See post on TDS

Further readings:

• Extracting Semi-Structured Data from PDFs on a large scale
• Extracting Tabular Data from PDFs