prefect-client 3.0.0rc18__py3-none-any.whl → 3.0.0rc20__py3-none-any.whl

prefect/records/filesystem.py

@@ -0,0 +1,207 @@
+import json
+import time
+from pathlib import Path
+from typing import Dict, Optional
+
+import pendulum
+from typing_extensions import TypedDict
+
+from prefect.logging.loggers import get_logger
+from prefect.records.base import RecordStore, TransactionRecord
+from prefect.results import BaseResult
+from prefect.transactions import IsolationLevel
+
+logger = get_logger(__name__)
+
+
+class _LockInfo(TypedDict):
+    """
+    A dictionary containing information about a lock.
+
+    Attributes:
+        holder: The holder of the lock.
+        expiration: Datetime when the lock expires.
+        path: Path to the lock file.
+    """
+
+    holder: str
+    expiration: Optional[pendulum.DateTime]
+    path: Path
+
+
+class FileSystemRecordStore(RecordStore):
+    """
+    A record store that stores data on the local filesystem.
+
+    Locking is implemented using a lock file with the same name as the record file,
+    but with a `.lock` extension.
+
+    Attributes:
+        records_directory: the directory where records are stored; defaults to
+            `{PREFECT_HOME}/records`
+    """
+
+    def __init__(self, records_directory: Path):
+        self.records_directory = records_directory
+        self._locks: Dict[str, _LockInfo] = {}
+
+    def _ensure_records_directory_exists(self):
+        self.records_directory.mkdir(parents=True, exist_ok=True)
+
+    def _lock_path_for_key(self, key: str) -> Path:
+        if (lock_info := self._locks.get(key)) is not None:
+            return lock_info["path"]
+        return self.records_directory.joinpath(key).with_suffix(".lock")
+
+    def _get_lock_info(self, key: str, use_cache=True) -> Optional[_LockInfo]:
+        if use_cache:
+            if (lock_info := self._locks.get(key)) is not None:
+                print("Got lock info from cache")
+                return lock_info
+
+        lock_path = self._lock_path_for_key(key)
+
+        try:
+            with open(lock_path, "r") as lock_file:
+                lock_info = json.load(lock_file)
+                lock_info["path"] = lock_path
+                expiration = lock_info.get("expiration")
+                lock_info["expiration"] = (
+                    pendulum.parse(expiration) if expiration is not None else None
+                )
+            self._locks[key] = lock_info
+            print("Got lock info from file")
+            return lock_info
+        except FileNotFoundError:
+            return None
+
+    def read(
+        self, key: str, holder: Optional[str] = None
+    ) -> Optional[TransactionRecord]:
+        if not self.exists(key):
+            return None
+
+        holder = holder or self.generate_default_holder()
+
+        if self.is_locked(key) and not self.is_lock_holder(key, holder):
+            self.wait_for_lock(key)
+        record_data = self.records_directory.joinpath(key).read_text()
+        return TransactionRecord(
+            key=key, result=BaseResult.model_validate_json(record_data)
+        )
+
+    def write(self, key: str, result: BaseResult, holder: Optional[str] = None) -> None:
+        self._ensure_records_directory_exists()
+
+        if self.is_locked(key) and not self.is_lock_holder(key, holder):
+            raise ValueError(
+                f"Cannot write to transaction with key {key} because it is locked by another holder."
+            )
+
+        record_path = self.records_directory.joinpath(key)
+        record_path.touch(exist_ok=True)
+        record_data = result.model_dump_json()
+        record_path.write_text(record_data)
+
+    def exists(self, key: str) -> bool:
+        return self.records_directory.joinpath(key).exists()
+
+    def supports_isolation_level(self, isolation_level: IsolationLevel) -> bool:
+        return isolation_level in {
+            IsolationLevel.READ_COMMITTED,
+            IsolationLevel.SERIALIZABLE,
+        }
+
+    def acquire_lock(
+        self,
+        key: str,
+        holder: Optional[str] = None,
+        acquire_timeout: Optional[float] = None,
+        hold_timeout: Optional[float] = None,
+    ) -> bool:
+        holder = holder or self.generate_default_holder()
+
+        self._ensure_records_directory_exists()
+        lock_path = self._lock_path_for_key(key)
+
+        if self.is_locked(key) and not self.is_lock_holder(key, holder):
+            lock_free = self.wait_for_lock(key, acquire_timeout)
+            if not lock_free:
+                return False
+
+        try:
+            Path(lock_path).touch(exist_ok=False)
+        except FileExistsError:
+            if not self.is_lock_holder(key, holder):
+                logger.debug(
+                    f"Another actor acquired the lock for record with key {key}. Trying again."
+                )
+                return self.acquire_lock(key, holder, acquire_timeout, hold_timeout)
+        expiration = (
+            pendulum.now("utc") + pendulum.duration(seconds=hold_timeout)
+            if hold_timeout is not None
+            else None
+        )
+
+        with open(Path(lock_path), "w") as lock_file:
+            json.dump(
+                {
+                    "holder": holder,
+                    "expiration": str(expiration) if expiration is not None else None,
+                },
+                lock_file,
+            )
+
+        self._locks[key] = {
+            "holder": holder,
+            "expiration": expiration,
+            "path": lock_path,
+        }
+
+        return True
+
+    def release_lock(self, key: str, holder: Optional[str] = None) -> None:
+        holder = holder or self.generate_default_holder()
+        lock_path = self._lock_path_for_key(key)
+        if not self.is_locked(key):
+            ValueError(f"No lock for transaction with key {key}")
+        if self.is_lock_holder(key, holder):
+            Path(lock_path).unlink(missing_ok=True)
+            self._locks.pop(key, None)
+        else:
+            raise ValueError(f"No lock held by {holder} for transaction with key {key}")
+
+    def is_locked(self, key: str, use_cache: bool = False) -> bool:
+        if (lock_info := self._get_lock_info(key, use_cache=use_cache)) is None:
+            return False
+
+        if (expiration := lock_info.get("expiration")) is None:
+            return True
+
+        expired = expiration < pendulum.now("utc")
+        if expired:
+            Path(lock_info["path"]).unlink()
+            self._locks.pop(key, None)
+            return False
+        else:
+            return True
+
+    def is_lock_holder(self, key: str, holder: Optional[str] = None) -> bool:
+        if not self.is_locked(key):
+            return False
+
+        holder = holder or self.generate_default_holder()
+        if not self.is_locked(key):
+            return False
+        if (lock_info := self._get_lock_info(key)) is None:
+            return False
+        return lock_info["holder"] == holder
+
+    def wait_for_lock(self, key: str, timeout: Optional[float] = None) -> bool:
+        seconds_waited = 0
+        while self.is_locked(key, use_cache=False):
+            if timeout and seconds_waited >= timeout:
+                return False
+            seconds_waited += 0.1
+            time.sleep(0.1)
+        return True

prefect/records/memory.py

@@ -2,6 +2,7 @@ import threading
 from typing import Dict, Optional, TypedDict
 
 from prefect.results import BaseResult
+from prefect.transactions import IsolationLevel
 
 from .base import RecordStore, TransactionRecord
 
@@ -38,10 +39,16 @@ class MemoryRecordStore(RecordStore):
         self._locks: Dict[str, _LockInfo] = {}
         self._records: Dict[str, TransactionRecord] = {}
 
-    def read(self, key: str) -> Optional[TransactionRecord]:
+    def read(
+        self, key: str, holder: Optional[str] = None
+    ) -> Optional[TransactionRecord]:
+        holder = holder or self.generate_default_holder()
+
+        if self.is_locked(key) and not self.is_lock_holder(key, holder):
+            self.wait_for_lock(key)
         return self._records.get(key)
 
-    def write(self, key: str, value: BaseResult, holder: Optional[str] = None) -> None:
+    def write(self, key: str, result: BaseResult, holder: Optional[str] = None) -> None:
         holder = holder or self.generate_default_holder()
 
         with self._locks_dict_lock:
@@ -49,11 +56,17 @@ class MemoryRecordStore(RecordStore):
                 raise ValueError(
                     f"Cannot write to transaction with key {key} because it is locked by another holder."
                 )
-        self._records[key] = TransactionRecord(key=key, result=value)
+        self._records[key] = TransactionRecord(key=key, result=result)
 
     def exists(self, key: str) -> bool:
         return key in self._records
 
+    def supports_isolation_level(self, isolation_level: IsolationLevel) -> bool:
+        return isolation_level in {
+            IsolationLevel.READ_COMMITTED,
+            IsolationLevel.SERIALIZABLE,
+        }
+
     def _expire_lock(self, key: str):
         """
         Expire the lock for the given key.

prefect/records/result_store.py

@@ -1,22 +1,26 @@
 from dataclasses import dataclass
-from typing import Any
+from typing import Any, Optional
 
 import pendulum
 
 from prefect.results import BaseResult, PersistedResult, ResultFactory
+from prefect.transactions import IsolationLevel
 from prefect.utilities.asyncutils import run_coro_as_sync
 
-from .base import RecordStore
+from .base import RecordStore, TransactionRecord
 
 
 @dataclass
 class ResultFactoryStore(RecordStore):
     result_factory: ResultFactory
-    cache: PersistedResult = None
+    cache: Optional[PersistedResult] = None
 
     def exists(self, key: str) -> bool:
         try:
-            result = self.read(key)
+            record = self.read(key)
+            if not record:
+                return False
+            result = record.result
             result.get(_sync=True)
             if result.expiration:
                 # if the result has an expiration,
@@ -29,25 +33,26 @@ class ResultFactoryStore(RecordStore):
         except Exception:
             return False
 
-    def read(self, key: str) -> BaseResult:
+    def read(self, key: str, holder: Optional[str] = None) -> TransactionRecord:
         if self.cache:
-            return self.cache
+            return TransactionRecord(key=key, result=self.cache)
         try:
             result = PersistedResult(
                 serializer_type=self.result_factory.serializer.type,
                 storage_block_id=self.result_factory.storage_block_id,
                 storage_key=key,
             )
-            return result
+            return TransactionRecord(key=key, result=result)
         except Exception:
             # this is a bit of a bandaid for functionality
             raise ValueError("Result could not be read")
 
-    def write(self, key: str, value: Any) -> BaseResult:
-        if isinstance(value, PersistedResult):
+    def write(self, key: str, result: Any, holder: Optional[str] = None) -> None:
+        if isinstance(result, PersistedResult):
             # if the value is already a persisted result, write it
-            value.write(_sync=True)
-            return value
-        elif isinstance(value, BaseResult):
-            return value
-        return run_coro_as_sync(self.result_factory.create_result(obj=value, key=key))
+            result.write(_sync=True)
+        elif not isinstance(result, BaseResult):
+            run_coro_as_sync(self.result_factory.create_result(obj=result, key=key))
+
+    def supports_isolation_level(self, isolation_level: IsolationLevel) -> bool:
+        return isolation_level == IsolationLevel.READ_COMMITTED