The Source transformation extracts the full structure of the document including text, tables, headings, and metadata. You can extract text from documents that have different document structures, such as invoices and reports, while preserving the order of the text.

To read a PDF, use the

Source

tab and select

Document

.

Data Integration

sets the input type to PDF automatically.

To read a directory of PDFs, change the

Source Type

in the advanced properties to

Directory

. For the

File Name Override

, enter

*.pdf

.

The

Fields

tab displays fields to store the text, file path, file type, and file name for each PDF.

You can pass the text to downstream Chunking and Vector Embedding transformations to build a RAG ingestion pipeline, or you can process the text, create structured data from it, and write it to a JSON file.