See README.txt for previous intro
This package has major enhancements in 2016-11...2017-02 [onwards] due to the AMI-EPPI project. The goal is to extract HTML tables with high precision / recall. we assume the input SVG is the putput of PDF2SVG. Currently we assume per-page and per-table input. The examples in current development are tables already excised (snipped manually with Inkscape), so the problem is reduced to something known to be a table.
The details are in TABLE.md