I am trying to join 2 spatial files that have some but not all attribution the same.

One of them (property_template.tab) has no data but has the schema that I want. The other one (cadastre_bpoly.tab) has the spatial info and only some required attributes.

In QGIS a simple merge vector gives the result I need where if an attribute matches it is filled but if not it's left.

In FME I have tried FeatureMerger where the requester/supplier has join on as $(SourceDataset_MAPINFO_5) but it doesn't give a result.

I have also tried ListBasedFeatureMerger.

ALL I need to do is the following:

  1. Get the previous output of the property table and use it as a template (with no data)
  2. Populate it with the cadastre and have it auto fill any attributes that are the same.
  3. Join by attributes with an xls sheet that has a key field but be able to build up the key field using existing columns.

I understand #3 is really a separate question but I included it as it may help in choosing the correct transformers.

For points 1&2 above the merge shouldn't try to do any type of join by attribute but it doesn't allow me to leave the Join Attributes Requester/Supplier blank.

===== UPDATE

Based on Mark Ireland's suggestion enter image description here

4 Answers 4


No experience with TAB files but in general a source file without any records will not read any feature. So using an empty file to merge the schema to the features is not going to work, this way.

You can read the schema from the file (SchemaReader or FeatureReader) and use it to write a dynamic schema (send Schema Feature directly to the writer, select Dynamic Schema Definition, "Schema from Schema Feature") but I doubt this is what you are looking for.

Other than that you are on the right track I think. You can easily put the same character, like 1, in the Requestor- and Supplier fields to simulate 1=1.

FeatureMerger Parameters

Connect the new input to the Requestor port and the old data to the Supplier port and there you go.

For #3 you do the same, now using the key field in the FeatureMerger.


I'd give the FeatureJoiner a look too -- it has very similar functionality to the FeatureMerger but uses database terminology and as such may be easier to understand. After the join you may need to merge the output from ports (with either transformer) to get what you wanted. Worst case, post a simple example here of input data and what you're doing and we can look more closely.


I think - if I understand correctly - that I would do a simple translation for parts 1 and 2. So, in an empty workspace, try following these steps...

  1. Choose Readers > Add Reader from the menubar.
  2. Set MapInfo tab as the reader format and select the Cadastre dataset.
  3. Click OK to add the reader.
  4. Choose Writers > Add Writer from the menubar.
  5. Set MapInfo tab as the writer format and choose a new folder to write to.
  6. Set Table Definition to Import From Dataset and click OK to add the writer.
  7. You will be prompted where to import the schema from. Set MapInfo Tab as the format and select the Property dataset. Click OK to finish adding the writer.

You now have a workspace that will write the Cadastre data to a new MapInfo dataset, using the Property dataset as a form of template.

If any attributes exist in the Cadastre data that also appear on the Property schema, they will get written out. Everything else will be ignored.

There are a few ways to do this, but I think my solution is about as good as you can get. You could use a Dynamic translation, but I don't see any point in that (unless you are going to run this multiple times and the Property dataset will change schema each time). The FeatureMerger/Joiner method mentioned in other answers isn't that simple because it doesn't define the output schema.

The one possible snag is that the Property dataset is empty. FME might not like that. Without features it might think there is no schema and you get an error when adding the writer. The simple solution would be to add a single record to that table. We aren't reading that dataset, just using the structure, so it doesn't matter whether the table is full or empty. That data won't get included, only features from the Cadastre data.

As you say, requirement #3 is really a different thing altogether. But, as everyone else here has mentioned, you can use a FeatureJoiner (or FeatureMerger) transformer to do this. Actually, given that the data is Excel, and you don't really need to read all of it into the workspace, I would use the DatabaseJoiner transformer. So create your key as a new attribute (StringConcatenator or AttributeManager > Text Editor dialog) and use that in the DatabaseJoiner to connect to the Excel sheet.

I found a reasonably suitable demo dataset so I put a short video together, to demonstrate the techniques above. I hope it is of use. You can find it here: https://www.screencast.com/t/tO1F3s3i

  • thank you for the details. I have tried this but the output data isn't formatting correctly, allthough it runs without errors. I will update the question shortly with what I tried. BTW you can get free data from data.qld.gov.au
    – GeorgeC
    Jun 25, 2019 at 23:11
  • 1
    I've figured it out -now working on the SQL connection as the excel was just to test if the core processes worked.
    – GeorgeC
    Jun 26, 2019 at 0:25

The feature merger should be okay to use but you may want to take the data coming from the different ports such as Not Merged to account for when a match was not found.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.