File size: 1,799 Bytes
c7df3c8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 | """Pull the actual sentences agents wrote around each deflation hit."""
import glob, json, os, re
BASE = r"C:\Users\acer\.claude\projects\C--Users-acer"
PATS = ["does not exist", "doesn't exist", "not real", "fabricated", "no evidence",
"placeholder", "decorative", "not materialized", "aspirational", "fictional",
"just a hash", "is a stub", "stubbed", "simulated", "mock", "no such",
"unverifiable", "does not actually"]
hits = []
for f in glob.glob(os.path.join(BASE, "**", "agent-*.jsonl"), recursive=True):
try:
raw = open(f, encoding="utf-8", errors="replace").read()
except Exception:
continue
for line in raw.splitlines():
try:
r = json.loads(line)
except Exception:
continue
if r.get("type") != "assistant":
continue
c = (r.get("message") or {}).get("content")
if not isinstance(c, list):
continue
for b in c:
if not (isinstance(b, dict) and b.get("type") == "text"):
continue
txt = b.get("text", "")
for sent in re.split(r"(?<=[.!?\n])\s+", txt):
s = sent.strip()
if not (20 < len(s) < 300):
continue
low = s.lower()
for p in PATS:
if p in low:
hits.append((os.path.basename(f)[:22], p, " ".join(s.split())))
break
seen, uniq = set(), []
for f, p, s in hits:
k = s.lower()
if k in seen:
continue
seen.add(k)
uniq.append((f, p, s))
print("HITS|raw=%d|unique_sentences=%d|json=0" % (len(hits), len(uniq)))
print()
for f, p, s in uniq:
print("HIT|pattern=%s|agent=%s|json=0" % (p, f))
print(" %s" % s)
|