| title | date | description |
|---|---|---|
Instrumentation |
2018-09-09 12:52:46 +1000 |
Instrumentation allows you to inject code that can observe the execution of a query and also change the runtime behaviours |
The graphql.execution.instrumentation.Instrumentation interface allows you to inject code that can observe the
execution of a query and also change the runtime behaviour.
The primary use case for this is to allow say performance monitoring and custom logging, but it could be used for many purposes.
When you build the Graphql object you can specify what Instrumentation to use (if any).
GraphQL.newGraphQL(schema)
.instrumentation(new TracingInstrumentation())
.build();An implementation of Instrumentation needs to implement the "begin" step methods that represent the execution of a graphql query.
Each step must give back a non-null graphql.execution.instrumentation.InstrumentationContext object which will be called back
when the step completes, and will be told that it succeeded or failed with a Throwable.
The following is a basic custom Instrumentation that measures overall execution time and puts it into a stateful object.
class CustomInstrumentationState implements InstrumentationState {
private Map<String, Object> anyStateYouLike = new HashMap<>();
void recordTiming(String key, long time) {
anyStateYouLike.put(key, time);
}
}
class CustomInstrumentation extends SimplePerformantInstrumentation {
@Override
public InstrumentationState createState() {
//
// instrumentation state is passed during each invocation of an Instrumentation method
// and allows you to put stateful data away and reference it during the query execution
//
return new CustomInstrumentationState();
}
@Override
public @Nullable InstrumentationContext<ExecutionResult> beginExecution(InstrumentationExecutionParameters parameters, InstrumentationState state) {
long startNanos = System.nanoTime();
return new SimpleInstrumentationContext<ExecutionResult>() {
@Override
public void onCompleted(ExecutionResult result, Throwable t) {
((CustomInstrumentationState) state).recordTiming(parameters.getQuery(), System.nanoTime() - startNanos);
}
};
}
@Override
public @NotNull DataFetcher<?> instrumentDataFetcher(DataFetcher<?> dataFetcher, InstrumentationFieldFetchParameters parameters, InstrumentationState state) {
//
// this allows you to intercept the data fetcher used to fetch a field and provide another one, perhaps
// that enforces certain behaviours or has certain side effects on the data
//
return dataFetcher;
}
@Override
public @NotNull CompletableFuture<ExecutionResult> instrumentExecutionResult(ExecutionResult executionResult, InstrumentationExecutionParameters parameters, InstrumentationState state) {
//
// this allows you to instrument the execution result somehow. For example the Tracing support uses this to put
// the `extensions` map of data in place
//
return CompletableFuture.completedFuture(executionResult);
}
}You can c
9447
ombine multiple Instrumentation objects together using the graphql.execution.instrumentation.ChainedInstrumentation class which
accepts a list of Instrumentation objects and calls them in that defined order.
List<Instrumentation> chainedList = new ArrayList<>();
chainedList.add(new FooInstrumentation());
chainedList.add(new BarInstrumentation());
ChainedInstrumentation chainedInstrumentation = new ChainedInstrumentation(chainedList);
GraphQL.newGraphQL(schema)
.instrumentation(chainedInstrumentation)
.build();graphql.execution.instrumentation.tracing.TracingInstrumentation is an Instrumentation implementation that creates tracing information
about the query that is being executed.
It follows the Apollo proposed tracing format defined at https://github.com/apollographql/apollo-tracing.
A detailed tracing map will be created and placed in the extensions section of the result.
So given a query like
query {
hero {
name
friends {
name
}
}
}It would return a result like
{
"data": {
"hero": {
"name": "R2-D2",
"friends": [
{
"name": "Luke Skywalker"
},
{
"name": "Han Solo"
},
{
"name": "Leia Organa"
}
]
}
},
"extensions": {
"tracing": {
"version": 1,
"startTime": "2017-08-14T23:13:39.362Z",
"endTime": "2017-08-14T23:13:39.497Z",
"duration": 135589186,
"execution": {
"resolvers": [
{
"path": [
"hero"
],
"parentType": "Query",
"returnType": "Character",
"fieldName": "hero",
"startOffset": 105697585,
"duration": 79111240
},
{
"path": [
"hero",
"name"
],
"parentType": "Droid",
"returnType": "String",
"fieldName": "name",
"startOffset": 125010028,
"duration": 20213
},
{
"path": [
"hero",
"friends"
],
"parentType": "Droid",
"returnType": "[Character]",
"fieldName": "friends",
"startOffset": 133352819,
"duration": 7927560
},
{
"path": [
"hero",
"friends",
0,
"name"
],
"parentType": "Human",
"returnType": "String",
"fieldName": "name",
"startOffset": 134105887,
"duration": 6783
},
{
"path": [
"hero",
"friends",
1,
"name"
],
"parentType": "Human",
"returnType": "String",
"fieldName": "name",
"startOffset": 134725922,
"duration": 7016
},
{
"path": [
"hero",
"friends",
2,
"name"
],
"parentType": "Human",
"returnType": "String",
"fieldName": "name",
"startOffset": 134875089,
"duration": 6342
}
]
}
}
}
}
graphql.execution.instrumentation.fieldvalidation.FieldValidationInstrumentation is an Instrumentation implementation that
can be used to validate fields and their arguments before query execution. If errors are returned during this process then
the query execution is aborted and the errors will be in the query result.
You can make you own custom implementation of FieldValidation or you can use the SimpleFieldValidation class
to add simple per field checks rules.
ExecutionPath fieldPath = ExecutionPath.parse("/user");
FieldValidation fieldValidation = new SimpleFieldValidation()
.addRule(fieldPath, new BiFunction<FieldAndArguments, FieldValidationEnvironment, Optional<GraphQLError>>() {
@Override
public Optional<GraphQLError> apply(FieldAndArguments fieldAndArguments, FieldValidationEnvironment environment) {
String nameArg = fieldAndArguments.getFieldArgument("name");
if (nameArg.length() > 255) {
return Optional.of(environment.mkError("Invalid user name", fieldAndArguments));
}
return Optional.empty();
}
});
FieldValidationInstrumentation instrumentation = new FieldValidationInstrumentation(
fieldValidation
);
GraphQL.newGraphQL(schema)
.instrumentation(instrumentation)
.build();graphql.analysis.MaxQueryComplexityInstrumentation in an Instrumentation implementation than can be used to abort a query if the total number of data
fields queried exceeds the defined limit.
GraphQL.newGraphQL(schema)
.instrumentation(new MaxQueryComplexityInstrumentation(10))
.build();With a query like:
query {
hero {
name
friends {
name
homeWorld {
name
climate
}
species {
name
}
}
homeWorld {
name
climate
}
}
}Would return a result like:
{
"errors": [
{
"message": "maximum query complexity exceeded 12 > 10",
"extensions": {
"classification": "ExecutionAborted"
}
}
]
}graphql.analysis.MaxQueryDepthInstrumentation in an Instrumentation implementation than can be used to abort a query if the total depth of
the query exceeds the defined limit.
GraphQL.newGraphQL(schema)
.instrumentation(new MaxQueryDepthInstrumentation(4))
.build();With a query like:
query {
hero {
name
friends {
name
friends {
name
friends {
name
}
}
}
}
}Would return a result like:
{
"errors": [
{
"message": "maximum query depth exceeded 4 > 3",
"extensions": {
"classification": "ExecutionAborted"
}
}
]
}