1. first we need to read the CSV as Text File;
2. then with counter we assign a unique id to each row;
3. then we distinquish the header row from non-header row by defining a attribute named "isHeader” and using REGEX 

   

    

4. we use python caller to assign a block-id to each group of line following a header
5. afterwards you can do separate rows based on the block-id and do whatever you intend to do

   

   python caller:

   import fme
   import fmeobjects

   
   class FeatureProcessor(object):
       """Template Class Interface:
       When using this class, make sure its name is set as the value of the 'Class to Process Features'
       transformer parameter.
       """
       
       def __init__(self):
           self.block_id=0
           """Base constructor for class members."""
           pass
           
       def has_support_for(self, support_type: int):
           """This method is called by FME to determine if the PythonCaller supports Bulk mode,
           which allows for significant performance gains when processing large numbers of features.
           Bulk mode cannot always be supported. 
           More information available in transformer help.
           """
           return support_type == fmeobjects.FME_SUPPORT_FEATURE_TABLE_SHIM
     
       def input(self, feature: fmeobjects.FMEFeature):
           is_header=feature.getAttribute('isHeader')
           cnt=feature.getAttribute('_count')
           if is_header=='Yes':
               self.block_id+=1
               
           feature.setAttribute('block_id', self.block_id)
           self.pyoutput(feature)

       def close(self):
           """This method is called once all the FME Features have been processed from input()."""
           pass

       def process_group(self):
           """This method is called by FME for each group when group processing mode is enabled.
           This implementation should reset any instance variables used for the next group. 
           Bulk mode should not be enabled when using group processing. 
           More information available in transformer help.
           """
           pass

Nice idea. You can also easily do this without python.

1. Read the CSV as Text File
2. Define the attribute "isHeader” as ​@mramezani suggests, and an attribute “GroupID".
3. Use an AttributeManager with Adjacent Feature Attributes enabled, looking at 1 prior feature. Set the default value for substitution to 0.
4. Set a Conditional Value in "GroupID”. When “isHeader"="Yes”, add 1 to the value of "GroupID” of the prior feature. When “isHeader"="No”, use the value of "GroupID” of the prior feature.

I'm not at a computer running FME at the moment, so you will have to look up the details yourself.

Thanks ​@mramezani it does give me a good direction to solve this issue.

​@geomancer I will check this as well, for now ​@mramezani method does solve it.

Hi ​@goldstein ,

Another way, you can also use a global variable to define "Block ID" for each header row and propagate it to subsequent data rows. This is an old work flow design before the Enable Adjacent Feature Attributes option have been introduced into AttributeCreator/AttributeManager, but still effective in some cases, I think.

Ah yes, the curious construction where the VariableRetriever is used before the VariableSetter.

Oh, this workflow is better.