Stats from .zsh_history

zsh-stats.py

import time
import re

hist = []
commands = []
count = 0
trueCount = 0
# by year
yearCounts = {}
monthCounts = {}
weekCounts = {}
dayCounts = {}
# total over all years
weekdayCounts = {}
hourCounts = {}

commandCounts = {}
longestCommand = ""

with open('.zsh_history', 'r', encoding="ISO-8859-1") as histFile:
	hist = histFile.readlines()
	histFile.close()

print("Analyzing...")
for line in hist:
	count += 1
	if line.find(": ") == 0:
		trueCount += 1
		epochTime = line[2:12]
		command = line[15:(len(line) - 1)]
		commands.append({
			"time": epochTime,
			"command": command
		})

		# time stats
		timeStruct = time.localtime(int(epochTime))
		year = time.strftime("%Y", timeStruct)
		month = time.strftime("%b %Y", timeStruct)
		week = time.strftime("%W-%Y", timeStruct)
		day = time.strftime("%Y-%m-%d", timeStruct)
		weekday = time.strftime("%a", timeStruct)
		hour = time.strftime("%H", timeStruct)

		if year not in yearCounts:
			newDict = {year: 0}
			yearCounts.update(newDict)
		yearCounts[year] += 1

		if month not in monthCounts:
			newDict = {month: 0}
			monthCounts.update(newDict)
		monthCounts[month] += 1

		if week not in weekCounts:
			newDict = {week: 0}
			weekCounts.update(newDict)
		weekCounts[week] += 1

		if day not in dayCounts:
			newDict = {day: 0}
			dayCounts.update(newDict)
		dayCounts[day] += 1

		if weekday not in weekdayCounts:
			newDict = {weekday: 0}
			weekdayCounts.update(newDict)
		weekdayCounts[weekday] += 1

		if hour not in hourCounts:
			newDict = {hour: 0}
			hourCounts.update(newDict)
		hourCounts[hour] += 1

		# command stats
		if len(command) > len(longestCommand):
			longestCommand = command
		# match with "(^[A-Za-z0-9./_\-+]+)|((?<=sudo )[A-Za-z0-9./_\-+]+)|((?<=\| )[A-Za-z0-9./_\-+]+)|((?<=&& )[A-Za-z0-9./_\-+]+)"gm
		# - command at the beginning of a line
		# - command after a '|'
		# - command after a '&&'
		# - after sudo (and including sudo)
		# it's not perfect, and misses a few commands + gets false positives

		# findall doesn't really work in a friendly way, so do all groups seperately
		commandsInLine = re.findall("(^[A-Za-z0-9./_\-+]+)", command)
		for comm in commandsInLine:
			if comm not in commandCounts and comm != '':
				newDict = {comm: 1}
				commandCounts.update(newDict)
			elif comm != '':
				commandCounts[comm] += 1
		commandsInLine = re.findall("((?<=sudo )[A-Za-z0-9./_\-+]+)", command)
		for comm in commandsInLine:
			if comm not in commandCounts and comm != '':
				newDict = {comm: 1}
				commandCounts.update(newDict)
			elif comm != '':
				commandCounts[comm] += 1
		commandsInLine = re.findall("((?<=\| )[A-Za-z0-9./_\-+]+)", command)
		for comm in commandsInLine:
			if comm not in commandCounts and comm != '':
				newDict = {comm: 1}
				commandCounts.update(newDict)
			elif comm != '':
				commandCounts[comm] += 1
		commandsInLine = re.findall("((?<=&& )[A-Za-z0-9./_\-+]+)", command)
		for comm in commandsInLine:
			if comm not in commandCounts and comm != '':
				newDict = {comm: 1}
				commandCounts.update(newDict)
			elif comm != '':
				commandCounts[comm] += 1

print(f"Total commands: {trueCount}")
print(f"Longest command: {longestCommand}")

print("\nBy year:")
for year in yearCounts:
	print(f"\t{year}\t{yearCounts[year]}")
print("\nBy month:")
for month in monthCounts:
	print(f"\t{month}\t{monthCounts[month]}")
print("\nBy week:")
for week in weekCounts:
	print(f"\t{week}\t{weekCounts[week]}")
print("\nBy day:")
for day in dayCounts:
	print(f"\t{day}\t{dayCounts[day]}")
print("\nBy weekday:")
for weekday in weekdayCounts:
	print(f"\t{weekday}\t{weekdayCounts[weekday]}")
print("\nBy hour:")
for hour in hourCounts:
	print(f"\t{hour}:00\t{hourCounts[hour]}")

print("\nCommand stats:")
for comm in commandCounts:
	print(f"\t{comm}\t{commandCounts[comm]}")