Skip to main content
Version: 0.4.10

Syntax Overview

While MistQL's syntactic footprint is kept relatively small, it does take a little time to get used to.

Initial Context#

The simplest query in MistQL is @, which simply returns the data that was passed into MistQL. @ refers to the current context, which initially is set to data, making @ the identity query in MistQL. As an example, we could have the following:

QueryDataResult
@["arbitrary", "data"] ["arbitrary", "data"]

In the above, we operate the @ query over the data ["arbitrary", "data"], returning the exact same data structure we passed in directly as output.

Context population#

The keys of the current context @ are populated in the namespace as variables, such that they can be accessed via bare names:

QueryDataResult
foo{"foo": "bar"}"bar"

In the above, the data has a foo key, which gets populated as variable foo in the above MistQL query.

Basic Operators#

MistQL supports basic numbers and arithmetic expressions, following standard algebraic notation. For example, consider the following queries.

ExampleExpressionResult
Simple1 + 56
Complicated10.5 * (30 % 7) - 5 / 820.375

These execute using PEMDAS exactly as expected.

MistQL also allows basic logical operations, using the ||, &&, and ! operators, as well as the true and false literals.

ExpressionResult
true β”‚β”‚ falsetrue
true && falsefalse
!falsetrue

MistQL allows short-circuiting:

ExpressionResult
0 β”‚β”‚ "foo""foo"
[1, 2, 3] && nullnull

Finally, MistQL supports standard comparisons and equivalences: 50 > 10 outputs false, 10 > 50 outputs false, and 10 == 10 outputs true.

ExpressionResult
50 > 10true
50 < 10false
10 == 10true

MistQL equality always returns false for disparate types, e.g. if comparing two different types, equality of the values will always result in false. For precise equality semantics, please refer to the Types section of the reference docs.

Literals#

JSON literals are all valid MistQL:

{ "foo": 1, "bar": "baz" }{ "zeep": [1, "zoop", false], "null": null }

This will hold true for all JSON values regardless of whatever data was provided, becuase MistQL expressions are a strict superset of JSON.

Similarly to JavaScript, you can omit the quotes around key names that are valid identifiers in object literals:

{"type": "cat"} == {type: "cat"}

JSON literals can contain complicated subexpressions, for example:

[2 * 3, 5 * 10, { food: "hot" +"dog"}] == [6, 50, {food: "hotdog"}]

Accessing Fields on Objects#

There are 2 main ways of accessing a field on an object in MistQL, namely Bracket Notation and Dot Notation.

Bracket Notation#

Accessing fields on objects in MistQL is similar to JavaScript. To access fields on objects in MistQL, you use index notation with square brackets using strings:

QueryDataResult
@["hello"]{"hello": "world"}"world"
foo["bar"]["baz"]{"foo": {"bar": { "baz": true}}}true

Dot Notation#

If the name of the key on an object is alphanumeric, you can use dot notation to access object keys as well. For example, given data of {"foo": "bar"}, you can validly rewrite @["foo"] as @.foo.

QueryDataResult
@["foo"]{"foo": "bar"}"bar"
@.foo{"foo": "bar"}"bar"

Nested object can be accessed as a series of dots:

QueryDataResult
@.foo.bar.baz{"foo": {"bar": {"baz": 5}}}5

Accessing fields on objects is null coalescing by default: if you access missing keys on an object @.these.keys.dont.exist, the expression will evaluate to null without erroring out.

Accessing strings and arrays#

Strings and arrays are accessed exclusively by bracket notation using zero-indexed numbers:

"hello"[0] == "h""hello"[4] == "o"[100, 200, 300, 400][0] == 100[100, 200, 300, 400][3] == 400

Using negative numbers allows indexing from the end of a string or array:

"world"[-1] == "d""world"[-2] == "l"[100, 200, 300, 400][0] == 100[100, 200, 300, 400][3] == 400

Additionally, you can use the : separator in a pythonic style to allow for selecting of ranges:

"china"[1:] == "hina""spain"[:-2] == "spa""portugal"[1:4] == "ort"

Similarly, we get the following for arrays:

[10, 20, 30, 40, 50][3:] == [40, 50][10, 20, 30, 40, 50][:2] == [10, 20][10, 20, 30, 40, 50][1:-1] == [20, 30, 40]

Calling Functions#

Functions in MistQL are called in a lisp-like syntax, with function name and arguments separated by whitespace:

functionname argument1 argument2 argument3

For example, the following query sorts the provided array using the single-argument function sort:

sort [1, 5, 3] results in the array [1, 3, 5]

Multi-argument functions follow the same pattern. For example, the following calls the index function using 3 arguments, namely 1, 3, and "foobar":

index 1 3 "foobar"

Caveat: Function arguments can look quite complicated#

Function arguments can be bare expressions. For example, take the following (fairly hard to read) expression:

index 54 * 2 1 - 3 / 5 @

In the above expression, index is taking 3 arguments:

  • 54 * 2
  • 1 - 3 / 5
  • @ We can make this fact more clear by putting parentheses around each of the arguments. Putting explicit parentheses yields the following:
index (54 * 2) (1 - 3 / 5) (@)

Piping#

Piping provides an easy way to pass the result from one function along to another function in a clean, readable manner. When an expression is piped, the result of the expression on the left-hand side of the pipe is passed as the last argument to the function on the right-hand side of the pipe.

For example, these two queries behave identically:

split "," "dog,cat,walrus""dog,cat,walrus" | split ","

This allows users to chain together a sequence of functions in an easy-to-read manner:

events | filter type == "purchase" | groupby email | keys

MistQL's standard library is built around enabling piping as often as possible. If something can be expressed with pipes, it’s generally clearer to do so.

The apply idiom#

You may be tempted to write the following to multiply 10 and 2:

10 | @ * 2         WRONG

This will error out, saying that the result of @ + 2 is not a function -- and it's right! Since piping relies on being able to pass functions in without arguments, allowing the above syntax would form an ambiguity:

10 | @ * 2         WRONG10 | apply @ * 2   RIGHT

The Root Variable $#

The root variable $ is an object containing:

  1. A reference to the root context variable (via the @ key.)
  2. All builtin functions in MistQL

Method overwriting#

Functions and variables share the same namespace in MistQL, which can lead to naming conflicts if the current context has keys with the same name as builtins.

As an example, consider the query map over the data { "map": "data" }. Since the key map exists in the data and is populated in the namespace via context population, the identifier map refers to the value "data".

What this means is that if you were to attempt to use the map builtin in this particular case, we'd end up with a name conflict, as the current context overwrites the map identifier. Running the query count | count over the data {"count": [1, 2, 3]} would result in a error stating that arrays cannot be called as functions.

In order to ensure we can always maintain a reference back to the original count function (or any other for that matter), MistQL provides a $ variable which has every builtin function as a key.

Rewriting count | count as count | $.count in the above case (such that $.count refers to the builtin count variable) resolves the issue.

$.@

The root variable $ also contains a key @, which refers directly to the data the query operates over. This is useful in deeply nested expressions, where the @ variable has been overwritten.