import pandas as pd
import numpy as np
# import matplotlib.pyplot as plt

timelog = pd.read_csv("timelog-titled.csv")

timelog["started"] = pd.to_datetime(timelog["started"]).dt.tz_convert("US/Eastern")
timelog["recorded"] = pd.to_datetime(timelog["recorded"]).dt.tz_convert("US/Eastern")
timelog["time"] = 30
# A pomodoro started before 3am Eastern time is considered to be a continuation
# of the day before, so we are, effectively, on West Coast time for determining
# the day we want to associate a time entry with.  PomodoroPrompt saves as UTC.
timelog["date"] = timelog["started"].dt.tz_convert("US/Pacific").dt.date
timelog["day_of_week"] = pd.to_datetime(timelog["date"]).dt.day_name()

# If a project has been specified (task prefixed with a colon), then put the
# project in its own column.
timelog['project'] = (np.where(timelog['description'].str.contains(': '), timelog['description'].str.split(': ', 1).str[0], None))
timelog['description'] = (np.where(timelog['description'].str.contains(': '), timelog['description'].str.split(': ', 1).str[1], timelog['description']))

# If a multiplier has been provided (an asterisk and an integer at the end of a
# task), then multiply the time by it and remove it from the description.
# Note that contains is regex by default, so we actually have to specify pattern
# and regex false to check if it contains an asterisk!
timelog['tmp_multiplier'] = (np.where(timelog['description'].str.contains(pat='*', regex=False), timelog['description'].str.split('*', 1).str[1], None))
timelog['description'] = (np.where(timelog['description'].str.contains(pat='*', regex=False), timelog['description'].str.split('*', 1).str[0], timelog['description']))

# Condense duplicate entries by date, summing the minutes spent, and listing
# the first started and last recorded times for each task.
tl = timelog.groupby(["date", "project", "description"]).agg({"time": 'sum', "started": 'min', "recorded": 'max'}).reset_index()