Well folks, I can't believe it's almost over. What a journey. I never imagined I'd be able to do this. I have an (almost) working project. I have a few more things that need to be done.

Let's highlight what IS done:

1. GroovyForm and its related metadata is stored
2. Model class is interrogated for its properties and those are stored in a container class.
3. From the data collected in the container class, I generate markup checking for a predefined set of data types.
   
4. The view and its related controller is generated -- sneaking in some groovy magic of course!
5. GroovyForms are successfully saved into the system, along with their respective, model, view and controller.
6. Forms are persisted when the module is shutdown and reloaded when it is restarted

What's left:

1. Editing metadata related to the forms currently in the system
2. Writing the servlet which handles submissions
3. Create some sample forms
   
4. Documentation

I did plan to add a lot of other stuff such as the ability to create custom form widgets, however there were a few problems which caused me to redirect my focus. These will come in after I finish with Summer of Code.

Okay, week 1 is nearing an end.I accomplished alot of things.

First, I completed the meta-data storage code. I wrote some tests, which revealed some bugs that needed fixing, and they were. I am happy to say, that the metadata storage works!

Writing the test was difficult due to the fact I store the forms in the system using a static List in my container class. I wound up updating to JUnit 4 to get at the @BeforeClass annotation so that the only one set of forms gets loaded into the container class. This helped me greatly. Additionally, the @AfterClass annotation was handy for cleaning up from the tests.

Secondly, Writing the code for interrogating the model was a piece of cake, thanks to groovy. All code for interrogating the model has 50 lines! That includes a model class I wrote to hold the field types and field names. I'll show you, but before I do, I should note that return statements are optional, and the final statement will be returned.

package org.openmrs.module.groovyforms.util

import java.lang.reflect.Field
import org.openmrs.module.groovyforms.metadata.model.GroovyFormsDomainModel

/*
 * Utility class containing methods for class interrogation.
 */
class GroovyFormsClassUtil {
    /**
     * Interrogates the class for all declared fields and
     * stores the type and name in a container class
     * @param the {@link Class#getCanonicalName() canonical name of the class}
     * @see Class#getCanonicalName()
     * @return a reference to a container class containing the type
     */
    static def getModel(fields) {
        def domainModel = new GroovyFormsDomainModel()
        def names = domainModel.fieldNames.&add;
        def types = domainModel.fieldTypes.&add;
        def f = fields.each {Field field ->
            names field.name
            types field.type.canonicalName
        }
        domainModel
    }

With groovy, it's so easy and concise (as you can see). Let's explain what's going on. First, I pass in the Field array I get from Field.getDeclaredFields(). Now, groovy adds methods onto the standard JDK classes, one of those methods is a method named each() which takes a closure. Now, back to the point, I pass a Field into the closure. That closure is executed for each element (in this case, a Field). Then I add the name, and the type to a List stored in a container class, now that container class is also written in groovy!

package org.openmrs.module.groovyforms.metadata.model

/*
 * Ths class holds information about the properties of the model. 
 */
class GroovyFormsDomainModel {

    /**
     * The field names
     */
    def fieldNames = []

    /**
     * The field types
     */
    def fieldTypes = []

}

Now, the fields are Lists, not arrays. Anyways, I've gotten off on a tangent here, so let me get back on track.

What I accomplished:


Week 1:

Code to generate the directory structure, serialization of metadata back/forth between XML and POJOs (Plain Old Java Objects), Wrote tests to ensure everything works in that regard. Additionally, I wrote the code to interrogate the domain model which I will generate the forms from.

Next Week

Write up the templates for the view/controller and write code to do the generation of the view/controller. Write some tests to ensure everything generates correctly.

Unrelated to that, my stored-value card from google came today. It feels nice to be $500 richer! This is going to be the best summer, I'm already having fun doing this. It's amazing seeing the whole project evolve into something amazing.

I have made a lot of progress on my Google Summer of Code Project.

First, I've completed the code for the generation of the directory structure for the forms. This is done using the form id, which is the name with all illegal characters removed. The only valid characters left are alpha-numeric and periods. Let me tell you, the regular expression to replace all of those characters was hell to write. But, thanks to the help of Jason Davis, I was able to make it much more readable and easier digest. You've gotta see it to get an idea of how bad it was, so here goes:

   /**
     * Removes all illegal characters, then removes all spaces and makes it all lower-case.
     *
     * @param str the form name
     * @return a viable form id.
     */
    public static String formNameToFormId(String str) {
        return str.replaceAll("[\\$\\(\\)%`~!@#&;:\\^=\\+\\?
\"\\*\\\\////\\\\{\\}'\\]\\[\\|\\s+]", "").toLowerCase();
   
    }

That is pretty much hell on earth to read, and it was hell on earth to type. Here's the *MUCH* simpler version:

   /**
     * Removes all illegal characters, then removes all spaces and makes it all lower-case.
     *
     * @param str the form name
     * @return a viable form id.
     */
    public static String formNameToFormId(String str) {
        return str.replaceAll("[^a-zA-Z0-9.]", "").toLowerCase();

    }

Note, you can clearly see that i'm negating that entire sequence. It does the SAME thing, but is MUCH more readable. That reads: "replace everything that isn't a letter from A to Z *OR* a to z(lower case letters) *OR* from 0 through 9 *OR* a period. Whereas the above just excluded the invalid characters explicitly but was a MONSTER of a regular expression.

Secondly, metadata persistence is written in, I'm using xstream. It's a great tool for XML serialization.

That's it. Oh, by the way, these are the things I earmarked TWO WEEKS FOR! Now, I push domain class model processing up, will work on that tomorrow.

Today was the official start of Google Summer of Code 2008 Hopefully everybody is finding it a pleasant experience. I have yet to start coding as today was a US Holiday (Memorial Day).

The project plan for the summer has been finalized. I decided to set two-week blocks, rather than do a weekly thing. I chose this, as it's extremely flexible and can scale items either back or forward.

The project plan can be found here. A Google Doc which my mentor and I used to plan the project can be found here. Finally, See my general plan written as a previous blog post.

Best of luck to all the summer of code students!