Repository files navigation

A simple wrapper for thePoppler PDF lib for Golang.

This go module exports a limited subset of Popplers functions, specifically only those needed to extract text from PDFs.It drops support for images in favor of also dropping thelibcairo dependency.

In order to use this module you need to install two packages. In Debian/Ubuntu this is done by:

# Debian-based OSapt-get install libpoppler-glib-dev

go get github.com/johbar/go-poppler

This is a fork oftimsat/go-poppler which is derived ofcheggaaa/go-poppler.

This fork fixes a couple of memory leaks that the upstream libs included.In addition it usesfinalizers as a safeguard against memory leaks.This enables you not to calldocument.Close() andpage.Close() but let the GC do the clean-up work instead.

This might add a significant memory overhead in long-running high-throughput processes (like a web service or batch processor), when a lot ofPoppler objects in unmanaged/off-heap memory are being created.They are evicted by the finalizer but the GC might run too late to prevent a OOM as it doesn't take these into account when it schedules the next cycle.

One advantage of relying on finalizes solely might be an improved CPU utilization because they run on their own goroutine.So your main routine doesn't need to handle the clean-up. (But you can achieve this in many other ways,I guess, e.g. by writing your own goroutine and using channels.)

Tl;dr: Be careful when relying on finalizers; do some (load) tests and watch your RAM!