code/land_fetch.py

import json, subprocess, time, sys
def run(kind, extra):
    Q = """query($c:String){repository(owner:"huggingface",name:"transformers"){%s(states:OPEN,first:40,after:$c){totalCount pageInfo{hasNextPage endCursor}
      nodes{number title createdAt updatedAt author{login} authorAssociation labels(first:10){nodes{name}} reactions{totalCount} comments{totalCount} %s}}}}""" % (kind, extra)
    out, c = [], None
    while True:
        a = ['gh','api','graphql','-f',f'query={Q}'] + (['-f',f'c={c}'] if c else [])
        for _ in range(5):
            r = subprocess.run(a, capture_output=True, text=True, timeout=120)
            if r.returncode == 0: break
            time.sleep(10)
        d = json.loads(r.stdout)['data']['repository'][kind]; out += d['nodes']
        print(f'[{kind} {len(out)}/{d["totalCount"]}]', flush=True)
        if not d['pageInfo']['hasNextPage']: return out
        c = d['pageInfo']['endCursor']
PR = 'isDraft mergeable reviewDecision additions deletions changedFiles files(first:60){nodes{path}} reviewRequests(first:5){totalCount}'
json.dump({'issues': run('issues', ''), 'prs': run('pullRequests', PR)}, open('land.json','w'))
print('done')