Graphing a Directory's File Statistics

graphing_directory_files_stats.py

import datetime
import math
import matplotlib.pyplot as plt # https://matplotlib.org
import mplcursors # https://mplcursors.readthedocs.io
import numpy as np # https://numpy.org/
import os
import sys

### usage notes
# Method to pull file statitics may depend on OS. 
# Tested on Mac OS. More info below.

### user input
CUSTOM_DIRECTORY = "Documents/code/notes"

### program
FILE_DIRECTORY = os.path.join(os.path.expanduser('~'),CUSTOM_DIRECTORY)
FILES = [x for x in os.listdir(FILE_DIRECTORY) if x.endswith(".md")]

EPOCH = datetime.datetime(1970,1,1)
NOW = int((datetime.datetime.utcnow()-EPOCH).total_seconds())

def days_between(first,second):
    return math.ceil((first-second)/86400)

files_stats = []
for file in FILES:
    file_path = os.path.join(FILE_DIRECTORY,file)
    
    # method to pull stats depends on computer OS, see
    # https://stackoverflow.com/a/39501288/15146924
    file_stats = os.stat(file_path)
    last_modified = int(file_stats.st_mtime)
    last_accessed = int(file_stats.st_atime)
    last_created = int(file_stats.st_birthtime)
    
    days_since_modified = days_between(NOW,last_modified)
    days_since_accessed = days_between(NOW,last_accessed)
    days_since_created = days_between(NOW,last_created)
    days_alive = days_between(last_accessed,last_created)
    days_worked = days_between(last_modified,last_created)

    files_stats.append([file, 
        last_modified, last_accessed, last_created,
        days_since_modified, days_since_accessed, days_since_created,
        days_alive, days_worked])

y = [i[2] for i in files_stats]     
x = [i[3] for i in files_stats]     
sizes = [i[1] for i in files_stats] 
colors = [i[2] for i in files_stats]

fig, ax = plt.subplots()

plt.ylabel("last_accessed")
plt.xlabel("created")

ax.scatter(x, y, c=colors)
ax.set(xlim=(min(x), max(x)), ylim=(min(y), max(y)))

cursor = mplcursors.cursor(ax, hover=True)
@cursor.connect("add")
def on_add(sel):
    sel.annotation.set(text=files_stats[sel.target.index][0][:-3])

plt.show()

Example described on Medium.