#!/usr/bin/env python3
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
██╗ ██╗ █████╗ ██████╗ ████████╗ █████╗ ███╗ ██╗██╗ █████╗
╚██╗██╔╝██╔══██╗██╔══██╗╚══██╔══╝██╔══██╗████╗ ██║██║██╔══██╗
╚███╔╝ ███████║██████╔╝ ██║ ███████║██╔██╗ ██║██║███████║
██╔██╗ ██╔══██║██╔══██╗ ██║ ██╔══██║██║╚██╗██║██║██╔══██║
██╔╝ ██╗██║ ██║██║ ██║ ██║ ██║ ██║██║ ╚████║██║██║ ██║
╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝╚═╝ ╚═══╝╚═╝╚═╝ ╚═╝
Edition:
## 29/10/2026 by Tsukini
File Name:
## sample.py
File Description:
## A sample of what the user typed himself (his prompts, his .info notes): counts of his habits, a random sample; local only
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
"""
Usage: sample.py [count] - his prompts (claude-cluster histories, the newest Claude Code transcripts: only what he typed) and
his .info notes. Local only: never paste the sample anywhere else.
"""
##### Imports #####
import collections # Used for Counter
import glob # Used for glob
import json # Used for loads
import os # Used for path
import random # Used for sample
import re # Used for findall, sub, match
import sys # Used for argv
##### Constants #####
HOME = os.path.expanduser("~")
TRANSCRIPTS = 120 # newest Claude Code transcripts read
TAIL = 3_000_000 # bytes read at the end of each
GENERATED = re.compile(r"\[(reply|from|claude-cluster|scheduled|a turn)|^\(a turn|^(You are session|Fix in|Feature in|Change in) ")
HABITS = ["???", "!!!", "...", "->", "/", "(", " ou autre", "ect", "tj", "genre", "truc", "fait en sorte", "rajoute", "check", "setup", "remove"]
##### Functions #####
def read_cluster() -> list:
"""Every prompt typed in claude-cluster (the main data folder and its instances)"""
texts = []
for path in glob.glob(f"{HOME}/.local/share/claude-cluster/**/history.jsonl", recursive=True):
for line in open(path, encoding="utf-8", errors="replace"):
try:
entry = json.loads(line)
except ValueError:
continue
text = entry.get("text") if isinstance(entry, dict) else entry
if isinstance(text, str):
texts.append(text)
return texts
def read_claude_code() -> list:
"""The messages the user typed in the newest Claude Code transcripts (no tool results, no commands)"""
texts = []
files = sorted(glob.glob(f"{HOME}/.claude/projects/*/*.jsonl"), key=os.path.getmtime, reverse=True)[:TRANSCRIPTS]
for path in files:
try:
with open(path, "rb") as f:
f.seek(max(0, os.path.getsize(path) - TAIL))
lines = f.read().decode("utf-8", "replace").splitlines()[1:]
except OSError:
continue
for line in lines:
if '"type":"user"' not in line:
continue
try:
content = (json.loads(line).get("message") or {}).get("content")
except ValueError:
continue
if isinstance(content, list):
content = " ".join(c.get("text", "") for c in content if isinstance(c, dict) and c.get("type") == "text")
if isinstance(content, str) and not content.startswith("<") and "tool_use_id" not in content:
texts.append(content)
return texts
def read_notes() -> list:
"""His .info notes (delivery / personal_delivery)"""
paths = glob.glob(f"{HOME}/delivery/**/.info", recursive=True) + glob.glob(f"{HOME}/personal_delivery/**/.info", recursive=True)
return [open(p, encoding="utf-8", errors="replace").read() for p in paths if os.path.isfile(p) and "/build" not in p]
def clean(texts: list) -> list:
"""His own words only: images / pasted blocks dropped, generated prompts and continuations skipped, no duplicate"""
seen = set()
out = []
for text in texts:
text = re.sub(r"\[(Image|Pasted text)[^\]]*\]", "", text).strip()
if len(text) < 15 or GENERATED.search(text) or "This session is being continued" in text or "SYSTEM NOTIFICATION" in text:
continue
if text[:200] not in seen:
seen.add(text[:200])
out.append(text)
return out
##### Main #####
def main() -> int:
count = int(sys.argv[1]) if len(sys.argv) > 1 and sys.argv[1].isdigit() else 40
messages = clean(read_cluster() + read_claude_code())
notes = clean(read_notes())
joined = "\n".join(messages).lower()
print(f"messages {len(messages)}, notes {len(notes)}, words {len(re.findall(r'[a-zà-ÿ]+', joined))}")
count_of = lambda habit: len(re.findall(r"\b" + re.escape(habit.strip()) + r"\b", joined)) if habit.strip()[0].isalpha() else joined.count(habit) # words whole
print("habits:", ", ".join(f"{h.strip()!r} {count_of(h)}" for h in HABITS))
print("accented letters / 1000 chars:", round(sum(joined.count(c) for c in "éèêàùç") * 1000 / max(1, len(joined)), 1))
print("first words:", collections.Counter(m.split()[0].lower() for m in messages if m.split()).most_common(15))
random.seed()
for text in random.sample(messages, min(count, len(messages))):
print("---")
print(text[:400])
for text in random.sample(notes, min(3, len(notes))):
print("--- note")
print(text[:600])
return 0
if __name__ == "__main__":
sys.exit(main())