runnable 0.25.0__py3-none-any.whl → 0.27.0__py3-none-any.whl

extensions/catalog/any_path.py

@@ -0,0 +1,201 @@
+import logging
+import os
+import shutil
+from abc import abstractmethod
+from pathlib import Path
+from typing import Any, Dict, List
+
+from cloudpathlib import CloudPath
+
+from runnable import defaults, utils
+from runnable.catalog import BaseCatalog
+from runnable.datastore import DataCatalog
+
+logger = logging.getLogger(defaults.LOGGER_NAME)
+
+
+class AnyPathCatalog(BaseCatalog):
+    """
+    A Catalog handler that uses the local file system for cataloging.
+
+    Note: Do not use this if the steps of the pipeline run on different compute environments.
+
+    Example config:
+
+    catalog:
+      type: file-system
+      config:
+        catalog_location: The location to store the catalog.
+        compute_data_folder: The folder to source the data from.
+
+    """
+
+    @abstractmethod
+    def get_summary(self) -> Dict[str, Any]: ...
+
+    @abstractmethod
+    def upload_to_catalog(self, file: Path) -> None: ...
+
+    @abstractmethod
+    def download_from_catalog(self, file: Path | CloudPath) -> None: ...
+
+    @abstractmethod
+    def get_catalog_location(self) -> Path | CloudPath:
+        """
+        For local file systems, this is the .catalog/run_id/compute_data_folder
+        For cloud systems, this is s3://bucket/run_id/compute_data_folder
+        """
+        ...
+
+    def get(self, name: str) -> List[DataCatalog]:
+        """
+        Get the file by matching glob pattern to the name
+
+        Args:
+            name ([str]): A glob matching the file name
+            run_id ([str]): The run id
+
+        Raises:
+            Exception: If the catalog location does not exist
+
+        Returns:
+            List(object) : A list of catalog objects
+        """
+        run_catalog = self.get_catalog_location()
+
+        # Iterate through the contents of the run_catalog and copy the files that fit the name pattern
+        # We should also return a list of data hashes
+        glob_files = run_catalog.glob(name)
+        logger.debug(
+            f"Glob identified {glob_files} as matches to from the catalog location: {run_catalog}"
+        )
+
+        data_catalogs = []
+        run_log_store = self._context.run_log_store
+        for file in glob_files:
+            if file.is_dir():
+                # Need not add a data catalog for the folder
+                continue
+
+            if str(file).endswith(".execution.log"):
+                continue
+
+            self.download_from_catalog(file)
+            relative_file_path = file.relative_to(run_catalog)  # type: ignore
+
+            data_catalog = run_log_store.create_data_catalog(str(relative_file_path))
+            data_catalog.catalog_relative_path = str(relative_file_path)
+            data_catalog.data_hash = utils.get_data_hash(str(relative_file_path))
+            data_catalog.stage = "get"
+            data_catalogs.append(data_catalog)
+
+        if not data_catalogs:
+            raise Exception(f"Did not find any files matching {name} in {run_catalog}")
+
+        return data_catalogs
+
+    def put(self, name: str) -> List[DataCatalog]:
+        """
+        Put the files matching the glob pattern into the catalog.
+
+        If previously synced catalogs are provided, and no changes were observed, we do not sync them.
+
+        Args:
+            name (str): The glob pattern of the files to catalog
+            run_id (str): The run id of the run
+            compute_data_folder (str, optional): The compute data folder to sync from. Defaults to settings default.
+            synced_catalogs (dict, optional): dictionary of previously synced catalogs. Defaults to None.
+
+        Raises:
+            Exception: If the compute data folder does not exist.
+
+        Returns:
+            List(object) : A list of catalog objects
+        """
+        run_id = self._context.run_id
+        logger.info(
+            f"Using the {self.service_name} catalog and trying to put {name} for run_id: {run_id}"
+        )
+
+        copy_from = Path(self.compute_data_folder)
+
+        if not copy_from.is_dir():
+            msg = (
+                f"Expected compute data folder to be present at: {copy_from} but not found. \n"
+                "Note: runnable does not create the compute data folder for you. Please ensure that the "
+                "folder exists.\n"
+            )
+            raise Exception(msg)
+
+        # Iterate through the contents of copy_from and if the name matches, we move them to the run_catalog
+        # We should also return a list of datastore.DataCatalog items
+        glob_files = copy_from.glob(name)
+        logger.debug(
+            f"Glob identified {glob_files} as matches to from the compute data folder: {copy_from}"
+        )
+
+        data_catalogs = []
+        run_log_store = self._context.run_log_store
+        for file in glob_files:
+            if file.is_dir():
+                # Need not add a data catalog for the folder
+                continue
+
+            relative_file_path = file.relative_to(copy_from)
+
+            data_catalog = run_log_store.create_data_catalog(str(relative_file_path))
+            data_catalog.catalog_relative_path = (
+                run_id + os.sep + str(relative_file_path)
+            )
+            data_catalog.data_hash = utils.get_data_hash(str(file))
+            data_catalog.stage = "put"
+            data_catalogs.append(data_catalog)
+
+            # TODO: Think about syncing only if the file is changed
+            self.upload_to_catalog(file)
+
+        if not data_catalogs:
+            raise Exception(f"Did not find any files matching {name} in {copy_from}")
+
+        return data_catalogs
+
+    def sync_between_runs(self, previous_run_id: str, run_id: str):
+        """
+        Given the previous run id, sync the catalogs between the current one and previous
+
+        Args:
+            previous_run_id (str): The previous run id to sync the catalogs from
+            run_id (str): The run_id to which the data catalogs should be synced to.
+
+        Raises:
+            Exception: If the previous run log does not exist in the catalog
+
+        """
+        logger.info(
+            f"Using the {self.service_name} catalog and syncing catalogs"
+            "between old: {previous_run_id} to new: {run_id}"
+        )
+
+        catalog_location = Path(self.get_catalog_location())
+        run_catalog = catalog_location / run_id
+        utils.safe_make_dir(run_catalog)
+
+        if not utils.does_dir_exist(catalog_location / previous_run_id):
+            msg = (
+                f"Catalogs from previous run : {previous_run_id} are not found.\n"
+                "Note: Please provision the catalog objects generated by previous run in the same catalog location"
+                " as the current run, even if the catalog handler for the previous run was different"
+            )
+            raise Exception(msg)
+
+        cataloged_files = list((catalog_location / previous_run_id).glob("*"))
+
+        for cataloged_file in cataloged_files:
+            if str(cataloged_file).endswith("execution.log"):
+                continue
+
+            if cataloged_file.is_file():
+                shutil.copy(cataloged_file, run_catalog / cataloged_file.name)
+            else:
+                shutil.copytree(cataloged_file, run_catalog / cataloged_file.name)
+            logger.info(f"Copied file from: {cataloged_file} to {run_catalog}")

extensions/catalog/file_system.py

@@ -1,253 +1,52 @@
 import logging
-import os
 import shutil
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Any
 
-from runnable import defaults, utils
-from runnable.catalog import BaseCatalog, is_catalog_out_of_sync
-from runnable.datastore import DataCatalog
+from cloudpathlib import CloudPath
+from pydantic import Field
 
-logger = logging.getLogger(defaults.LOGGER_NAME)
-
-
-class FileSystemCatalog(BaseCatalog):
-    """
-    A Catalog handler that uses the local file system for cataloging.
-
-    Note: Do not use this if the steps of the pipeline run on different compute environments.
+from extensions.catalog.any_path import AnyPathCatalog
+from runnable import defaults
 
-    Example config:
-
-    catalog:
-      type: file-system
-      config:
-        catalog_location: The location to store the catalog.
-        compute_data_folder: The folder to source the data from.
+logger = logging.getLogger(defaults.LOGGER_NAME)
 
-    """
 
+class FileSystemCatalog(AnyPathCatalog):
     service_name: str = "file-system"
-    catalog_location: str = defaults.CATALOG_LOCATION_FOLDER
 
-    def get_catalog_location(self):
-        return self.catalog_location
+    catalog_location: str = Field(default=defaults.CATALOG_LOCATION_FOLDER)
 
-    def get_summary(self) -> Dict[str, Any]:
-        summary = {
-            "Catalog Location": self.get_catalog_location(),
+    def get_summary(self) -> dict[str, Any]:
+        return {
+            "compute_data_folder": self.compute_data_folder,
+            "catalog_location": self.catalog_location,
         }
 
-        return summary
-
-    def get(
-        self, name: str, run_id: str, compute_data_folder: str = "", **kwargs
-    ) -> List[DataCatalog]:
-        """
-        Get the file by matching glob pattern to the name
+    def get_catalog_location(self) -> Path:
+        run_id = self._context.run_id
+        return Path(self.catalog_location) / run_id / self.compute_data_folder
 
-        Args:
-            name ([str]): A glob matching the file name
-            run_id ([str]): The run id
+    def download_from_catalog(self, file: Path | CloudPath) -> None:
+        assert isinstance(file, Path)
 
-        Raises:
-            Exception: If the catalog location does not exist
-
-        Returns:
-            List(object) : A list of catalog objects
-        """
-        logger.info(
-            f"Using the {self.service_name} catalog and trying to get {name} for run_id: {run_id}"
-        )
+        run_catalog = self.get_catalog_location()
+        relative_file_path = file.relative_to(run_catalog)
 
         copy_to = self.compute_data_folder
-        if compute_data_folder:
-            copy_to = compute_data_folder
-
-        copy_to = Path(copy_to)  # type: ignore
-
-        catalog_location = self.get_catalog_location()
-        run_catalog = Path(catalog_location) / run_id / copy_to
-
-        logger.debug(
-            f"Copying objects to {copy_to} from the run catalog location of {run_catalog}"
-        )
+        # Make the directory in the data folder if required
+        Path(copy_to / relative_file_path.parent).mkdir(parents=True, exist_ok=True)
+        shutil.copy(file, copy_to / relative_file_path)
 
-        if not utils.does_dir_exist(run_catalog):
-            msg = (
-                f"Expected Catalog to be present at: {run_catalog} but not found.\n"
-                "Note: Please make sure that some data was put in the catalog before trying to get from it.\n"
-            )
-            raise Exception(msg)
+    def upload_to_catalog(self, file: Path) -> None:
+        run_catalog = self.get_catalog_location()
+        run_catalog.mkdir(parents=True, exist_ok=True)
 
-        # Iterate through the contents of the run_catalog and copy the files that fit the name pattern
-        # We should also return a list of data hashes
-        glob_files = run_catalog.glob(name)
         logger.debug(
-            f"Glob identified {glob_files} as matches to from the catalog location: {run_catalog}"
+            f"Copying objects from {self.compute_data_folder} to the run catalog location of {run_catalog}"
         )
 
-        data_catalogs = []
-        run_log_store = self._context.run_log_store
-        for file in glob_files:
-            if file.is_dir():
-                # Need not add a data catalog for the folder
-                continue
-
-            if str(file).endswith(".execution.log"):
-                continue
-
-            relative_file_path = file.relative_to(run_catalog)
-
-            data_catalog = run_log_store.create_data_catalog(str(relative_file_path))
-            data_catalog.catalog_handler_location = catalog_location
-            data_catalog.catalog_relative_path = str(relative_file_path)
-            data_catalog.data_hash = utils.get_data_hash(str(file))
-            data_catalog.stage = "get"
-            data_catalogs.append(data_catalog)
-
-            # Make the directory in the data folder if required
-            Path(copy_to / relative_file_path.parent).mkdir(parents=True, exist_ok=True)
-            shutil.copy(file, copy_to / relative_file_path)
-
-            logger.info(f"Copied {file} from {run_catalog} to {copy_to}")
-
-        if not data_catalogs:
-            raise Exception(f"Did not find any files matching {name} in {run_catalog}")
-
-        return data_catalogs
-
-    def put(
-        self,
-        name: str,
-        run_id: str,
-        compute_data_folder: str = "",
-        synced_catalogs: Optional[List[DataCatalog]] = None,
-        **kwargs,
-    ) -> List[DataCatalog]:
-        """
-        Put the files matching the glob pattern into the catalog.
-
-        If previously synced catalogs are provided, and no changes were observed, we do not sync them.
-
-        Args:
-            name (str): The glob pattern of the files to catalog
-            run_id (str): The run id of the run
-            compute_data_folder (str, optional): The compute data folder to sync from. Defaults to settings default.
-            synced_catalogs (dict, optional): dictionary of previously synced catalogs. Defaults to None.
-
-        Raises:
-            Exception: If the compute data folder does not exist.
-
-        Returns:
-            List(object) : A list of catalog objects
-        """
-        logger.info(
-            f"Using the {self.service_name} catalog and trying to put {name} for run_id: {run_id}"
-        )
-
-        copy_from = self.compute_data_folder
-        if compute_data_folder:
-            copy_from = compute_data_folder
-        copy_from = Path(copy_from)  # type: ignore
-
-        catalog_location = self.get_catalog_location()
-        run_catalog = Path(catalog_location) / run_id
-        utils.safe_make_dir(run_catalog)
-
-        logger.debug(
-            f"Copying objects from {copy_from} to the run catalog location of {run_catalog}"
-        )
-
-        if not utils.does_dir_exist(copy_from):
-            msg = (
-                f"Expected compute data folder to be present at: {compute_data_folder} but not found. \n"
-                "Note: runnable does not create the compute data folder for you. Please ensure that the "
-                "folder exists.\n"
-            )
-            raise Exception(msg)
-
-        # Iterate through the contents of copy_from and if the name matches, we move them to the run_catalog
-        # We should also return a list of datastore.DataCatalog items
-
-        glob_files = copy_from.glob(name)  # type: ignore
-        logger.debug(
-            f"Glob identified {glob_files} as matches to from the compute data folder: {copy_from}"
-        )
-
-        data_catalogs = []
-        run_log_store = self._context.run_log_store
-        for file in glob_files:
-            if file.is_dir():
-                # Need not add a data catalog for the folder
-                continue
-
-            relative_file_path = file.relative_to(".")
-
-            data_catalog = run_log_store.create_data_catalog(str(relative_file_path))
-            data_catalog.catalog_handler_location = catalog_location
-            data_catalog.catalog_relative_path = (
-                run_id + os.sep + str(relative_file_path)
-            )
-            data_catalog.data_hash = utils.get_data_hash(str(file))
-            data_catalog.stage = "put"
-            data_catalogs.append(data_catalog)
-
-            if is_catalog_out_of_sync(data_catalog, synced_catalogs):
-                logger.info(f"{data_catalog.name} was found to be changed, syncing")
-
-                # Make the directory in the catalog if required
-                Path(run_catalog / relative_file_path.parent).mkdir(
-                    parents=True, exist_ok=True
-                )
-                shutil.copy(file, run_catalog / relative_file_path)
-            else:
-                logger.info(
-                    f"{data_catalog.name} was found to be unchanged, ignoring syncing"
-                )
-
-        if not data_catalogs:
-            raise Exception(f"Did not find any files matching {name} in {copy_from}")
-
-        return data_catalogs
-
-    def sync_between_runs(self, previous_run_id: str, run_id: str):
-        """
-        Given the previous run id, sync the catalogs between the current one and previous
-
-        Args:
-            previous_run_id (str): The previous run id to sync the catalogs from
-            run_id (str): The run_id to which the data catalogs should be synced to.
-
-        Raises:
-            Exception: If the previous run log does not exist in the catalog
-
-        """
-        logger.info(
-            f"Using the {self.service_name} catalog and syncing catalogs"
-            "between old: {previous_run_id} to new: {run_id}"
-        )
-
-        catalog_location = Path(self.get_catalog_location())
-        run_catalog = catalog_location / run_id
-        utils.safe_make_dir(run_catalog)
-
-        if not utils.does_dir_exist(catalog_location / previous_run_id):
-            msg = (
-                f"Catalogs from previous run : {previous_run_id} are not found.\n"
-                "Note: Please provision the catalog objects generated by previous run in the same catalog location"
-                " as the current run, even if the catalog handler for the previous run was different"
-            )
-            raise Exception(msg)
-
-        cataloged_files = list((catalog_location / previous_run_id).glob("*"))
-
-        for cataloged_file in cataloged_files:
-            if str(cataloged_file).endswith("execution.log"):
-                continue
+        relative_file_path = file.relative_to(self.compute_data_folder)
 
-            if cataloged_file.is_file():
-                shutil.copy(cataloged_file, run_catalog / cataloged_file.name)
-            else:
-                shutil.copytree(cataloged_file, run_catalog / cataloged_file.name)
-            logger.info(f"Copied file from: {cataloged_file} to {run_catalog}")
+        (run_catalog / relative_file_path.parent).mkdir(parents=True, exist_ok=True)
+        shutil.copy(file, run_catalog / relative_file_path)

extensions/catalog/minio.py

@@ -0,0 +1,72 @@
+import logging
+from functools import lru_cache
+from pathlib import Path
+from typing import Any
+
+from cloudpathlib import CloudPath, S3Client, S3Path
+from pydantic import Field, SecretStr
+
+from extensions.catalog.any_path import AnyPathCatalog
+from runnable import defaults
+
+logger = logging.getLogger(defaults.LOGGER_NAME)
+
+
+@lru_cache
+def get_minio_client(
+    endpoint_url: str, aws_access_key_id: str, aws_secret_access_key: str
+) -> S3Client:
+    return S3Client(
+        endpoint_url=endpoint_url,
+        aws_access_key_id=aws_access_key_id,
+        aws_secret_access_key=aws_secret_access_key,
+    )
+
+
+class MinioCatalog(AnyPathCatalog):
+    service_name: str = "minio"
+
+    endpoint_url: str = Field(default="http://localhost:9002")
+    aws_access_key_id: SecretStr = SecretStr(secret_value="minioadmin")
+    aws_secret_access_key: SecretStr = SecretStr(secret_value="minioadmin")
+    bucket: str = "runnable"
+
+    def get_summary(self) -> dict[str, Any]:
+        return {
+            "service_name": self.service_name,
+            "compute_data_folder": self.compute_data_folder,
+            "endpoint_url": self.endpoint_url,
+            "bucket": self.bucket,
+        }
+
+    def get_catalog_location(self) -> S3Path:
+        run_id = self._context.run_id
+
+        return S3Path(
+            f"s3://{self.bucket}/{run_id}/{self.compute_data_folder}".strip("."),
+            client=get_minio_client(
+                self.endpoint_url,
+                self.aws_access_key_id.get_secret_value(),
+                self.aws_secret_access_key.get_secret_value(),
+            ),
+        )
+
+    def download_from_catalog(self, file: Path | CloudPath) -> None:
+        assert isinstance(file, S3Path)
+
+        relative_file_path = file.relative_to(self.get_catalog_location())
+
+        file_to_download = Path(self.compute_data_folder) / relative_file_path
+        file_to_download.parent.mkdir(parents=True, exist_ok=True)
+
+        file.download_to(file_to_download)
+
+    def upload_to_catalog(self, file: Path) -> None:
+        run_catalog = self.get_catalog_location()
+
+        relative_file_path = file.relative_to(self.compute_data_folder)
+        (run_catalog / relative_file_path.parent).mkdir(parents=True, exist_ok=True)
+
+        file_in_cloud = run_catalog / file
+        assert isinstance(file_in_cloud, S3Path)
+        file_in_cloud.upload_from(file)

extensions/catalog/s3.py

@@ -0,0 +1,11 @@
+from cloudpathlib import S3Path
+
+from extensions.catalog.any_path import AnyPathCatalog
+
+
+class S3Catalog(AnyPathCatalog):
+    service_name: str = "s3"
+
+    def get_path(self, path: str) -> S3Path:
+        # TODO: Might need to assert the credentials are set
+        return S3Path(path)

extensions/pipeline_executor/__init__.py

@@ -151,54 +151,25 @@ class GenericPipelineExecutor(BasePipelineExecutor):
             # Nothing to get/put from the catalog
             return None
 
-        compute_data_folder = self.get_effective_compute_data_folder()
-
         data_catalogs = []
         for name_pattern in node_catalog_settings.get(stage) or []:
             if stage == "get":
                 data_catalog = self._context.catalog_handler.get(
                     name=name_pattern,
-                    run_id=self._context.run_id,
-                    compute_data_folder=compute_data_folder,
                 )
 
             elif stage == "put":
                 data_catalog = self._context.catalog_handler.put(
                     name=name_pattern,
-                    run_id=self._context.run_id,
-                    compute_data_folder=compute_data_folder,
-                    synced_catalogs=synced_catalogs,
                 )
+            else:
+                raise Exception(f"Stage {stage} not supported")
 
             logger.debug(f"Added data catalog: {data_catalog} to step log")
             data_catalogs.extend(data_catalog)
 
         return data_catalogs
 
-    def get_effective_compute_data_folder(self) -> str:
-        """
-        Get the effective compute data folder for the given stage.
-        If there is nothing to catalog, we return None.
-
-        The default is the compute data folder of the catalog but this can be over-ridden by the node.
-
-        Args:
-            stage (str): The stage we are in the process of cataloging
-
-
-        Returns:
-            str: The compute data folder as defined by the node defaulting to catalog handler
-        """
-        assert isinstance(self._context_node, BaseNode)
-        compute_data_folder = self._context.catalog_handler.compute_data_folder
-
-        catalog_settings = self._context_node._get_catalog_settings()
-        effective_compute_data_folder = (
-            catalog_settings.get("compute_data_folder", "") or compute_data_folder
-        )
-
-        return effective_compute_data_folder
-
     @property
     def step_attempt_number(self) -> int:
         """
@@ -219,9 +190,7 @@ class GenericPipelineExecutor(BasePipelineExecutor):
         )
         task_console.save_text(log_file_name)
         # Put the log file in the catalog
-        self._context.catalog_handler.put(
-            name=log_file_name, run_id=self._context.run_id
-        )
+        self._context.catalog_handler.put(name=log_file_name)
         os.remove(log_file_name)
 
     def _execute_node(