Generate realistic and reproducible Excel workbooks for automated testing.

xlsynth is an R package for generating synthetic Microsoft Excel workbooks.

It was originally developed to support testing of xlversion, an R package for comparing and fingerprinting Excel workbooks. During development it became clear that creating representative test files manually was slow, repetitive and difficult to automate.

xlsynth provides a declarative way to generate realistic Excel workbooks that can be used to develop, test and benchmark software that reads or processes .xlsx files.

Most software that processes Excel files is tested using clean, perfectly structured spreadsheets.

Real-world workbooks are rarely like that.

They often contain:

• blank rows
• hidden worksheets
• merged cells
• inconsistent date formats
• duplicated values
• mixed data types
• missing values
• inconsistent formatting

Testing only against ideal workbooks gives a false sense of robustness.

xlsynth helps developers generate reproducible workbooks that better represent the spreadsheets encountered in production.

• Declarative workbook schema
• Synthetic but realistic datasets
• Multi-sheet workbook generation
• Reproducible output using random seeds
• Optional workbook mutations for testing edge cases
• Designed for automated testing and CI pipelines

The first release focuses on three core capabilities.

Describe a workbook instead of manually creating it.

schema <- workbook_schema(

    sheets = list(

        sheet(

            name = "Sales",

            rows = 5000,

            columns = list(

                date(),

                customer(),

                product(),

                currency()

            )

        )

    )

)

Generate a complete Excel workbook from the schema.

generate_excel(

    schema,

    output = "sales.xlsx"

)

Introduce controlled imperfections into an existing workbook.

mutate_workbook(

    "sales.xlsx",

    actions = c(

        "blank_rows",

        "mixed_dates",

        "missing_values"

    ),

    seed = 42

)

Available mutations include:

• blank rows
• duplicated rows
• missing values
• mixed date formats
• merged cells
• hidden worksheets
• renamed worksheets
• inconsistent number formats

Generate a realistic workbook.

schema <- workbook_schema(

    sheets = list(

        sheet(

            name = "Employees",

            rows = 1000,

            columns = list(

                employee_id(),

                person_name(),

                department(),

                hire_date(),

                salary()

            )

        )

    )

)

generate_excel(

    schema,

    output = "employees.xlsx"

)

xlsynth and xlversion are complementary.

• xlsynth generates realistic Excel workbooks for testing.
• xlversion analyses, fingerprints and compares those workbooks.

Together they provide a reproducible workflow for developing and testing software that processes Excel files.

• Data engineers
• ETL developers
• QA engineers
• Software developers
• R package developers
• Anyone building software that reads Excel workbooks

• Reproducible
• Declarative
• Modular
• Extensible

• Workbook schema
• Workbook generation
• Basic synthetic data generators
• Workbook mutations

Potential future extensions include:

• Industry-specific templates
• Additional mutation strategies
• Large workbook generation
• Performance benchmarking
• Workbook fuzz testing

xlsynth started as an internal tool while developing xlversion.

Creating realistic Excel files by hand quickly became one of the biggest bottlenecks when writing automated tests.

Rather than maintaining dozens of manually crafted spreadsheets, xlsynth generates reproducible workbooks on demand, making automated testing simpler and more representative of real-world scenarios.

🚧 Early development (MVP)

Contributions, suggestions and feedback are welcome.