7.15.6. geo_distance

7.15.6.1. Summary

geo_distance calculates the value of distance between specified two points.

7.15.6.2. Syntax

geo_distance requires two point. The parameter approximate_type is optional:

geo_distance(point1, point2)
geo_distance(point1, point2, approximate_type)

The default value of approximate_type is "rectangle". If you omit approximate_type, geo_distance calculates the value of distance as if "rectangle" was specified.

7.15.6.3. Usage

geo_distance is one of the Groonga builtin functions.

You can call a builtin function in grn_expr

geo_distance function calculates the value of distance (approximate value) between the coordinate of point1 and the coordinate of point2.

Note

Groonga provides three built in functions for calculating the value of distance. There are geo_distance(), geo_distance2() and geo_distance3(). The difference of them is the algorithm of calculating distance. geo_distance2() and geo_distance3() were deprecated since version 1.2.9. Use geo_distance(point1, point2, "sphere") instead of geo_distance2(point1, point2). Use geo_distance(point1, point2, "ellipsoid") instead of geo_distance3(point1, point2).

Lets’s learn about geo_distance usage with examples. This section shows simple usages.

Here are two schema definition and sample data to show the difference according to the usage. Those samples show how to calculate the value of distance between New York City and London.

  1. Using the column value of location for calculating the distance (Cities table)

  2. Using the explicitly specified coordinates for calculating the distance (Geo table)

7.15.6.3.1. Using the column value of location

Here are a schema definition of Cities table and sample data to show usage.

table_create Cities TABLE_HASH_KEY ShortText
column_create Cities location COLUMN_SCALAR WGS84GeoPoint
load --table Cities
[
["_key", "location"],
["New York City", "146566000x-266422000"]
]

Execution example:

table_create Cities TABLE_HASH_KEY ShortText
# [[0, 1337566253.89858, 0.000355720520019531], true]
column_create Cities location COLUMN_SCALAR WGS84GeoPoint
# [[0, 1337566253.89858, 0.000355720520019531], true]
load --table Cities
[
["_key", "location"],
["New York City", "146566000x-266422000"]
]
# [[0, 1337566253.89858, 0.000355720520019531], 1]

This execution example creates a table named Cities which has one column named location. location column stores the value of coordinate. The coordinate of Tokyo is stored as sample data.

Execution example:

select Cities --output_columns _score --filter 1 --scorer '_score = geo_distance(location, "185428000x-461000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "_score",
#           "Int32"
#         ]
#       ],
#       [
#         5807750
#       ]
#     ]
#   ]
# ]

This sample shows that geo_distance use the value of location column and the value of coordinate to calculate distance.

The value (“185428000x-461000”) passed to geo_distance as the second argument is the coordinate of London.

7.15.6.3.2. Using the explicitly specified value of location

Here are a schema definition of Geo table and sample data to show usage.

table_create Geo TABLE_HASH_KEY ShortText
column_create Geo distance COLUMN_SCALAR Int32
load --table Geo
[
  {
    "_key": "the record for geo_distance() result"
  }
]

Execution example:

table_create Geo TABLE_HASH_KEY ShortText
# [[0, 1337566253.89858, 0.000355720520019531], true]
column_create Geo distance COLUMN_SCALAR Int32
# [[0, 1337566253.89858, 0.000355720520019531], true]
load --table Geo
[
  {
    "_key": "the record for geo_distance() result"
  }
]
# [[0, 1337566253.89858, 0.000355720520019531], 1]

This execution example creates a table named Geo which has one column named distance. distance column stores the value of distance.

Execution example:

select Geo --output_columns distance --scorer 'distance = geo_distance("146566000x-266422000", "185428000x-461000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "distance",
#           "Int32"
#         ]
#       ],
#       [
#         5807750
#       ]
#     ]
#   ]
# ]

This sample shows that geo_distance use the coordinate of London and the coordinate of New York to calculate distance.

7.15.6.4. Parameters

7.15.6.4.1. Required parameters

There are two required parameter, point1 and point2.

7.15.6.4.1.1. point1

Specifies the start point that you want to calculate the value of distance between two points.

You can specify the value of GeoPoint type. [1]

See Data types about GeoPoint.

7.15.6.4.1.2. point2

Specifies the end point that you want to calculate the value of distance between two points.

You can specify the value of GeoPoint type or the string indicating the coordinate.

See Data types about GeoPoint and the coordinate.

7.15.6.4.2. Optional parameter

There is a optional parameter, approximate_type.

7.15.6.4.2.1. approximate_type

Specifies how to approximate the geographical features for calculating the value of distance.

You can specify the value of approximate_type by one of the followings.

  • rectangle

  • sphere

  • ellipsoid

Note

There is a limitation about geo_distance. geo_distance can not calculate the value of distance between two points across meridian, equator or the date line if you use sphere or ellipsoid as approximate type. There is not such a limitation for rectangle. This is temporary limitation according to the implementation of Groonga, but it will be fixed in the future release.

7.15.6.4.2.1.1. rectangle

This parameter require to approximate the geographical features by square approximation for calculating the distance.

Since the value of distance is calculated by simple formula, you can calculate the value of distance fast. But, the error of distance increases as it approaches the pole.

You can also specify rect as abbrev expression.

Here is a sample about calculating the value of distance with column value.

Execution example:

select Cities --output_columns _score --filter 1 --scorer '_score = geo_distance(location, "185428000x-461000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "_score",
#           "Int32"
#         ]
#       ],
#       [
#         5807750
#       ]
#     ]
#   ]
# ]

Here is a sample about calculating the value of distance with explicitly specified point.

Execution example:

select Geo --output_columns distance --scorer 'distance = geo_distance("146566000x-266422000", "185428000x-461000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "distance",
#           "Int32"
#         ]
#       ],
#       [
#         5807750
#       ]
#     ]
#   ]
# ]

Here are samples about calculating the value of distance with explicitly specified point across meridian, equator, the date line.

Execution example:

select Geo --output_columns distance --scorer 'distance = geo_distance("175904000x8464000", "145508000x-13291000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "distance",
#           "Int32"
#         ]
#       ],
#       [
#         1051322
#       ]
#     ]
#   ]
# ]

This sample shows the value of distance across meridian. The return value of geo_distance("175904000x8464000", "145508000x-13291000", "rectangle") is the value of distance from Paris, Flance to Madrid, Spain.

Execution example:

select Geo --output_columns distance --scorer 'distance = geo_distance("146566000x-266422000", "-56880000x-172310000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "distance",
#           "Int32"
#         ]
#       ],
#       [
#         6880439
#       ]
#     ]
#   ]
# ]

This sample shows the value of distance across equator. The return value of geo_distance("146566000x-266422000", "-56880000x-172310000", "rectangle") is the value of distance from New York, The United Status to Brasillia, Brasil.

Execution example:

select Geo --output_columns distance --scorer 'distance = geo_distance("143660000x419009000", "135960000x-440760000", "rectangle")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "distance",
#           "Int32"
#         ]
#       ],
#       [
#         10475660
#       ]
#     ]
#   ]
# ]

This sample shows the value of distance across the date line. The return value of geo_distance("143660000x419009000", "135960000x-440760000", "rectangle") is the value of distance from Beijin, China to San Francisco, The United States.

Note

geo_distance uses square approximation as default. If you omit approximate_type, geo_distance behaves like rectangle was specified.

Note

geo_distance accepts the string indicating the coordinate as the value of point1 when the value of approximate_type is "rectangle". If you specified the string indicating the coordinate as the value of point1 with sphere or ellipsoid, geo_distance returns 0 as the value of distance.

7.15.6.4.2.1.2. sphere

This parameter require to approximate the geographical features by spherical approximation for calculating the distance.

It is slower than rectangle, but the error of distance becomes smaller than rectangle.

You can also specify sphr as abbrev expression.

Here is a sample about calculating the value of distance with column value.

Execution example:

select Cities --output_columns _score --filter 1 --scorer '_score = geo_distance(location, "185428000x-461000", "sphere")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "_score",
#           "Int32"
#         ]
#       ],
#       [
#         5558225
#       ]
#     ]
#   ]
# ]
7.15.6.4.2.1.3. ellipsoid

This parameter require to approximate the geographical features by ellipsoid approximation for calculating the distance.

It uses the calculation of distance by the formula of Hubeny. It is slower than sphere, but the error of distance becomes smaller than sphere.

You can also specify ellip as abbrev expression.

Here is a sample about calculating the value of distance with column value.

Execution example:

select Cities --output_columns _score --filter 1 --scorer '_score = geo_distance(location, "185428000x-461000", "ellipsoid")'
# [
#   [
#     0,
#     1337566253.89858,
#     0.000355720520019531
#   ],
#   [
#     [
#       [
#         1
#       ],
#       [
#         [
#           "_score",
#           "Int32"
#         ]
#       ],
#       [
#         5836138
#       ]
#     ]
#   ]
# ]

7.15.6.5. Return value

geo_distance returns the value of distance in float type. The unit of return value is meter.

Footnote