Built by Maven

Datasets

Overview

The IDataSet interface represents a collection of tables — the primary abstraction dbUnit uses to manipulate tabular data, whether it comes from a file, a live database, or is built programmatically. Its core contract:

  • getTableNames() / getTables() / getTable(String) — the tables in the dataset.
  • iterator() / reverseIterator() — insertion-order (and reverse) access, used by e.g. DatabaseOperation.INSERT/DELETE_ALL to respect foreign-key ordering.
  • isCaseSensitiveTableNames() — whether table name lookups are case-sensitive.

ITable represents one table: its ITableMetaData (columns), row count, and getValue(row, column) cell access. See ITable & Table Metadata for the full class reference.

Building Datasets Programmatically

DefaultDataSet is the basic mutable IDataSet implementation used to construct datasets in code rather than loading them from a file.

DataSetBuilder is a fluent builder on top of it for the common case, avoiding hand-built DefaultTable/Column wiring. Column data types default to UNKNOWN and are resolved at runtime by dbUnit.

Single-table example:

IDataSet dataSet = new DataSetBuilder()
    .table("FOO")
        .columns("ID", "NAME")
        .row(1, "Alice")
        .row(2, "Bob")
    .build();

Multi-table example:

IDataSet dataSet = new DataSetBuilder()
    .table("FOO")
        .columns("ID", "NAME")
        .row(1, "Alice")
    .table("BAR")
        .columns("X")
        .row(42)
    .build();

Dataset Formats and Decorators

Each of the following is a focused page — file-based formats first, then the programmatic/live sources, then the decorators that wrap any IDataSet/ ITable to add behavior, and finally the cross-cutting rules for how a field’s text becomes a typed value:

Page Covers
Flat XML FlatXmlDataSet — one XML element per row, the most commonly used file format.
XML XmlDataSet — the original, more verbose XML format.
CSV CsvDataSet — one CSV file per table, plus a table-ordering.txt manifest.
JSON JsonDataSet — one top-level key per table, values as row-object arrays.
YAML YamlDataSet — one top-level key per table.
Excel (XLS) XlsDataSet — one sheet per table.
SQL*Loader SqlLoaderControlDataSet — builds a dataset from Oracle SQL*Loader control files.
Query-Based QueryDataSet — tables populated from arbitrary SQL queries against a live connection.
Streaming Datasets StreamingDataSet — forward-only, low-memory access over a producer.
Data File Loader DataFileLoader implementations — load a dataset file from the classpath, with ReplacementDataSet substitution built in.
Decorators CompositeDataSet, FilteredDataSet, SortedTable/SortedDataSet, ReplacementDataSet, TableDecoratorDataSet, CachedDataSet.
Dataset Values Not a format — how a field’s text becomes a typed column value in any of the above: nulls, empty strings, [TEXT|BASE64|FILE|URL] and [now…​] syntax, UUID literals, booleans/numbers from strings, and Flat XML column sensing.

For a live database as a dataset source (IDatabaseConnection.createDataSet()), see Connections & Configuration.