Cast int cell to string in excel file

Let’s say I want to parse the attached file. It has in the A column string or int, but I want every cells of that column to be string.

I tried with this package.yaml

resources:
  - name: example
    type: table
    path: example.xlsx
    scheme: file
    format: xlsx
    encoding: utf-8
    mediatype: application/vnd.ms-excel
    schema:
      fields:
        - name: A
          type: string
        - name: B
          type: string
    pipeline:
      steps:
        - type: table-normalize
        - type: field-merge
          fromNames:
            - A
            - B
          name: AB

and runned this python script

from frictionless import Package

p = Package.from_descriptor('package-issue.yaml')

p.extract()

for r in p.resources:
    r.transform()
    print(r.to_view())

I have this error when I try to tranform it with field-merge

frictionless.exception.FrictionlessException: [step-error] Step is not valid: "field_merge" raises "sequence item 0: expected str instance, NoneType found" 

Is there a way to cast all ints in string ?

example.xlsx

[yohanboniface]

@shashigharti we have an ETL where we ingest data that we do not produce. The dataset file and format is known, but we cannot change it.

In other words, we want to be able to import any dataset from anywhere on the Internet, “just” by defining a “schema file” on our side that would give to our ETL the rules needed to “normalize” the data before importing it in our database. By normalizing, I mean making sure for example that an int column only contains int, or mapping “pseudo booleans” to a real boolean column, or making sure that some administrative codes are valid, etc.

The ETL output is a normalized table in a postgresql database (each resource ends in a separate table).

To give a bit more details: we currently have our own code base, very similar to Frictionless, and we are evaluating the opportunity to use Frictionless, instead of reinventing the wheel.

To have an example of how we describe our ETL steps currently: https://datahub.incubateur.tech/infrastructure/insitu/-/blob/main/insitu/datasets/numerique.yml

Hope this helps! 😃

[roll]

We’re releasing frictionless@5.0.0b11 that now supports formats.ExcelControl.stringified:

resources:
  - name: example
    type: table
    path: example.xlsx
    scheme: file
    format: xlsx
    encoding: utf-8
    mediatype: application/vnd.ms-excel
    dialect:
      excel:
        stringified: true
    schema:
      fields:
        - name: A
          type: string
        - name: B
          type: string
    pipeline:
      steps:
        - type: table-normalize
        - type: field-merge
          fromNames:
            - A
            - B
          name: AB