import json
import sys
import re
import subprocess
from pathlib import Path

sys.stdout.reconfigure(encoding='utf-8')

# Run git diff for const_work_item_master_2026-01-01.json in commit 32c4c5cf
cmd = ["git", "diff", "32c4c5cf~1", "32c4c5cf", "--", "resources/data_work_item_master/const_work_item_master_2026-01-01.json"]
res = subprocess.run(cmd, capture_output=True, text=True, encoding='utf-8')

diff_lines = res.stdout.splitlines()
print(f"Total diff lines in const master: {len(diff_lines)}")

# Extract lines starting with '-' (removed lines) inside "notes": [...]
removed_notes = []
current_section = "미식별"
for l in diff_lines:
    if '"section":' in l:
        current_section = l.strip()
    if l.startswith('-') and not l.startswith('---'):
        cleaned = l[1:].strip()
        if cleaned.startswith('"') and cleaned.endswith('",'):
            cleaned = cleaned[1:-2]
        elif cleaned.startswith('"') and cleaned.endswith('"'):
            cleaned = cleaned[1:-1]
        if cleaned and cleaned not in ['notes": [', ']']:
            removed_notes.append((current_section, cleaned))

print(f"Total removed notes lines: {len(removed_notes)}")

# Filter lines that look like actual rules (containing %, 할증, 손료, 가산, 공제, etc.)
rule_keywords = ['%', '가산', '할증', '공제', '손료', '손율', '감한다', '할인', '포함한다', 'Q=', '적용한다']
removed_rules = []
for sec, line in removed_notes:
    if any(k in line for k in rule_keywords):
        # check if it was not just a title
        if not re.match(r'^[0-9]+[\.\)]\s*[^:%]+$', line):
            removed_rules.append((sec, line))

print(f"Removed lines containing pricing keywords in const master: {len(removed_rules)}")
print("\n--- Sample 20 Removed Rules in Const Master ---")
for sec, r in removed_rules[:20]:
    print(f"{sec} -> {r}")