the bottom left. This feature is described in the user guide.
Please note that adding table of contents support is a bit tricky (and optional).
When Dash displays a documentation page, it also looks for special anchors inside the HTML and generates a table of contents. To add table of contents support, you need to go through all the HTML pages and insert anchors in a format that Dash understands. The anchors need to be inserted in the HTML pages themselves.
The format for the anchors is:
<a name="//apple_ref/cpp/Entry Type/Entry Name" class="dashAnchor"></a>
The only things that you need to change in the format above are:
Entry type- one of the supported entry types.
Entry name- the name that is shown by Dash in the table of contents. Preferably, it should be percent escaped.
You can see an example of how to insert anchors at https://github.com/jkozera/zeal/blob/master/gendocsets/extjs/parse.py (in Python).
You'll also need to add this entry in the docset's Info.plist:
Dash displays an index page when you highlight a docset in the documentation browser. If Dash can't find a suitable index page automatically, you can specify it inside the Info.plist by adding the following entry:
The path should be relative to the
Documents folder inside the docset. It can also be a
After adding the index, remove and re-add the docset in Dash's Preferences.
To set a custom icon for your docset, simply add a
icon.png file directly inside the docset bundle. For example, the file path would be
The size of the icon should be 16x16 or 32x32. For Retina display support, you can either use a single 32x32 icon or 2 separate icons:
icon.png (16x16) and
After adding the icon, remove and re-add the docset in Dash's Preferences.
Starting with version 3.0, Dash users can open the online version of pages inside docsets. To do that, Dash needs to know where to find the online pages for your docset.
You have 2 options to support online redirection:
DashDocSetFallbackURLkey in your docset's Info.plist. The value should be the base URL of the docs. Example:
Alternatively, add a HTML comment inside each and every HTML file of your docset. The comment needs to be added next to the
<html> tag of the pages and should look like this:
<html><!-- Online page at https://docs.python.org/3/library/intro.html -->
Starting with version 4.0, Dash now shows a play button next to docsets which takes users to a playground for the language/framework of the docset. To support it you need to:
DashDocSetPlayURLkey in your docset's Info.plist. The value should be the URL of the docset playground. Example:
https://repl.it/F9J7/1for the Swift docset.
By default, Dash does not allow external
.js scripts. You can enable them by adding this entry to the docset's Info.plist:
After adding this entry, remove and re-add the docset in Dash's Preferences.
Important: It is highly recommended that you contribute your docset to Dash instead of setting up a docset feed. Docset feeds should only be used for docsets which only you, your team or very few users find useful.
Documentation feeds allow Dash users to conveniently install and update docsets. If you want to distribute a docset you should set up a feed for it, so that you'll be able to update it in the future.
Dash documentation feeds are very simple, check one out: NodeJS_Sample.xml.
It's a XML file that contains the following:
<version>element. You can use any versioning system you want. Dash will use string comparison to determine whether or not to download an update.
<url>elements. These point to the URL of the archived docset.
To archive your docset, use the following command:
tar --exclude='.DS_Store' -cvzf <docset name>.tgz <docset name>.docset
Currently only one docset per feed is supported (only one
You can share docset feeds using a custom URL scheme, which will allow Dash to subscribe to that feed with a single click.
dash-feed://<URL encoded feed URL>
By URL encoded, I mean percent-encoding (e.g. what you’d get by clicking the encode button here).