GitHub GraphQL API about replacing existing REST API

Currently, our GitHub API are all REST API, and the REST API has the problem of over-fetching or multiple requests.

We want to use the GraphQL API to retrieve the key information we need. With the data obtained this way, we can calculate related metrics.

The GraphQL API endpoint is https://api.github.com/graphql. All requests must be HTTP POST requests with application/json encoded bodies.

Pass the token in your GraphQL request using the Authorization HTTP header with a value token <token>.

• Replaced api

GET：https://api.github.com/octocat

• Method local

GitHubFeignClient.verifyToken

query {
  viewer {
    login
  }
}

• Once query cost total node

1

• The meaning of response status

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'content-type: application/json' \
  --data '{"query": "{viewer {login}}"}'

• Replaced api

GET：https://api.github.com/repos/{repository}/branches/{branchName}

• Method local

GitHubFeignClient.verifyCanReadTargetBranch

query {
  repository(owner: string, name: string) {
    ref(qualifiedName: "refs/heads/branchName: string") {
      name
    }
  }
}

• Once query cost total node

1

• Required value and demo

if you want to get main branch for au-heartbeat/Heartbeat, you can use as following.

query {
  repository(owner: "au-heartbeat", name: "Heartbeat") {
    ref(qualifiedName: "refs/heads/main") {
      name
    }
  }
}

• The meaning of response status

if branch verification success，you will get response with ref have branch name.

status: 200

response:
{
	"data": {
		"repository": {
			"ref": {
				"name": "main"
			}
		}
	}
}

if the token can’t get target branch，you will get response with ref is null.

status: 200

response:
{
	"data": {
		"repository": {
			"ref": null
		}
	}
}

if the token can’t get repository，you will get response with repository is null.

status: 200

response:
{
	"data": {
		"repository": null
	}
}

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'content-type: application/json' \
  --data '{
    "query": "{repository(owner: \"au-heartbeat\", name: \"Heartbeat\") { ref(qualifiedName: \"refs/heads/main\") { name }}}"
}'

Replaced api

GET：https://api.github.com/repos/{repository}/commits/{commitId}

Method local

GitHubFeignClient.getCommitInfo

To get information about a specific configuration in the {OWNER}/{REPO} repository, you can use the following GraphQL query.

query {
  repository (owner: string, name: string) {
		commit: object(oid: string) {
			... on Commit {
				message
				author {
					name
					email
					date
				}
				committer {
					name
					email
					date
				}
			}
		}
  }
}

• Once query cost total node

1

• Required value and demo

The value of first must be between 1 and 100.

if you want to get target commit for au-heartbeat/Heartbeat, you can use as following.

query {
  repository (owner: "au-heartbeat", name: "Heartbeat") {
		commit: object(oid: "414e*********************************ae6") {
			... on Commit {
				message
				author {
					name
					email
					date
				}
				committer {
					name
					email
					date
				}
			}
		}
  }
}

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'Content-Type: application/json' \
  --header 'User-Agent: insomnia/8.6.0' \
  --data '{"query":"{ repository (owner: \"au-heartbeat\", name: \"Heartbeat\") { commit: object(oid: \"414e*********************************ae6\") { ... on Commit { message author { name email date } committer { name email date }}}}}"}'

Replaced api

GET：https://api.github.com/repos/{repository}/pulls/{mergedPullNumber}/commits

Method local

GitHubFeignClient.getPullRequestCommitInfo

To get information about a specific configuration in the {OWNER}/{REPO} repository, you can use the following GraphQL query.

query {
  repository(owner: string, name: string) {
    pullRequest(number: int) {
      commits(first: int, after: string) {
        edges {
          node {
            commit {
              message
              author {
                name
                email
                date
              }
						  committer {
                name
                email
                date
              }
            }
          }
        }
				pageInfo {
					endCursor
          hasNextPage
        }
      }
    }
  }
}

• Once query cost total node

1

• Required value and demo

The value of first must be between 1 and 100.

You can put endCursor in after to get next page.

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'Content-Type: application/json' \
  --header 'User-Agent: insomnia/8.6.1' \
  --data '{"query":"{ repository(owner: \"au-heartbeat\", name: \"Heartbeat\") { pullRequest(number: 1037) { commits(first: 100, after: \"\") { edges { node { commit { oid message author { name email date } committer { name email date }}}} pageInfo { endCursor hasNextPage }}}}}"}'

• Replaced api

GET：https://api.github.com/repos/{repository}/commits/{deployId}/pulls

• Method local

GitHubFeignClient.getPullRequestListInfo

To get information about a specific configuration in the {OWNER}/{REPO} repository, you can use the following GraphQL query.

query {
  repository (owner: string, name: string) {
		commit: object(oid: string) {
			... on Commit {
				associatedPullRequests(first: int, after: string) {
					nodes {
						number
						createdAt
						mergedAt
						mergeCommit {
							oid
						}
					}
				pageInfo {
					endCursor
          hasNextPage
        }
				}
				}
			}
		}
  }

• Required value and demo

The value of first must be between 1 and 100.

You can put endCursor in after to get next page.

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'Content-Type: application/json' \
  --header 'User-Agent: insomnia/8.6.0' \
  --data '{"query":"{ repository (owner: \"au-heartbeat\", name: \"Heartbeat\") { commit: object(oid: \"414e*********************************ae6\") { ... on Commit { associatedPullRequests(first: 100, after: \"\") { nodes { number createdAt mergedAt mergeCommit { oid }} pageInfo { endCursor hasNextPage }}}}}}"}'

• Call api by curl

By the demo, if you want to call graphQL, you need to pass a member variable to the api as a query object.

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'content-type: application/json' \
  --data '{"query": "{viewer {login}}"}'

getPullRequestListInfo api and getPullRequestCommitInfo can merge to one api.

we can use commit id to get all pull request and then get all commit list by pull request.

query {
  repository(owner: string, name: string) {
    object(expression: string) {
      ... on Commit {
        associatedPullRequests(first: int, after: int) {
          edges {
            node {
							number
							createdAt
							mergedAt
							mergeCommit {
								oid
							}
              commits(first: int, after: int) {
                edges {
                  node {
                    commit {
                      message
                      author {
                        name
                        email
                        date
                      }
											committer {
												name
												email
												date
											}
                    }
                  }
                }
								pageInfo {
									endCursor
									hasNextPage
								}
              }
            }
          }
					pageInfo {
						endCursor
						hasNextPage
					}
				}
      }
    }
  }
}

• Once query cost total node

1

• Required value and demo

The value of first must be between 1 and 100.

You can put endCursor in after to get next page.

• Test GraphQL API curl

curl --request POST \
  --url https://api.github.com/graphql \
  --header 'Authorization: token ghp_***...***' \
  --header 'Content-Type: application/json' \
  --header 'User-Agent: insomnia/8.6.1' \
  --data '{"query":"{ repository(owner: \"au-heartbeat\", name: \"Heartbeat\") { object(expression: \"414e*********************************ae6\") { ... on Commit { associatedPullRequests(first: 100, after: \"\") { edges { node { number createdAt mergedAt mergeCommit { oid } commits(first: 100, after: \"\") { edges { node { commit { oid message author { name email date } committer { name email date }}}} pageInfo { endCursor hasNextPage }}}} pageInfo { endCursor hasNextPage }}}}}}"}'

Individual calls cannot request more than 500,000 total nodes.

• How to calculate node count

For example

query {
  viewer {
    repositories(first: 50) {
      edges {
        repository:node {
          name
          issues(first: 10) {
            totalCount
            edges {
              node {
                title
                bodyHTML
              }
            }
          }
        }
      }
    }
  }
}

IS

50 + 50 * 10 = 550

5,000 points per hour per user.

GitHub Enterprise Cloud organization have a higher rate limit of 10,000 points per hour.

• How to calculate points count

You can use following query in target gql.

  rateLimit {
    limit
    remaining
    used
    resetAt
    cost
  }

• Required value and demo

For example

query {
  rateLimit {
    limit
    remaining
    used
    resetAt
    cost
  }
  viewer {
    login
    repositories(first: 100) {
      edges {
        node {
          id
          issues(first: 50) {
            edges {
              node {
                id
                labels(first: 60) {
                  edges {
                    node {
                      id
                      name
                    }
                  }
                }
              }
            }
          }
        }
      }
    }
  }
}

You can get response, the cost is every query count.

{
	"data": {
		"rateLimit": {
			"limit": 5000,
			"remaining": 4786,
			"used": 214,
			"resetAt": "2024-04-03T08:47:28Z",
			"cost": 51
		},
		"viewer": {
      ...
		}
	}
}

You can switch Rest api to replace graph ql

The following is a flow chart of the program that interacts with the GitHub API

We define

We can get

Verify stage:（ V + B ）
+
Compute data stage: N * R *（ C * (P +（I + T）)）
+
Report generation stage：N * R * ( D * I )
=
（ V + B ）+ N * R *（ C * (P +（I + T）)）+  N * R * ( D * I ) < 5000

And then introduce some constants

1 + 1 + N * R * ( C * (1 + (1 + 1) ) + N * R * ( D * 1) < 5000
NR * (3C + D) ~ 4NRD < 4998
NRD < 1249

We can get some combination