This example demonstrates how to extract tabular data from PDF documents. The lib/totable.jar file uses Computer Vision techniques to recognize tables and outputs XML markup. The DeepDive extractor parses the XML markup and populates a database relation with table data. The schema of the relation is as follows:

CREATE TABLE table_cells(
  id bigint,
  filename text,
  page int,
  table_id int,
  row int,
  column_start int,
  column_end int,
  content text,
  pdf_coordinates text
)

The TableCell in udf/table_cell.py can be reused in your own applications to deal with tabular data.

1. Edit the db.url file according to your database configuration.
2. Run deepdive initdb, then deepdive run

This example application does not perform any inference. It's purpose is to show how to use extractors to extract tables from PDF data.