Further to Kruncher's suggestion, have you experimented with the MONOSPACED indicator?
Bear in mind that PDF documents are, by comparison with 'old style' reports written for line printers, more likely to have been produced with 'Office' products and so are susceptible to the invasion of proportionally spaced fonts mixed with fixed width fonts. (Indeed just 'fonts' can add greatly to the complexity as part of their attempts to beautify ugly reports!)
This can make life especially challenging for the PDF interpreter.
I sometimes have to remind myself that the objective is simply to extract information no matter how ugly the method gets. Sometimes you have to extract what you can in a chunk and slice and dice it later in the table window using calculated fields. Potentially that is quicker and safer, being less prone, hopefully, to the vagaries of future versions of the report.
If you have any control over the creation of the PDFs you might be able to edit the output settings, so that a regular delimiter appears between columns in your data. Then you can use the floating trap and break it up using Lsplit() on the Column() value.
In SAP, we can export some reports as "unconverted" spool files, others we output as "raw" and these options give us different delimiters. We're not, mercifully, having to use PDFs yet.
Of course, if you've no control over the generation of the PDFs, then things are trickier. Have you tried printing the PDF as an XPS and using that?