sun_position/geo.py

   1 #!/usr/bin/env python
   2 # SPDX-FileCopyrightText: 2010 `Maximilian Hoegner <hp.maxi@hoegners.de>`
   3 #
   4 # SPDX-License-Identifier: GPL-3.0-or-later
   5
   6 # geo.py is a python module with no dependencies on extra packages,
   7 # providing some convenience functions for working with geographic
   8 # coordinates
   9
  10 ### Part one - Functions for dealing with points on a sphere ###
  11
  12 ### Part two - A tolerant parser for position strings ###
  13 import re
  14
  15
  16 class Parser:
  17     """ A parser class using regular expressions. """
  18
  19     def __init__(self):
  20         self.patterns = {}
  21         self.raw_patterns = {}
  22         self.virtual = {}
  23
  24     def add(self, name, pattern, virtual=False):
  25         """ Adds a new named pattern (regular expression) that can reference previously added patterns by %(pattern_name)s.
  26         Virtual patterns can be used to make expressions more compact but don't show up in the parse tree. """
  27         self.raw_patterns[name] = "(?:" + pattern + ")"
  28         self.virtual[name] = virtual
  29
  30         try:
  31             self.patterns[name] = ("(?:" + pattern + ")") % self.patterns
  32         except KeyError as e:
  33             raise (Exception, "Unknown pattern name: %s" % str(e))
  34
  35     def parse(self, pattern_name, text):
  36         """ Parses 'text' with pattern 'pattern_name' and returns parse tree """
  37
  38         # build pattern with subgroups
  39         sub_dict = {}
  40         subpattern_names = []
  41         for s in re.finditer(r"%\(.*?\)s", self.raw_patterns[pattern_name]):
  42             subpattern_name = s.group()[2:-2]
  43             if not self.virtual[subpattern_name]:
  44                 sub_dict[subpattern_name] = "(" + self.patterns[
  45                     subpattern_name] + ")"
  46                 subpattern_names.append(subpattern_name)
  47             else:
  48                 sub_dict[subpattern_name] = self.patterns[subpattern_name]
  49
  50         pattern = "^" + (self.raw_patterns[pattern_name] % sub_dict) + "$"
  51
  52         # do matching
  53         m = re.match(pattern, text)
  54
  55         if m is None:
  56             return None
  57
  58         # build tree recursively by parsing subgroups
  59         tree = {"TEXT": text}
  60
  61         for i in range(len(subpattern_names)):
  62             text_part = m.group(i + 1)
  63             if text_part is not None:
  64                 subpattern = subpattern_names[i]
  65                 tree[subpattern] = self.parse(subpattern, text_part)
  66
  67         return tree
  68
  69
  70 position_parser = Parser()
  71 position_parser.add("direction_ns", r"[NSns]")
  72 position_parser.add("direction_ew", r"[EOWeow]")
  73 position_parser.add("decimal_separator", r"[\.,]", True)
  74 position_parser.add("sign", r"[+-]")
  75
  76 position_parser.add("nmea_style_degrees", r"[0-9]{2,}")
  77 position_parser.add("nmea_style_minutes",
  78                     r"[0-9]{2}(?:%(decimal_separator)s[0-9]*)?")
  79 position_parser.add(
  80     "nmea_style", r"%(sign)s?\s*%(nmea_style_degrees)s%(nmea_style_minutes)s")
  81
  82 position_parser.add(
  83     "number",
  84     r"[0-9]+(?:%(decimal_separator)s[0-9]*)?|%(decimal_separator)s[0-9]+")
  85
  86 position_parser.add("plain_degrees", r"(?:%(sign)s\s*)?%(number)s")
  87
  88 position_parser.add("degree_symbol", r"°", True)
  89 position_parser.add("minutes_symbol", r"'|′|`|´", True)
  90 position_parser.add("seconds_symbol",
  91                     r"%(minutes_symbol)s%(minutes_symbol)s|″|\"",
  92                     True)
  93 position_parser.add("degrees", r"%(number)s\s*%(degree_symbol)s")
  94 position_parser.add("minutes", r"%(number)s\s*%(minutes_symbol)s")
  95 position_parser.add("seconds", r"%(number)s\s*%(seconds_symbol)s")
  96 position_parser.add(
  97     "degree_coordinates",
  98     r"(?:%(sign)s\s*)?%(degrees)s(?:[+\s]*%(minutes)s)?(?:[+\s]*%(seconds)s)?|(?:%(sign)s\s*)%(minutes)s(?:[+\s]*%(seconds)s)?|(?:%(sign)s\s*)%(seconds)s"
  99 )
 100
 101 position_parser.add(
 102     "coordinates_ns",
 103     r"%(nmea_style)s|%(plain_degrees)s|%(degree_coordinates)s")
 104 position_parser.add(
 105     "coordinates_ew",
 106     r"%(nmea_style)s|%(plain_degrees)s|%(degree_coordinates)s")
 107
 108 position_parser.add(
 109     "position", (
 110         r"\s*%(direction_ns)s\s*%(coordinates_ns)s[,;\s]*%(direction_ew)s\s*%(coordinates_ew)s\s*|"
 111         r"\s*%(direction_ew)s\s*%(coordinates_ew)s[,;\s]*%(direction_ns)s\s*%(coordinates_ns)s\s*|"
 112         r"\s*%(coordinates_ns)s\s*%(direction_ns)s[,;\s]*%(coordinates_ew)s\s*%(direction_ew)s\s*|"
 113         r"\s*%(coordinates_ew)s\s*%(direction_ew)s[,;\s]*%(coordinates_ns)s\s*%(direction_ns)s\s*|"
 114         r"\s*%(coordinates_ns)s[,;\s]+%(coordinates_ew)s\s*"
 115     ))
 116
 117
 118 def get_number(b):
 119     """ Takes appropriate branch of parse tree and returns float. """
 120     s = b["TEXT"].replace(",", ".")
 121     return float(s)
 122
 123
 124 def get_coordinate(b):
 125     """ Takes appropriate branch of the parse tree and returns degrees as a float. """
 126
 127     r = 0.
 128
 129     if b.get("nmea_style"):
 130         if b["nmea_style"].get("nmea_style_degrees"):
 131             r += get_number(b["nmea_style"]["nmea_style_degrees"])
 132         if b["nmea_style"].get("nmea_style_minutes"):
 133             r += get_number(b["nmea_style"]["nmea_style_minutes"]) / 60.
 134         if b["nmea_style"].get(
 135                 "sign") and b["nmea_style"]["sign"]["TEXT"] == "-":
 136             r *= -1.
 137     elif b.get("plain_degrees"):
 138         r += get_number(b["plain_degrees"]["number"])
 139         if b["plain_degrees"].get(
 140                 "sign") and b["plain_degrees"]["sign"]["TEXT"] == "-":
 141             r *= -1.
 142     elif b.get("degree_coordinates"):
 143         if b["degree_coordinates"].get("degrees"):
 144             r += get_number(b["degree_coordinates"]["degrees"]["number"])
 145         if b["degree_coordinates"].get("minutes"):
 146             r += get_number(b["degree_coordinates"]["minutes"]["number"]) / 60.
 147         if b["degree_coordinates"].get("seconds"):
 148             r += get_number(
 149                 b["degree_coordinates"]["seconds"]["number"]) / 3600.
 150         if b["degree_coordinates"].get(
 151                 "sign") and b["degree_coordinates"]["sign"]["TEXT"] == "-":
 152             r *= -1.
 153
 154     return r
 155
 156
 157 def parse_position(s):
 158     """ Takes a (utf8-encoded) string describing a position and returns a tuple of floats for latitude and longitude in degrees.
 159     Tries to be as tolerant as possible with input. Returns None if parsing doesn't succeed. """
 160
 161     parse_tree = position_parser.parse("position", s)
 162     if parse_tree is None:
 163         return None
 164
 165     lat_sign = +1.
 166     if parse_tree.get(
 167             "direction_ns") and parse_tree["direction_ns"]["TEXT"] in ("S",
 168                                                                        "s"):
 169         lat_sign = -1.
 170
 171     lon_sign = +1.
 172     if parse_tree.get(
 173             "direction_ew") and parse_tree["direction_ew"]["TEXT"] in ("W",
 174                                                                        "w"):
 175         lon_sign = -1.
 176
 177     lat = lat_sign * get_coordinate(parse_tree["coordinates_ns"])
 178     lon = lon_sign * get_coordinate(parse_tree["coordinates_ew"])
 179
 180     return lat, lon