Export GML with Date attribut using Dynamic Reader and Writer

From the oracle data, are you able to identify what fields are meant to be datetimes?

In the DateTimeConverter you can specify either the attribute (_creation_instance) or to use the value that _creation_instance (@Value(_creation_instance)) contains

Hi @hkingsbury ,

Sorry, I don’t get “_creation_instance” values from the Reader. I get the input parameters and “fme_feature_type_name” value. Could the problem be in the configuration of Reader?

Tkanks!

Well, my solution abuses BulkAttributeRenamer, RegEx and List Attribute naming convention, and is complex…..but it works.

Frankly though, I would probably just write something in PythonCaller, but the below works for out-of-the-box Transformers.

Step 1 is to get Attributes that are Date/Time from the Schema Features from a FeatureReader

 

Step 2 is to Join these with the Dynamic Feature Reader Output.  In this case, there are 11 Date/Time Attributes for the 100 Sample Features, so this gives 11 x 100 Joins.  This gives a “Grid” of Values, where each of the 1,100 Features represents 1 Date/Time Attribute “Cell” 11 columns of date/time Attributes x 100 Features to give 1,100 Features to operate on.

Step 3 is to temporarily rename each Date/Time Dynamic Attribute to static/known Name by abusing BulkAttributeRenamer, which can work well as a type of  “Dynamic Schema AttributeCreator”

 

So now we have a “Table” from this that is _count (Workspace Unique Feature ID), fme_feature_type (Table Name), name (of date/time Attribute) and “_date_col” which is the temporary known and exposed name of the Date/Time Attribute

Step 4 is to find which Attribute “Cells” have a valid FME Date/Time value, and convert them to Y-m-d
 

Step 5 is to know  a bit about how Lists work in being just a collection of Attributes that happen to have their Name as in a format like _list{1}.Attribute Name .  If we remove the _list{1}. part of the List Attribute Name by again abusing BulkAttributeRenamer, then they become just standard Attributes as far as the Workspace is concerned.

 

So now we use a FeatureJoiner with Override from the Right set up to replace the original FME format Date/Time Attribute Values with the new DateTimeConverter String values.

So waaay more complex than I expected to “Dynamically” find Date/Time Attributes in a Dynamic Reader situation.  It would be a lot simpler if could have a Transformer like AttributeCreator that could perform in a dynamic “Attribute Scan” type mode that could traverse all Attributes and use Eg. Conditional Values. 🤨

Or as above, I would probably just write a Custom PythonCaller...

The easiest way is probably to get a SchemaScanner to generate a valid schema feature?

I tried a bit and this might be harder than I suspected initially. The SchemaScanner does not seem to work as I hoped it would.

hkingsbury wrote:

From the oracle data, are you able to identify what fields are meant to be datetimes?

In the DateTimeConverter you can specify either the attribute (_creation_instance) or to use the value that _creation_instance (@Value(_creation_instance)) contains

 

I would try what is suggested here.

_creation_instance is just an example of an attribute, substitute it with the attribute that contains the field names