Tablib is a Python library for working with tabular datasets in a format-agnostic way. It provides a core `Dataset` class that stores rows and optional headers, and a `Databook` class for collections of datasets. The library's primary functionality is converting between many structured data formats: CSV, TSV, JSON, YAML, Excel (XLS/XLSX), OpenDocument (ODS), dBase (DBF), HTML, Jira, LaTeX, reStructuredText, SQL INSERT statements, and pandas DataFrames. It also supports CLI table rendering via the `tabulate` package. The architecture is modular: each format is implemented as a class with `export_set`, `import_set`, `export_book`, `import_book`, and optionally `detect` methods. These are registered with a central `Registry` that also sets up dynamic property access on `Dataset` and `Databook` objects (e.g., `dataset.csv`, `dataset.json`). Lazy loading through importlib and string-based class references minimizes startup cost for optional formats like XLSX or YAML. The core `Dataset` object provides methods for row/column manipulation, sorting, filtering by tags, transposing, stacking, and applying formatters for custom cell transformations. Data flows are straightforward: users populate a `Dataset`, optionally set headers, then export to any registered format with a single `export()` call. Import reverses this process. Key entry points are `tablib.Dataset()`, `tablib.Databook()`, `tablib.import_set()`, `tablib.import_book()`, and `tablib.detect_format()`. The library has no hard runtime dependencies; all format-specific packages (openpyxl, xlrd, xlwt, odfpy, pyyaml, pandas, tabulate) are optional and requested via extras like `tablib[xlsx]` or `tablib[all]`. The vendored `dbfpy` module handles DBF format without external dependencies. Exception handling is structured with custom exception classes for invalid dimensions, unsupported formats, and missing headers.

The LLM identified 3 files worth highlighting for reuse: