21 KiB
(see also licenses for dev. deps.)
JSONPath Plus
Analyse, transform, and selectively extract data from JSON documents (and JavaScript objects).
jsonpath-plus
expands on the original specification to add some
additional operators and makes explicit some behaviors the original
did not spell out.
Try the browser demo or Runkit (Node).
Features
- Compliant with the original jsonpath spec
- Convenient additions or elaborations not provided in the original spec:
^
for grabbing the parent of a matching item~
for grabbing property names of matching items (as array)- Type selectors for obtaining:
- Basic JSON types:
@null()
,@boolean()
,@number()
,@string()
,@array()
,@object()
@integer()
- The compound type
@scalar()
(which also acceptsundefined
and non-finite numbers when querying JavaScript objects as well as all of the basic non-object/non-function types) @other()
usable in conjunction with a user-definedotherTypeCallback
- Non-JSON types that can nevertheless be used when querying
non-JSON JavaScript objects (
@undefined()
,@function()
,@nonFinite()
)
- Basic JSON types:
@path
/@parent
/@property
/@parentProperty
/@root
shorthand selectors within filters- Escaping
`
for escaping remaining sequence@['...']
/?@['...']
syntax for escaping special characters within property names in filters
- Documents
$..
(getting all parent components)
- ESM and UMD export formats
- In addition to queried values, can return various meta-information including paths or pointers to the value, as well as the parent object and parent property name (to allow for modification).
- Utilities for converting between paths, arrays, and pointers
- Option to prevent evaluations permitted in the original spec or supply a sandbox for evaluated values.
- Option for callback to handle results as they are obtained.
Benchmarking
jsonpath-plus
is consistently performant with both large and small datasets compared to other json querying libraries per json-querying-performance-testing. You can verify these findings by running the project yourself and adding more perf cases.
Install
npm install jsonpath-plus
Setup
Node.js
const {JSONPath} = require('jsonpath-plus');
const result = JSONPath({path: '...', json});
Browser
For browser usage you can directly include dist/index-browser-umd.js
; no Browserify
magic is necessary:
<!-- Polyfill recommended by Babel for items not covered for older
browsers in dist -->
<script src="node_modules/core-js-bundle/minified.js"></script>
<script src="node_modules/jsonpath-plus/dist/index-browser-umd.js"></script>
<script>
const result = JSONPath.JSONPath({path: '...', json: {}});
</script>
ESM (Modern browsers)
You may also use ES6 Module imports (for modern browsers):
<script type="module">
import {
JSONPath
} from './node_modules/jsonpath-plus/dist/index-browser-esm.js';
const result = JSONPath({path: '...', json: {}});
</script>
ESM (Bundlers)
Or if you are bundling your JavaScript (e.g., with Rollup), just use,
noting that mainFields
should include browser
for browser builds (for Node, the default, which
checks module
, should be fine):
import {JSONPath} from 'jsonpath-plus';
const result = JSONPath({path: '...', json});
Usage
The full signature available is:
const result = JSONPath([options,] path, json, callback, otherTypeCallback);
The arguments path
, json
, callback
, and otherTypeCallback
can alternatively be expressed (along with any other of the
available properties) on options
.
Note that result
will contain all items found (optionally
wrapped into an array) whereas callback
can be used if you
wish to perform some operation as each item is discovered, with
the callback function being executed 0 to N times depending
on the number of independent items to be found in the result.
See the docs below for more on JSONPath
's available arguments.
See also the API docs.
Properties
The properties that can be supplied on the options object or evaluate method (as the first argument) include:
- path (required) - The JSONPath expression as a (normalized or unnormalized) string or array
- json (required) - The JSON object to evaluate (whether of null, boolean, number, string, object, or array type).
- autostart (default: true) - If this is supplied as
false
, one may call theevaluate
method manually. - flatten (default: false) - Whether the returned array of results will be flattened to a single dimension array.
- resultType (default: "value") - Can be case-insensitive form of "value", "path", "pointer", "parent", or "parentProperty" to determine respectively whether to return results as the values of the found items, as their absolute paths, as JSON Pointers to the absolute paths, as their parent objects, or as their parent's property name. If set to "all", all of these types will be returned on an object with the type as key name.
- sandbox (default: {}) - Key-value map of variables to be available to code evaluations such as filtering expressions. (Note that the current path and value will also be available to those expressions; see the Syntax section for details.)
- wrap (default: true) - Whether or not to wrap the results
in an array. If
wrap
is set tofalse
, and no results are found,undefined
will be returned (as opposed to an empty array whenwrap
is set to true). Ifwrap
is set tofalse
and a single non-array result is found, that result will be the only item returned (not within an array). An array will still be returned if multiple results are found, however. To avoid ambiguities (in the case where it is necessary to distinguish between a result which is a failure and one which is an empty array), it is recommended to switch the default tofalse
. - preventEval (default: false) - Although JavaScript evaluation
expressions are allowed by default, for security reasons (if one is
operating on untrusted user input, for example), one may wish to
set this option to
true
to throw exceptions when these expressions are attempted. - parent (default: null) - In the event that a query could be made to return the root node, this allows the parent of that root node to be returned within results.
- parentProperty (default: null) - In the event that a query
could be made to return the root node, this allows the
parentProperty
of that root node to be returned within results. - callback (default: (none)) - If supplied, a callback will be
called immediately upon retrieval of an end point value. The three arguments
supplied will be the value of the payload (according to
resultType
), the type of the payload (whether it is a normal "value" or a "property" name), and a full payload object (with allresultType
s). - otherTypeCallback (default: <A function that throws an error
when @other() is encountered>) - In the current absence of JSON
Schema support, one can determine types beyond the built-in types by
adding the operator
@other()
at the end of one's query. If such a path is encountered, theotherTypeCallback
will be invoked with the value of the item, its path, its parent, and its parent's property name, and it should return a boolean indicating whether the supplied value belongs to the "other" type or not (or it may handle transformations and return false).
Instance methods
- evaluate(path, json, callback, otherTypeCallback) OR
evaluate({path: <path>, json: <json object>, callback:
<callback function>, otherTypeCallback:
<otherTypeCallback function>}) - This method is only
necessary if the
autostart
property is set tofalse
. It can be used for repeated evaluations using the same configuration. Besides the listed properties, the latter method pattern can accept any of the other allowed instance properties (except forautostart
which would have no relevance here).
Class properties and methods
- JSONPath.cache - Exposes the cache object for those who wish to preserve and reuse it for optimization purposes.
- JSONPath.toPathArray(pathAsString) - Accepts a normalized or
unnormalized path as string and converts to an array: for
example,
['$', 'aProperty', 'anotherProperty']
. - JSONPath.toPathString(pathAsArray) - Accepts a path array and
converts to a normalized path string. The string will be in a form
like:
$['aProperty']['anotherProperty][0]
. The JSONPath terminal constructions~
and^
and type operators like@string()
are silently stripped. - JSONPath.toPointer(pathAsArray) - Accepts a path array and
converts to a JSON Pointer.
The string will be in a form like:
/aProperty/anotherProperty/0
(with any~
and/
internal characters escaped as per the JSON Pointer spec). The JSONPath terminal constructions~
and^
and type operators like@string()
are silently stripped.
Syntax through examples
Given the following JSON, taken from http://goessner.net/articles/JsonPath/:
{
"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": 19.95
}
}
}
and the following XML representation:
<store>
<book>
<category>reference</category>
<author>Nigel Rees</author>
<title>Sayings of the Century</title>
<price>8.95</price>
</book>
<book>
<category>fiction</category>
<author>Evelyn Waugh</author>
<title>Sword of Honour</title>
<price>12.99</price>
</book>
<book>
<category>fiction</category>
<author>Herman Melville</author>
<title>Moby Dick</title>
<isbn>0-553-21311-3</isbn>
<price>8.99</price>
</book>
<book>
<category>fiction</category>
<author>J. R. R. Tolkien</author>
<title>The Lord of the Rings</title>
<isbn>0-395-19395-8</isbn>
<price>22.99</price>
</book>
<bicycle>
<color>red</color>
<price>19.95</price>
</bicycle>
</store>
Please note that the XPath examples below do not distinguish between
retrieving elements and their text content (except where useful for
comparisons or to prevent ambiguity). Note: to test the XPath examples
(including 2.0 ones), this demo
may be helpful (set to xml
or xml-strict
).
XPath | JSONPath | Result | Notes |
---|---|---|---|
/store/book/author | $.store.book[*].author | The authors of all books in the store | Can also be represented without the $. as store.book[*].author (though this is not present in the original spec); note that some character literals ($ and @ ) require escaping, however |
//author | $..author | All authors | |
/store/* | $.store.* | All things in store, which are its books (a book array) and a red bicycle (a bicycle object). | |
/store//price | $.store..price | The price of everything in the store. | |
//book[3] | $..book[2] | The third book (book object) | |
//book[last()] | ..book\[(@.length-1)]<br> ..book[-1:] |
The last book in order. | To access a property with a special character, utilize [(@['...'])] for the filter (this particular feature is not present in the original spec) |
//book[position()<3] | ..book\[0,1]<br> ..book[:2] |
The first two books | |
//book/*[self::category|self::author] or //book/(category,author) in XPath 2.0 | $..book[0][category,author] | The categories and authors of all books | |
//book[isbn] | $..book[?(@.isbn)] | Filter all books with an ISBN number | To access a property with a special character, utilize [?@['...']] for the filter (this particular feature is not present in the original spec) |
//book[price<10] | $..book[?(@.price<10)] | Filter all books cheaper than 10 | |
//*[name() = 'price' and . != 8.95] | $..*[?(@property === 'price' && @ !== 8.95)] | Obtain all property values of objects whose property is price and which does not equal 8.95 | With the bare @ allowing filtering objects by property value (not necessarily within arrays), you can add ^ after the expression to get at the object possessing the filtered properties |
/ | $ | The root of the JSON object (i.e., the whole object itself) | To get a literal $ (by itself or anywhere in the path), you must use the backtick escape |
//*/*|//*/*/text() | $..* | All Elements (and text) beneath root in an XML document. All members of a JSON structure beneath the root. | |
//* | $.. | All Elements in an XML document. All parent components of a JSON structure including root. | This behavior was not directly specified in the original spec |
//*[price>19]/.. | $..[?(@.price>19)]^ | Parent of those specific items with a price greater than 19 (i.e., the store value as the parent of the bicycle and the book array as parent of an individual book) | Parent (caret) not present in the original spec |
/store/*/name() (in XPath 2.0) | $.store.*~ | The property names of the store sub-object ("book" and "bicycle"). Useful with wildcard properties. | Property name (tilde) is not present in the original spec |
/store/book[not(. is /store/book[1])] (in XPath 2.0) | $.store.book[?(@path !== "$['store']['book'][0]")] | All books besides that at the path pointing to the first | @path not present in the original spec |
//book[parent::*/bicycle/color = "red"]/category | $..book[?(@parent.bicycle && @parent.bicycle.color === "red")].category | Grabs all categories of books where the parent object of the book has a bicycle child whose color is red (i.e., all the books) | @parent is not present in the original spec |
//book/*[name() != 'category'] | $..book.*[?(@property !== "category")] | Grabs all children of "book" except for "category" ones | @property is not present in the original spec |
//book[position() != 1] | $..book[?(@property !== 0)] | Grabs all books whose property (which, being that we are reaching inside an array, is the numeric index) is not 0 | @property is not present in the original spec |
/store/*/*[name(parent::*) != 'book'] | $.store.*[?(@parentProperty !== "book")] | Grabs the grandchildren of store whose parent property is not book (i.e., bicycle's children, "color" and "price") | @parentProperty is not present in the original spec |
//book[count(preceding-sibling::*) != 0]/*/text() | $..book.*[?(@parentProperty !== 0)] | Get the property values of all book instances whereby the parent property of these values (i.e., the array index holding the book item parent object) is not 0 | @parentProperty is not present in the original spec |
//book[price = /store/book[3]/price] | $..book[?(@.price === @root.store.book[2].price)] | Filter all books whose price equals the price of the third book | @root is not present in the original spec |
//book/../*[. instance of element(*, xs:decimal)] (in XPath 2.0) | $..book..*@number() | Get the numeric values within the book array | @number(), the other basic types (@boolean(), @string()), other low-level derived types (@null(), @object(), @array()), the JSONSchema-added type, @integer(), the compound type @scalar() (which also accepts undefined and non-finite numbers for JavaScript objects as well as all of the basic non-object/non-function types), the type, @other(), to be used in conjunction with a user-defined callback (see otherTypeCallback ) and the following non-JSON types that can nevertheless be used with JSONPath when querying non-JSON JavaScript objects (@undefined(), @function(), @nonFinite()) are not present in the original spec |
//book/*[name() = 'category' and matches(., 'tion$')] (XPath 2.0) | $..book.*[?(@property === "category" && @.match(/TION$/i))] | All categories of books which match the regex (end in 'TION' case insensitive) | @property is not present in the original spec. |
//book/[matches(name(), 'bn$')]/parent:: (XPath 2.0) | $..book.*[?(@property.match(/bn$/i))]^ | All books which have a property matching the regex (end in 'TION' case insensitive) | @property is not present in the original spec. Note: Uses the parent selector ^ at the end of the expression to return to the parent object; without the parent selector, it matches the two isbn key values. |
` (e.g., `$ to match a property literally named $ ) |
Escapes the entire sequence following (to be treated as a literal) | ` is not present in the original spec; to get a literal backtick, use an additional backtick to escape |
Any additional variables supplied as properties on the optional "sandbox" object option are also available to (parenthetical-based) evaluations.
Potential sources of confusion for XPath users
- In JSONPath, a filter expression, in addition to its
@
being a reference to its children, actually selects the immediate children as well, whereas in XPath, filter conditions do not select the children but delimit which of its parent nodes will be obtained in the result. - In JSONPath, array indexes are, as in JavaScript, 0-based (they begin from 0), whereas in XPath, they are 1-based.
- In JSONPath, equality tests utilize (as per JavaScript) multiple equal signs whereas in XPath, they use a single equal sign.
Ideas
- Support OR outside of filters (as in XPath
|
) and grouping. - Create syntax to work like XPath filters in not selecting children?
- Allow option for parentNode equivalent (maintaining entire chain of parent-and-parentProperty objects up to root)
Development
Running the tests on Node:
npm test
For in-browser tests:
- Serve the js/html files:
npm run browser-test
- Visit http://localhost:8082/test/.