Trying to fetch all commit PRs using GraphQL API

get_comits.py

#!/usr/bin/env  python3

# Trying to use graphQL to fetch all commits associated with PR https://github.com/pytorch/pytorch/pull/77471
# Expect to get 700+ commits for for some reason pagination ends after first 250
# Needs valid PAT passed via GITHUB_TOKEN environment variable
GH_GET_COMMITS = """
query($cursor: String) {
  repository(owner: "pytorch", name: "pytorch") {
    pullRequest(number: 77471) {
      commits(first: 100, after: $cursor) {
        nodes {
          commit {
            oid
          }
        }
        pageInfo {
          hasNextPage
          endCursor
        }
        totalCount
      }
    }
  }
}
"""


def gh_graphql(query, **kwargs):
    import os
    import json
    from urllib.request import urlopen, Request
    token = os.getenv("GITHUB_TOKEN")
    headers = {"Accept": "application/vnd.github.v3+json", "Authorization": f"token {token}"}
    data = json.dumps({"query": query, "variables": kwargs}).encode()
    with urlopen(Request("https://api.github.com/graphql", data=data, headers=headers)) as conn:
        rc = json.load(conn)
        if "errors" in rc:
          raise RuntimeError(f"GraphQL query {query}, args {kwargs} failed: {rc['errors']}")
        return rc


if __name__ == "__main__":
    cursor = ""
    for pg_idx in range(10):
      rc = gh_graphql(GH_GET_COMMITS, cursor=cursor)["data"]["repository"]["pullRequest"]["commits"]
      print(pg_idx, len(rc["nodes"]), rc["pageInfo"], rc["totalCount"])
      if not bool(rc["pageInfo"]["hasNextPage"]): break
      cursor = rc["pageInfo"]["endCursor"]