z230

Mazání Library/20 MazáníPrádnýchAdresářů.py

@@ -0,0 +1,38 @@
+import os
+
+
+def cleanup_empty_folders(target_path, dry_run=True):
+    # Kontrola, zda cesta existuje
+    if not os.path.exists(target_path):
+        print(f"Chyba: Cesta {target_path} neexistuje.")
+        return
+
+    print(f"Prohledávám: {target_path}")
+    if dry_run:
+        print("POZOR: Spuštěno v režimu DRY RUN (nic se nemaže)\n")
+
+    # topdown=False je klíčové - začínáme od nejhlubších složek
+    for root, dirs, files in os.walk(target_path, topdown=False):
+        for name in dirs:
+            folder_path = os.path.join(root, name)
+
+            # Kontrola, zda je složka prázdná
+            # Listdir vypíše vše v adresáři (včetně skrytých souborů)
+            if not os.listdir(folder_path):
+                if dry_run:
+                    print(f"[DRY RUN] Složka by byla smazána: {folder_path}")
+                else:
+                    try:
+                        os.rmdir(folder_path)
+                        print(f"Smazáno: {folder_path}")
+                    except OSError as e:
+                        print(f"Chyba při mazání {folder_path}: {e}")
+
+
+if __name__ == "__main__":
+    # Cesta k vašemu síťovému sdílení
+    # V Pythonu na Windows použijte r"" (raw string) kvůli zpětným lomítkům
+    path_to_clean = r"\\tower1\#library"
+
+    # Prvně spusťte s dry_run=True, abyste viděli, co se stane
+    cleanup_empty_folders(path_to_clean, dry_run=True)

WalkFilesOnBackupHDD/20 WalkBackupHDD.py

@@ -15,11 +15,15 @@ import platform
 import sys
 from blake3 import blake3
 
+def get_path_hash(path_str: str) -> bytes:
+    """Calculates MD5 hash of the path and returns raw 16 bytes for BINARY(16)."""
+    return hashlib.md5(path_str.encode('utf-8')).digest()
+
 # ==============================
 # ⚙️ USER CONFIGURATION
 # ==============================
-DISK_DRIVE_LETTER = "p"  # (e.g., "E", "F", "P")
+DISK_DRIVE_LETTER = "z"  # (e.g., "E", "F", "P")
-DISK_HOSTNAME = "#HD05"  # (e.g., "#HD015")
+DISK_HOSTNAME = "TW22"  # (e.g., "#HD015")
 
 # 🔒 SAFETY SWITCH
 # True  = LIST ONLY (No DB changes). "Simulates" the run.

WalkFilesOnBackupHDD/30 WalkBackupHDD.py

@@ -0,0 +1,313 @@
+#!/usr/bin/python3
+# -*- coding: utf-8 -*-
+
+r"""
+FAST FILE HASH INDEXER – WINDOWS CLIENT (HARDCODED CONFIG)
+- Mode: PHYSICAL BACKUP
+- Hostname in DB = Disk Label (e.g., #HD015)
+- Path in DB     = Relative path (e.g., /Movies/Film.mkv)
+"""
+
+import os
+import time
+import pymysql
+import socket
+import platform
+import sys
+import hashlib
+from blake3 import blake3
+
+# ==============================
+# ⚙️ USER CONFIGURATION
+# ==============================
+DISK_DRIVE_LETTER = "z"  # (e.g., "E", "F", "P")
+DISK_HOSTNAME = "TW22"  # (e.g., "#HD015")
+
+# 🔒 SAFETY SWITCH
+DRY_RUN = False
+
+# ==============================
+# TECHNICAL CONFIG
+# ==============================
+CHUNK_SIZE = 5 * 1024 * 1024  # 5 MB
+PROGRESS_MIN_SIZE = 500 * 1024 * 1024  # 500 MB
+PROGRESS_INTERVAL = 1.0  # seconds
+
+EXCLUDED_DIRS = {"$RECYCLE.BIN", "System Volume Information", "RECYCLER", "msdownld.tmp"}
+
+# --- File Size Limits ---
+FILE_MIN_SIZE = 0
+FILE_MAX_SIZE = 1024 * 1024 * 1024 * 1024  # 1TB
+
+# --- DB Config ---
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+    "autocommit": True,
+}
+
+PRINT_SKIPPED = False  # Set True to see files that were already in DB
+
+# ==============================
+# SYSTEM INFO
+# ==============================
+REAL_PC_HOSTNAME = socket.gethostname()
+OS_NAME = platform.system()
+
+
+# ==============================
+# FUNCTIONS
+# ==============================
+
+def get_path_hash(path_str: str) -> bytes:
+    """Calculates MD5 hash of the path and returns raw 16 bytes for BINARY(16)."""
+    return hashlib.md5(path_str.encode('utf-8')).digest()
+
+
+def compute_blake3(path: str) -> bytes:
+    h = blake3()
+    total_size = os.path.getsize(path)
+    show_progress = total_size >= PROGRESS_MIN_SIZE
+
+    processed = 0
+    start_time = time.time()
+    last_report = start_time
+
+    try:
+        with open(path, "rb") as f:
+            while True:
+                chunk = f.read(CHUNK_SIZE)
+                if not chunk:
+                    break
+
+                h.update(chunk)
+                processed += len(chunk)
+
+                if show_progress:
+                    now = time.time()
+                    if now - last_report >= PROGRESS_INTERVAL:
+                        elapsed = now - start_time
+                        speed = processed / elapsed if elapsed > 0 else 0
+                        percent = processed / total_size * 100
+                        remaining = total_size - processed
+                        eta = remaining / speed if speed > 0 else 0
+
+                        print(
+                            f"   ⏳ {percent:6.2f}% | "
+                            f"{processed / 1024 / 1024:8.1f} / {total_size / 1024 / 1024:.1f} MB | "
+                            f"{speed / 1024 / 1024:6.1f} MB/s | "
+                            f"ETA {time.strftime('%H:%M:%S', time.gmtime(eta))}",
+                            flush=True
+                        )
+                        last_report = now
+
+        if show_progress:
+            total_time = time.time() - start_time
+            avg_speed = total_size / total_time if total_time > 0 else 0
+            print(
+                f"   ✅ DONE | "
+                f"{total_size / 1024 / 1024:.1f} MB | "
+                f"avg {avg_speed / 1024 / 1024:.1f} MB/s | "
+                f"time {time.strftime('%H:%M:%S', time.gmtime(total_time))}",
+                flush=True
+            )
+
+        return h.digest()
+
+    except Exception as e:
+        print(f"⚠️ HASH ERROR: {path} - {e}")
+        raise
+
+
+def size_allowed(size: int) -> bool:
+    if FILE_MIN_SIZE is not None and size < FILE_MIN_SIZE: return False
+    if FILE_MAX_SIZE is not None and size > FILE_MAX_SIZE: return False
+    return True
+
+
+def normalize_db_path(scan_root, disk_path):
+    """
+    Converts a physical Windows path to the standardized DB format.
+    E:\Movies\File.mkv -> /Movies/File.mkv
+    """
+    try:
+        rel_path = os.path.relpath(disk_path, scan_root)
+    except ValueError:
+        return None
+
+    clean_path = rel_path.replace("\\", "/")
+    if not clean_path.startswith("/"):
+        clean_path = "/" + clean_path
+
+    return clean_path
+
+
+# ==============================
+# MAIN
+# ==============================
+
+def main():
+    print("🚀 BLAKE3 External Disk Indexer (MySQL 9 Compatible)", flush=True)
+    print(f"🖥 Running on PC: {REAL_PC_HOSTNAME}", flush=True)
+
+    if DRY_RUN:
+        print("🛡️  DRY RUN MODE ACTIVE: No changes will be made to DB.", flush=True)
+    else:
+        print("⚠️  LIVE MODE: Changes WILL be committed to DB.", flush=True)
+
+    scan_root = f"{DISK_DRIVE_LETTER}:\\"
+
+    if not os.path.isdir(scan_root):
+        print(f"❌ ERROR: Drive '{scan_root}' not found!")
+        return
+
+    try:
+        db = pymysql.connect(**DB_CONFIG)
+        cur = db.cursor()
+    except Exception as e:
+        print(f"❌ DB Connection failed: {e}")
+        return
+
+    print(f"📥 Loading DB index for: '{DISK_HOSTNAME}'...", flush=True)
+
+    # === LOAD EXISTING DB RECORDS ===
+    # We load path_hash as well for precise deletion
+    cur.execute("""
+        SELECT full_path, file_size, UNIX_TIMESTAMP(mtime)
+        FROM file_md5_index
+        WHERE host_name = %s
+    """, (DISK_HOSTNAME,))
+
+    indexed_map = {row[0]: (row[1], row[2]) for row in cur.fetchall()}
+    print(f"✅ Found {len(indexed_map):,} files in DB for this disk.", flush=True)
+
+    # =========================================================
+    # PHASE 1: CLEANUP (DELETE MISSING FILES)
+    # =========================================================
+    print("======================================", flush=True)
+    print("🧹 PHASE 1: Checking for deleted files...", flush=True)
+
+    current_disk_paths = set()
+    for root, dirs, files in os.walk(scan_root):
+        dirs[:] = [d for d in dirs if d not in EXCLUDED_DIRS]
+        for fname in files:
+            disk_path = os.path.join(root, fname)
+            clean_path = normalize_db_path(scan_root, disk_path)
+            if clean_path:
+                current_disk_paths.add(clean_path)
+
+    paths_to_delete = set(indexed_map.keys()) - current_disk_paths
+
+    if paths_to_delete:
+        print(f"🗑️  Found {len(paths_to_delete):,} files to delete from DB.")
+        if DRY_RUN:
+            for p in sorted(list(paths_to_delete))[:20]:
+                print(f"   - {p}")
+        else:
+            # Delete using path_hash for index efficiency
+            batch_size = 500
+            to_delete_list = list(paths_to_delete)
+            for i in range(0, len(to_delete_list), batch_size):
+                batch_paths = to_delete_list[i: i + batch_size]
+                # Map paths to their MD5 hashes
+                batch_hashes = [get_path_hash(p) for p in batch_paths]
+
+                format_strings = ','.join(['%s'] * len(batch_hashes))
+                query = f"DELETE FROM file_md5_index WHERE host_name = %s AND path_hash IN ({format_strings})"
+
+                try:
+                    cur.execute(query, [DISK_HOSTNAME] + batch_hashes)
+                except Exception as e:
+                    print(f"❌ Error deleting batch: {e}")
+
+            for p in paths_to_delete:
+                del indexed_map[p]
+            print("✅ Cleanup complete.")
+    else:
+        print("✅ No deleted files detected.")
+
+    # =========================================================
+    # PHASE 2: SCAN & UPDATE (HASHING)
+    # =========================================================
+    print("======================================", flush=True)
+    print("🚀 PHASE 2: Scanning for changes & new files...", flush=True)
+
+    new_files = 0
+    skipped = 0
+    errors = 0
+    seen_paths = set()
+
+    for root, dirs, files in os.walk(scan_root):
+        dirs[:] = [d for d in dirs if d not in EXCLUDED_DIRS]
+        for fname in files:
+            disk_path = os.path.join(root, fname)
+            try:
+                stat = os.stat(disk_path)
+            except OSError:
+                errors += 1
+                continue
+
+            size = stat.st_size
+            if not size_allowed(size):
+                continue
+
+            clean_path = normalize_db_path(scan_root, disk_path)
+            if not clean_path or clean_path in seen_paths:
+                continue
+            seen_paths.add(clean_path)
+
+            mtime = int(stat.st_mtime)
+
+            # Match Check
+            if clean_path in indexed_map:
+                db_size, db_mtime = indexed_map[clean_path]
+                if size == db_size and mtime == db_mtime:
+                    skipped += 1
+                    continue
+
+            # Compute Hashes
+            try:
+                b3_hash = compute_blake3(disk_path)
+                p_hash = get_path_hash(clean_path)  # Essential for MySQL 9 Unique Index
+            except Exception:
+                errors += 1
+                continue
+
+            if DRY_RUN:
+                print(f"🛡️  [DRY RUN] NEW/UPDATE: {clean_path}")
+                new_files += 1
+            else:
+                cur.execute("""
+                    INSERT INTO file_md5_index
+                        (os_name, host_name, full_path, path_hash, file_name, directory,
+                        file_size, mtime, blake3)
+                    VALUES (%s, %s, %s, %s, %s, %s, %s, FROM_UNIXTIME(%s), %s)
+                    ON DUPLICATE KEY UPDATE
+                        file_size  = VALUES(file_size),
+                        mtime      = VALUES(mtime),
+                        blake3     = VALUES(blake3),
+                        updated_at = CURRENT_TIMESTAMP
+                """, (
+                    OS_NAME, DISK_HOSTNAME, clean_path, p_hash, fname,
+                    os.path.dirname(clean_path), size, mtime, b3_hash
+                ))
+                new_files += 1
+                print(f"➕ ADDED: {clean_path} | {b3_hash.hex()[:8]}...")
+
+    print("======================================", flush=True)
+    print(f"✅ Processed  : {new_files}")
+    print(f"⏭ Skipped    : {skipped}")
+    print(f"🗑 Deleted    : {len(paths_to_delete)}")
+    print(f"⚠️ Errors     : {errors}")
+    print("🏁 Done.")
+
+    cur.close()
+    db.close()
+
+
+if __name__ == "__main__":
+    main()

WalkFilesOnBackupHDD/40 TestPathNormalizedinTable.py

@@ -0,0 +1,158 @@
+#!/opt/bin/python3
+# -*- coding: utf-8 -*-
+
+import pymysql
+import hashlib
+import posixpath
+import unicodedata
+from binascii import hexlify
+
+# ============================================================
+# CONFIG
+# ============================================================
+
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+}
+
+HOST_FILTER = "tower1"   # None = all hosts
+LIMIT = None               # e.g. 50000 for testing
+SHOW_EXAMPLES = 20
+
+# ============================================================
+# CANONICAL PATH
+# ============================================================
+
+def canonical_path(path_str: str) -> str:
+    if not path_str:
+        return path_str
+
+    path_str = path_str.replace("\\", "/")
+    path_str = posixpath.normpath(path_str)
+    path_str = unicodedata.normalize("NFC", path_str)
+
+    return path_str
+
+def md5_bytes(path_str: str) -> bytes:
+    return hashlib.md5(path_str.encode("utf-8")).digest()
+
+# ============================================================
+# MAIN
+# ============================================================
+
+def main():
+
+    db = pymysql.connect(**DB_CONFIG)
+    cur = db.cursor(pymysql.cursors.SSCursor)
+
+    sql = """
+        SELECT id, full_path, path_hash
+        FROM file_md5_index
+    """
+
+    params = []
+
+    if HOST_FILTER:
+        sql += " WHERE host_name = %s"
+        params.append(HOST_FILTER)
+
+    if LIMIT:
+        sql += " LIMIT %s"
+        params.append(LIMIT)
+
+    cur.execute(sql, params)
+
+    total = 0
+    ok = 0
+    path_change = 0
+    hash_change = 0
+
+    examples_path = []
+    examples_hash = []
+
+    for rec_id, full_path, stored_hash in cur:
+
+        total += 1
+
+        canonical = canonical_path(full_path)
+
+        raw_hash = md5_bytes(full_path)
+        canonical_hash = md5_bytes(canonical)
+
+        # ---------------------------------------------------
+        # CASE 1: fully OK
+        # ---------------------------------------------------
+        if full_path == canonical and stored_hash == canonical_hash:
+            ok += 1
+
+        # ---------------------------------------------------
+        # CASE 2: path string would change
+        # ---------------------------------------------------
+        if full_path != canonical:
+            path_change += 1
+
+            if len(examples_path) < SHOW_EXAMPLES:
+                examples_path.append((rec_id, full_path, canonical))
+
+        # ---------------------------------------------------
+        # CASE 3: hash would change
+        # ---------------------------------------------------
+        if stored_hash != canonical_hash:
+            hash_change += 1
+
+            if len(examples_hash) < SHOW_EXAMPLES:
+                examples_hash.append(
+                    (rec_id, full_path,
+                     hexlify(stored_hash).decode(),
+                     hexlify(canonical_hash).decode())
+                )
+
+        if total % 100000 == 0:
+            print(f"Checked {total:,} rows...")
+
+    # ============================================================
+    # REPORT
+    # ============================================================
+
+    print("\n" + "=" * 70)
+    print("AUDIT SUMMARY")
+    print("=" * 70)
+
+    print(f"Total rows checked : {total:,}")
+    print(f"OK (already canonical + hash OK) : {ok:,}")
+    print(f"Paths that would change         : {path_change:,}")
+    print(f"Hashes that would change        : {hash_change:,}")
+
+    print("=" * 70)
+
+    # ------------------------------------------------------------
+    # SHOW EXAMPLES
+    # ------------------------------------------------------------
+
+    if examples_path:
+        print("\n⚠ PATH CHANGE EXAMPLES:")
+        for rec_id, old, new in examples_path:
+            print(f"[id={rec_id}]")
+            print(" DB :", old)
+            print(" NEW:", new)
+            print()
+
+    if examples_hash:
+        print("\n❌ HASH CHANGE EXAMPLES:")
+        for rec_id, path, old_hash, new_hash in examples_hash:
+            print(f"[id={rec_id}] {path}")
+            print(" Stored :", old_hash)
+            print(" New    :", new_hash)
+            print()
+
+    cur.close()
+    db.close()
+
+
+if __name__ == "__main__":
+    main()

WalkFilesOnBackupHDD/50 Onetimepathnormalization.py

@@ -0,0 +1,188 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+"""
+ONE-TIME MIGRATION: Normalize full_path (NFC, forward slashes) + recompute path_hash
+- Targets ONLY one host_name (Tower1 by default)
+- Safe with UNIQUE(host_name, path_hash)
+- Handles collisions by skipping conflicting rows and logging them
+- DRY_RUN supported
+"""
+
+import sys
+import time
+import hashlib
+import posixpath
+import unicodedata
+import pymysql
+from pymysql.err import IntegrityError
+
+# =========================
+# CONFIG
+# =========================
+HOST_TO_FIX = "Tower"     # <-- set your Unraid host_name exactly as stored in DB
+DRY_RUN = True             # <-- first run True; then switch to False to apply
+BATCH_SELECT_FETCH = 5000  # server-side cursor fetch size (streaming)
+COMMIT_EVERY = 2000        # commit after N successful updates (when DRY_RUN=False)
+LOG_EVERY = 50000          # progress print
+
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+    "autocommit": False,
+}
+
+# =========================
+# CANONICALIZATION
+# =========================
+def canonical_path(path_str: str) -> str:
+    if not path_str:
+        return path_str
+    path_str = path_str.replace("\\", "/")
+    path_str = posixpath.normpath(path_str)
+    path_str = unicodedata.normalize("NFC", path_str)
+    return path_str
+
+def md5_bytes(path_str: str) -> bytes:
+    return hashlib.md5(path_str.encode("utf-8")).digest()  # 16 raw bytes for BINARY(16)
+
+# =========================
+# MAIN
+# =========================
+def main():
+    print(f"[{time.strftime('%Y-%m-%d %H:%M:%S')}] 🚀 Tower path_hash migration")
+    print(f"Host: {HOST_TO_FIX}")
+    print(f"DRY_RUN: {DRY_RUN}")
+    sys.stdout.flush()
+
+    db = pymysql.connect(**DB_CONFIG)
+
+    # streaming cursor for reading
+    read_cur = db.cursor(pymysql.cursors.SSCursor)
+    read_cur.execute(
+        """
+        SELECT id, full_path, path_hash
+        FROM file_md5_index
+        WHERE host_name = %s
+        """,
+        (HOST_TO_FIX,),
+    )
+
+    # normal cursor for updates
+    upd_cur = db.cursor()
+
+    total = 0
+    needs_change = 0
+    updated_ok = 0
+    collisions = 0
+    other_errors = 0
+
+    start = time.time()
+    pending_commits = 0
+
+    # Optional: make server-side cursor fetch a bit larger
+    # (PyMySQL streams regardless; this just makes loop smoother)
+    # Not strictly necessary.
+
+    while True:
+        rows = read_cur.fetchmany(BATCH_SELECT_FETCH)
+        if not rows:
+            break
+
+        for rec_id, full_path, stored_hash in rows:
+            total += 1
+
+            new_path = canonical_path(full_path)
+            new_hash = md5_bytes(new_path)
+
+            # already canonical & correct
+            if new_path == full_path and new_hash == stored_hash:
+                if total % LOG_EVERY == 0:
+                    elapsed = time.time() - start
+                    print(f"Checked {total:,} | needs_change {needs_change:,} | updated {updated_ok:,} | collisions {collisions:,} | {elapsed:.1f}s")
+                    sys.stdout.flush()
+                continue
+
+            needs_change += 1
+
+            if DRY_RUN:
+                # in dry-run we just count; no DB writes
+                continue
+
+            # Update with collision handling via UNIQUE(host_name, path_hash)
+            try:
+                # Use a savepoint so a duplicate-key error doesn't kill the whole transaction
+                upd_cur.execute("SAVEPOINT sp_one;")
+
+                upd_cur.execute(
+                    """
+                    UPDATE file_md5_index
+                    SET full_path = %s,
+                        path_hash = %s
+                    WHERE id = %s
+                    """,
+                    (new_path, new_hash, rec_id),
+                )
+
+                upd_cur.execute("RELEASE SAVEPOINT sp_one;")
+
+                updated_ok += 1
+                pending_commits += 1
+
+                if pending_commits >= COMMIT_EVERY:
+                    db.commit()
+                    pending_commits = 0
+
+            except IntegrityError as e:
+                # Duplicate key = collision on (host_name, path_hash)
+                # This means some OTHER row in the same host already has this new_hash.
+                upd_cur.execute("ROLLBACK TO SAVEPOINT sp_one;")
+                upd_cur.execute("RELEASE SAVEPOINT sp_one;")
+                collisions += 1
+
+                # Print a short line occasionally (avoid huge spam)
+                if collisions <= 50 or collisions % 1000 == 0:
+                    print(f"⚠ COLLISION id={rec_id} | {e}")
+                    sys.stdout.flush()
+
+            except Exception as e:
+                upd_cur.execute("ROLLBACK TO SAVEPOINT sp_one;")
+                upd_cur.execute("RELEASE SAVEPOINT sp_one;")
+                other_errors += 1
+                if other_errors <= 50 or other_errors % 1000 == 0:
+                    print(f"❌ ERROR id={rec_id} | {e}")
+                    sys.stdout.flush()
+
+            if total % LOG_EVERY == 0:
+                elapsed = time.time() - start
+                print(f"Checked {total:,} | needs_change {needs_change:,} | updated {updated_ok:,} | collisions {collisions:,} | {elapsed:.1f}s")
+                sys.stdout.flush()
+
+    # finalize
+    if not DRY_RUN:
+        if pending_commits:
+            db.commit()
+        print("✅ Migration finished (committed).")
+    else:
+        print("⚠ DRY_RUN finished (no changes written).")
+
+    elapsed = time.time() - start
+    print("=" * 70)
+    print(f"Total rows checked   : {total:,}")
+    print(f"Rows needing change  : {needs_change:,}")
+    print(f"Rows updated         : {updated_ok:,}")
+    print(f"Collisions (skipped) : {collisions:,}")
+    print(f"Other errors         : {other_errors:,}")
+    print(f"Elapsed              : {elapsed:.1f}s")
+    print("=" * 70)
+
+    read_cur.close()
+    upd_cur.close()
+    db.close()
+
+if __name__ == "__main__":
+    main()

WalkFilesOnBackupHDD/51 testthoseneedchangewhetherok.py

@@ -0,0 +1,146 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import os
+import hashlib
+import posixpath
+import unicodedata
+import pymysql
+import time
+
+# =========================
+# CONFIG
+# =========================
+
+HOST_TO_CHECK = "Tower"
+WINDOWS_UNC_BASE = r"\\tower"
+
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+}
+
+PRINT_FIRST_CHANGES = 20
+LOG_EVERY = 5000
+
+# =========================
+# CANONICAL
+# =========================
+
+def canonical_path(path_str):
+    path_str = path_str.replace("\\", "/")
+    path_str = posixpath.normpath(path_str)
+    path_str = unicodedata.normalize("NFC", path_str)
+    return path_str
+
+def md5_bytes(path_str):
+    return hashlib.md5(path_str.encode("utf-8")).digest()
+
+# =========================
+# PATH MAP
+# =========================
+
+def linux_to_windows_unc(linux_path):
+    rel = linux_path[len("/mnt/user/"):]
+    return os.path.join(WINDOWS_UNC_BASE, *rel.split("/"))
+
+# =========================
+# MAIN
+# =========================
+
+def main():
+
+    print("=" * 70)
+    print("🔍 Tower Canonical Path SMB Verification")
+    print(f"Host: {HOST_TO_CHECK}")
+    print(f"UNC Base: {WINDOWS_UNC_BASE}")
+    print("=" * 70)
+
+    db = pymysql.connect(**DB_CONFIG)
+    cur = db.cursor(pymysql.cursors.SSCursor)
+
+    cur.execute("""
+        SELECT id, full_path, path_hash
+        FROM file_md5_index
+        WHERE host_name = %s
+    """, (HOST_TO_CHECK,))
+
+    total = 0
+    needs_change = 0
+    exists_ok = 0
+    missing = 0
+
+    printed_changes = 0
+
+    start = time.time()
+
+    for rec_id, full_path, stored_hash in cur:
+
+        total += 1
+
+        new_path = canonical_path(full_path)
+        new_hash = md5_bytes(new_path)
+
+        # Already canonical
+        if new_path == full_path and new_hash == stored_hash:
+            continue
+
+        needs_change += 1
+
+        win_path = linux_to_windows_unc(new_path)
+        exists = os.path.exists(win_path)
+
+        if exists:
+            exists_ok += 1
+        else:
+            missing += 1
+
+        # ---- Print first examples ----
+        if printed_changes < PRINT_FIRST_CHANGES:
+            print("\n🔧 CHANGE DETECTED")
+            print(f"ID       : {rec_id}")
+            print(f"DB PATH  : {full_path}")
+            print(f"NEW PATH : {new_path}")
+            print(f"WIN PATH : {win_path}")
+            print(f"Exists   : {exists}")
+            printed_changes += 1
+
+        # ---- Progress ----
+        if total % LOG_EVERY == 0:
+            elapsed = time.time() - start
+            rate = total / elapsed if elapsed else 0
+
+            print(
+                f"📊 Checked {total:,} rows | "
+                f"Needs change {needs_change:,} | "
+                f"Exists {exists_ok:,} | "
+                f"Missing {missing:,} | "
+                f"{rate:,.0f} rows/sec"
+            )
+
+    # =========================
+    # SUMMARY
+    # =========================
+
+    elapsed = time.time() - start
+
+    print("\n" + "=" * 70)
+    print("✅ FINAL SUMMARY")
+    print("=" * 70)
+    print(f"Total scanned     : {total:,}")
+    print(f"Needs change     : {needs_change:,}")
+    print(f"Exists on Tower  : {exists_ok:,}")
+    print(f"Missing on Tower : {missing:,}")
+    print(f"Runtime          : {elapsed:.1f}s")
+    print("=" * 70)
+
+    cur.close()
+    db.close()
+
+
+if __name__ == "__main__":
+    main()

WalkFilesOnBackupHDD/53 towerpathcorrection.py

@@ -0,0 +1,139 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+"""
+TOWER PATH NORMALIZATION MIGRATION
+----------------------------------
+✔ Normalizes full_path → NFC canonical
+✔ Recalculates path_hash
+✔ Uses two DB connections (streaming safe)
+✔ Idempotent (safe to rerun)
+✔ Production safe
+"""
+
+import pymysql
+import hashlib
+import posixpath
+import unicodedata
+import time
+
+# =========================
+# CONFIG
+# =========================
+
+HOST_TO_FIX = "tower1"
+BATCH_FETCH = 5000
+COMMIT_EVERY = 2000
+
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+    "autocommit": False,
+}
+
+# =========================
+# CANONICALIZATION
+# =========================
+
+def canonical_path(path_str: str) -> str:
+    path_str = path_str.replace("\\", "/")
+    path_str = posixpath.normpath(path_str)
+    path_str = unicodedata.normalize("NFC", path_str)
+    return path_str
+
+def md5_bytes(path_str: str) -> bytes:
+    return hashlib.md5(path_str.encode("utf-8")).digest()
+
+# =========================
+# MAIN
+# =========================
+
+def main():
+
+    print("=" * 70)
+    print("🚀 TOWER PATH NORMALIZATION MIGRATION")
+    print(f"Host: {HOST_TO_FIX}")
+    print("=" * 70)
+
+    start = time.time()
+
+    # --- TWO CONNECTIONS ---
+    db_read = pymysql.connect(**DB_CONFIG)
+    db_write = pymysql.connect(**DB_CONFIG)
+
+    read_cur = db_read.cursor(pymysql.cursors.SSCursor)
+    write_cur = db_write.cursor()
+
+    read_cur.execute("""
+        SELECT id, full_path, path_hash
+        FROM file_md5_index
+        WHERE host_name = %s
+    """, (HOST_TO_FIX,))
+
+    total = 0
+    updated = 0
+    skipped = 0
+    pending_commit = 0
+
+    while True:
+
+        rows = read_cur.fetchmany(BATCH_FETCH)
+
+        if not rows:
+            break
+
+        for rec_id, full_path, stored_hash in rows:
+
+            total += 1
+
+            new_path = canonical_path(full_path)
+            new_hash = md5_bytes(new_path)
+
+            if new_path == full_path and new_hash == stored_hash:
+                skipped += 1
+                continue
+
+            write_cur.execute("""
+                UPDATE file_md5_index
+                SET full_path = %s,
+                    path_hash = %s
+                WHERE id = %s
+            """, (new_path, new_hash, rec_id))
+
+            updated += 1
+            pending_commit += 1
+
+            if pending_commit >= COMMIT_EVERY:
+                db_write.commit()
+                pending_commit = 0
+
+        print(
+            f"Checked {total:,} | Updated {updated:,} | Skipped {skipped:,}"
+        )
+
+    if pending_commit:
+        db_write.commit()
+
+    elapsed = time.time() - start
+
+    print("\n" + "=" * 70)
+    print("✅ MIGRATION FINISHED")
+    print("=" * 70)
+    print(f"Total checked : {total:,}")
+    print(f"Rows updated  : {updated:,}")
+    print(f"Rows skipped  : {skipped:,}")
+    print(f"Runtime       : {elapsed:.1f}s")
+    print("=" * 70)
+
+    read_cur.close()
+    write_cur.close()
+    db_read.close()
+    db_write.close()
+
+
+if __name__ == "__main__":
+    main()

WalkFilesOnBackupHDD/54 53_aktualizovanýspolemhotovo.py

@@ -0,0 +1,123 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import pymysql
+import hashlib
+import posixpath
+import unicodedata
+import time
+from pymysql.err import IntegrityError
+
+HOST_TO_FIX = "tower1"
+BATCH_FETCH = 5000
+COMMIT_EVERY = 2000
+
+DB_CONFIG = {
+    "host": "192.168.1.50",
+    "port": 3306,
+    "user": "root",
+    "password": "Vlado9674+",
+    "database": "torrents",
+    "charset": "utf8mb4",
+    "autocommit": False,
+}
+
+def canonical_path(path_str: str) -> str:
+    path_str = path_str.replace("\\", "/")
+    path_str = posixpath.normpath(path_str)
+    path_str = unicodedata.normalize("NFC", path_str)
+    return path_str
+
+def md5_bytes(path_str: str) -> bytes:
+    return hashlib.md5(path_str.encode("utf-8")).digest()
+
+def main():
+
+    print("🚀 PATH NORMALIZATION WITH TEMP FLAG")
+
+    db_read = pymysql.connect(**DB_CONFIG)
+    db_write = pymysql.connect(**DB_CONFIG)
+
+    read_cur = db_read.cursor(pymysql.cursors.SSCursor)
+    write_cur = db_write.cursor()
+
+    read_cur.execute("""
+        SELECT id, full_path, path_hash
+        FROM file_md5_index
+        WHERE host_name=%s
+          AND path_corrected=0
+    """, (HOST_TO_FIX,))
+
+    total = updated = skipped = deleted = 0
+    pending_commit = 0
+
+    while True:
+
+        rows = read_cur.fetchmany(BATCH_FETCH)
+        if not rows:
+            break
+
+        for rec_id, full_path, stored_hash in rows:
+
+            total += 1
+
+            new_path = canonical_path(full_path)
+            new_hash = md5_bytes(new_path)
+
+            try:
+
+                # už je canonical → jen označit
+                if new_path == full_path and new_hash == stored_hash:
+
+                    write_cur.execute("""
+                        UPDATE file_md5_index
+                        SET path_corrected=1
+                        WHERE id=%s
+                    """, (rec_id,))
+
+                    skipped += 1
+
+                else:
+
+                    write_cur.execute("""
+                        UPDATE file_md5_index
+                        SET full_path=%s,
+                            path_hash=%s,
+                            path_corrected=1
+                        WHERE id=%s
+                    """, (new_path, new_hash, rec_id))
+
+                    updated += 1
+
+            except IntegrityError as e:
+
+                # duplicate canonical path → smazat
+                if e.args[0] == 1062:
+                    write_cur.execute(
+                        "DELETE FROM file_md5_index WHERE id=%s",
+                        (rec_id,)
+                    )
+                    deleted += 1
+                else:
+                    raise
+
+            pending_commit += 1
+
+            if pending_commit >= COMMIT_EVERY:
+                db_write.commit()
+                pending_commit = 0
+
+        print(f"Checked {total:,} | Updated {updated:,} | Skipped {skipped:,} | Deleted {deleted:,}")
+
+    if pending_commit:
+        db_write.commit()
+
+    read_cur.close()
+    write_cur.close()
+    db_read.close()
+    db_write.close()
+
+    print("✅ FINISHED")
+
+if __name__ == "__main__":
+    main()