Three word location

From Rosetta Code
Revision as of 21:17, 6 February 2022 by Jjuanhdez (talk | contribs) (Three word location en FreeBASIC)
Three word location is a draft programming task. It is not yet considered ready to be promoted as a complete task, for reasons that should be found in its talk page.

If one were to enter the words: 'softly affiliate activation' at the what3words.com site, the response would be a location in Walt Disney World. The latitude and longitude for that spot is 28.3852 -81.5638. Using that service enables anyone to specify any place on the Earth with three words.

Task: Provide a similar solution - Display a location on the Earth with three words derived from a latitude longitude pair.

For: latitude = 28.3852 longitude = -81.5638 Display: W18497 W11324 W01322

Implementation:

Build an array of 28126, 6 character words in the form W00000 thru W28125.

Convert latitude and longitude into positive integers.

Build a 43 bit integer containing latitude (21 bits) and longitude (22 bits).

Isolate most significant 15 bits for word 1 index.

Isolate next 14 bits for word 2 index.

Isolate next 14 bits for word 3 index.

Fetch each word from the word array.

Display the words.

Reverse the procedure and display the original latitude and longitude.

Extra credit: Find a way to populate the word array with actual words.


AppleScript

Index words

When the words are index-based as in the task description, it's not necessary to generate all 28126. <lang applescript>on locationToWords({latitude, longitude})

   -- "Convert" the coordinates to positive integers by adding enough degrees to ensure positive results,
   -- multiplying by enough to left shift by four decimal places, and rounding.
   set intLat to ((latitude + 90) * 10000) as integer
   set intLong to ((longitude + 180) * 10000) as integer
   -- Derive a 15-bit and two 14-bit values from the two results' 43 bits.
   set output to {intLat div 64, intLat mod 64 * 256 + intLong div 16384, intLong mod 16384}
   -- Coerce the three values to text "words" beginning with "W" and any necessary leading zeros.
   repeat with thisIndex in output
       set thisIndex's contents to "W" & text 2 thru 6 of ((100000 + thisIndex) as text)
   end repeat
   
   return output

end locationToWords

on wordsToLocation(threeWords)

   set indices to {}
   repeat with thisWord in threeWords
       set end of indices to (text 2 thru -1 of thisWord) as integer
   end repeat
   set intLat to (beginning of indices) * 64 + (item 2 of indices) div 256 mod 64
   set intLong to (item 2 of indices) mod 256 * 16384 + (end of indices)
   
   return {intLat / 10000 - 90, intLong / 10000 - 180}

end wordsToLocation

-- Task code: local location, threeWords, checkLocation set location to {28.3852, -81.5638} set threeWords to locationToWords(location) set checkLocation to wordsToLocation(threeWords) return {location, threeWords, checkLocation}</lang>

Output:

<lang applescript>{{28.3852, -81.5638}, {"W18497", "W11324", "W01322"}, {28.3852, -81.5638}}</lang>

Actual words

<lang applescript>on locationToWords({latitude, longitude}, listOfWords)

   script o
       property wordList : listOfWords
   end script
   
   set intLat to ((latitude + 90) * 10000) as integer
   set intLong to ((longitude + 180) * 10000) as integer
   set output to {intLat div 64, intLat mod 64 * 256 + intLong div 16384, intLong mod 16384}
   repeat with thisIndex in output
       set thisIndex's contents to item (thisIndex + 1) of o's wordList -- AppleScript indices are 1-based.
   end repeat
   
   return output

end locationToWords

on wordsToLocation(threeWords, listOfWords)

   script o
       property wordList : listOfWords
   end script
   
   set indices to {}
   repeat with thisWord in threeWords
       set thisWord to thisWord's contents
       set i to 1
       repeat until (item i of o's wordList is thisWord)
           set i to i + 1
           if (i > 28126) then error "wordsToLocation() handler: The word “" & thisWord & "” isn't in the word list."
       end repeat
       set end of indices to i - 1 -- Converted to 0-based index.
   end repeat
   set intLat to (beginning of indices) * 64 + (item 2 of indices) div 256 mod 64
   set intLong to (item 2 of indices) mod 256 * 16384 + (end of indices)
   
   return {intLat / 10000 - 90, intLong / 10000 - 180}

end wordsToLocation

-- Task code: local o, location, threeWords, checkLocation -- Use the words in unixdict.txt. It only has 25110 of them by AppleScript's count, -- so make up the shortfall with invented plurals and third-persons-singular. script

   property wordList : words of (read file ((path to desktop as text) & "unixdict.txt") as «class utf8»)
   property additionalWords : {}

end script set o to result repeat with i from 1 to (28126 - (count o's wordList))

   tell item i of o's wordList
       if (it ends with "s") then
           set end of o's additionalWords to it & "es"
       else
           set end of o's additionalWords to it & "s"
       end if
   end tell

end repeat set o's wordList to o's wordList & o's additionalWords

set location to {28.3852, -81.5638} set threeWords to locationToWords(location, o's wordList) set checkLocation to wordsToLocation(threeWords, o's wordList) return {location, threeWords, checkLocation}</lang>

Output:

<lang applescript>{{28.3852, -81.5638}, {"quote", "hygiene", "aristotelean"}, {28.3852, -81.5638}}</lang>

AutoHotkey

Conversion based on Wren
WordList From link suggested by Symsyn <lang AutoHotkey>URLDownloadToFile, http://www-personal.umich.edu/~jlawler/wordlist, % A_Temp "\wordlist.txt" FileRead, wordList, % A_Temp "\wordlist.txt"

LL := [28.3852, -81.5638] num := LL2num(LL) words := LL2words(wordList, LL) LL2 := words2LL(wordList, words)

MsgBox, 262144, , % result := " ( LL = " LL.1 ", " LL.2 " LL2num = " num.1 ", " num.2 ", " num.3 " LL2words = " words.1 ", " words.2 ", " words.3 " words2LL = " LL2.1 ", " LL2.2 " )" return

-----------------------------------------------

LL2words(wordList, LL){  ; Latitude/Longitude to 3 words

   num := LL2num(LL)
   wli := wordList(wordList).1
   return [wli[num.1], wli[num.2], wli[num.3]]

}

-----------------------------------------------

words2LL(wordList, w){  ; 3 words to Latitude/Longitude

   iow := wordList(wordList).2
   LL := num2LL([iow[w.1], iow[w.2], iow[w.3]])
   return [ll.1, ll.2]

}

-----------------------------------------------

wordList(wordList){  ; word list to two arrays

   wli:=[], iow:=[]    ; word list index, index of word
   for i, word in StrSplit(wordList, "`n", "`r")
       if (word ~= "^[a-z]+$") && (StrLen(word) <= 8) && (StrLen(word) > 3)
           wli.Push(word), iow[word] := wli.MaxIndex()
   return [wli, iow]

}

-----------------------------------------------

LL2num(LL){  ; Latitude/Longitude to 3 numbers

   ilat := LL.1*10000 + 900000
   ilon := LL.2*10000 + 1800000
   latlon := (ilat << 22) + ilon
   return [(latlon >> 28) & 0x7fff, (latlon >> 14) & 0x3fff, latlon & 0x3fff]

}

-----------------------------------------------

num2LL(w){  ; 3 numbers to Latitude/Longitude

   latlon := (w.1 << 28) | (w.2 << 14) | w.3
   ilat := latlon >> 22
   ilon := latlon & 0x3fffff
   return [(ilat-900000)/10000, (ilon-1800000)/10000]

}

-----------------------------------------------</lang>
Output:
LL = 28.3852, -81.5638
LL2num = 18497, 11324, 1322
LL2words = malleus, fasten, analytic
words2LL = 28.385200, -81.563800

C

Translation of: Go

<lang c>#include <stdio.h>

  1. include <stdlib.h>

typedef long long int64;

void to_word(char *ws, int64 w) {

   sprintf(ws, "W%05lld", w);

}

int64 from_word(char *ws) {

   return atoi(++ws);

}

int main() {

   double lat, lon;
   int64 latlon, ilat, ilon, w1, w2, w3;
   char w1s[7], w2s[7], w3s[7];
   printf("Starting figures:\n");
   lat = 28.3852;
   lon = -81.5638;
   printf("  latitude = %0.4f, longitude = %0.4f\n", lat, lon);

   // convert lat and lon to positive integers
   ilat = (int64)(lat*10000 + 900000);
   ilon = (int64)(lon*10000 + 1800000);

   // build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon)
   latlon = (ilat << 22) + ilon;
   // isolate relevant bits
   w1 = (latlon >> 28) & 0x7fff;
   w2 = (latlon >> 14) & 0x3fff;
   w3 = latlon & 0x3fff;
   // convert to word format
   to_word(w1s, w1);
   to_word(w2s, w2);
   to_word(w3s, w3);

   // and print the results
   printf("\nThree word location is:\n");
   printf("  %s %s %s\n", w1s, w2s, w3s);
   /* now reverse the procedure */
   w1 = from_word(w1s);
   w2 = from_word(w2s);
   w3 = from_word(w3s);
   latlon = (w1 << 28) | (w2 << 14) | w3;
   ilat = latlon >> 22;
   ilon = latlon & 0x3fffff;
   lat = (double)(ilat-900000) / 10000;
   lon = (double)(ilon-1800000) / 10000;
   // and print the results
   printf("\nAfter reversing the procedure:\n");
   printf("  latitude = %0.4f, longitude = %0.4f\n", lat, lon);
   return 0;

}</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Delphi

Translation of: Go

<lang Delphi> program Three_word_location;

{$APPTYPE CONSOLE}

{$R *.res}

uses

 System.SysUtils;

type

 TThreeWordLocation = array of string;
 TGlobalPosition = record
 private
   FLatitude: Double;
   FLongitude: Double;
   FWs: TThreeWordLocation;
   class function toWord(w: int64): string; static;
   class function fromWord(ws: string): int64; static;
   procedure SetLatitude(const Value: Double);
   procedure SetLongitude(const Value: Double);
   procedure Recalculate;
   function GetTWLocationAsStr: string;
 public
   constructor Create(_lat, _lon: Double); overload;
   constructor Create(Ws: TThreeWordLocation); overload;
   procedure Assign(Ws: TThreeWordLocation);
   property Latitude: Double read FLatitude write SetLatitude;
   property Longitude: Double read FLongitude write SetLongitude;
   property TWLocation: TThreeWordLocation read FWs;
   property TWLocationAsStr: string read GetTWLocationAsStr;
 end;

{ TGlobalPosition }

constructor TGlobalPosition.Create(_lat, _lon: Double); begin

 FLatitude := _lat;
 FLongitude := _lon;
 Recalculate;

end;

constructor TGlobalPosition.Create(ws: TThreeWordLocation); begin

 Assign(ws);

end;

class function TGlobalPosition.fromWord(ws: string): int64; begin

 Result := StrToInt(ws.Substring(1));

end;

function TGlobalPosition.GetTWLocationAsStr: string; var

 i: Integer;

begin

 Result := ;
 for i := 0 to 2 do
   Result := Result + ' ' + FWs[i];
 Result := Result.Trim;

end;

procedure TGlobalPosition.Recalculate; var

 i: Integer;
 w: array[0..2] of int64;
 ilat, ilon, latlon: Int64;

begin

 SetLength(FWs, 3);
 // convert lat and lon to positive integers
 ilat := Round(FLatitude * 10000 + 900000);
 ilon := Round(FLongitude * 10000 + 1800000);
  // build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon)
 latlon := (ilat shl 22) + ilon;
 // isolate relevant bits
 w[0] := (latlon shr 28) and $7fff;
 w[1] := (latlon shr 14) and $3fff;
 w[2] := latlon and $3fff;
 // convert to word format
 for i := 0 to 2 do
   FWs[i] := toWord(w[i]);

end;

procedure TGlobalPosition.SetLatitude(const Value: Double); begin

 FLatitude := Value;
 Recalculate;

end;

procedure TGlobalPosition.SetLongitude(const Value: Double); begin

 FLongitude := Value;
 Recalculate;

end;

class function TGlobalPosition.toWord(w: int64): string; begin

 result := format('W%.5d', [w]);

end;

procedure TGlobalPosition.Assign(Ws: TThreeWordLocation); var

 i: Integer;
 w: array[0..2] of int64;
 ilat, ilon, latlon: Int64;

begin

 SetLength(FWs, 3);
 for i := 0 to 2 do
 begin
   FWs[i] := Ws[i];
   w[i] := fromWord(Ws[i]);
 end;
 latlon := (w[0] shl 28) or (w[1] shl 14) or w[2];
 ilat := latlon shr 22;
 ilon := latlon and $3fffff;
 FLatitude := (ilat - 900000) / 10000;
 FLongitude := (ilon - 1800000) / 10000;

end;

var

 pos: TGlobalPosition;

begin

 pos.Create(28.3852, -81.5638);
 Writeln('Starting figures:');
 Writeln(Format('  latitude = %0.4f, longitude = %0.4f', [pos.Latitude, pos.Longitude]));
 Writeln(#10'Three word location is:');
 Writeln('  ', pos.TWLocationAsStr);
 Writeln(#10'After reversing the procedure:');
 // pos.Create(['W18497','W11324','W01322']);
 pos.Create(pos.TWLocation);
 Writeln(Format('  latitude = %0.4f, longitude = %0.4f', [pos.Latitude, pos.Longitude]));
 Readln;

end.

</lang>

Output:
Starting figures:
  latitude = 28,3852, longitude = -81,5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28,3852, longitude = -81,5638


FreeBASIC

Translation of: Nim

<lang freebasic>Print "Starting figures:" Dim As Double lat = 28.3852, longi = -81.5638 Print Using " latitude = &, longitude = &"; lat; longi

' Convert "lat" and "long" to positive integers. Dim As Integer ilat = Cint(lat * 10000 + 900000) Dim As Integer ilong = Cint(longi * 10000 + 1800000)

' Build 43 bit int comprising 21 bits (lat) and 22 bits (lon). Dim As Double latlong = ilat Shl 22 + ilong

' Isolate relevant bits. Dim As Integer w1 = latlong Shr 28 And &H7fff Dim As Integer w2 = latlong Shr 14 And &H3fff Dim As Integer w3 = latlong And &H3fff

' Convert to word format. Dim As String*5 w1s = String(5, "0"), w2s = String(5, "0"), w3s = String(5, "0") Mid(w1s, 6-Len(Str(w1))) = Str(w1) Mid(w2s, 6-Len(Str(w2))) = Str(w2) Mid(w3s, 6-Len(Str(w3))) = Str(w3)

' Print the results. Print !"\nThree word location is:" Print Using " W\ \ W\ \ W\ \"; w1s; w2s; w3s

latlong = w1 Shl 28 Or w2 Shl 14 Or w3 ilat = latlong Shr 22 ilong = latlong And &H3fffff lat = (ilat - 900000) / 10000 longi = (ilong - 1800000) / 10000

' Print the results. Print !"\nAfter reversing the procedure:" Print Using " latitude = &, longitude = &"; lat; longi Sleep</lang>

Output:
Igual que la entrada de Nim


Go

Translation of: Wren

Though no need for big integers as we have int64 built in. <lang go>package main

import (

   "fmt"
   "strconv"

)

func toWord(w int64) string { return fmt.Sprintf("W%05d", w) }

func fromWord(ws string) int64 {

   var u, _ = strconv.ParseUint(ws[1:], 10, 64)
   return int64(u)

}

func main() {

   fmt.Println("Starting figures:")
   lat := 28.3852
   lon := -81.5638
   fmt.Printf("  latitude = %0.4f, longitude = %0.4f\n", lat, lon)
   // convert lat and lon to positive integers
   ilat := int64(lat*10000 + 900000)
   ilon := int64(lon*10000 + 1800000)
   // build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon)
   latlon := (ilat << 22) + ilon
   // isolate relevant bits
   w1 := (latlon >> 28) & 0x7fff
   w2 := (latlon >> 14) & 0x3fff
   w3 := latlon & 0x3fff
   // convert to word format
   w1s := toWord(w1)
   w2s := toWord(w2)
   w3s := toWord(w3)
   // and print the results
   fmt.Println("\nThree word location is:")
   fmt.Printf("  %s %s %s\n", w1s, w2s, w3s)
   /* now reverse the procedure */
   w1 = fromWord(w1s)
   w2 = fromWord(w2s)
   w3 = fromWord(w3s)
   latlon = (w1 << 28) | (w2 << 14) | w3
   ilat = latlon >> 22
   ilon = latlon & 0x3fffff
   lat = float64(ilat-900000) / 10000
   lon = float64(ilon-1800000) / 10000
   // and print the results
   fmt.Println("\nAfter reversing the procedure:")
   fmt.Printf("  latitude = %0.4f, longitude = %0.4f\n", lat, lon)

}</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Julia

Direct translation from the SymSyn example given by the task creator, though note that idiomatic Julia would usually code this as two small encode() and decode() functions. <lang julia>

  1. Three Word Location - convert latitude and longitude to three words

LAT = 28.3852 LON = -81.5638

  1. build word array W00000 ... W28125

wordarray = ["W" * string(x, pad=5) for x in 0:28125]

  1. make latitude and longitude positive integers

ILAT = Int(LAT * 10000 + 900000) ILON = Int(LON * 10000 + 1800000)

  1. build 43 bit integer containing latitude (21 bits) and longitude (22 bits)

LATLON = (ILAT << 22) + ILON

  1. isolate most significant 15 bits for word 1 index
  2. next 14 bits for word 2 index
  3. next 14 bits for word 3 index

W1 = (LATLON >> 28) & 0x7fff W2 = (LATLON >> 14) & 0x3fff W3 = LATLON & 0x3fff

  1. fetch each word from word array

w1 = wordarray[W1 + 1] w2 = wordarray[W2 + 1] w3 = wordarray[W3 + 1]

  1. display words

println("$w1 $w2 $w3")


  1. reverse the procedure
  1. look up each word

(w1index, w2index, w3index) = indexin([w1, w2, w3], wordarray) .- 1

  1. build the latlon integer from the word indexes

latlon = (w1index << 28) | (w2index << 14) | w3index


  1. isolate the latitude and longitude

ilon = latlon & 0xfffff ilat = latlon >> 22

  1. convert back to floating point values

lon = (ilon - 1800000) / 10000 lat = (ilat - 900000) / 10000

  1. display values

println("latitude = $lat longitude = $lon")

</lang>

Output:
W18497 W11324 W01322
latitude = 28.3852 longitude = -81.5638

Idiomatic version with scrambling

<lang julia>using Random

const LAT = 28.3852 const LON = -81.5638

  1. build word array W00000 ... W28125

const wordarray = ["W" * string(x, pad=5) for x in 0:28125]

function threewordencode(lat, lon, seed=0) # returns vector of 3 strings

   arr = wordarray
   if seed != 0
       rng = MersenneTwister(seed)
       arr = shuffle(rng, deepcopy(wordarray))
   end
   i = (Int(lat * 10000 + 900000) << 22) | Int(lon * 10000 + 1800000)
   return map(x -> arr[x + 1], [(i >> 28) & 0x7fff, (i >> 14) & 0x3fff, i & 0x3fff])

end

function threeworddecode(w1, w2, w3, seed=0) # returns pair of Float64

   arr = wordarray
   if seed != 0
       rng = MersenneTwister(seed)
       arr = shuffle(rng, deepcopy(wordarray))
   end
   (i1, i2, i3) = indexin([w1, w2, w3], arr) .- 1
   latlon = (i1 << 28) | (i2 << 14) | i3
   ilon, ilat = latlon & 0xfffff, latlon >> 22
   return  (ilon - 1800000) / 10000, (ilat - 900000) / 10000

end

words = threewordencode(LAT, LON) println(join(words, " "))

lat, lon = threeworddecode(words...) println("latitude = $lat longitude = $lon")

println("\nWith scramble using key 12345678:") words = threewordencode(LAT, LON, 12345678) println(join(words, " ")) lat, lon = threeworddecode(words..., 12345678) println("latitude = $lat longitude = $lon")

</lang>

Output:
W18497 W11324 W01322
latitude = -81.5638 longitude = 28.3852

With scramble using key 12345678:
W20242 W23427 W16215
latitude = -81.5638 longitude = 28.3852

Kotlin

Translation of: Go

<lang scala>fun toWord(w: Long): String {

   return "W%05d".format(w)

}

fun fromWord(ws: String): Long {

   return ws.substring(1).toUInt().toLong()

}

fun main() {

   println("Starting figures:")
   var lat = 28.3852
   var lon = -81.5638
   println("  latitude = %.4f, longitude = %.4f".format(lat, lon))
   println()
   // convert lat and lon to positive integers
   var ilat = (lat * 10000 + 900000).toLong()
   var ilon = (lon * 10000 + 1800000).toLong()
   // build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon)
   var latlon = (ilat shl 22) + ilon
   // isolate relevant bits
   var w1 = (latlon shr 28) and 0x7fff
   var w2 = (latlon shr 14) and 0x3fff
   var w3 = latlon and 0x3fff
   // convert to word format
   val w1s = toWord(w1)
   val w2s = toWord(w2)
   val w3s = toWord(w3)
   // and print the results
   println("Three word location is:")
   println("  $w1s $w2s $w3s")
   println()
   /* now reverse the procedure */
   w1 = fromWord(w1s)
   w2 = fromWord(w2s)
   w3 = fromWord(w3s)
   latlon = (w1 shl 28) or (w2 shl 14) or w3
   ilat = latlon shr 22
   ilon = latlon and 0x3fffff
   lat = (ilat - 900000).toDouble() / 10000
   lon = (ilon - 1800000).toDouble() / 10000
   // and print the results
   println("After reversing the procedure:")
   println("  latitude = %.4f, longitude = %.4f".format(lat, lon))

}</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Lua

Translation of: C

<lang lua>function toWord(w)

   return string.format("W%05d", w)

end

function fromWord(ws)

   return tonumber(string.sub(ws, 2, -1))

end


print("Starting figures:") lat = 28.3852 lon = -81.5638 print(string.format(" latitude = %0.4f, longitude = %0.4f", lat, lon)) print()

-- convert from lat and lon to positive integers ilat = lat * 10000 + 900000 ilon = lon * 10000 + 1800000

-- build 43 bit number comprising 21 bits (lat) and 22 bits (lon) latlon = math.floor((ilat << 22) + ilon)

-- isloate relevant bits w1 = (latlon >> 28) & 0x7fff w2 = (latlon >> 14) & 0x3fff w3 = latlon & 0x3fff

-- convert to word format w1s = toWord(w1) w2s = toWord(w2) w3s = toWord(w3)

-- and print the results print("Three word location is:") print(" " .. w1s .. " " .. w2s .. " " .. w3s) print()


-- now reverse the procedure w1 = fromWord(w1s) w2 = fromWord(w2s) w3 = fromWord(w3s)

latlon = (w1 << 28) | (w2 << 14) | w3 ilat = latlon >> 22 ilon = latlon & 0x3fffff lat = (ilat - 900000) / 10000 lon = (ilon - 1800000) / 10000

-- and print the results print("After reversing the procedure:") print(string.format(" latitude = %0.4f, longitude = %0.4f", lat, lon))</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Nim

Translation of: Go

<lang Nim>import strformat, strutils

func toWord(w: int64): string = &"W{w:05}"

func fromWord(ws: string): int64 = ws[1..5].parseInt()

echo "Starting figures:" var

 lat = 28.3852
 long = -81.5638

echo &" latitude = {lat:0.4f}, longitude = {long:0.4f}"

  1. Convert "lat" and "long" to positive integers.

var

 ilat = int64(lat * 10_000 + 900_000)
 ilong = int64(long * 10_000 + 1_800_000)
  1. Build 43 bit int comprising 21 bits (lat) and 22 bits (lon).

var latlong = ilat shl 22 + ilong

  1. Isolate relevant bits.

var

 w1 = latlong shr 28 and 0x7fff
 w2 = latlong shr 14 and 0x3fff
 w3 = latlong and 0x3fff
  1. Convert to word format.

let

 w1s = w1.toWord
 w2s = w2.toWord
 w3s = w3.toWord
  1. Print the results.

echo "\nThree word location is:" echo &" {w1s} {w2s} {w3s}"

  1. Reverse the procedure.

w1 = w1s.fromWord w2 = w2s.fromWord w3 = w3s.fromWord

latlong = w1 shl 28 or w2 shl 14 or w3 ilat = latlong shr 22 ilong = latlong and 0x3fffff lat = float(ilat - 900_000) / 10_000 long = float(ilong - 1_800_000) / 10_000

  1. Print the results.

echo "\nAfter reversing the procedure:" echo &" latitude = {lat:0.4f}, longitude = {long:0.4f}"</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Perl

Translation of: Raku

<lang perl>use strict; use warnings; use feature 'say'; use bignum; # without this, round-trip results not exact

use Math::AnyNum 'polymod';

  1. SYNTHETICS HANDLING

my @synth; push @synth, join , @$_ for map { [split /:/] } glob '{b,d,f,h,j,k,l,m,n,p,r,s,t,w,y,z}:{a,e,i,o,u}'; my(%htnys,$c); $htnys{$_} = $c++ for @synth; my $exp = @synth; my $prec = 10_000;

sub bin2dec { unpack('N', pack('B32', substr('0' x 32 . shift, -32))) }

sub synth { join , reverse @synth[polymod(shift() + int(rand 18) * 28126, $exp, $exp) ] }

sub thnys {

   my @n = @htnys{ shift() =~ /(..)(..)(..)/ };  # NB notation on hash slice: % -> @
   ($n[2] + $n[1]*$exp + $n[0]*$exp**2) % 28126

}

  1. ENCODE / DECODE

sub w_encode {

   my($lat, $lon, $f) = @_;
   $f = \&synth unless $f;
   my @words;
   my $bits = sprintf '%021b%022b', int(($lat+90)*$prec), int(($lon+180)*$prec);
   push @words, &$f(bin2dec($_)) for $bits =~ / (.{15}) (.{14}) (.{14}) /x;
   @words

}

sub w_decode {

   my($w, $f) = @_;
   $f = \&thnys unless $f;
   my $s = '%015b';
   my $bin = sprintf($s, &$f($$w[0])) . substr(sprintf($s, &$f($$w[1])), 1) . substr(sprintf($s, &$f($$w[2])), 1);
   (bin2dec(substr($bin,0,21))/$prec - 90), (bin2dec(substr($bin,21))/$prec - 180)

}

  1. TESTING

for ([ 51.4337, -0.2141, 'Wimbledon'],

    [ 21.2596,   -157.8117,   'Diamond Head crater'],
    [-55.9652,    -67.2256,   'Monumento Cabo De Hornos'],
    [ 71.170924,   25.782998, 'Nordkapp, Norway'],
    [ 45.762983,    4.834520, 'Café Perl, Lyon'],
    [ 48.391541, -124.736731, 'Cape Flattery Lighthouse, Tatoosh Island'],
   ) {
   my($lat, $lon, $description) = @$_;
   my @words = w_encode $lat, $lon;
   my @index = w_encode $lat, $lon, sub { shift };
   printf "Coordinates: %s, %s (%s)\n   To Index: %s\n  To 3-word: %s\nFrom 3-word: %s, %s\n From Index: %s, %s\n\n",
     $lat, $lon, $description, join(' ',@index), join(' ',@words), w_decode(\@words), w_decode(\@index, sub { shift() });

}</lang>

Output:
Coordinates: 51.4337, -0.2141 (Wimbledon)
   To Index: 22099 365 12003
  To 3-word: yotema ritomi rahiku
From 3-word: 51.4337, -0.2141
 From Index: 51.4337, -0.2141

Coordinates: 21.2596, -157.8117 (Diamond Head crater)
   To Index: 17384 5133 8891
  To 3-word: hayibi batufo jokube
From 3-word: 21.2596, -157.8117
 From Index: 21.2596, -157.8117

Coordinates: -55.9652, -67.2256 (Monumento Cabo De Hornos)
   To Index: 5317 15428 13632
  To 3-word: fubeha zidura nerupe
From 3-word: -55.9652, -67.2256
 From Index: -55.9652, -67.2256

Coordinates: 71.170924, 25.782998 (Nordkapp, Norway)
   To Index: 25182 15741 9829
  To 3-word: zorenu jaboda kiyika
From 3-word: 71.1709, 25.7829
 From Index: 71.1709, 25.7829

Coordinates: 45.762983, 4.83452 (Café Perl, Lyon)
   To Index: 21212 15728 13337
  To 3-word: ludefu bimepo demojo
From 3-word: 45.7629, 4.8345
 From Index: 45.7629, 4.8345

Coordinates: 48.391541, -124.736731 (Cape Flattery Lighthouse, Tatoosh Island)
   To Index: 21623 11041 11960
  To 3-word: kakofo radaki habuho
From 3-word: 48.3915, -124.7368
 From Index: 48.3915, -124.7368

Phix

Translation of: Go

<lang Phix>requires("0.8.2") -- (new || && << and >> operators, pre-0.8.2 code left in as comments)

function toWord(integer w)

   return sprintf("W%05d", w)

end function

function fromWord(string ws)

   sequence r = scanf(ws,"W%05d")
   integer res = r[1][1]
   return res

end function

printf(1,"Starting figures:\n") atom lat = 28.3852,

    lon = -81.5638

printf(1," latitude = %0.4f, longitude = %0.4f\n", {lat, lon})

-- convert lat and lon to positive integers integer ilat := floor((lat+90)*10000),

       ilon := floor((lon+180)*10000)

-- build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon) -- (Std phix atoms have 53/64 bits of precision on 32/64 bit, both plenty) --atom latlon := ilat*power(2,22) + ilon -- (pre-0.8.2) atom latlon := (ilat << 22) + ilon

-- isolate relevant bits --integer w1 = and_bits(floor(latlon/power(2,28)),0x7fff), --("") -- w2 = and_bits(floor(latlon/power(2,14)),0x3fff), -- w3 = and_bits(latlon,0x3fff) integer w1 = (latlon >> 28) && 0x7fff,

       w2 = (latlon >> 14) && 0x3fff,
       w3 = latlon && 0x3fff

-- convert to word format string w1s = toWord(w1),

      w2s = toWord(w2),
      w3s = toWord(w3)

-- and print the results printf(1,"\nThree word location is:\n") printf(1," %s %s %s\n", {w1s, w2s, w3s})

-- now reverse the procedure w1 = fromWord(w1s) w2 = fromWord(w2s) w3 = fromWord(w3s)

-- NB: or_bits (likewise ||), being expressly 32-bit, is NOT appropriate here... --latlon = w1*power(2,28) + w2*power(2,14) + w3 -- (pre-0.8.2) --ilat = floor(latlon/power(2,22)) --ilon = and_bits(latlon,0x3fffff) latlon = (w1 << 28) + (w2 << 14) + w3 ilat = latlon >> 22 ilon = latlon && 0x3fffff lat = ilat/10000 - 90 lon = ilon/10000 - 180

-- and print the results printf(1,"\nAfter reversing the procedure:\n") printf(1," latitude = %0.4f, longitude = %0.4f\n", {lat, lon})</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638

Raku

Works with: Rakudo version 2020.07

The task

In large part due to the complete lack of specification, reference implementation, or guidance from the task creator, came up with my own bespoke synthetic word list.

Words always consist of a series of consonant/vowel pairs. Uses a cut down alphabet to reduce possible confusion from overlapping pronunciation.

Some letters with overlapping pronunciation are removed: c: confusable with k or s, g: overlaps with j, x: overlaps with z, q: just because, v: similar to w and we have way more than enough characters anyway.

As it is, with this alphabet we can form 512000 different 6 character "words"; 28126 is a drop in the bucket. To spread out the the words a bit, add a bit of randomness. 28126 fits into 512000 18 and a bit times. Add a random multiple of 28126 to the encoder then modulus it back out on decode. Will get different results on different runs.

We don't bother to pre-calculate and store the words, just generate them on the fly.

Official pronunciation guide:

  • a - long a (say may day)
  • e - long e (he me see)
  • i - long i (hi sigh die)
  • o - long o (go so low)
  • u - long u (due boo moo)

<lang perl6># SYNTHETICS HANDLING my @synth = flat < b d f h j k l m n p r s t w y z > X~ < a e i o u >; my %htnys = @synth.antipairs; my $exp = @synth.elems;

sub synth (Int $v) { @synth[($v + (^18).pick * 28126).polymod($exp xx *).reverse || 0].join }

sub thnys (Str $v) { (sum %htnys{$v.comb(2).reverse} Z* 1, $exp, $exp**2) % 28126 }


  1. ENCODE / DECODE

sub w-encode ( Rat(Real) $lat, Rat(Real) $lon, :&f = &synth ) {

   $_ = (($lat +  90) * 10000).round.fmt('%021b') ~ (($lon + 180) * 10000).round.fmt('%022b');
   (:2(.substr(0,15)), :2(.substr(15,14)),:2(.substr(29)))».&f

}

sub w-decode ( *@words, :&f = &thnys ) {

   my $bin = (@words».&f Z, <0 1 1>).map({.[0].fmt('%015b').substr(.[1])}).join;
   (:2($bin.substr(0,21))/10000) - 90, (:2($bin.substr(21))/10000) - 180

}


  1. TESTING

for 51.4337, -0.2141, # Wimbledon

   21.2596,-157.8117, # Diamond Head crater
  -55.9652, -67.2256, # Monumento Cabo De Hornos
   59.3586,  24.7447, # Lake Raku
   29.2021,  81.5324, # Village Raku
   -7.1662,  53.9470, # The Indian ocean, south west of Seychelles
   28.3852, -81.5638  # Walt Disney World
 -> $lat, $lon {
   my @words = w-encode $lat, $lon;
   my @index = w-encode $lat, $lon, :f( { $_ } );
   printf "Coordinates: %s, %s\n   To Index: %s\n  To 3-word: %s\nFrom 3-word: %s, %s\n From Index: %s, %s\n\n",
     $lat, $lon, @index.Str, @words.Str, w-decode(@words), w-decode @index, :f( { $_ } );

}</lang>

Output:
Coordinates: 51.4337, -0.2141
   To Index: 22099 365 12003
  To 3-word: zofobe fohujo habute
From 3-word: 51.4337, -0.2141
 From Index: 51.4337, -0.2141

Coordinates: 21.2596, -157.8117
   To Index: 17384 5133 8891
  To 3-word: nijemo zanaza fupawu
From 3-word: 21.2596, -157.8117
 From Index: 21.2596, -157.8117

Coordinates: -55.9652, -67.2256
   To Index: 5317 15428 13632
  To 3-word: zanohu julaso husese
From 3-word: -55.9652, -67.2256
 From Index: -55.9652, -67.2256

Coordinates: 59.3586, 24.7447
   To Index: 23337 4732 15831
  To 3-word: kapupi hokame supoku
From 3-word: 59.3586, 24.7447
 From Index: 59.3586, 24.7447

Coordinates: 29.2021, 81.5324
   To Index: 18625 5535 10268
  To 3-word: dijule nutuza nefini
From 3-word: 29.2021, 81.5324
 From Index: 29.2021, 81.5324

Coordinates: -7.1662, 53.947
   To Index: 12942 12942 12942
  To 3-word: rakudo rakudo rakudo
From 3-word: -7.1662, 53.947
 From Index: -7.1662, 53.947

Coordinates: 28.3852, -81.5638
   To Index: 18497 11324 1322
  To 3-word: tabesa nekaso bupodo
From 3-word: 28.3852, -81.5638
 From Index: 28.3852, -81.5638

(Ok, I admit I manipulated that second to last one, but it is a correct and valid 3-word location in this implementation. There is less than 1 chance in 5000 that it will produce that specific word group though.)

A thought experiment

A little thought experiment... Latitude, longitude to four decimal places is accurate to about 11.1 meters at the equator, smaller the further from the equator you get. What would it take to support five decimal places? (Accurate to 1.11 meters.)

360 * 100000 == 36000000; ceiling 36000000.log(2) == 26;

So we need 26 bits to cover 360.00000; half of that for 180.00000, or 26 bits + 25 bits == 51 bits. 51 / 3 == 17. 2**17 == 131072 indices. The previous synthetics routine provides much more than enough.

How many sylabics will we need to minimally cover it?

∛131072 == 50.7968...

So at least 51. The synthetics routine provide sylabics in blocks of 5, so we would need at least 11 consonants.

Capriciously and somewhat randomly cutting down the list we arrive at this.

10 times better accuracy in the same three, 6-letter word space.

<lang perl6># SYNTHETICS HANDLING my @synth = flat < b d f j k n p r s t w > X~ < a e i o u >; my %htnys = @synth.antipairs; my $exp = @synth.elems; my $prec = 100_000;


sub synth (Int $v) { @synth[$v.polymod($exp xx *).reverse || 0].join }

sub thnys (Str $v) { sum %htnys{$v.comb(2).reverse} Z× 1, $exp, $exp² }


  1. ENCODE / DECODE

sub w-encode ( Rat(Real) $lat, Rat(Real) $lon, :&f = &synth ) {

   $_ = (($lat +  90) × $prec).round.fmt('%025b') ~ (($lon + 180) × $prec).round.fmt('%026b');
   (:2(.substr(0,17)), :2(.substr(17,17)), :2(.substr(34)))».&f

}

sub w-decode ( *@words, :&f = &thnys ) {

   my $bin = @words».&f.map({.fmt('%017b')}).join;
   (:2($bin.substr(0,25))/$prec) - 90, (:2($bin.substr(25))/$prec) - 180

}


  1. TESTING

for 51.43372, -0.21412, # Wimbledon center court

   21.25976,-157.81173, # Diamond Head summit
  -55.96525, -67.22557, # Monumento Cabo De Hornos
   28.3852,  -81.5638,  # Walt Disney World
   89.99999, 179.99999, # test some
  -89.99999,-179.99999  # extremes
 -> $lat, $lon {
   my @words = w-encode $lat, $lon;
   printf "Coordinates: %s, %s\n   To Index: %s\n  To 3-word: %s\nFrom 3-word: %s, %s\n\n",
     $lat, $lon, w-encode($lat, $lon, :f({$_})).Str, @words.Str, w-decode(@words);

}</lang>

Output:
Coordinates: 51.43372, -0.21412
   To Index: 55247 71817 21724
  To 3-word: jofuni kosasi diduwu
From 3-word: 51.43372, -0.21412

Coordinates: 21.25976, -157.81173
   To Index: 43460 110608 121675
  To 3-word: fukafa repebo safija
From 3-word: 21.25976, -157.81173

Coordinates: -55.96525, -67.22557
   To Index: 13294 108118 5251
  To 3-word: bukeru rasaso besane
From 3-word: -55.96525, -67.22557

Coordinates: 28.3852, -81.5638
   To Index: 46244 28747 13220
  To 3-word: jajasu duniri bukaka
From 3-word: 28.3852, -81.5638

Coordinates: 89.99999, 179.99999
   To Index: 70312 65298 86271
  To 3-word: kofoki kepifo nonope
From 3-word: 89.99999, 179.99999

Coordinates: -89.99999, -179.99999
   To Index: 0 512 1
  To 3-word: ba duji be
From 3-word: -89.99999, -179.99999

Symsyn

<lang Symsyn> | Three Word Location - convert latitude and longitude to three words

lat : 28.3852 lon : -81.5638

| build word array W00000 ... W28125

i
if i <= 28125
   ~ i $r
   #$r szr
   'W00000' $t
   (6-szr) szr 
   szr #$t
   + $r $t
   + $t $wordarray 
   + i
   goif
endif

| make latitude and longitude positive integers

{lat * 10000 + 900000} ilat
{lon * 10000 + 1800000} ilon

| build 43 bit integer containing latitude (21 bits) and longitude (22 bits)

ilat latlon
shl latlon 22
+ ilon latlon

| isolate most significant 15 bits for word 1 index | next 14 bits for word 2 index | next 14 bits for word 3 index

latlon:42:15 w1
latlon:27:14 w2
latlon:13:14 w3

| fetch each word from word array

 (w1*6+1) w1
 $wordarray.w1 $w1 6
 (w2*6+1) w2
 $wordarray.w2 $w2 6
 (w3*6+1) w3
 $wordarray.w3 $w3 6

| display words

"$w1 ' ' $w2 ' ' $w3" []


| reverse the procedure


| look up each word

call bsearch 0 28125 $w1
result w1index
call bsearch 0 28125 $w2
result w2index
call bsearch 0 28125 $w3
result w3index

| build the latlon integer from the word indexes

w1index latlon
shl latlon 14
+ w2index latlon
shl latlon 14
+ w3index latlon

| isolate the latitude and longitude

latlon:21:22 ilon
latlon:42:21 ilat

| convert back to floating point values

{(ilon - 1800000) / 10000} lon
{(ilat - 900000) / 10000} lat

| display values

"'latitude = ' lat ' longitude = ' lon" [] 
stop

bsearch

param L H $word
if L <= H 
   ((L + H) shr 1) M
   (M*6+1) I
   $wordarray.I $w 6
   if $w > $word 
      - 1 M H
   else
      if $w < $word 
         + 1 M L
      else      
         return M
      endif
   endif
   goif
endif
return -1 

</lang>

Output:
W18497 W11324 W01322
latitude =          28.3852 longitude =         -81.5638

Using Real Words

<lang Symsyn>

| Three Word Location - convert latitude and longitude to three words


lat : 28.3852 lon : -81.5638

| Build real word array of the first 28126 words of 8 | or less characters from list of 69905 words sorted alphabetically | at http://www-personal.umich.edu/~jlawler/wordlist. | There are 36282 words of 8 or less characters here.


opentext 'LargeWordList.txt' wf
if ioresult <> 0
   stop
endif
if i <= 28125
   [wf] $s
   if ioresult <> 0
      go closefile
   endif
   #$s wsz
   if wsz <= 8
      + $s $wordarray
      (8-wsz) wsz
      + '        ' $wordarray wsz
      + i
   endif
   goif
endif

closefile close wf

| make latitude and longitude positive integers

{lat * 10000 + 900000} ilat
{lon * 10000 + 1800000} ilon

| build 43 bit integer containing latitude (21 bits) and longitude (22 bits)

ilat latlon
shl latlon 22
+ ilon latlon

| isolate most significant 15 bits for word 1 index | next 14 bits for word 2 index | next 14 bits for word 3 index

latlon:42:15 w1
latlon:27:14 w2
latlon:13:14 w3

| fetch each word from word array

 (w1*8+1) w1
 $wordarray.w1 $w1 8
 (w2*8+1) w2
 $wordarray.w2 $w2 8
 (w3*8+1) w3
 $wordarray.w3 $w3 8

| display words

"$w1 ' ' $w2 ' ' $w3" []


| reverse the procedure


| look up each word

call bsearch 0 28125 $w1
result w1index
call bsearch 0 28125 $w2
result w2index
call bsearch 0 28125 $w3
result w3index

| build the latlon integer from the word indexes

w1index latlon
shl latlon 14
+ w2index latlon
shl latlon 14
+ w3index latlon

| isolate the latitude and longitude

latlon:21:22 ilon
latlon:42:21 ilat

| convert back to floating point values

{(ilon - 1800000) / 10000} lon
{(ilat - 900000) / 10000} lat

| display values

"'latitude = ' lat ' longitude = ' lon" [] 
stop

bsearch

param L H $word
if L <= H 
   ((L + H) shr 1) M
   (M*8+1) I
   $wordarray.I $w 8
   if $w > $word 
      - 1 M H
   else
      if $w < $word 
         + 1 M L
      else      
         return M
      endif
   endif
   goif
endif
return -1 

</lang>

Output:
ling     everyone amoral  
latitude =          28.3852 longitude =         -81.5638

Wren

Library: Wren-fmt
Library: Wren-big

This just follows the steps in the task description though I couldn't see any point in creating a 28,126 element array when two simple functions will do.

Note that bitwise operations are limited to 32-bit unsigned integers in Wren which isn't big enough here so we use BigInts instead. <lang ecmascript>import "/fmt" for Fmt import "/big" for BigInt

// functions to convert to and from the word format 'W00000' var toWord = Fn.new { |w| Fmt.swrite("W$05d", w) } var fromWord = Fn.new { |w| Num.fromString(w[1..-1]) }

// set latitude and longitude and print them System.print("Starting figures:") var lat = 28.3852 var lon = -81.5638 Fmt.print(" latitude = $0.4f, longitude = $0.4f", lat, lon)

// convert lat and lon to positive BigInts var ilat = BigInt.new(lat * 10000 + 900000) var ilon = BigInt.new(lon * 10000 + 1800000)

// build 43 bit BigInt comprising 21 bits (lat) and 22 bits (lon) var latlon = (ilat << 22) + ilon

// isolate relevant bits and convert back to 'normal' ints var w1 = ((latlon >> 28) & 0x7fff).toSmall var w2 = ((latlon >> 14) & 0x3fff).toSmall var w3 = (latlon & 0x3fff).toSmall

// convert to word format w1 = toWord.call(w1) w2 = toWord.call(w2) w3 = toWord.call(w3)

// and print the results System.print("\nThree word location is:") Fmt.print(" $s $s $s", w1, w2, w3)

/* now reverse the procedure */ w1 = BigInt.new(fromWord.call(w1)) w2 = BigInt.new(fromWord.call(w2)) w3 = BigInt.new(fromWord.call(w3)) latlon = (w1 << 28) | (w2 << 14) | w3 ilat = (latlon >> 22).toSmall ilon = (latlon & 0x3fffff).toSmall lat = (ilat - 900000) / 10000 lon = (ilon - 1800000) / 10000

// and print the results System.print("\nAfter reversing the procedure:") Fmt.print(" latitude = $0.4f, longitude = $0.4f", lat, lon)</lang>

Output:
Starting figures:
  latitude = 28.3852, longitude = -81.5638

Three word location is:
  W18497 W11324 W01322

After reversing the procedure:
  latitude = 28.3852, longitude = -81.5638