In dsidavis/ReadPDF: Extract Information from PDF Documents

Finding Tables.

Tables can take many forms. However, tables in a given document often have the same basic structure and appearance.

In several papers, + tables have "Table number" either as a centered title or start of a sentence, + some text underneath this which servese as a caption + a line that spans the entire column or page + some column header information, potentially with smaller lines + another column/page-wide line separating the header and body + body content + a line that ends the table + possibly some text coresponding to footnotes from the table.

The content is also separated into columns. We can use any lines in the header or the positions of the text within the header to try to identify the column start and ends.

See the PDFTables package for how to read the tables. Our goal here is to find the tables. These two are related.

So how do we find a table corresponding to the description above.

First we find the word Table. This might be Table, TABLE, or actually separated into T and then ABLE with different fonts.

getNodeSet(ag, "//text[. = 'Table' or . = 'TABLE' or (. = 'T' and following-sibling::text[1] ='ABLE')]")

getTables() does this.

For each of these nodes, we + check to see if it is in the center of a column or the center of a page. It could also be at the start of the line.

• Find the lines and see if +

Next we look for lines near by. We also find the text

dsidavis/ReadPDF documentation built on June 12, 2025, 6:39 a.m.