merge_writeback.py

skills/client-enrich/scripts/merge_writeback.py

1. #!/usr/bin/env python3
2. """
3. merge_writeback.py: unified write-back combining Apollo person/company data and
4. web-search signals for client-enrich. Calls enrich_writeback.py so the vault
5. contract holds (writes only under ## Research + frontmatter; never Human Notes).
7. Per stub:
8. - Person = Apollo person lines if matched, else from existing frontmatter.
9. - Company = Apollo company lines if matched, else minimal (name + LinkedIn + TBD firmographics).
10. - Signals = Apollo company_signals (funding/growth) + web-search signals; [TBD] if none.
11. - Sources = web sources + Apollo sources + provenance lines.
12. - Frontmatter = filled from Apollo (empty fields only; enrich_writeback enforces that).
14. Usage:
15. python3 merge_writeback.py --batch no_li_40.json --web web_signals_40.json \
16. --apollo apollo_results.json --stubdir "/path/03-Clients/Check-In" \
17. --script "/path/enrich_writeback.py" [--date YYYY-MM-DD]
18. """
19. import argparse, json, os, subprocess, sys, datetime, tempfile
21. def main():
22. ap = argparse.ArgumentParser()
23. ap.add_argument("--batch", required=True)
24. ap.add_argument("--web", required=True)
25. ap.add_argument("--apollo", required=True)
26. ap.add_argument("--stubdir", required=True)
27. ap.add_argument("--script", required=True)
28. ap.add_argument("--date", default=datetime.date.today().isoformat())
29. args = ap.parse_args()
31. batch = {r["file"]: r for r in json.load(open(args.batch, encoding="utf-8"))}
32. web = json.load(open(args.web, encoding="utf-8"))
33. apollo = json.load(open(args.apollo, encoding="utf-8")) if os.path.exists(args.apollo) else {}
35. rows = []
36. for fname, r in batch.items():
37. ap_e = apollo.get(fname, {})
38. matched = bool(ap_e.get("matched"))
39. web_e = web.get(fname, {})
41. if matched and ap_e.get("person"):
42. person = ap_e["person"]
43. else:
44. person = []
45. if r.get("title"): person.append(f"- Title: {r['title']}")
46. if r.get("seniority"): person.append(f"- Seniority: {r['seniority']}")
47. loc = ", ".join([x for x in [r.get("city"), r.get("state")] if x])
48. if loc: person.append(f"- Location: {loc}")
49. person = person or ["[TBD]"]
51. if matched and ap_e.get("company"):
52. company = ap_e["company"]
53. else:
54. company = []
55. if r.get("company"): company.append(f"- {r['company']}")
56. if r.get("company_li"): company.append(f"- LinkedIn: {r['company_li']}")
57. company.append("- [TBD: firmographics not available (Apollo miss or no corporate domain)]")
59. signals = list(ap_e.get("company_signals") or []) + list(web_e.get("signals") or [])
60. if not signals:
61. signals = ["[TBD: no recent signals found via Apollo or web search]"]
63. sources = list(web_e.get("sources") or []) + list(ap_e.get("sources") or [])
64. if web_e.get("signals"):
65. sources.append(f"- Web search, {args.date}")
66. if not matched:
67. sources.append("- Stub frontmatter (prior enrichment)")
69. payload = {"frontmatter": ap_e.get("frontmatter", {}) if matched else {},
70. "person": person, "company": company,
71. "signals": signals, "sources": sources}
72. with tempfile.NamedTemporaryFile("w", suffix=".json", delete=False) as tf:
73. json.dump(payload, tf); pj = tf.name
74. res = subprocess.run([sys.executable, args.script, "--stub",
75. os.path.join(args.stubdir, fname), "--payload", pj,
76. "--date", args.date], capture_output=True, text=True)
77. os.unlink(pj)
78. rows.append((fname, matched, len([s for s in signals if not s.startswith("[TBD")]),
79. res.returncode == 0, res.stderr.strip()[:120]))
81. okc = sum(1 for _,_,_,k,_ in rows if k)
82. mc = sum(1 for _,m,_,_,_ in rows if m)
83. sc = sum(1 for _,_,n,_,_ in rows if n > 0)
84. print(f"WROTE {okc}/{len(rows)} OK | Apollo-matched: {mc} | with >=1 signal: {sc}")
85. for f, m, n, k, err in rows:
86. print(f" {f[:30]:32} | {'APOLLO' if m else 'frontmtr':8} | {n} sig | {'OK' if k else 'FAIL '+err}")
88. if __name__ == "__main__":
89. main()