10 comments

[ 2.9 ms ] story [ 30.6 ms ] thread
Is Excel the new religion?
This joke would make more sense if the title was something like “How to convert from PDF to Excel”.
That depends on whether PDF is the heathen religion, or the heathen.
This line confused me, "We are going to see that you do need developers for this, but sysadmins."

I'm guessing it meant to read, "We are going to see that you do not need developers for this, but sysadmins."

This is exactly how we do it for few of our clients. Opencv + Python is provides a pretty successful solution for this.
As someone who has now done this exact thing a couple of times for work, I'm going to say that I disagree with the article about one point: you probably do need developers/data scientists unless your extraction/standardisation exercise is trivial in size and complexity. Obviously that depends on your subject matter a bit, but generally people won't approach you if its small scale and already in a good structure.

It's all well and good sucking out tables from pdfs in something roughly akin to the format they were entered, this doesn't make the tables magically analysable by machine, even under the assumption of perfect extraction (which rarely holds in real life, at least in my experience).

So fundamentally, you need someone that can define the problem, parse the text and information from the pdf, munge the data from its original form into something useful, while writing some tests and procedures to verify and quantify the quality/success of doing so.

I'm not saying that's sysadmin work or not (since job titles don't really mean anything much to me), but it definitely sounds like developer/data science work, and it's beyond the abilities of the non-tech crew...ymmv

If you're looking for quick and easy, rather than an OSS solution, https://pdftables.com/ might be of interest.

(Disclaimer: I was working at the company when they were developing this)