hebdobot/review/stats.py

from dataclasses import dataclass
from datetime import datetime
import re


@dataclass
class ReviewData:
    date: datetime
    user_count: int
    duration: int


class Board:
    def __init__(self, datas):
        self.datas = {}
        self.init(datas)

    def init(self, datas):
        pass

    def __repr__(self):
        return " ".join([f"{key} ({value})" for key, value in self.datas.items()])

    def __getitem__(self, key):
        if key in self.datas:
            return self.datas[key]
        return 0

    def position(self, key):
        if key in self.datas:
            return sorted(
                self.datas, key=lambda x: self.datas[x]
            ).index(key) + 1

    @property
    def min(self):
        if self.datas == []:
            return 0
        return min(self.datas.values())

    @property
    def max(self):
        if self.datas == []:
            return 0
        return max(self.datas.values())

    @property
    def avg(self):
        if self.datas == []:
            return 0
        return self.sum / len(self.datas)

    @property
    def sum(self):
        if self.datas == []:
            return 0
        return sum(self.datas.values())


class UserBoard(Board):
    def init(self, datas):
        for data in datas:
            if data.user_count not in self.datas:
                self.datas[data.user_count] = 0
            self.datas[data.user_count] += 1


class DurationBoard(Board):
    def init(self, datas):
        for data in datas:
            if data.duration not in self.datas:
                self.datas[data.duration] = 0
            self.datas[data.duration] += 1


class ReviewStats:
    def __init__(self, filepath):
        self.filepath = filepath
        self.datas = []
        self.users_board = None
        self.durations_board = None

    @property
    def size(self):
        return len(self.datas)

    @property
    def first(self):
        return sorted(self.datas, key=lambda data: data.date)[0] if self.datas else None

    @property
    def biggest(self):
        return (
            sorted(self.datas, key=lambda data: data.user_count)[-1]
            if self.datas
            else None
        )

    @property
    def min_users(self):
        if self.datas == []:
            return 0
        return min(self.datas, key=lambda data: data.user_count).user_count

    @property
    def max_users(self):
        if self.datas == []:
            return 0
        return max(self.datas, key=lambda data: data.user_count).user_count

    @property
    def avg_users(self):
        if self.datas == []:
            return 0
        return sum([data.user_count for data in self.datas]) / self.size

    @property
    def min_duration(self):
        if self.datas == []:
            return 0
        datas = [data for data in self.datas if data.duration is not None]
        return min(datas, key=lambda data: data.duration).duration

    @property
    def max_duration(self):
        if self.datas == []:
            return 0
        datas = [data for data in self.datas if data.duration is not None]
        return max(datas, key=lambda data: data.duration).duration

    @property
    def avg_duration(self):
        if self.datas == []:
            return 0
        datas = [data for data in self.datas if data.duration is not None]
        return sum([data.duration for data in datas]) / len(datas)

    def year_review(self, year):
        return len([data for data in self.datas if data.date.year == year])

    def load(self):
        self.datas = []
        try:
            with open(self.filepath) as file_handle:
                lines = file_handle.read().splitlines()
                for line in lines:
                    if line.strip() != "":
                        datas = re.split(r"\s+", line)
                        if len(datas) == 2:
                            self.datas.append(
                                ReviewData(
                                    datetime.strptime(datas[0], "%Y%m%d-%Hh%M"),
                                    int(datas[1]),
                                    None,
                                )
                            )
                        else:
                            self.datas.append(
                                ReviewData(
                                    datetime.strptime(datas[0], "%Y%m%d-%Hh%M"),
                                    int(datas[1]),
                                    int(datas[2]),
                                )
                            )
        except FileNotFoundError:
            # no file, no stats
            pass
        self.users_board = UserBoard(self.datas)
        self.durations_board = DurationBoard(self.datas)

    def save(self):
        with open(self.filepath, "wt") as file_handle:
            for data in self.datas:
                file_handle.write(
                    f"{data.date.strftime('%Y%m%d-%Hh%M')}\t"
                    f"{data.user_count}\t{data.duration}\n"
                )