This project scrapes game logs and player data from nfl.com. Check out the get_all.py script for a demonstration of how to pull all points from 2001 to present. If you just want the data you can download a SQLite database containing those records here (75MB).
If you're interested in working with this data, also definitely check out the NFL's own API, which serves up this same data that powers nfl.com/stats, but only offering records starting in 2009 and semi-limited access to the public.
The program can produce two tables. The Game Logs table stores statistics for each player for each game played (yards, touchdowns, tackles, etc), along with metadata about the game (date, score, teams, etc.) for about 500,000 data points in 78 dimensions. A unique PlayerID stored with each entry can be used to identify the player referenced. The Players table is indexed by these PlayerIDs and contains records on the position, height, weight, name, etc. of each player.
The getter module contains the LinkGetter and PlayerGetter classes used to retrieve links to player pages and data from those pages respectively.
The formatter module contains the GameFormatter and PlayerFormatter classes for cleaning and storing raw game log and player profile data.