Files
medevio/10ReadPozadavky/PRAVIDELNE_5_SaveToFileSystem incremental.py
2025-11-30 20:21:01 +01:00

243 lines
6.6 KiB
Python

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import os
import shutil
import pymysql
import re
from pathlib import Path
from datetime import datetime
import time
import sys
# ==============================
# 🛡 SAFE PRINT FOR CP1250 / EMOJI
# ==============================
def safe_print(text: str = ""):
enc = sys.stdout.encoding or ""
if not enc.lower().startswith("utf"):
# Strip emoji and characters outside BMP for Task Scheduler
text = ''.join(ch for ch in text if ord(ch) < 65536)
try:
print(text)
except UnicodeEncodeError:
# ASCII fallback
text = ''.join(ch for ch in text if ord(ch) < 128)
print(text)
# ==============================
# ⚙️ CONFIGURATION
# ==============================
DB_CONFIG = {
"host": "192.168.1.76",
"port": 3307,
"user": "root",
"password": "Vlado9674+",
"database": "medevio",
"charset": "utf8mb4",
}
BASE_DIR = Path(r"z:\Dropbox\Ordinace\Dokumentace_ke_zpracování\MP")
BASE_DIR.mkdir(parents=True, exist_ok=True)
def sanitize_name(name: str) -> str:
"""Replace invalid filename characters with underscore."""
return re.sub(r'[<>:"/\\|?*\x00-\x1F]', "_", name).strip()
def make_abbrev(title: str) -> str:
if not title:
return ""
words = re.findall(r"[A-Za-zÁ-Žá-ž0-9]+", title)
abbr = ""
for w in words:
if w.isdigit():
abbr += w
else:
abbr += w[0]
return abbr.upper()
# ==============================
# 🧹 DELETE UNEXPECTED FILES
# ==============================
def clean_folder(folder: Path, valid_files: set):
if not folder.exists():
return
for f in folder.iterdir():
if f.is_file():
if f.name.startswith(""):
continue
sanitized = sanitize_name(f.name)
if sanitized not in valid_files:
safe_print(f"🗑️ Removing unexpected file: {f.name}")
try:
f.unlink()
except Exception as e:
safe_print(f"⚠️ Could not delete {f}: {e}")
# ==============================
# 📦 DB CONNECTION
# ==============================
conn = pymysql.connect(**DB_CONFIG)
cur_meta = conn.cursor(pymysql.cursors.DictCursor)
cur_blob = conn.cursor()
safe_print("🔍 Loading metadata from DB (FAST)…")
cur_meta.execute("""
SELECT d.id AS download_id,
d.request_id,
d.filename,
d.created_at,
p.updatedAt AS req_updated_at,
p.pacient_jmeno AS jmeno,
p.pacient_prijmeni AS prijmeni,
p.displayTitle
FROM medevio_downloads d
JOIN pozadavky p ON d.request_id = p.id
WHERE d.downloaded_at IS NULL
ORDER BY p.updatedAt DESC
""")
rows = cur_meta.fetchall()
safe_print(f"📋 Found {len(rows)} attachment records.\n")
# ==============================
# 🧠 MAIN LOOP WITH PROGRESS
# ==============================
unique_request_ids = []
seen = set()
for r in rows:
req_id = r["request_id"]
if req_id not in seen:
unique_request_ids.append(req_id)
seen.add(req_id)
total_requests = len(unique_request_ids)
safe_print(f"🔄 Processing {total_requests} unique requests...\n")
processed_requests = set()
current_index = 0
for r in rows:
req_id = r["request_id"]
if req_id in processed_requests:
continue
processed_requests.add(req_id)
current_index += 1
percent = (current_index / total_requests) * 100
safe_print(f"\n[ {percent:5.1f}% ] Processing request {current_index} / {total_requests}{req_id}")
# ========== FETCH VALID FILENAMES ==========
cur_meta.execute(
"SELECT filename FROM medevio_downloads WHERE request_id=%s",
(req_id,)
)
valid_files = {sanitize_name(row["filename"]) for row in cur_meta.fetchall()}
# ========== BUILD FOLDER NAME ==========
updated_at = r["req_updated_at"] or datetime.now()
date_str = updated_at.strftime("%Y-%m-%d")
prijmeni = sanitize_name(r["prijmeni"] or "Unknown")
jmeno = sanitize_name(r["jmeno"] or "")
title = r.get("displayTitle") or ""
abbr = make_abbrev(title)
clean_folder_name = sanitize_name(
f"{date_str} {prijmeni}, {jmeno} [{abbr}] {req_id}"
)
# ========== DETECT EXISTING FOLDER ==========
existing_folder = None
for f in BASE_DIR.iterdir():
if f.is_dir() and req_id in f.name:
existing_folder = f
break
main_folder = existing_folder if existing_folder else BASE_DIR / clean_folder_name
# ========== MERGE DUPLICATES ==========
possible_dups = [
f for f in BASE_DIR.iterdir()
if f.is_dir() and req_id in f.name and f != main_folder
]
for dup in possible_dups:
safe_print(f"♻️ Merging duplicate folder: {dup.name}")
clean_folder(dup, valid_files)
main_folder.mkdir(parents=True, exist_ok=True)
for f in dup.iterdir():
if f.is_file():
target = main_folder / f.name
if not target.exists():
f.rename(target)
shutil.rmtree(dup, ignore_errors=True)
# ========== CLEAN MAIN FOLDER ==========
clean_folder(main_folder, valid_files)
# ========== DOWNLOAD MISSING FILES ==========
added_new_file = False
main_folder.mkdir(parents=True, exist_ok=True)
for filename in valid_files:
dest_plain = main_folder / filename
dest_marked = main_folder / ("" + filename)
if dest_plain.exists() or dest_marked.exists():
continue
added_new_file = True
cur_blob.execute(
"SELECT file_content FROM medevio_downloads "
"WHERE request_id=%s AND filename=%s",
(req_id, filename)
)
row = cur_blob.fetchone()
if not row:
continue
content = row[0]
if not content:
continue
with open(dest_plain, "wb") as f:
f.write(content)
safe_print(f"💾 Wrote: {dest_plain.relative_to(BASE_DIR)}")
# ========== REMOVE ▲ FLAG IF NEW FILES ADDED ==========
if added_new_file and "" in main_folder.name:
new_name = main_folder.name.replace("", "").strip()
new_path = main_folder.parent / new_name
if new_path != main_folder:
try:
main_folder.rename(new_path)
safe_print(f"🔄 Folder flag ▲ removed → {new_name}")
main_folder = new_path
except Exception as e:
safe_print(f"⚠️ Could not rename folder: {e}")
safe_print("\n🎯 Export complete.\n")
cur_blob.close()
cur_meta.close()
conn.close()