Refactoring DbEngine

This commit is contained in:
2025-05-10 20:40:03 +02:00
parent 2daff83e67
commit e1c10183eb
12 changed files with 515 additions and 210 deletions

View File

@@ -25,8 +25,8 @@ class DbException(Exception):
class RefHelper:
def __init__(self, get_obj_path):
self.get_obj_path = get_obj_path
def __init__(self, get_ref_path):
self.get_ref_path = get_ref_path
def save_ref(self, obj):
"""
@@ -40,12 +40,12 @@ class RefHelper:
digest = get_stream_digest(buffer)
target_path = self.get_obj_path(digest)
target_path = self.get_ref_path(digest)
if not os.path.exists(os.path.dirname(target_path)):
os.makedirs(os.path.dirname(target_path))
buffer.seek(0)
with open(self.get_obj_path(digest), "wb") as file:
with open(self.get_ref_path(digest), "wb") as file:
while chunk := buffer.read(BUFFER_SIZE):
file.write(chunk)
@@ -58,7 +58,7 @@ class RefHelper:
:param digest:
:return:
"""
with open(self.get_obj_path(digest), 'rb') as file:
with open(self.get_ref_path(digest), 'rb') as file:
return pickle.load(file)
@@ -73,40 +73,50 @@ class DbEngine:
def __init__(self, root: str = None):
self.root = root or ".mytools_db"
self.serializer = Serializer(RefHelper(self._get_obj_path))
self.debug_serializer = DebugSerializer(RefHelper(self.debug_load))
self.serializer = Serializer(RefHelper(self._get_ref_path))
self.debug_serializer = DebugSerializer(RefHelper(self._get_ref_path))
self.lock = RLock()
def is_initialized(self):
def is_initialized(self, user_id: str):
"""
:return:
"""
return os.path.exists(self.root)
return os.path.exists(self._get_user_root(user_id))
def init(self):
def init(self, user_id: str):
"""
Make sure that the DbEngine is properly initialized
:return:
"""
if not os.path.exists(self.root):
logger.debug(f"Creating root folder in {os.path.abspath(self.root)}.")
os.mkdir(self.root)
if not os.path.exists(self._get_user_root(user_id)):
logger.debug(f"Creating root folder in {os.path.abspath(self._get_user_root(user_id))}.")
os.makedirs(self._get_user_root(user_id))
def save(self, user_id: str, entry: str, obj: object) -> str:
def save(self, user_id: str, user_email: str, entry: str, obj: object) -> str:
"""
Save a snapshot of an entry
:param user_id:
:param user_email:
:param entry:
:param obj: snapshot to save
:return:
"""
with self.lock:
logger.info(f"Saving {user_id=}, {entry=}, {obj=}")
if not user_id:
raise DbException("user_id is None")
if not user_email:
raise DbException("user_email is None")
if not entry:
raise DbException("entry is None")
# prepare the data
as_dict = self._serialize(obj)
as_dict[TAG_PARENT] = [self._get_entry_digest(entry)]
as_dict[TAG_USER] = user_id
as_dict[TAG_PARENT] = [self._get_entry_digest(user_id, entry)]
as_dict[TAG_USER] = user_email
as_dict[TAG_DATE] = datetime.datetime.now().strftime('%Y%m%d %H:%M:%S %z')
# transform into a stream
@@ -117,7 +127,7 @@ class DbEngine:
# compute the digest to know where to store it
digest = hashlib.sha256(byte_stream).hexdigest()
target_path = self._get_obj_path(digest)
target_path = self._get_obj_path(user_id, digest)
if os.path.exists(target_path):
# the same object is already saved. Noting to do
return digest
@@ -128,8 +138,8 @@ class DbEngine:
with open(target_path, "wb") as file:
file.write(byte_stream)
# update the head to remember where is the latest entry
self._update_head(entry, digest)
# update the head to remember where the latest entry is
self._update_head(user_id, entry, digest)
logger.debug(f"New head for entry '{entry}' is {digest}")
return digest
@@ -144,24 +154,25 @@ class DbEngine:
with self.lock:
logger.info(f"Loading {user_id=}, {entry=}, {digest=}")
digest_to_use = digest or self._get_entry_digest(entry)
digest_to_use = digest or self._get_entry_digest(user_id, entry)
logger.debug(f"Using digest {digest_to_use}.")
if digest_to_use is None:
raise DbException(entry)
target_file = self._get_obj_path(digest_to_use)
target_file = self._get_obj_path(user_id, digest_to_use)
with open(target_file, 'r', encoding='utf-8') as file:
as_dict = json.load(file)
return self._deserialize(as_dict)
def put(self, user_id: str, entry, key: str, value: object):
def put(self, user_id: str, user_email, entry, key: str, value: object):
"""
Save a specific record.
This will create a new snapshot is the record is new or different
You should not mix the usage of put_many() and save() as it's two different way to manage the db
:param user_email:
:param user_id:
:param entry:
:param key:
@@ -182,16 +193,17 @@ class DbEngine:
return False
entry_content[key] = value
self.save(user_id, entry, entry_content)
self.save(user_id, user_email, entry, entry_content)
return True
def put_many(self, user_id: str, entry, items: list):
def put_many(self, user_id: str, user_email, entry, items: list):
"""
Save a list of item as one single snapshot
A new snapshot will not be created if all the items already exist
You should not mix the usage of put_many() and save() as it's two different way to manage the db
:param user_id:
:param user_email:
:param entry:
:param items:
:return:
@@ -213,12 +225,12 @@ class DbEngine:
is_dirty = True
if is_dirty:
self.save(user_id, entry, entry_content)
self.save(user_id, user_email, entry, entry_content)
return True
return False
def exists(self, entry: str):
def exists(self, user_id, entry: str):
"""
Tells if an entry exist
:param user_id:
@@ -226,7 +238,7 @@ class DbEngine:
:return:
"""
with self.lock:
return self._get_entry_digest(entry) is not None
return self._get_entry_digest(user_id, entry) is not None
def get(self, user_id: str, entry: str, key: str | None = None, digest=None):
"""
@@ -247,9 +259,19 @@ class DbEngine:
return entry_content[key]
def debug_head(self):
def debug_root(self):
"""
Lists all folders in the root directory
:return: List of folder names
"""
with self.lock:
head_path = os.path.join(self.root, self.HeadFile)
if not os.path.exists(self.root):
return []
return [f for f in os.listdir(self.root) if os.path.isdir(os.path.join(self.root, f))]
def debug_head(self, user_id):
with self.lock:
head_path = os.path.join(self.root, user_id, self.HeadFile)
# load
try:
with open(head_path, 'r') as file:
@@ -259,14 +281,17 @@ class DbEngine:
return head
def debug_load(self, digest):
def debug_load(self, user_id, digest):
with self.lock:
target_file = self._get_obj_path(digest)
target_file = self._get_obj_path(user_id, digest)
with open(target_file, 'r', encoding='utf-8') as file:
as_dict = json.load(file)
return self.debug_serializer.deserialize(as_dict)
def debug_get_digest(self, user_id, entry):
return self._get_entry_digest(user_id, entry)
def _serialize(self, obj):
"""
Just call the serializer
@@ -280,14 +305,14 @@ class DbEngine:
def _deserialize(self, as_dict):
return self.serializer.deserialize(as_dict)
def _update_head(self, entry, digest):
def _update_head(self, user_id, entry, digest):
"""
Actually dumps the snapshot in file system
:param entry:
:param digest:
:return:
"""
head_path = os.path.join(self.root, self.HeadFile)
head_path = os.path.join(self.root, user_id, self.HeadFile)
# load
try:
with open(head_path, 'r') as file:
@@ -302,13 +327,16 @@ class DbEngine:
with open(head_path, 'w') as file:
json.dump(head, file)
def _get_entry_digest(self, entry):
def _get_user_root(self, user_id):
return os.path.join(self.root, user_id)
def _get_entry_digest(self, user_id, entry):
"""
Search for the latest digest, for a given entry
:param entry:
:return:
"""
head_path = os.path.join(self.root, self.HeadFile)
head_path = os.path.join(self._get_user_root(user_id), self.HeadFile)
try:
with open(head_path, 'r') as file:
head = json.load(file)
@@ -319,17 +347,25 @@ class DbEngine:
except KeyError:
return None
def _get_head_path(self):
def _get_head_path(self, user_id: str):
"""
Location of the Head file
:return:
"""
return os.path.join(self.root, self.HeadFile)
return os.path.join(self._get_user_root(user_id), self.HeadFile)
def _get_obj_path(self, digest):
def _get_obj_path(self, user_id, digest):
"""
Location of objects
:param digest:
:return:
"""
return os.path.join(self.root, "objects", digest[:24], digest)
return os.path.join(self._get_user_root(user_id), "objects", digest[:24], digest)
def _get_ref_path(self, digest):
"""
Location of reference. They are not linked to the user folder
:param digest:
:return:
"""
return os.path.join(self.root, "refs", digest[:24], digest)

View File

@@ -169,6 +169,24 @@ class SettingsManager:
else:
return default
def remove(self, session: dict, key: str):
user_id = session["user_id"] if session else NO_SESSION
user_email = session["user_email"] if session else NOT_LOGGED
return self._db_engine.remove(user_email, user_id, key)
def update(self, session: dict, old_key: str, key: str, value: object):
user_id = session["user_id"] if session else NO_SESSION
user_email = session["user_email"] if session else NOT_LOGGED
def _update_helper(_old_key, _key, _value):
pass
if hasattr(self._db_engine, "lock"):
with self._db_engine.lock:
_update_helper(old_key, key, value)
else:
_update_helper(old_key, key, value)
def init_user(self, user_id: str, user_email: str):
"""
Init the settings block space for a user