PlantUML defines a standardized way to encode diagram text description to a simple string of characters that contains only digits, letters, underscore and minus character.
The goal of such an encoding is to facilitate communication of diagrams through URL (see server).
This encoding includes compression to keep encoded strings as short as possible.
The encoded metadata is stored in the generated PNG, so the diagram source can be extracted from the diagram itself! (see server#metadata).
The following compression algorithms are available:
Starting in version 1.2017.20, PlantUML also supports the Brotli algorithm (issue #117) that gives better results for larger diagrams. An initial 0 character is added to the encoded string to indicate Brotli (Deflated data never starts with 0).
You can also use simple HEX encoding, see below. An initial ~h is added to indicate this encoding.
For example, the following uml text description:
Alice -> Bob: Authentication Request
Bob --> Alice: Authentication Response
Reencoded in ASCII using a transformation close to base64
Why not use Base64?
The main reason is historic: this format was not created to be public at first. Now, it's too late to change it. However, the only difference is in character order.
Where in base64 the mapping array for values 0-63 is:
To indicate the use of HEX format, you must add ~h at the start of the data sent to PlantUML server.
Since there is no compression here, the URL will become very long as the diagram grows.