I've opened up nfo files and it looks like its filled with nothing but random characters.
nfo files as created for Kodi don't look the quite the same on the inside as the scene nfo files that come with a download. None of the art, release group crap, etc. Basically for the purpose of what we are talking about it is just a glorified xml file with media data.
When I get new content I don't use the nfo files that come with the media. My library managing software creates a new nfo file from scratch right then.
I guess I still don't see what this extra step buys you.
Without nfo files any time I want to update my library the scraper has to go through and check all the media files themselves (that have to be named perfect) and then gets the needed information off a website. With nfo files it simply scans though the file system for a single file type, finds those files, and then extracts the media information from them.
So if I ever need to rebuild my library in the future (which happens) without nfo files I have to go and rescrape EVERYTHING off the TVDB right then and there. With nfo files I already downloaded all that information locally ONCE when I added that file to the server so I can put it all together in a library in a fraction of the time.
Scraping, as a concept, is a hack. It is a piece of software that goes to a web page designed for human consumption and tries to copy that information off like a human would. That means if the website changes the formatting your scraper is broken because it is looking in the wrong place for the info as it lacks the intelligence a human has to look elsewhere in the page.
nfo files on the other hand are like having a xml settings file for a program. The media program will ALWAYS be able to extract the needed information from it because it will always be formatted in the same way.
You can't get around an initial scrape to get your information. But with nfo files you only have to do that process once- when its most convenient to do it- and not every time you want to redo your library.
It seems like all this does it make it a manual step when plex does it automatically, with a reasonably high degree of accuracy, assuming the name is correct to begin with.
None of it is manual, it is all automatic. I have a program that gets the media file, renames it to the proper naming convention, scrapes the media information to a nfo file, and then puts it in the proper place on my server. My setup basically puts the media all on a platter for Kodi. I do jack crap nowadays.
I mean, the Plex scraper isn't magic. Plex is a fork of XBMC, and from what I understand a lot of that scraper code is what XBMC had when the fork happened. The Plex scraper has the same limitation any scraper does- aka it wants to scrape directly to its library all at once. This sucks if you just got a lot of content recently and you want to update it all at once.
With nfo files basically the time spent scraping is shifted from when you tell the software to update the library to when the file is acquired. That is a huge advantage, and makes it the only solution out there that can handle a large library. All nfo files do is remove a bottleneck by putting a needed task elsewhere in the chain.
Another piece of this we aren't even mentioning is how the software not only creates an nfo file, but it also downloads locally all the fanart associated with the tv show. This is a huge benefit because honestly all that image downloading is what most often breaks scrapers trying to scrape a large library.