The diagram_text_file is processed in paragraph mode, each paragraph is
separated from the next by a blank line. The order of the paragraphs is
important (that is a minor bug). The order is:
You may include as many edge paragraphs as you like, but each one should
have a different label.
At least two example input files are included in the distribution. Look
in the samples subdirectory for files ending in .input. The README file
in that directory explains what these input files represent. The
corresponding .svg and .png files hold the resulting images. The .png
images were produced with the freely available Batik rasterizer,
which is part of the Batik project from Apache.
See http://xml.apache.org/batik/ for information on Apache Batik.
In case you have trouble finding those files, one is included here:
(0) () (5)
() () (10)
0,0,S 0,2,N Counter
2,0,E 2,3,E Clock
This represents the states in a coin operated vending machine which accepts
coins with values of 5, 10, and 25. The purchase price for the machine is
25, so 25 and 30 are accepting states. Zero is the start state. Each
deposited coin increases the total, which is used to label the nodes.
Edges represent deposits and are labeled with the value of the deposited
coin. (The diagram is not complete. It is meant to show features of
diagrams. If you have a coin-op machine you need to build, feel free
to extend it as needed.)
Each paragraph except the one for node layout has a label. The labels are
only used for edge paragraphs. All coordinates are in terms of the node
layout which is numbered from 0. Columns are listed first, then rows
(think x and y from algebra). Sides of the nodes (where edges begin and
end) are labeled with compass points N(orth), S(outh), E(ast), and W(est).
This input format is idiosyncratic. If you dont like it, you should design
your own and interface directly with UML::State. If you dont like listing
columns first, or numbering from zero, you will need to account for that
in your new script or change UML::State.