usno-sunrise-table-parser.py

#!/usr/bin/env python3
"""Parse sunrise/sunset tables as provided by the USNO website at https://aa.usno.navy.mil/data/RS_OneYear"""

import re
import argparse


def parse_table(filename: str):
    def to_decimal_deg(latlon_tup) -> float:
        assert len(latlon_tup) == 3 and re.match("[NESW]", latlon_tup[0])
        deg = int(latlon_tup[1]) + int(latlon_tup[2]) / 60.0
        if latlon_tup[0] in ("S", "W"):
            return -deg
        return deg

    def time_colon(timestr: str) -> str:
        if timestr.isdigit():
            return timestr[:2] + ":" + timestr[2:]
        return timestr

    months = [{} for i in range(12)]
    year, location = None, None
    with open(filename, encoding="utf8") as f:
        for line in f:
            if line.startswith("Location: "):
                m = re.search(r"\s+([EW])(\d+)\s+(\d+)[\s,]+([NS])(\d+)\s+(\d+)", line)
                assert m, "can't parse location as expected"
                lon, lat = m.groups()[:3], m.groups()[3:]
                location = (to_decimal_deg(lat), to_decimal_deg(lon))

                m = re.search("Rise and Set for the Sun for (\d+)", line)
                assert m, "can't parse year as expected"
                year = int(m.group(1))
            elif re.match(r"^\d\d ", line):
                # this is a fixed tabular format with occasional empty "cells", so we'll just use indexes
                day = int(line[:2])
                for month, slot in enumerate(range(4, 135, 11)):
                    sunrise, sunset = (
                        line[slot : slot + 4].strip(),
                        line[slot + 5 : slot + 5 + 4].strip(),
                    )
                    if (sunrise and sunset):  # ignore nonexistent or weird days with only one value
                        assert len(sunrise) == len(sunset) == 4, "unexpected format"
                        months[month][day]=(time_colon(sunrise), time_colon(sunset))

    assert year and location
    return location, year, months


if __name__ == "__main__":

    location, year, months = parse_table("xyz.txt")

    for month, days in enumerate(months, start=1):
        for day, values in days.items():
            print(year, month, day, values)