Introduction

Glumly is a simple Python script that I wrote to help me generate UML diagrams to embed in a presentation. It parses an input file with a few basic rules, and it outputs a DOT file that is interpreted with Graphviz.

Before developing Glumly, I had been testing yUML and I had liked it, so I wanted to make my UML graphs looks as yUML ones. To achieve a similar result, I embellished the SVG obtained from Graphviz with xsltproc using the notugly-full.xsl stylesheet by Michael Kennedy wich was written as an extension of notugly.xls by Vidar Hokstad.

Requisites

To make UML diagrams with Glumly you need to:

• Install Python, to run the script;
• Install Graphviz, to compile the graph;
• Download notugly-full.xsl, to make the output less ugly;
• Install xsltproc, to apply the stylesheet;
• Install Inkscape, if you want the output in a format different from SVG.

Download

Download Glumly v0.2 from here:

Usage

The Glumly script takes two optional arguments: an input and an output file. If no input or output file is provided, stdin and stdout are used.

$ ./glumly.py --help
Glumly, a free yUML-like UML creator for DOT.
Usage: glumly.py {inputfile|- {outputfile}}

The suggested piping is using Graphviz to compile the DOT output and xsltproc to apply the notugly-full.xsl stylesheet. The piping looks like:

$ ./glumly.py inputfile.txt | dot -Tsvg | xsltproc notugly-full.xsl - > outputfile.svg

The pipe step using xsltproc and notugly-full.xsl could be omitted... but the output will be UGLY!.

File format

The input is an ASCII tab delimited file. Each line may begin with an identifier or with a numeral for comments. The identifiers fall into two categories: connections and other properties.

All connections have a single letter identifier and share the same syntax. In a connection row, the first col is the identifier and the second and third cols are the names of the connected nodes. If a fourth col is found, it is used as a label at the end of the connection, in the same way a fifth col is used as a label at the begin of the connection. In labels the * and .. symbols are treated as special ones and replaced by Unicode infinity (∞) and ellipsis (‥).

The connection identifiers are:

• S, for Single association;
• A, for Aggregation;
• C, for Composition;
• R, for caRdinality;
• I, for Interface;
• D, for Double association;
• H, for inHeritance;
• E, for dEpendency.

Other properties are:

• CO, for COlor. The second col is the node name and third is the Graphviz color.
• LA, for LAbel. The second col is the node name and third is the label string. If \ is found within the string, the box is made using a record node, then you can use | to split with lines and \l, \n and \r to break the line and align left, center or right. If more than three cols are found, the fourth and fifth ones are used as attributes and members list, where elements must be separated by commas.
• IN, for INterface. The second col is the name of the node that is an interface.
• RA, for RAnks. The cols are the names of the nodes that are in the same rank.

Example

This input, contained in a file called example.txt:

# A color
CO	Customer	green
CO	Delivery	yellow

# Connections:
# ID	Begin   	End     	End label	Begin label
A	Customer	Order   	orders 0..*	1
S	Order   	Product 	*       	*
S	Order   	Delivery	1
C	Order   	LineItem	*       	*
D	Category	Product
H	Delivery	International
H	Delivery	National

Compiled with:

$ ./glumly.py example.txt | dot -Tsvg | xsltproc -o example.svg notugly-full.xsl -
$ inkscape example.svg -e example.png

Produces this PNG graph:

(Here you can compare with the yUML representation of the same UML diagram.)

The same example compiled with Graphviz but omitting the notugly step looks ugly as you can see here.

License

Glumly is distributed under the BOLA license, which means public domain (if applies).

All feedback is welcome.