Learn best practices and considerations for migrating from GitHubScope's REST API to GitHubScope's GraphQL API.
In this article
...
...
Example: Getting the data you need and nothing more
...
...
Table of Contents |
---|
Differences in API logic
Migrating from REST to GraphQL represents a significant shift in API logic. The differences between REST as a style and GraphQL as a specification make it difficult—and often undesirable—to replace REST API calls with GraphQL API queries on a one-to-one basis. We've included specific examples of migration below.
To migrate your code from the REST API to the REST API to the GraphQL API:
Review the GraphQL the GraphQL spec
Review GitHubScope's GraphQL schema
Consider how any existing code you have currently interacts with the GitHub Scope REST API
Use Global Node IDs to reference objects between API versions
Significant advantages of GraphQL include:
Here are examples of each.
Example #1: Getting the data you need, and nothing more
A single REST API call retrieves a list of your organization's members:
Code Block |
---|
curl -v https://apicms.githubscopear.com/orgsapi/:orgv2/membersusers |
The REST payload contains excessive data if your goal is to retrieve only member names and links to avatars. However, a GraphQL query returns only what you specify:
Code Block |
---|
query { organization(login:"github")viewer { membersWithRole(first: 100) {organization { edgesmembers(first: 10) { nodeedges { namenode { avatarUrl name } } } } } |
Consider another example: retrieving a list of pull requests and checking if each one is mergeable. A call to the REST API retrieves a list of pull requests and their summary representations:
Code Block |
---|
curl -v https://api.github.com/repos/:owner/:repo/pulls |
Determining if a pull request is mergeable requires retrieving each pull request individually for its detailed representation (a large payload) and checking whether its mergeable
attribute is true or false:
Code Block |
---|
curl -v https://api.github.com/repos/:owner/:repo/pulls/:number |
With GraphQL, you could retrieve only the number
and mergeable
attributes for each pull request:
Code Block |
---|
query avatar { repository(owner:"octocat", name:"Hello-World") { pullRequests(last: 10) { fileUrl edges { } node { } number mergeable } } } } } |
Example #2: Nesting
Querying with nested fields lets you replace multiple REST calls with fewer GraphQL queries. For example, retrieving a pull request scenario along with its commits, non-review comments, and reviews ar contents and associated groups using the REST API requires four separate calls:
Code Block |
---|
curl -v https://apicms.githubscopear.com/reposapi/:ownerv2/:repo/pullsscenarios/:numberid curl -v https://apicms.githubscopear.com/reposapi/:ownerv2/:reposcenarios/pulls/:numberid/commitsattachments curl -v https://apicms.githubscopear.com/reposapi/:ownerv2/:reposcenarios/issues/:numberid/commentsauthoring_groups curl -v https://apicms.githubscopear.com/reposapi/:ownerv2/:reposcenarios/pulls/:numberid/reviewsgroups |
Using the GraphQL API, you can retrieve the data with a single query using nested fields:
Code Block |
---|
{ repositorynode(ownerid: "octocat", name: "Hello-WorldID_OF_SCENARIO") { pullRequest(number: 1) {... on Scenario name commits(first: 10) { content edges { fileUrl node { fileContentType commit { oid message fileUrlExpiresAt } } } } comments(first: 10) groups { edges { node nodes { body id author { name login } } } } reviews(first: 10) { edges { node { state } } } } } } |
You can also extend the power of this query by substituting by substituting a variable for variable for the pull request number.
Example #3: Strong typing
GraphQL schemas are strongly typed, making data handling safer.
Consider an example of adding a comment to an issue or pull request renaming a scenario using a GraphQL mutationGraphQL mutation, and mistakenly specifying an integer rather than a string for the value of clientMutationId
name
:
Code Block |
---|
mutation { addComment(input:{clientMutationIdupdateScenario(id: 1234, subjectId: "MDA6SXNzdWUyMjcyMDA2MTT=", bodyinput:{name: "Looks good to me!"1234}) { clientMutationIdid commentEdgename { groups node { nodes { body repository { id name nameWithOwner } issue { number } } } } } |
Executing this query returns errors specifying the expected types for the operation:
Code Block |
---|
{ "data": null, "errors": [ { "message": "Argument 'input' on Field 'addCommentname' has an invalid value. Expected type 'AddCommentInputString!'.", "locations": [ { "line": 3, "column": 3 } ] }, { "message": "Argument 'clientMutationId' on InputObject 'AddCommentInput' has an invalid value. Expected type 'String'.", "locations": [ { "line": 3, "column": 20 } ] } ] } |
Wrapping 1234
in quotes transforms the value from an integer into a string, the expected type:
Code Block |
---|
mutation { addComment(input:{clientMutationId: "1234", subjectId: "MDA6SXNzdWUyMjcyMDA2MTT=", body: "Looks good to me!"}) { clientMutationId commentEdge { node { body repository { id name nameWithOwner } issue { number } } } } } |