Hi,
You could use the csv_line_number. Expose this attribute on the format attribute tab.
Add the last csv_line_number to the linenumbers from the manipulated file.
This, after u have skipped (or removed) the headerline(s) of the manipulated file.
Route both to a sorter, sort and write. (this last step might not be needed, but to make sure..)
Hi Matt,
I would use the fanout option of the CSV writer feature type. You can control reading order for source datasets in Navigator window. That is, upper writer in Navigator will be run first.
Takashi
When doing it that way you may want to skip header, using the csv reader parameter panel.
Thanks Gio and Takashi for your rapid reponses. I've been working on another problem for the last couple of days so have only just checked back.
I've attempted the approach Takashi has provided in his screenshot. The problem I have is that the 1111_AB.csv and 1111_AB_type47.csv have records of different schemas so I receive datatype mismatch messages such as
Attribute of type decimal has an illegal value of '2014-04-30'. Value must be < 1000 or > -1000. Value set to missing value
String value `2014-04-30' contains invalid characters and could not be converted into a float
String value `%0' contains invalid characters and could not be converted into a float
String value `%0' contains invalid characters and could not be converted into a float
when I run the workspace.
How do I accommodate this?
If it's allowed that the fields contain mixed type data, specify "text" to types for the fields in the CSV writer feature type. Otherwise, use the AttributeClassifier to determine data type, and modify Failed values appropriately.
I ended up using two separate workbenches. 1 to perform the manipulation of attributes in the *type47.csv files and then a workspace with text line readers and writers with the Read Whole File at Once parameter set to Yes to put the files together in exactly the way they're formatted in the source files.
