Azure AI 搜尋的 OData 運算式語法參考

Azure AI 搜尋會使用 OData 運算式 作為整個 API 的參數。 最常用於和 $filter 參數的 $orderby OData 表達式。 這些表達式可能相當複雜,包含多個子句、函式和運算符。 不過,即使是像屬性路徑這樣的簡單 OData 運算式,也會在 Azure AI 搜尋 REST API 的許多部分中使用。 例如,路徑表達式可用來參考 API 中任何地方的複雜字段子欄位,例如在建議工具中列出子字段、評分函式、$select參數,甚至是 Lucene 查詢中的欄位搜尋時。

本文說明所有這些形式的 OData 運算式使用正式文法。 還有互動式 圖表 可協助以可視化方式探索文法。

正式文法

我們可以使用 EBNF (Extended Backus-Naur Form) 文法來描述 Azure AI 搜尋所支援的 OData 語言子集。 規則會以最複雜的表達式開頭列出「由上而下」,並將它們細分為更基本表達式。 頂端是對應至 Azure AI 搜尋服務 REST API 特定參數的文法規則:

  • $filter,由 filter_expression 規則定義。
  • $orderby,由 order_by_expression 規則定義。
  • $select,由 select_expression 規則定義。
  • 規則所 field_path 定義的欄位路徑。 整個 API 都會使用欄位路徑。 它們可以參考索引的最上層欄位,或具有一或多個 複雜欄位上階的子欄位

在EBNF是可 瀏覽的語法圖表 之後,可讓您以互動方式探索其規則之間的文法和關聯性。

/* Top-level rules */

filter_expression ::= boolean_expression

order_by_expression ::= order_by_clause(',' order_by_clause)*

select_expression ::= '*' | field_path(',' field_path)*

field_path ::= identifier('/'identifier)*


/* Shared base rules */

identifier ::= [a-zA-Z_][a-zA-Z_0-9]*


/* Rules for $orderby */

order_by_clause ::= (field_path | sortable_function) ('asc' | 'desc')?

sortable_function ::= geo_distance_call | 'search.score()'


/* Rules for $filter */

boolean_expression ::=
    collection_filter_expression
    | logical_expression
    | comparison_expression
    | boolean_literal
    | boolean_function_call
    | '(' boolean_expression ')'
    | variable

/* This can be a range variable in the case of a lambda, or a field path. */
variable ::= identifier | field_path

collection_filter_expression ::=
    field_path'/all(' lambda_expression ')'
    | field_path'/any(' lambda_expression ')'
    | field_path'/any()'

lambda_expression ::= identifier ':' boolean_expression

logical_expression ::=
    boolean_expression ('and' | 'or') boolean_expression
    | 'not' boolean_expression

comparison_expression ::= 
    variable_or_function comparison_operator constant | 
    constant comparison_operator variable_or_function

variable_or_function ::= variable | function_call

comparison_operator ::= 'gt' | 'lt' | 'ge' | 'le' | 'eq' | 'ne'


/* Rules for constants and literals */

constant ::=
    string_literal
    | date_time_offset_literal
    | integer_literal
    | float_literal
    | boolean_literal
    | 'null'

string_literal ::= "'"([^'] | "''")*"'"

date_time_offset_literal ::= date_part'T'time_part time_zone

date_part ::= year'-'month'-'day

time_part ::= hour':'minute(':'second('.'fractional_seconds)?)?

zero_to_fifty_nine ::= [0-5]digit

digit ::= [0-9]

year ::= digit digit digit digit

month ::= '0'[1-9] | '1'[0-2]

day ::= '0'[1-9] | [1-2]digit | '3'[0-1]

hour ::= [0-1]digit | '2'[0-3]

minute ::= zero_to_fifty_nine

second ::= zero_to_fifty_nine

fractional_seconds ::= integer_literal

time_zone ::= 'Z' | sign hour':'minute

sign ::= '+' | '-'

/* In practice integer literals are limited in length to the precision of
the corresponding EDM data type. */
integer_literal ::= sign? digit+

float_literal ::=
    sign? whole_part fractional_part? exponent?
    | 'NaN'
    | '-INF'
    | 'INF'

whole_part ::= integer_literal

fractional_part ::= '.'integer_literal

exponent ::= 'e' sign? integer_literal

boolean_literal ::= 'true' | 'false'


/* Rules for functions */

function_call ::=
    geo_distance_call |
    boolean_function_call

geo_distance_call ::=
    'geo.distance(' variable ',' geo_point ')'
    | 'geo.distance(' geo_point ',' variable ')'

geo_point ::= "geography'POINT(" lon_lat ")'"

lon_lat ::= float_literal ' ' float_literal

boolean_function_call ::=
    geo_intersects_call |
    search_in_call |
    search_is_match_call

geo_intersects_call ::=
    'geo.intersects(' variable ',' geo_polygon ')'

/* You need at least four points to form a polygon, where the first and
last points are the same. */
geo_polygon ::=
    "geography'POLYGON((" lon_lat ',' lon_lat ',' lon_lat ',' lon_lat_list "))'"

lon_lat_list ::= lon_lat(',' lon_lat)*

search_in_call ::=
    'search.in(' variable ',' string_literal(',' string_literal)? ')'

/* Note that it is illegal to call search.ismatch or search.ismatchscoring
from inside a lambda expression. */
search_is_match_call ::=
    'search.ismatch'('scoring')?'(' search_is_match_parameters ')'

search_is_match_parameters ::=
    string_literal(',' string_literal(',' query_type ',' search_mode)?)?

query_type ::= "'full'" | "'simple'"

search_mode ::= "'any'" | "'all'"

語法圖表

若要以可視化方式探索 Azure AI 搜尋所支援的 OData 語言文法,請嘗試互動式語法圖表:

另請參閱