GraphQL, gqlXPath, and Enhanced Transformer

I recently worked on a project that required manipulating GraphQL documents for queries and mutations.
I discovered that certain techniques for working with JSON and XML could also be useful when working with GraphQL documents.

GraphQL documents are explicitly used for sending queries and mutations to the GraphQL service rather than primarily for data storage and exchange like JSON.

To modify a GraphQL document, the following steps need to be followed:

1. Traverse through the GraphQL document.
2. Identify the relevant GraphQL node or nodes.
3. Manipulate the GraphQL node or nodes as required.
4. Create a new GraphQL document with the manipulated node or nodes.
5. Pass the new GraphQL document to the GraphQL server to execute the query or mutation.

Let's exclude point (5) from our discussion, as it can be accomplished through multiple tools and code snippets.

gqlex will an advanced solution for the aforementioned statements.

gqlex will offer a new way to select GraphQL node

This article will explain how ti use gqlex to traverse a GraphQL document, select the relevant nodes, and manipulate the document according to your needs.
We'll also play with the code and elaborate on relevant use cases.

The library is part of the Intuit open-source community named gqlex: gqlex-opensource-intuit, which aims to provide a polyglot implementation for path-selection and transformation for GraphQL with advanced techniques to optimize the traversal over GraphQL document.

Traverse over GraphQL document

Every element in GraphQL, such as document, query, mutation, fragment, inline fragment, directive, etc., is derived from a node with unique attributes and behavior.

gqlex uses the observer pattern in which the GraphQL document acts as a subject, traverses the entire document and notifies relevant observers on the node and context.

This observer pattern separates the traversal over the GraphQL document from the execution part that the observer consumer code would like to perform.

Selection of Node

XML has XPath.
JSON has JSONPath.

Now ... GraphQL document has gqlXPath

gqlXPath can be used to navigate through nodes in a GraphQL document.
gqlXPath uses path expressions to select nodes or node on the GraphQL document.

Behind the scenes, the gqlXPath utilizes the traversal module
and selects the node according to the required expression.

gqlXPath syntax

gqlXPath uses path expressions to select nodes in GraphQL document. The node is selected by following a path or steps.

The most useful path expressions are listed below:

Expression	Description
//	Path prefix: Select all nodes from the root node and use a slash as a separator between path elements.
/	Path prefix: Select the first node from the root node and use a slash as a separator between path elements. The range is not supported when the first node is selected.
{x:y}/	Path prefix, Select path node(s), between a range of x path node and y path node (inclusion), use of slash as a separator between path elements. x and y are positive integers. All nodes are selected if no x and y are not set.
{:y}//	Path prefix, Select path node(s), between a range of first path node result to y, using a slash as a separator between path elements. x and y are positive integers. All nodes are selected if no x and y are not set.
{x:}/	Path prefix, Select path node(a), between range of x path node result to the end of path nodes result, use slash as a separator between path elements. x and y are integers. if no x and y are not set, select all path nodes.
{:}//	Path prefix, Select node(s) from the root node, use of slash as a separator between path elements.
...	Support of relative path "any" selection e.g. {x:y}//a/b/.../f any can be set anywhere in the gqlXPath, except at the end of the gqlXPath, You can set many any as you request, this will help you while selecting node in large GraphQL structure, so you won't be required to mention/build the entire node structure.

The library also provides an equivalent code-named SyntaxPath that provides gqlXPath expression abilities use of code, mainly used by automation code.

Transformer

The transformer provides the ability to transform (Manipulate) GraphQL document simply.
The transformer uses the abilities provided by gqlex such as: gqlXPath, SyntaxPath etc.

The gqlex provides the following transform methods:

1. Add Children - Add children node to selected GraphQL node or nodes
2. Add Sibling - Add sibling node to selected GraphQL node or nodes
3. Duplicate - Duplicate selected node by duplication number; multi nodes cannot be duplicated
4. Remove Children - Remove selected nodes or node
5. Update name - Update selected node names or node names; for inline fragments, it will update the typeCondition name
6. Update alias value - update field alias value and fragment spread alias value.

Code Play

Let's start with traversal over the GrqphQL document.

Start by creating my observer. Let's name it StringBuilderObserver.

The observer will append the GraphQL node to some StringBuilder.

This way, we achieve separation of concern:

• Traverse over the GraphQL document nodes
• Append node values with StringBuilder ....

public class StringBuilderObserver implements TraversalObserver {
    private final List<StringBuilderElem> stringBuilderElems = new ArrayList<>();

    private final boolean isIgnoreCollection = true;
    @Override
    public void updateNodeEntry(Node node, Node parentNode, Context context, ObserverAction observerAction) {

        String  message = "";
        DocumentElementType documentElementType = context.getDocumentElementType();
        switch (documentElementType) {

            case DOCUMENT:
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "Document", documentElementType.name());

                break;

            case DIRECTIVE:
                message = MessageFormat.format("Name : {0} ||  Type : {1}", ((Directive) node).getName(), documentElementType.name());
                break;
            case FIELD:
                Field field = (Field) node;
                message = MessageFormat.format("Name : {0} || Alias : {1} ||  Type : {2}",
                        field.getName(),
                        field.getAlias(),
                        documentElementType.name());
                break;
            case OPERATION_DEFINITION:
                message = MessageFormat.format("Name : {0} ||  Type : {1}",
                        ((OperationDefinition) node).getOperation().toString(), documentElementType.name());
                break;
            case INLINE_FRAGMENT:
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "InlineFragment",
                        documentElementType.name());

                break;
            case FRAGMENT_DEFINITION:
                message = MessageFormat.format("Name : {0} ||  Type : {1}",
                        ((FragmentDefinition) node).getName(), documentElementType.name());

                break;
            case FRAGMENT_SPREAD:
                message = MessageFormat.format("Node : {0} ||  Type : {1}", ((FragmentSpread) node).getName(), documentElementType.name());
                break;
            case VARIABLE_DEFINITION:
                message = MessageFormat.format("Name : {0} || Default Value : {1} ||  Type : {2}",
                        ((VariableDefinition) node).getName(), ((VariableDefinition) node).getDefaultValue(), documentElementType.name());

                break;
            case ARGUMENT:
                message = MessageFormat.format("Name : {0} || Value : {1} ||  Type : {2}",
                        ((Argument) node).getName(),
                        ((Argument) node).getValue(),
                        documentElementType.name());
                break;
            case ARGUMENTS:
                if(isIgnoreCollection) return;
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "Arguments", documentElementType.name());
                break;
            case SELECTION_SET:
                if(isIgnoreCollection) return;
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "SelectionSet", documentElementType.name());
                break;
            case VARIABLE_DEFINITIONS:
                if(isIgnoreCollection) return;
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "VariableDefinitions", documentElementType.name());
                break;
            case DIRECTIVES:
                if(isIgnoreCollection) return;
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "Directives", documentElementType.name());
                break;
            case DEFINITIONS:
                if(isIgnoreCollection) return;
                message = MessageFormat.format("Node : {0} ||  Type : {1}", "Definitions", documentElementType.name());

                break;
        }

        if(Strings.isNullOrEmpty(message)){
            return;
        }

        stringBuilderElems.add(new StringBuilderElem(message, context.getLevel()));

        levels.add(context.getLevel());
        //spaces++;
    }
    private final List<Integer> levels = new ArrayList<>();

    public String getGqlBrowsedPrintedString() {
        return getStringAs(true);
    }

    private String getStringAs(boolean isIdent) {
        int j=0;
        StringBuilder stringBuilder = new StringBuilder();
        for (StringBuilderElem stringBuilderElem : stringBuilderElems) {
            j++;
            String spaceStr = "";
            if( isIdent) {
                for (int i = 0; i < stringBuilderElem.getDepth(); i++) {
                    spaceStr += " ";
                }
                stringBuilder.append(spaceStr + stringBuilderElem.getName() + "\n");
            }else{
                stringBuilder.append( stringBuilderElem.getName() + (j+1<stringBuilderElems.size()? " " : "") );
            }
        }
        return stringBuilder.toString();
    }

    public String getGqlBrowsedString(){
        return getStringAs(false);
    }

//    int spaces = 0;
    @Override
    public void updateNodeExit( Node node,Node parentNode, Context context, ObserverAction observerAction) {
    }
}

GqlTraversal traversal = new GqlTraversal();

StringBuilderObserver gqlStringBuilderObserver = new StringBuilderObserver();

traversal.getGqlTraversalObservable().addObserver(gqlStringBuilderObserver);
traversal.traverse(file);

System.out.println( gqlStringBuilderObserver.getGqlBrowsedString());

After we saw how the traversal is working, let's digg with some examples of gqlXPath selection GraphQL nodes

gqlXPath defines an expression language as defined above, in addition, the expression language contains more terms to familiar with, Element Name and types abbreviations.
Why it's important: A GraphQL document is more than a structure similar to JSON; It also provides a DSL exposed by the GraphQL server and GraphQL language.
The types and the element name will assist the gqlXPath in selecting the exact node or nodes.

Element Names

element_name	Description
type=	Select element by type abbreviate
name=	Select element by name
alias=	Select element by alias name

Available types and abbreviation for _type _element_name

Type abbreviate	Description
doc	DOCUMENT
frag	FRAGMENT_DEFINITION
direc	DIRECTIVE
fld	FIELD
mutation	MUTATION_DEFINITION
query	OPERATION_DEFINITION
infrag	INLINE_FRAGMENT
var	VARIABLE_DEFINITION
arg	ARGUMENT

Let's practice the gqlXPath expression,

GraphQL Document

In this document, we have 2 nodes named: 'name', but of different types: argument and field.

query {
    Instrument(Name: "1234") {
        Reference {
            Name
            title
        }
    }
}

• Select all nodes (double slash) 'name', which is an argument type //query/Instrument/name[type=arg]

• Select first node (single slash) 'name', which is an argument type /query/Instrument/name[type=arg]

• //query/.../name[type=arg] Same as query //query/Instrument/name[type=arg]

• Select of 'name' node, which is a field under Reference, reside under Instrument, reside under query //query/Instrument/Reference/name

• //query/Instrument/.../name Same as //query/Instrument/Reference/name

• //.../name Same as //query/Instrument/Reference/name

GraphQl Document

query Hero($episode: Episode, $withFriends: Boolean!) {
  hero(episode: $episode) {
    name
    friends @include(if: $withFriends) {
      name
    }
    friends @include(if: $withFriends) {
      name
    }
    friends @include(if: $withFriends) {
      name
    }
    friends @include(if: $withFriends) {
      name
    }
  }
}

• Select all query nodes named hero //query[name=hero]

• Select first query (single slash) node named hero /query[name=hero]

• Select all nodes named 'name', reside under friends node //query[name=hero]/hero/friends/name

• Select all nodes named 'name' within a range of index 0 and index 2 (inclusion), reside under friends node {0:2}//query[name=hero]/hero/friends/name

• Select node named 'name', reside under any node, which reside under hero node /query[name=hero]/hero/.../name

• Select $withFriends variable reside directly under the root node named hero //query[name=hero]/withFriends[type=var]

• Select include directive first node resides under friends, which resides under root query node named hero /query[name=hero]/hero/friends/include[type=direc]

• Select the 'if' argument node, reside under the @include directive //.../include[type=direc]/if[type=arg]

• Select episode variable //.../episode[type=var]

How to use gqlXPath in the code:

SelectorFacade selectorFacade = new SelectorFacade();

String queryString = Files.readString(file.toPath());

// query {  Instrument(id: "1234") }
GqlNodeContext select = selectorFacade.select(queryString, "//query/Instrument /   Reference  /");

Use of SyntaxPath

String queryString = Files.readString(file.toPath());

SyntaxBuilder gqlexBuilder = new SyntaxBuilder();

gqlexBuilder.appendQuery();
gqlexBuilder.appendField("Instrument");
gqlexBuilder.appendField("Reference");

// query {  Instrument(id: "1234") }
GqlNodeContext select = selectorFacade.select(queryString, gqlexBuilder.build());

And finally we will dwell on an example that will illustrate the use of gqlXPath node selection GraphQL manipulation (AKA transform).

Here a mutation GraphQL document

mutation CreateReviewForEpisode($ep: Episode!, $review: ReviewInput!) {
  createReview(episode: $ep, review: $review) {
    stars
    commentary
    stars
    commentary
  }
}

The following Java code will select and transform the mutation document,

String queryString = Files.readString(sourceFile.toPath());

TransformBuilder transformBuilder = new TransformBuilder();
transformBuilder.addChildrenNode("//mutation[name=CreateReviewForEpisode]/createReview/stars",new Field("child_of_stars"))
        .addSiblingNode("//mutation[name=CreateReviewForEpisode]/createReview/stars",new Field("sibling_of_stars"))
        .updateNodeName("//mutation[name=CreateReviewForEpisode]/createReview/stars","star_new_name")
        .removeNode("//mutation[name=CreateReviewForEpisode]/createReview/commentary")
        .duplicateNode("//mutation[name=CreateReviewForEpisode]/createReview/sibling_of_stars", 10);

TransformExecutor transformExecutor = new TransformExecutor(transformBuilder);

RawPayload rawPayload = new RawPayload();
rawPayload.setQueryValue(queryString);

RawPayload executeRawPayload = transformExecutor.execute(rawPayload);

Description:

add new children node named: child_of_stars under gqlXPath: //mutation[name=CreateReviewForEpisode]/createReview/stars

Add new sibling node named: sibling_of_stars under gqlXPath selected node: //mutation[name=CreateReviewForEpisode]/createReview/stars

Set new node name: star_new_name value to the selected node by gqlXPath:
//mutation[name=CreateReviewForEpisode]/createReview/stars

Remove selected node by gqlXPath //mutation[name=CreateReviewForEpisode]/createReview/commentary

Duplicate selected gqlXPath node 10 times //mutation[name=CreateReviewForEpisode]/createReview/sibling_of_stars

Use TransformBuilder to build the transform plan, with selected node use of gqlXPath and the command to execute.
The transform plan is loaded to the TransformExecutor with the GraphQL payload.

The execution will result in a new GraphQL document,

mutation CreateReviewForEpisode($ep: Episode!, $review: ReviewInput!) {
  createReview(episode: $ep, review: $review) {
    sibling_of_stars
    star_new_name {
      child_of_stars
    }
    star_new_name {
      child_of_stars
    }
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
    sibling_of_stars
  }
}

Last example, will demonstrate the manipulation of directive from include to exclude

Here the query GraphQL document,

query Hero($episode: Episode, $withFriends: Boolean!) {
  hero(episode: $episode) {
    name
    friends **@include**(if: $withFriends) {
      name
    }
  }
}

Here the code:

        String queryString = Files.readString(file.toPath());

        // query {  Instrument(id: "1234") }
        GqlNodeContext includeDirectiveNode = selectorFacade.selectSingle(queryString, "//query[name=hero]/hero/friends/include[type=direc]");

        assertNotNull(includeDirectiveNode);

        assertTrue(includeDirectiveNode.getType().equals(DocumentElementType.DIRECTIVE));
        System.out.println("\nBefore manipulation:\n\n" + queryString);

        // Node newNode = new Field("new_name");
        Node excludeDirectiveNode = TransformUtils.updateNodeName(includeDirectiveNode, "exclude");

        String newGqlValue = gqlexWriter.writeToString(excludeDirectiveNode);

        System.out.println("\nAfter manipulation:\n\n" + newGqlValue);

        GqlNodeContext excludeUpdateNode = selectorFacade.selectSingle(newGqlValue, "//query[name=hero]/hero/friends/exclude[type=direc]");

        assertTrue(excludeUpdateNode.getType().equals(DocumentElementType.DIRECTIVE));

query Hero($episode: Episode, $withFriends: Boolean!) {
  hero(episode: $episode) {
    name
    friends @exclude(if: $withFriends) {
      name
    }
  }
}

gqlex Use Cases

The gqlex can be used during base code when the developer is required to enrich the GraphQL document with more fields while querying the server for data, or during manipulation of data in the server so the code can articulate the relevant fields to manipulate in the service side.
gqlex can also be utilized during integration or E2E testing, generating synthetic GraphQL data with a high velocity and managed solution.

I elaborate the following use cases with more details:

Synthetic GraphQL document creation
Testing, mostly the integration testing part, will demand the ability to query the GraphQL server with different queries and mutations.

Of course, the developer can maintain large lists of example files to send to the server or to find and replace the
relevant string in GraphQL document, but it is a cumbersome solution, hard to maintain etc.

gqlex gives you the ability to manipulate the query or the mutation with ease and use configuration-wise to list your
plan of action (gqlXPath, Transform Commands, and Argument to execute], and execute the plan:

Configuration
  Plan
    steps
      step 1
         gqlXPath (String)
         transform_commands
           transform_command
               command
               argument_object_definition
      step n
         gqlXPath (String)
         transform_commands
           transform_command
               command
               argument_object_definition
  origin_file_to_manipulate

Configuration config = read_configuration_plan(plan_file);

config_verification()

build_plan -> ... use of TransformBuilder

new_graphql_document = execute_plan -> ... use of TransformExecuter

gqlex gives you versatility and the ability to produce synthetic GraphQL data and verify the integrity of a GraphQL service.

Articulate GraphQL document on-the-fly

Sometimes, you may need to dynamically build a query or mutation based on business logic or configuration and send it to the GraphQL server.
The gqlex library can assist with this.

The gqlex library only allows manipulation of the GraphQL skeleton file, not its creation.

The developer can create the skeleton GraphQL file, store the file in resource folder.
skeleton file, means file with structure but without field only.
with the of syntaxPath, gqlex can assemble the gqlXPath and set the plan strategy on the fly,
then use of TransformBuilder to build the plan, then use the TransformExecutor to run the plan.