old_ _Creating_a_converter

• • WARNING**: I believe this to be old now. It should be modified or maybe eliminated.

---

This is a rough guide to creating a JUMBOConverter

Notes added as PMR created a GAMESSUK-PUNCH converter. We assume you are creating a converter for the BarComp system - replace "BarComp" by your own system. Be consistent in the use of classNames and variables

Planning

You will need:

 * a set of exemplar files
 * JUMBOConverters in a Development environment (e.g. Eclipse, NetBeans, IntelliJ...) with maven enabled
 * the program manual. This is optional and I have written converters without manuals but it helps to know what the output actually means and how it is structured.
 * a modest unerstanding of Mercurial and Maven
 * a reasonable understanding of Java, regexes,
 * a long day

Getting started

This is written with an Eclipse flavour - others should work by analogy.

We have written a set of templates for a fictitious format "foo" in . You should copy all resources into a new package . Then rename the classes (*using the refactoring tools*) and you should now have:

 * BarComp2XMLConverter (converts BarComp format to rawCML)
 * BarCompBlock.java (holds the blocks in the BarComp2XML parse
 * BarCompProcessor (helper application for BarComp2XMLConverter to create and process the blocks)
 * BarCompXML2CMLConverter (converts rawCML to completeCML)
 * BarCompXMLProcessor (helper application for BarCompXML2CMLConverter to add CML structure and semantics )

This will fail to compile and you have to make some manual changes:

 * globally replace the string "FOO_" by "BARCOMP_"
 * in BarComp2XMLConverter find the two lines: {{{

public static final String BARCOMP_PREFIX = "foo"; public static final String BARCOMP_URI = "http://wwmm.ch.cam.ac.uk/dict/foo"; }}} and change them to: This establishes the dictionary namespace. (We may change this to use properties files or central dictionary locations later)

Now you should only have two compile errors with undeclared variables:

You will need to edit them into the class. *This must be done even more carefully as if you corrupt this file you'll screw up other people*. There are THREE edits required in three different locations. (Yes, we'll probably change this to a *.properties file...). Please keep the code in structured form.

 * Clone the FOO declarations to create and add (in alphabetical order) {{{

public static final Type BARCOMP; public static final Type BARCOMP_XML; }}}

 * Clone the FOO assignments to create and add (in alphabetical order): {{{

BARCOMP = new Type("chemical/x-barcomp", ObjectType.TEXT, "barcomp"); BARCOMP_XML = new Type("chemical/x-barcomp-xml", ObjectType.XML, "barcomp.xml"); }}}

• If there is a chemical MIME type for either or both of these, please use them.* If there isn't, liaise with PMR or Henry Rzepa to have one registered. You may also need to liaise with the program developers.

The "barcomp" is the lexical type. It may also the a file suffix. This is a messy area because some many people use non-standard suffixes. The "barcomp" and "barcomp.xml" should be unique in the file. (Yes, we should rewrite this with properties files and so on...).

 * Then clone and edit the lines and add (in alphabetical order) : {{{

typeList.add(BARCOMP); typeList.add(BARCOMP_XML); }}}

Now it should compile...

Editing the Code modules

to be continued