JSONPath: Query expressions for JSON

1.1. Terminology

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶

The grammatical rules in this document are to be interpreted as ABNF, as described in [RFC5234]. ABNF terminal values in this document define Unicode scalar values rather than their UTF-8 encoding. For example, the Unicode PLACE OF INTEREST SIGN (U+2318) would be defined in ABNF as %x2318.¶

Functions are referred to using the function name followed by a pair of parentheses, as in fname().¶

The terminology of [RFC8259] applies except where clarified below. The terms "Primitive" and "Structured" are used to group different kinds of values as in Section 1 of [RFC8259]; JSON Objects and Arrays are structured, all other values are primitive. Definitions for "Object", "Array", "Number", and "String" remain unchanged. Importantly "object" and "array" in particular do not take on a generic meaning, such as they would in a general programming context.¶

Additional terms used in this document are defined below.¶

Value:

As per [RFC8259], a structure conforming to the generic data model of JSON, i.e., composed of constituents such as structured values, namely JSON objects and arrays, and primitive data, namely numbers and text strings as well as the special values null, true, and false. [RFC8259] focuses on the textual representation of JSON values and does not fully define the value abstraction assumed here.¶

Member:

A name/value pair in an object. (A member is not itself a value.)¶

Name:

The name (a string) in a name/value pair constituting a member. This is also used in [RFC8259], but that specification does not formally define it. It is included here for completeness.¶

Element:

A value in a JSON array.¶

Index:

An integer that identifies a specific element in an array.¶

Query:

Short name for a JSONPath expression.¶

Query Argument:

Short name for the value a JSONPath expression is applied to. (Also used for actual parameters of function-expressions.)¶

Location:

the position of a value within the query argument. This can be thought of as a sequence of names and indexes navigating to the value through the objects and arrays in the query argument, with the empty sequence indicating the query argument itself. A location can be represented as a Normalized Path (defined below).¶

Node:

The pair of a value along with its location within the query argument.¶

Root Node:

The unique node whose value is the entire query argument.¶

Root Node Identifier:

The expression $ which refers to the root node of the query argument.¶

Current Node Identifier:

The expression @ which refers to the current node in the context of the evaluation of a filter expression (described later).¶

Children (of a node):

If the node is an array, the nodes of its elements. If the node is an object, the nodes of its member values. If the node is neither an array nor an object, it has no children.¶

Descendants (of a node):

The children of the node, together with the children of its children, and so forth recursively. More formally, the "descendants" relation between nodes is the transitive closure of the "children" relation.¶

Depth (of a descendant node within a value):

The number of ancestors of the node within the value. The root node of the value has depth zero, the children of the root node have depth one, their children have depth two, and so forth.¶

Nodelist:

A list of nodes. While a nodelist can be represented in JSON, e.g. as an array, this document does not require or assume any particular representation.¶

Parameter:

Formal parameter (of a function) that can take a function argument (an actual parameter) in a function-expression.¶

Normalized Path:

A form of JSONPath expression that identifies a node in a value by providing a query that results in exactly that node. Each node in a query argument is identified by exactly one Normalized Path (we say, the Normalized Path is "unique" for that node), and, to be a Normalized Path for a specific query argument, the Normalized Path needs to identify exactly one node. Similar to, but syntactically different from, a JSON Pointer [RFC6901]. Note: This definition is based on the syntactical definition in Section 2.7; JSONPath expressions that identify a node in a value but do not conform to that syntax are not Normalized Paths.¶

Unicode Scalar Value:

Any Unicode [UNICODE] code point except high-surrogate and low-surrogate code points. In other words, integers in either of the inclusive base 16 ranges 0 to D7FF and E000 to 10FFFF. JSONPath queries are sequences of Unicode scalar values.¶

Segment:

One of the constructs which select children ([<selectors>]) or descendants (..[<selectors>]) of an input value.¶

Selector:

A single item within a segment that takes the input value and produces a nodelist consisting of child nodes of the input value.¶

Singular Query:

A JSONPath expression built from segments that have been syntactically restricted in a certain way (Section 2.3.5.1) so that, regardless of the input value, the expression produces a nodelist containing at most one node. Note: JSONPath expressions that always produce a singular nodelist but do not conform to the syntax in Section 2.3.5.1 are not Singular Queries.¶

1.2. History

This document is based on Stefan Gössner's popular JSONPath proposal dated 2007-02-21 [JSONPath-orig], builds on the experience from the widespread deployment of its implementations, and provides a normative specification for it.¶

Appendix B describes how JSONPath was inspired by XML's XPath [XPath].¶

JSONPath was intended as a light-weight companion to JSON implementations in programming languages such as PHP and JavaScript, so instead of defining its own expression language, like XPath did, JSONPath delegated parts of a query to the underlying runtime, e.g., JavaScript's eval() function. As JSONPath was implemented in more environments, JSONPath expressions became decreasingly portable. For example, regular expression processing was often delegated to a convenient regular expression engine.¶

This document aims to remove such implementation-specific dependencies and serve as a common JSONPath specification that can be used across programming languages and environments. This means that backwards compatibility is not always achieved; a design principle of this document is to go with a "consensus" between implementations even if it is rough, as long as that does not jeopardize the objective of obtaining a usable, stable JSON query language.¶

The term JSONPath was chosen because of the XPath inspiration and also because the outcome of a query consists of paths identifying nodes in the JSON query argument.¶

1.3. JSON Values

The JSON value a JSONPath query is applied to is, by definition, a valid JSON value. A JSON value is often constructed by parsing a JSON text.¶

The parsing of a JSON text into a JSON value and what happens if a JSON text does not represent valid JSON are not defined by this document. Sections 4 and 8 of [RFC8259] identify specific situations that may conform to the grammar for JSON texts but are not interoperable uses of JSON, as they may cause unpredictable behavior. This document does not attempt to define predictable behavior for JSONPath queries in these situations.¶

Specifically, the "Semantics" subsections of Sections 2.3.1, 2.3.2, 2.3.5, and 2.5.2 describe behavior that becomes unpredictable when the JSON value for one of the objects under consideration was constructed out of JSON text that exhibits multiple members for a single object that share the same member name ("duplicate names", see Section 4 of [RFC8259]). Also, selecting a child by name (Section 2.3.1) and comparing strings (Section 2.3.5.2.2 in Section 2.3.5) assume these strings are sequences of Unicode scalar values, becoming unpredictable if they are not (Section 8.2 of [RFC8259]).¶

1.4. Overview of JSONPath Expressions

A JSONPath expression is applied to a JSON value, known as the query argument. The output is a nodelist.¶

A JSONPath expression consists of an identifier followed by a series of zero or more segments each of which contains one or more selectors.¶

$['store']['book'][0]['title']

$.store.book[0].title

$.store.book[[email protected] < 10].title

1.5. JSONPath Examples

This section is informative. It provides examples of JSONPath expressions.¶

The examples are based on the simple JSON value shown in Figure 1, representing a bookstore (that also has a bicycle).¶

{ "store": {
    "book": [
      { "category": "reference",
        "author": "Nigel Rees",
        "title": "Sayings of the Century",
        "price": 8.95
      },
      { "category": "fiction",
        "author": "Evelyn Waugh",
        "title": "Sword of Honour",
        "price": 12.99
      },
      { "category": "fiction",
        "author": "Herman Melville",
        "title": "Moby Dick",
        "isbn": "0-553-21311-3",
        "price": 8.99
      },
      { "category": "fiction",
        "author": "J. R. R. Tolkien",
        "title": "The Lord of the Rings",
        "isbn": "0-395-19395-8",
        "price": 22.99
      }
    ],
    "bicycle": {
      "color": "red",
      "price": 399
    }
  }
}

Table 2 shows some JSONPath queries that might be applied to this example and their intended results.¶

2.1. Overview

A JSONPath expression is a string which, when applied to a JSON value, the query argument, selects zero or more nodes of the argument and outputs these nodes as a nodelist.¶

A query MUST be encoded using UTF-8. The grammar for queries given in this document assumes that its UTF-8 form is first decoded into Unicode scalar values as described in [RFC3629]; implementation approaches that lead to an equivalent result are possible.¶

A string to be used as a JSONPath query needs to be well-formed and valid. A string is a well-formed JSONPath query if it conforms to the ABNF syntax in this document. A well-formed JSONPath query is valid if it also fulfills all semantic requirements posed by this document, which are:¶

1. Integer numbers in the JSONPath query that are relevant to the JSONPath processing (e.g., index values and steps) MUST be within the range of exact integer values defined in I-JSON (see Section 2.2 of [RFC7493]), namely within the interval [−(2⁵³)+1, (2⁵³)−1].¶
2. Uses of function extensions MUST be well-typed, as described in Section 2.4.¶

A JSONPath implementation MUST raise an error for any query which is not well-formed and valid. The well-formedness and the validity of JSONPath queries are independent of the JSON value the query is applied to. No further errors relating to the well-formedness and the validity of a JSONPath query can be raised during application of the query to a value. This clearly separates well-formedness/validity errors in the query from mismatches that may actually stem from flaws in the data.¶

Mismatches between the structure expected by a valid query and the structure found in the data can lead to empty query results, which may be unexpected and indicate bugs in either. JSONPath implementations might therefore want to provide diagnostics to the application developer that aid in finding the cause of empty results.¶

Obviously, an implementation can still fail when executing a JSONPath query, e.g., because of resource depletion, but this is not modeled in this document. However, the implementation MUST NOT silently malfunction. Specifically, if a valid JSONPath query is evaluated against a structured value whose size is too large to process the query correctly (for instance requiring the processing of numbers that fall outside the range of exact values), the implementation MUST provide an indication of overflow.¶

(Readers familiar with the HTTP error model may be reminded of 400 type errors when pondering well-formedness and validity, while resource depletion and related errors are comparable to 500 type errors.)¶

jsonpath-query      = root-identifier segments
segments            = *(S segment)

B                   = %x20 /    ; Space
                      %x09 /    ; Horizontal tab
                      %x0A /    ; Line feed or New line
                      %x0D      ; Carriage return
S                   = *B        ; optional blank space

2.2. Root Identifier

root-identifier     = "$"

{"k": "v"}

2.3. Selectors

Selectors appear only inside child segments (Section 2.5.1) and descendant segments (Section 2.5.2).¶

A selector produces a nodelist consisting of zero or more children of the input value.¶

There are various kinds of selectors which produce children of objects, children of arrays, or children of either objects or arrays.¶

selector            = name-selector  /
                      wildcard-selector /
                      slice-selector /
                      index-selector /
                      filter-selector

The syntax and semantics of each kind of selector are defined below.¶

name-selector       = string-literal

string-literal      = %x22 *double-quoted %x22 /     ; "string"
                      %x27 *single-quoted %x27       ; 'string'

double-quoted       = unescaped /
                      %x27      /                    ; '
                      ESC %x22  /                    ; \"
                      ESC escapable

single-quoted       = unescaped /
                      %x22      /                    ; "
                      ESC %x27  /                    ; \'
                      ESC escapable

ESC                 = %x5C                           ; \  backslash

unescaped           = %x20-21 /                      ; see RFC 8259
                         ; omit 0x22 "
                      %x23-26 /
                         ; omit 0x27 '
                      %x28-5B /
                         ; omit 0x5C \
                      %x5D-D7FF /   ; skip surrogate code points
                      %xE000-10FFFF

escapable           = %x62 / ; b BS backspace U+0008
                      %x66 / ; f FF form feed U+000C
                      %x6E / ; n LF line feed U+000A
                      %x72 / ; r CR carriage return U+000D
                      %x74 / ; t HT horizontal tab U+0009
                      "/"  / ; / slash (solidus) U+002F
                      "\"  / ; \ backslash (reverse solidus) U+005C
                      (%x75 hexchar) ;  uXXXX      U+XXXX

hexchar             = non-surrogate /
                      (high-surrogate "\" %x75 low-surrogate)
non-surrogate       = ((DIGIT / "A"/"B"/"C" / "E"/"F") 3HEXDIG) /
                       ("D" %x30-37 2HEXDIG )
high-surrogate      = "D" ("8"/"9"/"A"/"B") 2HEXDIG
low-surrogate       = "D" ("C"/"D"/"E"/"F") 2HEXDIG

HEXDIG              = DIGIT / "A" / "B" / "C" / "D" / "E" / "F"

{
  "o": {"j j": {"k.k": 3}},
  "'": {"@": 2}
}

wildcard-selector   = "*"

{
  "o": {"j": 1, "k": 2},
  "a": [5, 3]
}

index-selector      = int                        ; decimal integer

int                 = "0" /
                      (["-"] DIGIT1 *DIGIT)      ; - optional
DIGIT1              = %x31-39                    ; 1-9 non-zero digit

["a","b"]

slice-selector      = [start S] ":" S [end S] [":" [S step ]]

start               = int       ; included in selection
end                 = int       ; not included in selection
step                = int       ; default: 1

FUNCTION Normalize(i, len):
  IF i >= 0 THEN
    RETURN i
  ELSE
    RETURN len + i
  END IF

FUNCTION Bounds(start, end, step, len):
  n_start = Normalize(start, len)
  n_end = Normalize(end, len)

  IF step >= 0 THEN
    lower = MIN(MAX(n_start, 0), len)
    upper = MIN(MAX(n_end, 0), len)
  ELSE
    upper = MIN(MAX(n_start, -1), len-1)
    lower = MIN(MAX(n_end, -1), len-1)
  END IF

  RETURN (lower, upper)

IF step > 0 THEN

  i = lower
  WHILE i < upper:
    SELECT a(i)
    i = i + step
  END WHILE

ELSE if step < 0 THEN

  i = upper
  WHILE lower < i:
    SELECT a(i)
    i = i + step
  END WHILE

END IF

["a", "b", "c", "d", "e", "f", "g"]

filter-selector     = "?" S logical-expr

logical-expr        = logical-or-expr
logical-or-expr     = logical-and-expr *(S "||" S logical-and-expr)
                        ; disjunction
                        ; binds less tightly than conjunction
logical-and-expr    = basic-expr *(S "&&" S basic-expr)
                        ; conjunction
                        ; binds more tightly than disjunction

basic-expr          = paren-expr /
                      comparison-expr /
                      test-expr

paren-expr          = [logical-not-op S] "(" S logical-expr S ")"
                                        ; parenthesized expression
logical-not-op      = "!"               ; logical NOT operator

test-expr           = [logical-not-op S]
                     (filter-query / ; existence/non-existence
                      function-expr) ; LogicalType or NodesType
filter-query        = rel-query / jsonpath-query
rel-query           = current-node-identifier segments
current-node-identifier = "@"

comparison-expr     = comparable S comparison-op S comparable
literal             = number / string-literal /
                      true / false / null
comparable          = literal /
                      singular-query / ; singular query value
                      function-expr    ; ValueType
comparison-op       = "==" / "!=" /
                      "<=" / ">=" /
                      "<"  / ">"

singular-query      = rel-singular-query / abs-singular-query
rel-singular-query  = current-node-identifier singular-query-segments
abs-singular-query  = root-identifier singular-query-segments
singular-query-segments = *(S (name-segment / index-segment))
name-segment        = ("[" name-selector "]") /
                      ("." member-name-shorthand)
index-segment       = "[" index-selector "]"

number              = (int / "-0") [ frac ] [ exp ] ; decimal number
frac                = "." 1*DIGIT                  ; decimal fraction
exp                 = "e" [ "-" / "+" ] 1*DIGIT    ; decimal exponent
true                = %x74.72.75.65                ; true
false               = %x66.61.6c.73.65             ; false
null                = %x6e.75.6c.6c                ; null

{
  "obj": {"x": "y"},
  "arr": [2, 3]
}

{
  "a": [3, 5, 1, 2, 4, 6,
        {"b": "j"},
        {"b": "k"},
        {"b": {}},
        {"b": "kilo"}
       ],
  "o": {"p": 1, "q": 2, "r": 3, "s": 5, "t": {"u": 6}},
  "e": "f"
}

2.4. Function Extensions

Beyond the filter expression functionality defined in the preceding subsections, JSONPath defines an extension point that can be used to add filter expression functionality: "Function Extensions".¶

This section defines the extension point as well as some function extensions that use this extension point. While these mechanisms are designed to use the extension point, they are an integral part of the JSONPath specification and are expected to be implemented like any other integral part of this specification.¶

A function extension defines a registered name (see Section 3.2) that can be applied to a sequence of zero or more arguments, producing a result. Each registered function name is unique.¶

A function extension MUST be defined such that its evaluation is side-effect free, i.e., all possible orders of evaluation and choices of short-circuiting or full evaluation of an expression containing it MUST lead to the same result. (Note: memoization or logging are not side effects in this sense as they are visible at the implementation level only — they do not influence the result of the evaluation.)¶

function-name       = function-name-first *function-name-char
function-name-first = LCALPHA
function-name-char  = function-name-first / "_" / DIGIT
LCALPHA             = %x61-7A  ; "a".."z"

function-expr       = function-name "(" S [function-argument
                         *(S "," S function-argument)] S ")"
function-argument   = literal /
                      filter-query / ; (includes singular-query)
                      logical-expr /
                      function-expr

Any function expressions in a query must be well-formed (by conforming to the above ABNF) and well-typed, otherwise the JSONPath implementation MUST raise an error (see Section 2.1). To define which function expressions are well-typed, a type system is first introduced.¶

$[?length(@.authors) >= 5]

$[?count(@.*.author) >= 5]

$[?match(@.date, "1974-05-..")]

$[?search(@.author, "[BR]ob")]

$[?value(@..color) == "red"]

2.5. Segments

For each node in an input nodelist, segments apply one or more selectors to the node and concatenate the results of each selector into per-input-node nodelists, which are then concatenated in the order of the input nodelist to form a single segment result nodelist.¶

It turns out that the more segments there are in a query, the greater the depth in the input value of the nodes of the resultant nodelist:¶

• A query with N segments, where N >= 0, produces a nodelist consisting of nodes at depth in the input value of N or greater.¶
• A query with N segments, where N >= 0, all of which are child segments (Section 2.5.1), produces a nodelist consisting of nodes precisely at depth N in the input value.¶

There are two kinds of segment: child segments and descendant segments.¶

segment             = child-segment / descendant-segment

The syntax and semantics of each kind of segment are defined below.¶

child-segment       = bracketed-selection /
                      ("."
                       (wildcard-selector /
                        member-name-shorthand))

bracketed-selection = "[" S selector *(S "," S selector) S "]"

member-name-shorthand = name-first *name-char
name-first          = ALPHA /
                      "_"   /
                      %x80-D7FF /   ; skip surrogate code points
                      %xE000-10FFFF
name-char           = DIGIT / name-first

DIGIT               = %x30-39              ; 0-9
ALPHA               = %x41-5A / %x61-7A    ; A-Z / a-z

["a", "b", "c", "d", "e", "f", "g"]

descendant-segment  = ".." (bracketed-selection /
                            wildcard-selector /
                            member-name-shorthand)

{
  "o": {"j": 1, "k": 2},
  "a": [5, 3, [{"j": 4}, {"k": 6}]]
}

2.6. Semantics of null

Note: JSON null is treated the same as any other JSON value: it is not taken to mean "undefined" or "missing".¶

{"a": null, "b": [null], "c": [{}], "null": 1}

2.7. Normalized Paths

A Normalized Path is a unique representation of the location of a node in a value which uniquely identifies the node in the value. Specifically, a Normalized Path is a JSONPath query with restricted syntax (defined below), e.g., $['book'][3], which when applied to the value results in a nodelist consisting of just the node identified by the Normalized Path. Note: a Normalized Path represents the identity of a node in a specific value. There is precisely one Normalized Path identifying any particular node in a value.¶

A nodelist may be represented compactly in JSON as an array of strings, where the strings are Normalized Paths.¶

Normalized Paths provide a predictable format that simplifies testing and post-processing of nodelists, e.g., to remove duplicate nodes. Normalized Paths are used in this document as result paths in examples.¶

Normalized Paths use the canonical bracket notation, rather than dot notation.¶

Single quotes are used in Normalized Paths to delimit string member names. This reduces the number of characters that need escaping when Normalized Paths appear in double quote-delimited strings, e.g., in JSON texts.¶

Certain characters are escaped in Normalized Paths, in one and only one way; all other characters are unescaped.¶

Note: Normalized Paths are singular queries, but not all singular queries are Normalized Paths. For example, $[-3] is a singular query, but is not a Normalized Path. The Normalized Path equivalent to $[-3] would have an index equal to the array length minus 3. (The array length must be at least 3 if $[-3] is to identify a node.)¶

normalized-path      = root-identifier *(normal-index-segment)
normal-index-segment = "[" normal-selector "]"
normal-selector      = normal-name-selector / normal-index-selector
normal-name-selector = %x27 *normal-single-quoted %x27 ; 'string'
normal-single-quoted = normal-unescaped /
                       ESC normal-escapable
normal-unescaped     =    ; omit %x0-1F control codes
                       %x20-26 /
                          ; omit 0x27 '
                       %x28-5B /
                          ; omit 0x5C \
                       %x5D-D7FF /   ; skip surrogate code points
                       %xE000-10FFFF

normal-escapable     = %x62 / ; b BS backspace U+0008
                       %x66 / ; f FF form feed U+000C
                       %x6E / ; n LF line feed U+000A
                       %x72 / ; r CR carriage return U+000D
                       %x74 / ; t HT horizontal tab U+0009
                       "'" /  ; ' apostrophe U+0027
                       "\" /  ; \ backslash (reverse solidus) U+005C
                       (%x75 normal-hexchar)
                                       ; certain values u00xx U+00XX
normal-hexchar       = "0" "0"
                       (
                          ("0" %x30-37) / ; "00"-"07"
                             ; omit U+0008-U+000A BS HT LF
                          ("0" %x62) /    ; "0b"
                             ; omit U+000C-U+000D FF CR
                          ("0" %x65-66) / ; "0e"-"0f"
                          ("1" normal-HEXDIG)
                        )
normal-HEXDIG        = DIGIT / %x61-66    ; "0"-"9", "a"-"f"
normal-index-selector = "0" / (DIGIT1 *DIGIT)
                        ; non-negative decimal integer

Since there can only be one Normalized Path identifying a given node, the syntax stipulates which characters are escaped and which are not. So the definition of normal-hexchar is designed for hex escaping of characters which are not straightforwardly printable, for example U+000B LINE TABULATION, but for which no standard JSON escape, such as \n, is available.¶

3.1. Registration of Media Type application/jsonpath

IANA is requested to register the following media type [RFC6838]:¶

Type name:

application¶

Subtype name:

jsonpath¶

Required parameters:

N/A¶

Optional parameters:

N/A¶

Encoding considerations:

binary (UTF-8)¶

Security considerations:

See the Security Considerations section of RFCXXXX.¶

Interoperability considerations:

N/A¶

Published specification:

RFCXXXX¶

Applications that use this media type:

Applications that need to convey queries in JSON data¶

Fragment identifier considerations:

N/A¶

Additional information:

Deprecated alias names for this type:

N/A¶

Magic number(s):

N/A¶

File extension(s):

N/A¶

Macintosh file type code(s):

N/A¶

Person & email address to contact for further information: [email protected]¶

Intended usage:

COMMON¶

Restrictions on usage:

N/A¶

Author:

JSONPath WG¶

Change controller:

IETF¶

Provisional registration? (standards tree only):

no¶

3.2. Function Extensions

This specification defines a new "Function Extensions sub-registry" in a new "JSONPath Parameters registry", with the policy "expert review" (Section 4.5 of [BCP26]).¶

Each entry in the sub-registry must include:¶

Function Name:

a lower case ASCII [STD80] string that starts with a letter and can contain letters, digits and underscore characters afterwards ([a-z][_a-z0-9]*). No other entry in the sub-registry can have the same function name.¶

Brief description:

a brief description¶

Parameters:

A comma-separated list of zero or more declared types, one for each of the arguments expected for this function extension¶

Result:

The declared type of the result for this function extension¶

Change Controller:

(see Section 2.3 of [BCP26])¶

Reference:

a reference document that provides a description of the function extension¶

Initial entries in this sub-registry are as listed in Table 19; the entries in the Column "Change Controller" all have the value "IETF" and the entries in the column "Reference" all have the value "Section 2.4 of RFCXXXX":¶

4.1. Attack Vectors on JSONPath Implementations

Historically, JSONPath has often been implemented by feeding parts of the query to an underlying programming language engine, e.g., JavaScript's eval() function. This approach is well known to lead to injection attacks and would require perfect input validation to prevent these attacks (see Section 12 of [RFC8259] for similar considerations for JSON itself). Instead, JSONPath implementations need to implement the entire syntax of the query without relying on the parsers of programming language engines.¶

Attacks on availability may attempt to trigger unusually expensive runtime performance exhibited by certain implementations in certain cases. (See Section 10 of [RFC8949] for issues in hash-table implementations, and Section 8 of [I-D.draft-ietf-jsonpath-iregexp] for performance issues in regular expression implementations.) Implementers need to be aware that good average performance is not sufficient as long as an attacker can choose to submit specially crafted JSONPath queries or query arguments that trigger surprisingly high, possibly exponential, CPU usage or, for example via a naive recursive implementation of the descendant segment, stack overflow. Implementations need to have appropriate resource management to mitigate these attacks.¶

4.2. Attack Vectors on How JSONPath Queries are Formed

JSONPath queries are often not static, but formed from variables that provide index values, member names, or values to compare with in a filter expression. These variables need to be validated (e.g., only allowing specific constructs such as .name to be formed when the given values allow that) and translated (e.g., by escaping string delimiters). Not performing these validations and translations correctly can lead to unexpected failures, which can lead to Availability, Confidentiality, and Integrity breaches, in particular if an adversary has control over the values (e.g., by entering them into a Web form). The resulting class of attacks, injections (e.g., SQL injections), is consistently found among the top causes of application security vulnerabilities and requires particular attention.¶

4.3. Attacks on Security Mechanisms that Employ JSONPath

Where JSONPath is used as a part of a security mechanism, attackers can attempt to provoke unexpected or unpredictable behavior, or take advantage of differences in behavior between JSONPath implementations.¶

Unexpected or unpredictable behavior can arise from a query argument with certain constructs described as unpredictable by [RFC8259]. Predictable behavior can be expected, except in relation to the ordering of objects, for any query argument conforming with [RFC7493].¶

Other attacks can target the behavior of underlying technologies such as UTF-8 (see Section 10 of [RFC3629]) and the Unicode character set.¶

B.1. JSONPath and XPath

JSONPath expressions apply to JSON values in the same way as XPath expressions are used in combination with an XML document. JSONPath uses $ to refer to the root node of the query argument, similar to XPath's / at the front.¶

JSONPath expressions move further down the hierarchy using dot notation ($.store.book[0].title) or the bracket notation ($['store']['book'][0]['title']), a lightweight/limited, and a more heavyweight syntax replacing XPath's / within query expressions.¶

Both JSONPath and XPath use * for a wildcard. The descendant operators, starting with .., borrowed from [E4X], are similar to XPath's //. The array slicing construct [start:end:step] is unique to JSONPath, inspired by [SLICE] from ECMASCRIPT 4.¶

Filter expressions are supported via the syntax ?<logical-expr> as in¶

$.store.book[[email protected] < 10].title

Table 20 extends Table 1 by providing a comparison with similar XPath concepts.¶

For further illustration, Table 21 shows some XPath expressions and their JSONPath equivalents.¶

XPath has a lot more functionality (location paths in unabbreviated syntax, operators and functions) than listed in this comparison. Moreover, there are significant differences in how the subscript operator works in XPath and JSONPath:¶

• Square brackets in XPath expressions always operate on the node set resulting from the previous path fragment. Indices always start at 1.¶
• With JSONPath, square brackets operate on each of the nodes in the nodelist resulting from the previous query segment. Array indices always start at 0.¶