How to produce dynamic schema output from an output port from FeatureWriter?

Hi Ashish,

I'm not sure of any setting that would help but I do have a suggestion. What you could do is add a FeatureReader after the Writer to read in the data with the correct Schema. It's an extra step but should solve the issue.

Hi @brianatsafe,

Thanks for your suggestion.

I tried adding a FeatureReader after the Writer. However, I could not set the FeatureReader properly so that it dynamically reads the schema. It seems the attributes for each feature is read in, but not exposed (However, I can't use AttributeExposer as I would not know the attribute names beforehand).

Also, the filenames of the csv file that I'm writing (and reading again) are based on an attribute value. (If I explicitly select the csv file, then FeatureReader correctly produces the schema.)

The output from the FeatureReader is as follows:

My workspace is as follows:

Thanks @ashish_man. Excellent context here. I understand the issue a bit deeper and I'd like to know more about what you would like to do.

Question: downstream from the FeatureWriter, what specifically do you want to achieve with FME? Let us know what you are trying to do and I'll provide any suggestions. (for example, are you trying to edit Attribute Names if a specific name is in the file...or anther type or process)

Thanks for the support @brianatsafe. My objective is to automate reading tables from a number of Autocad drawings from different locations, and present it in a way useful to other user groups. The final output format could be a database containing info from all locations. Mostly the tables in the Autocad drawings are just a bunch of lines and text put together that appear as a table. Moreover, the title, attribute names, content, location or the number of the tables on the dwg are not consistent. The only thing common are few key words of the attribute names of the table.

There may be many different way to solve this problem, but I started by searching these keywords, their neighbour lines and after a bunch of text location based transformations, I was able to reproduce the tables in a CSV file by setting dynamic schema definition. The CSV file is just an temporary file that had to be created for changing the schema of the table dynamically. My next step is to generalise the tables (renaming attributes to standard format, merging tables etc) so that it can be put into a database. But for it, I have to dynamically read the tables.

I might need to avoid dynamic reading of CSV, by doing all necessary transformations before writing to the CSV. Please suggest.

Thanks,

Ashish

Thank you for that context. Super helpful and this sounds like an interesting project! Creating those tables from drawings is a huge accomplishment in my books. Anyways, I think part of the problem is that you need some user interaction between reading the original data and then once the table is understood you want to further process it. Unless you have a number of specific rules that you can specify in the workspace, it may be easier to have a user drive the process.

Anyone else have other ideas?