前几天群里有人问ABAP有没有Geohash函数,用来帮助SAP存储门店位置、实现查找附近门店的功能。因为没有查到,所以我动手写了一个。
Geohash是什么
Geohash是一种公共域地理编码系统,它将一个地理位置编码成一串字母和数字。字符串越长,表示的范围越精确。两个Geohash字符串的相同前缀越多,表示它们所代表的地点的距离越近,这样就可以利用字符串的前缀匹配来快速查询附近的地点信息。
关于Geohash的更多介绍,可以参考:
本文链接:https://www.cnblogs.com/hhelibeb/p/11426826.html
原创内容,转载请注明
实现
我在Github创建了一个repo,包含了自己写的编码、解码方法。地址是:https://github.com/hhelibeb/geohash-abap
代码如下,目前还有更多功能在实现中,有兴趣的朋友可以来一起写。
(注:github上面的代码会不定时更新,博客中贴的只是初始代码,请前往github查看最新代码)
class zcl_geohash definition
public
final
create public . public section. types: begin of ty_hash,
hash type string,
end of ty_hash.
types: ty_hash_t type standard table of ty_hash with empty key. types:
ty_tude type p length decimals . constants c_max_hash_length type i value ##NO_TEXT. class-methods class_constructor .
class-methods encode
importing
!longitude type ty_tude
!latitude type ty_tude
!length type i default
returning
value(r_geo_hash) type string .
class-methods decode
importing
!geohash type string
exporting
!longitude type ty_tude
!latitude type ty_tude . class-methods neighbors importing geohash type string
returning value(neighbors) type ty_hash_t. class-methods validate importing geohash type string
returning value(valid) type abap_bool. private section. types:
begin of ty_base32,
decimals type i,
base32 type string,
end of ty_base32 .
types:
ty_base32_t1 type hashed table of ty_base32 with unique key decimals .
types:
ty_base32_t2 type hashed table of ty_base32 with unique key base32 . types: begin of ty_neighbors_odd,
f1 type string,
f2 type string,
f3 type string,
f4 type string,
f5 type string,
f6 type string,
f7 type string,
f8 type string,
end of ty_neighbors_odd.
types: ty_neighbors_odd_t type standard table of ty_neighbors_odd with empty key. types: begin of ty_neighbors_even,
f1 type string,
f2 type string,
f3 type string,
f4 type string,
end of ty_neighbors_even.
types: ty_neighbors_even_t type standard table of ty_neighbors_even with empty key. class-data mt_base32_code1 type ty_base32_t1 .
class-data mt_base32_code2 type ty_base32_t2 . class-data mt_neighbors_odd type ty_neighbors_odd_t.
class-data mt_neighbors_even type ty_neighbors_even_t. constants c_longitude_min type ty_tude value '-180.00' ##NO_TEXT.
constants c_longitude_max type ty_tude value '180.00' ##NO_TEXT.
constants c_latitude_min type ty_tude value '-90.00' ##NO_TEXT.
constants c_latitude_max type ty_tude value '90.00' ##NO_TEXT.
constants c_zero type c value '' ##NO_TEXT.
constants c_one type c value '' ##NO_TEXT. class-methods bin_to_dec
importing
!i_bin type string default ''
returning
value(r_dec) type int4 .
class-methods dec_to_bin
importing
!i_dec type int4
returning
value(r_bin) type string . class-methods get_bin
importing
!i_left type ty_tude
!i_right type ty_tude
!i_tude type ty_tude
exporting
!e_left type ty_tude
!e_right type ty_tude
!e_bin type char1 .
class-methods get_tude
importing
!i_left type ty_tude
!i_right type ty_tude
!i_bin type string
exporting
!e_left type ty_tude
!e_right type ty_tude
!e_tude type ty_tude . class-methods: get_index importing index type i
offset type i
max_index type i
returning value(r_index) type i.
class-methods: get_code_neighbor importing i_table type standard table
i_member type string
returning value(r_table) type ty_hash_t. endclass.
class zcl_geohash implementation. method bin_to_dec. if contains( val = i_bin regex = `[^]` ).
return.
endif. data(length) = strlen( i_bin ). data(l_index) = . do length times. data(temp) = i_bin+l_index(). if temp = .
r_dec = r_dec + ** ( length - l_index - ).
endif. l_index = l_index + . enddo. endmethod. method class_constructor. mt_base32_code1 = value #(
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = '' )
( decimals = base32 = 'b' )
( decimals = base32 = 'c' )
( decimals = base32 = 'd' )
( decimals = base32 = 'e' )
( decimals = base32 = 'f' )
( decimals = base32 = 'g' )
( decimals = base32 = 'h' )
( decimals = base32 = 'j' )
( decimals = base32 = 'k' )
( decimals = base32 = 'm' )
( decimals = base32 = 'n' )
( decimals = base32 = 'p' )
( decimals = base32 = 'q' )
( decimals = base32 = 'r' )
( decimals = base32 = 's' )
( decimals = base32 = 't' )
( decimals = base32 = 'u' )
( decimals = base32 = 'v' )
( decimals = base32 = 'w' )
( decimals = base32 = 'x' )
( decimals = base32 = 'y' )
( decimals = base32 = 'z' )
). mt_base32_code2 = mt_base32_code1. mt_neighbors_odd = value #(
( f1 = 'b' f2 = 'c' f3 = 'f' f4 = 'g' f5 = 'u' f6 = 'v' f7 = 'y' f8 = 'z' )
( f1 = '' f2 = '' f3 = 'd' f4 = 'e' f5 = 's' f6 = 't' f7 = 'w' f8 = 'x' )
( f1 = '' f2 = '' f3 = '' f4 = '' f5 = 'k' f6 = 'm' f7 = 'q' f8 = 'r' )
( f1 = '' f2 = '' f3 = '' f4 = '' f5 = 'h' f6 = 'j' f7 = 'n' f8 = 'p' )
). mt_neighbors_even = value #(
( f1 = 'p' f2 = 'r' f3 = 'x' f4 = 'z' )
( f1 = 'n' f2 = 'q' f3 = 'w' f4 = 'y' )
( f1 = 'j' f2 = 'm' f3 = 't' f4 = 'v' )
( f1 = 'h' f2 = 'k' f3 = 's' f4 = 'u' )
( f1 = '' f2 = '' f3 = 'e' f4 = 'g' )
( f1 = '' f2 = '' f3 = 'd' f4 = 'f' )
( f1 = '' f2 = '' f3 = '' f4 = 'c' )
( f1 = '' f2 = '' f3 = '' f4 = 'b' )
). endmethod. method decode. types: numc5 type n length . data(length) = strlen( geohash ). if length <= .
return.
endif. if length > c_max_hash_length.
length = c_max_hash_length.
endif. data(geo_hash_internal) = to_lower( geohash ). data(hash_index) = . do length times. data(base32) = geo_hash_internal+hash_index(). data(decimals) = value #( mt_base32_code2[ base32 = base32 ]-decimals optional ). data(bin5) = conv numc5( dec_to_bin( decimals ) ). data: mix_bin type string,
longitude_bin type string,
latitude_bin type string. mix_bin = mix_bin && bin5. hash_index = hash_index + . enddo. data(bin_index) = . do strlen( mix_bin ) times. data(bin) = mix_bin+bin_index(). if bin_index mod = .
longitude_bin = longitude_bin && bin.
else.
latitude_bin = latitude_bin && bin.
endif. bin_index = bin_index + . enddo. data(longitude_left) = c_longitude_min.
data(longitude_right) = c_longitude_max.
data(latitude_left) = c_latitude_min.
data(latitude_right) = c_latitude_max. data(longitude_index) = . do strlen( longitude_bin ) times. data(bin_longitude) = longitude_bin+longitude_index(). get_tude(
exporting
i_left = longitude_left
i_right = longitude_right
i_bin = bin_longitude
importing
e_left = longitude_left
e_right = longitude_right
e_tude = longitude
). longitude_index = longitude_index + . enddo. data(latitude_index) = . do strlen( latitude_bin ) times. data(bin_latitude) = latitude_bin+latitude_index(). get_tude(
exporting
i_left = latitude_left
i_right = latitude_right
i_bin = bin_latitude
importing
e_left = latitude_left
e_right = latitude_right
e_tude = latitude
). latitude_index = latitude_index + . enddo. endmethod. method dec_to_bin. "ignore negative number
data(temp) = .
data(dec) = i_dec. while dec > .
temp = dec mod .
dec = dec / - temp.
r_bin = r_bin && conv char1( temp ).
endwhile. r_bin = reverse( r_bin ). endmethod. method encode. if length < .
return.
endif. if length > c_max_hash_length.
data(hash_length) = c_max_hash_length.
else.
hash_length = length.
endif. data(loop_times) = hash_length * / + . data: longitude_bin type string,
latitude_bin type string,
mix_bin type string. data(longitude_left) = c_longitude_min.
data(longitude_right) = c_longitude_max.
data(latitude_left) = c_latitude_min.
data(latitude_right) = c_latitude_max. do loop_times times. get_bin(
exporting
i_left = longitude_left
i_right = longitude_right
i_tude = longitude
importing
e_left = longitude_left
e_right = longitude_right
e_bin = data(longitude_bin_temp)
). get_bin(
exporting
i_left = latitude_left
i_right = latitude_right
i_tude = latitude
importing
e_left = latitude_left
e_right = latitude_right
e_bin = data(latitude_bin_temp)
). mix_bin = mix_bin && longitude_bin_temp && latitude_bin_temp. enddo. data(code_index) = . do hash_length times. data(offset) = code_index * .
data(bin) = mix_bin+offset(). r_geo_hash = r_geo_hash && value #(
mt_base32_code1[ decimals = bin_to_dec( i_bin = bin ) ]-base32 optional ). code_index = code_index + . enddo. endmethod. method get_bin. data(mid) = conv ty_tude( ( i_left + i_right ) / ). if i_tude <= mid.
e_bin = c_zero.
e_left = i_left.
e_right = mid.
else.
e_bin = c_one.
e_left = mid.
e_right = i_right.
endif. endmethod. method get_code_neighbor. data(table_descr) = cast cl_abap_tabledescr( cl_abap_tabledescr=>describe_by_data( i_table ) ). data(column_count) = lines(
cast cl_abap_structdescr( table_descr->get_table_line_type( ) )->components ). data(col_index) = . loop at i_table assigning field-symbol(<line>). data(row_index) = sy-tabix. col_index = . while col_index <= column_count. assign component col_index of structure <line> to field-symbol(<field>).
if sy-subrc = .
if <field> = i_member.
data(found) = abap_true.
exit.
endif.
endif. col_index = col_index + . endwhile. if found = abap_true.
exit.
endif. endloop. if found = abap_false.
return.
endif. types: begin of ty_direction,
row type i,
col type i,
end of ty_direction. data: direction_index_table type standard table of ty_direction. direction_index_table = value #(
( row = - col = - )
( row = - col = )
( row = - col = + )
( row = col = - )
( row = col = + )
( row = col = - )
( row = col = )
( row = col = + )
). data(row_count) = lines( i_table ). loop at direction_index_table assigning field-symbol(<direction_index>). data(row_result) = get_index( index = row_index offset = <direction_index>-row max_index = row_count ).
data(col_result) = get_index( index = col_index offset = <direction_index>-col max_index = column_count ). read table i_table assigning <line> index row_result.
if sy-subrc = .
assign component col_result of structure <line> to <field>.
if sy-subrc = .
r_table = value #( base r_table ( hash = <field> ) ).
endif.
endif. endloop. endmethod. method get_index. if abs( offset ) >= max_index.
return.
endif. r_index = index + offset. if r_index > max_index .
r_index = offset.
endif. if r_index <= .
r_index = max_index + r_index.
endif. endmethod. method get_tude. data(mid) = conv ty_tude( ( i_left + i_right ) / ). if i_bin = c_zero.
e_left = i_left.
e_right = mid.
e_tude = ( i_left + mid ) / .
else.
e_left = mid.
e_right = i_right.
e_tude = ( mid + i_right ) / .
endif. endmethod. method neighbors. if geohash is initial.
return.
endif. data(geohash_internal) = to_lower( geohash ). data(length) = strlen( geohash_internal ). data(offset) = length - . data(suffix) = geohash_internal+offset(). if length mod = .
data(code_table) = get_code_neighbor( i_table = mt_neighbors_even i_member = suffix ).
else.
code_table = get_code_neighbor( i_table = mt_neighbors_odd i_member = suffix ).
endif. data(prefix) = geohash_internal(offset). loop at code_table assigning field-symbol(<hash>).
neighbors = value #( base neighbors ( hash = prefix && <hash>-hash ) ).
endloop. endmethod. method validate. valid = abap_false. if geohash is initial .
return.
endif. if strlen( geohash ) > c_max_hash_length.
return.
endif. data(geohash_internal) = to_lower( geohash ). data(geohash_index) = . do strlen( geohash ) times. data(hash) = geohash_internal+geohash_index(). if not line_exists( mt_base32_code2[ base32 = hash ] ).
return.
endif. geohash_index = geohash_index + . enddo. valid = abap_true. endmethod.
endclass.
使用
本节包含一些使用示例。
编码
以浙江省丽水中学的经纬度坐标 (28.4751600000, 119.9314500000) 为例,
编码代码如下,
report ztest_qq1. data(hash) = zcl_geohash=>encode(
i_latitude = '28.4751600000'
i_longitude = '119.9314500000'
). cl_demo_output=>display( hash ).
可以得到结果wtj3cper。
默认的geohash长度是8位,也可以使用更长的编码提高精度,比如,
data(hash) = zcl_geohash=>encode(
i_latitude = '28.4751600000'
i_longitude = '119.9314500000'
i_length =
).
可以得到wtj3cperv6d9。12是geohash-abap支持的最大长度。
解码
对上面得到的geohash编码结果wtj3cperv6d9进行解码,
zcl_geohash=>decode(
exporting
i_geo_hash = hash
importing
e_latitude = data(latitude)
e_longitude = data(longitude)
). cl_demo_output=>display( latitude && ',' && longitude ).
得到的结果是 (28.475159956144, 119.931449834260) 可以看到是一个近似结果,和原值有微小的差距。
查询
将地点的geohash存储在数据库中之后,可以方便地用SQL中的like关键字,查找到附近的地点。
比如select * from table where geohash like 'wtj3cperv%'等等...
相邻区域编码
为了准群找到最近的地址,需要找到一个geohash所代表的区域的周围8个区域。
为什么?
比如,下图中边缘附近的红点。
黄色的点要比黑色的点更加靠近红点,但是由于黑点跟红点的GeoHash前缀匹配数目更多,因此直接用like查询,会得到黑点。
(参考文章:https://blog.csdn.net/youhongaa/article/details/78816700)
获取附近8个区域编码的方法:
data(neighbors) = zcl_geohash=>neighbors( 'wtj3cper' ).