Can you help me to compare fields in two tables? I have two tables, which contain ID and Measure fields. I need to take stroke by stroke values from field1(table1) and find all values in field1(table2) which are less than sought-for value. And then take largest of them and put it on stroke field2(table1). Example showed on picture below.
[GIS] Compare values of two features by FME
attribute-tablefields-attributesfme
Related Solutions
Just use the Change Detector. There's nothing more to it. You can check that the geometry is the same, or just check attributes.
See the description from FME help:
ChangeDetector
Detects changes between two sets of input features.
This transformer is often used with multiple readers, to identify changed features in the two files. It can identify all features that two input files have in common, and those which are in one file and not the other, such as the additions and the deletions.
As usual with FME, they have already implemented something, so doing it yourself with a CRC Calculator, or feature merging is unnecessary.
I have two ways to do this.
One is to rename the attributes in both tables to unique names, merge the records with the same ID into one record and create new, original, fieldnames. This way you are fully in control and you are able to handle each field its own way. Performance will be quite good as this way of working is not hard on internal memory. Downside is a lot of fields means a lot of work.
- Read both tables.
- Prefix attribute names with BulkAttributeRenamers. For example "Table1.FieldA" and "Table2.FieldA".
- Merge records from tables with FeatureMerger based on ID.
- Use AttributeCreator with conditional value to recreate fields based on conditional per field. For example "FieldA" : If "Table1.FieldA" = "Table2.FieldA" then "Table1.FieldA", Else If "Table1.FieldA" is null or "Table1.FieldA" = 0 then "Table2.FieldA", Else If "Table2.FieldA" is null or "Table2.FieldA" = 0 then "Table1.FieldA", Else "DUMMY" (I think you can create one and duplicate if you want. If not you could use an AttributeCreator for each field to prevent doing double work.)
- Use Tester to check if values are DUMMY.
- If fields are correct use Keeper to remove the "Table1." and "Table2." fields.
- Write output.
The other, more generic, solution is to create a feature for each field, merge the field features with the same ID and fieldname into one record and create new, definitive, fieldnames. Then aggregate the field features back into the original records. This way you only have to use one AttributeCreator for each variant. A bit harder to design, harder on performance as you will get rows x columns number of features. But you wont have to rebuild each different fieldname. Which is a big plus with this much attributes :)
Tip: If the tables have geometry, the GeometryExtractor and GeometryReplacer are a nice way to control geometry flow.
Best Answer
OK, so you basically need to compare each of the records in table 2 against each record in table 1. To me that suggests you are going to need a loop of some sort. You could either use a loop (or nested loops) inside a Custom Transformer, or you could do an alternative type of loop using a list.
For the Custom Transformer, check out this FMEpedia example, or read chapter three of the FME Desktop Advanced training course.
For the list-type looping, check out my blog post on the subject.
Hope this helps. It's a tricky problem, but one that should be solvable with a little work.