code/candidates.py

import json, re, sys
d = json.load(open(sys.argv[1]))
KW = re.compile(r"(supersed|in favou?r of|duplicate of|already (been )?(fixed|merged|resolved)|fixed (by|in|with) #|resolved (by|in) #|closed by #|done in #|addressed (in|by) #)", re.I)

def merged_refs(n, created):
    out = []
    for e in n["timelineItems"]["nodes"]:
        s = e.get("source") or e.get("subject")
        if s and s.get("__typename") == "PullRequest" and s.get("merged") and s["number"] != n["number"]:
            out.append({"pr": s["number"], "title": s["title"], "mergedAt": s["mergedAt"],
                        "closes": bool(e.get("willCloseTarget")) or e["__typename"] == "ConnectedEvent",
                        "base": s.get("baseRefName")})
        c = e.get("canonical")
        if c: out.append({"dup": c["number"], "state": c["state"]})
    return out

def hints(n):
    return [{"by": c["author"]["login"] if c["author"] else None, "at": c["createdAt"], "text": c["body"][:400]}
            for c in n["comments"]["nodes"] if KW.search(c["body"])]

cands = []
for kind in ("issues", "prs"):
    for n in d[kind]:
        refs, h = merged_refs(n, n["createdAt"]), hints(n)
        sup = []
        if kind == "prs":
            for iss in n["closingIssuesReferences"]["nodes"]:
                for p in iss["closedByPullRequestsReferences"]["nodes"]:
                    if p["merged"] and p["number"] != n["number"]:
                        sup.append({"issue": iss["number"], "issue_state": iss["state"], "merged_pr": p["number"]})
        if refs or h or sup:
            score = 3 if (any(r.get("closes") for r in refs) or sup) else 2 if h else 1
            cands.append({"kind": kind[:-1].replace("pr", "pr"), "number": n["number"], "title": n["title"],
                          "author": (n["author"] or {}).get("login"), "created": n["createdAt"], "updated": n["updatedAt"],
                          "draft": n.get("isDraft"), "merged_refs": refs, "hint_comments": h, "sibling_merged": sup, "score": score})
cands.sort(key=lambda c: -c["score"])
json.dump(cands, open(sys.argv[2], "w"), indent=1)
from collections import Counter
print(Counter((c["kind"], c["score"]) for c in cands))