"""SA data functions"""

import os
import gzip
import re
from datetime import datetime, timedelta


class SaUsage:
    """Holds a record from sadatarunner.sh or sa -cm as an object
    with attributes for each column

    Args:
        line (str): line from sa -cm to parse
    Hint:
        Use vars() on this to get a dict of the columns
    Attributes:
        user (str): username
        procs (int): num of processes executed
        perc_procs (float): percent of total processes
        real (float): real execution time
        perc_real (float): percent of real execution time from total
        cpu (float): cpu time in cp
        avg_iops (int): average iops per process
        iops (int): total iops (units of 1000)
    """

    __module__ = 'rads'

    sa_re = re.compile(
        r'(?P<user>[^\s]+)\s+'
        r'(?P<procs>\d+)\s+'
        r'(?P<perc_procs>\d+\.?\d*)%\s+'
        r'(?P<real>\d+\.?\d*)re\s+'
        r'(?P<perc_real>\d+\.?\d*)%\s+'
        r'(?P<cpu>\d+\.?\d*)cp\s+'
        r'(?P<perc_cp>\d+\.?\d*)%\s+'
        r'(?P<avg_iops>\d+)avio\s+'
        r'(?P<iops>\d+)k\s*$'
    )

    def __init__(self, line: str):
        match = SaUsage.sa_re.match(line)
        if not match:
            raise ValueError('Invalid sa line')  # could be the header line
        gdict = match.groupdict()
        self.user = gdict['user']
        self.procs = int(gdict['procs'])
        self.perc_procs = float(gdict['perc_procs'])
        self.real = float(gdict['real'])
        self.perc_real = float(gdict['perc_real'])
        self.cpu = float(gdict['cpu'])
        self.perc_cp = float(gdict['perc_cp'])
        self.avg_iops = int(gdict['avg_iops'])
        self.iops = int(gdict['iops'])


def get_cp_usage(offset: str = 0) -> dict[str, SaUsage]:
    """Returns archived usage data collected by RADS in sadatarunner.sh

    Args:
        offset: number of days ago (default of 0 means today)
    Returns:
        mapping of usernames to ``SaUsage`` objects
    """
    offset += 1
    # get the time offset days ago
    # month needs to be the short string representation of the month (i.e. Jun)
    when = datetime.now() - timedelta(days=offset)
    month, day = when.strftime('%b %d').split()
    sadatadir = os.path.join('/var/log/sa/rads', month, day)
    try:
        histfiles = [
            os.path.join(sadatadir, x)
            for x in os.listdir(sadatadir)  # may raise FileNotFoundError
            if x.endswith('-avg.gz')
        ]
    except FileNotFoundError:
        return {}  # nothing recorded for this day
    try:
        # get the latest
        histfile = sorted(histfiles, key=os.path.getmtime, reverse=True)[0]
    except IndexError:
        return {}  # sadatadir contained no items
    try:
        with gzip.open(histfile, 'rt', encoding='ascii') as filehandle:
            contents = filehandle.read().splitlines()
    except OSError:
        return {}  # invalid file?
    ret = {}
    for line in contents:
        try:
            usage = SaUsage(line)
        except ValueError:
            continue
        ret[usage.user] = usage
    return ret


get_cp_usage.__module__ = 'rads'