notebook

indexer/db.py

@@ -0,0 +1,91 @@
+import pymysql
+import hashlib
+from indexer.config import DB_CONFIG, ROOT_NAME
+
+
+def get_connection():
+    return pymysql.connect(**DB_CONFIG)
+
+
+def preload_mark_all_missing():
+    """
+    Na začátku běhu:
+    označí všechny soubory jako neexistující.
+    Ty, které skener znovu najde, se přepnou zpět na exists_now = 1.
+    """
+    conn = get_connection()
+    try:
+        with conn.cursor() as cur:
+            cur.execute("UPDATE files SET exists_now = 0")
+        conn.commit()
+    finally:
+        conn.close()
+
+
+def path_hash(path: str) -> bytes:
+    """
+    MD5 hash cesty – pouze identifikátor, ne bezpečnostní hash
+    """
+    return hashlib.md5(path.encode("utf-8")).digest()
+
+
+def find_file_by_path(cur, path_hash_bytes):
+    cur.execute(
+        """
+        SELECT id, file_size, mtime, content_hash
+        FROM files
+        WHERE path_hash = %s
+        """,
+        (path_hash_bytes,)
+    )
+    return cur.fetchone()
+
+
+def insert_file(cur, file):
+    cur.execute(
+        """
+        INSERT INTO files (
+            root_name, full_path, path_hash,
+            file_name, directory,
+            file_size, mtime, content_hash,
+            first_seen, last_seen, exists_now
+        )
+        VALUES (
+            %s, %s, %s,
+            %s, %s,
+            %s, %s, %s,
+            NOW(), NOW(), 1
+        )
+        """,
+        (
+            ROOT_NAME,
+            file["full_path"],
+            path_hash(file["full_path"]),
+            file["file_name"],
+            file["directory"],
+            file["size"],
+            file["mtime"],
+            file["content_hash"],
+        )
+    )
+    return cur.lastrowid
+
+
+def update_file(cur, file_id, file):
+    cur.execute(
+        """
+        UPDATE files
+        SET file_size = %s,
+            mtime = %s,
+            content_hash = %s,
+            last_seen = NOW(),
+            exists_now = 1
+        WHERE id = %s
+        """,
+        (
+            file["size"],
+            file["mtime"],
+            file["content_hash"],
+            file_id,
+        )
+    )