fractal-server 2.14.15__py3-none-any.whl → 2.15.0a0__py3-none-any.whl

fractal_server/tasks/v2/local/collect_pixi.py

@@ -0,0 +1,222 @@
+import json
+import logging
+import shutil
+import time
+from pathlib import Path
+from tempfile import TemporaryDirectory
+
+from ..utils_database import create_db_tasks_and_update_task_group_sync
+from ..utils_pixi import parse_collect_stdout
+from ..utils_pixi import SOURCE_DIR_NAME
+from fractal_server.app.db import get_sync_db
+from fractal_server.app.models.v2 import TaskGroupActivityV2
+from fractal_server.app.models.v2 import TaskGroupV2
+from fractal_server.app.schemas.v2 import FractalUploadedFile
+from fractal_server.app.schemas.v2 import TaskGroupActivityActionV2
+from fractal_server.app.schemas.v2 import TaskGroupActivityStatusV2
+from fractal_server.app.schemas.v2.manifest import ManifestV2
+from fractal_server.config import get_settings
+from fractal_server.logger import reset_logger_handlers
+from fractal_server.logger import set_logger
+from fractal_server.syringe import Inject
+from fractal_server.tasks.utils import get_log_path
+from fractal_server.tasks.v2.local._utils import _customize_and_run_template
+from fractal_server.tasks.v2.local._utils import check_task_files_exist
+from fractal_server.tasks.v2.utils_background import add_commit_refresh
+from fractal_server.tasks.v2.utils_background import fail_and_cleanup
+from fractal_server.tasks.v2.utils_background import get_current_log
+from fractal_server.tasks.v2.utils_background import prepare_tasks_metadata
+from fractal_server.tasks.v2.utils_templates import SCRIPTS_SUBFOLDER
+from fractal_server.utils import get_timestamp
+
+
+def collect_local_pixi(
+    *,
+    task_group_activity_id: int,
+    task_group_id: int,
+    tar_gz_file: FractalUploadedFile,
+) -> None:
+    settings = Inject(get_settings)
+
+    LOGGER_NAME = f"{__name__}.ID{task_group_activity_id}"
+
+    with TemporaryDirectory() as tmpdir:
+        log_file_path = get_log_path(Path(tmpdir))
+        logger = set_logger(
+            logger_name=LOGGER_NAME,
+            log_file_path=log_file_path,
+        )
+
+        with next(get_sync_db()) as db:
+            activity = db.get(TaskGroupActivityV2, task_group_activity_id)
+            task_group = db.get(TaskGroupV2, task_group_id)
+            if activity is None or task_group is None:
+                logging.error(
+                    "Cannot find database rows with "
+                    f"{task_group_id=} and {task_group_activity_id=}:\n"
+                    f"{task_group=}\n{activity=}. Exit."
+                )
+                return
+
+            logger.info("START")
+            for key, value in task_group.model_dump().items():
+                logger.debug(f"task_group.{key}: {value}")
+
+            if Path(task_group.path).exists():
+                error_msg = f"{task_group.path} already exists."
+                logger.error(error_msg)
+                fail_and_cleanup(
+                    task_group=task_group,
+                    task_group_activity=activity,
+                    logger_name=LOGGER_NAME,
+                    log_file_path=log_file_path,
+                    exception=FileExistsError(error_msg),
+                    db=db,
+                )
+                return
+
+            # Set `pixi_bin` and check that it exists
+            pixi_home = settings.pixi.versions[task_group.pixi_version]
+            pixi_bin = Path(pixi_home, "bin/pixi").as_posix()
+
+            try:
+                Path(task_group.path).mkdir(parents=True)
+                logger.info(f"Created {task_group.path}")
+                archive_path = Path(
+                    task_group.path, tar_gz_file.filename
+                ).as_posix()
+                logger.info(f"Write tar.gz-file contents into {archive_path}.")
+                with open(archive_path, "wb") as f:
+                    f.write(tar_gz_file.contents)
+                task_group.archive_path = archive_path
+                task_group = add_commit_refresh(obj=task_group, db=db)
+
+                replacements = {
+                    ("__PIXI_HOME__", pixi_home),
+                    ("__PACKAGE_DIR__", task_group.path),
+                    ("__TAR_GZ_PATH__", archive_path),
+                    (
+                        "__IMPORT_PACKAGE_NAME__",
+                        task_group.pkg_name.replace("-", "_"),
+                    ),
+                    ("__SOURCE_DIR_NAME__", SOURCE_DIR_NAME),
+                }
+
+                activity.status = TaskGroupActivityStatusV2.ONGOING
+                activity.log = get_current_log(log_file_path)
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                stdout = _customize_and_run_template(
+                    template_filename="pixi_1_collect.sh",
+                    replacements=replacements,
+                    script_dir=Path(
+                        task_group.path, SCRIPTS_SUBFOLDER
+                    ).as_posix(),
+                    prefix=(
+                        f"{int(time.time())}_"
+                        f"{TaskGroupActivityActionV2.COLLECT}_"
+                    ),
+                    logger_name=LOGGER_NAME,
+                )
+                activity.log = get_current_log(log_file_path)
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                # Parse stdout
+                parsed_output = parse_collect_stdout(stdout)
+                package_root = parsed_output["package_root"]
+                venv_size = parsed_output["venv_size"]
+                venv_file_number = parsed_output["venv_file_number"]
+
+                # Read and validate manifest
+                # NOTE: we are only supporting the manifest path being relative
+                # to the top-level folder
+                manifest_path = f"{package_root}/__FRACTAL_MANIFEST__.json"
+                with open(manifest_path) as json_data:
+                    pkg_manifest_dict = json.load(json_data)
+                logger.info(f"loaded {manifest_path=}")
+                logger.info("now validating manifest content")
+                pkg_manifest = ManifestV2(**pkg_manifest_dict)
+                logger.info("validated manifest content")
+                activity.log = get_current_log(log_file_path)
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                logger.info("_prepare_tasks_metadata - start")
+                task_list = prepare_tasks_metadata(
+                    package_manifest=pkg_manifest,
+                    package_version=task_group.version,
+                    package_root=Path(package_root),
+                    pixi_bin=pixi_bin,
+                    pixi_manifest_path=(
+                        Path(
+                            task_group.path,
+                            SOURCE_DIR_NAME,
+                            "pyproject.toml",
+                        ).as_posix()
+                    ),
+                )
+                check_task_files_exist(task_list=task_list)
+                logger.info("_prepare_tasks_metadata - end")
+                activity.log = get_current_log(log_file_path)
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                logger.info("create_db_tasks_and_update_task_group - start")
+                create_db_tasks_and_update_task_group_sync(
+                    task_list=task_list,
+                    task_group_id=task_group.id,
+                    db=db,
+                )
+                logger.info("create_db_tasks_and_update_task_group - end")
+
+                # Update task_group data
+                logger.info(
+                    "Add env_info, venv_size and venv_file_number "
+                    "to TaskGroupV2 - start"
+                )
+                with Path(
+                    task_group.path,
+                    SOURCE_DIR_NAME,
+                    "pixi.lock",
+                ).open() as f:
+                    pixi_lock_contents = f.read()
+
+                # NOTE: see issue 2626 about whether to keep `pixi.lock` files
+                # in the database
+                task_group.env_info = pixi_lock_contents
+                task_group.venv_size_in_kB = int(venv_size)
+                task_group.venv_file_number = int(venv_file_number)
+                task_group = add_commit_refresh(obj=task_group, db=db)
+                logger.info(
+                    "Add env_info, venv_size and venv_file_number "
+                    "to TaskGroupV2 - end"
+                )
+
+                # Finalize (write metadata to DB)
+                logger.info("finalising - START")
+                activity.status = TaskGroupActivityStatusV2.OK
+                activity.timestamp_ended = get_timestamp()
+                activity = add_commit_refresh(obj=activity, db=db)
+                logger.info("finalising - END")
+                logger.info("END")
+
+                reset_logger_handlers(logger)
+
+            except Exception as collection_e:
+                # Delete corrupted package dir
+                try:
+                    logger.info(f"Now delete folder {task_group.path}")
+                    shutil.rmtree(task_group.path)
+                    logger.info(f"Deleted folder {task_group.path}")
+                except Exception as rm_e:
+                    logger.error(
+                        "Removing folder failed.\n"
+                        f"Original error:\n{str(rm_e)}"
+                    )
+
+                fail_and_cleanup(
+                    task_group=task_group,
+                    task_group_activity=activity,
+                    logger_name=LOGGER_NAME,
+                    log_file_path=log_file_path,
+                    exception=collection_e,
+                    db=db,
+                )

fractal_server/tasks/v2/local/deactivate.py

@@ -87,10 +87,10 @@ def deactivate_local(
                 activity.status = TaskGroupActivityStatusV2.ONGOING
                 activity = add_commit_refresh(obj=activity, db=db)
 
-                if task_group.pip_freeze is None:
+                if task_group.env_info is None:
                     logger.warning(
                         "Recreate pip-freeze information, since "
-                        f"{task_group.pip_freeze=}. NOTE: this should only "
+                        f"{task_group.env_info=}. NOTE: this should only "
                         "happen for task groups created before 2.9.0."
                     )
                     # Prepare replacements for templates
@@ -120,7 +120,7 @@ def deactivate_local(
                     logger.info("Add pip freeze stdout to TaskGroupV2 - start")
                     activity.log = get_current_log(log_file_path)
                     activity = add_commit_refresh(obj=activity, db=db)
-                    task_group.pip_freeze = pip_freeze_stdout
+                    task_group.env_info = pip_freeze_stdout
                     task_group = add_commit_refresh(obj=task_group, db=db)
                     logger.info("Add pip freeze stdout to TaskGroupV2 - end")
 
@@ -131,15 +131,15 @@ def deactivate_local(
                         f"Handle specific cases for {task_group.origin=}."
                     )
 
-                    # Blocking situation: `wheel_path` is not set or points
+                    # Blocking situation: `archive_path` is not set or points
                     # to a missing path
                     if (
-                        task_group.wheel_path is None
-                        or not Path(task_group.wheel_path).exists()
+                        task_group.archive_path is None
+                        or not Path(task_group.archive_path).exists()
                     ):
                         error_msg = (
                             "Invalid wheel path for task group with "
-                            f"{task_group_id=}. {task_group.wheel_path=} is "
+                            f"{task_group_id=}. {task_group.archive_path=} is "
                             "unset or does not exist."
                         )
                         logger.error(error_msg)
@@ -153,48 +153,52 @@ def deactivate_local(
                         )
                         return
 
-                    # Recoverable situation: `wheel_path` was not yet copied
+                    # Recoverable situation: `archive_path` was not yet copied
                     # over to the correct server-side folder
-                    wheel_path_parent_dir = Path(task_group.wheel_path).parent
-                    if wheel_path_parent_dir != Path(task_group.path):
+                    archive_path_parent_dir = Path(
+                        task_group.archive_path
+                    ).parent
+                    if archive_path_parent_dir != Path(task_group.path):
                         logger.warning(
-                            f"{wheel_path_parent_dir.as_posix()} differs from "
-                            f"{task_group.path}. NOTE: this should only "
+                            f"{archive_path_parent_dir.as_posix()} differs "
+                            f"from {task_group.path}. NOTE: this should only "
                             "happen for task groups created before 2.9.0."
                         )
 
-                        if task_group.wheel_path not in task_group.pip_freeze:
+                        if task_group.archive_path not in task_group.env_info:
                             raise ValueError(
-                                f"Cannot find {task_group.wheel_path=} in "
+                                f"Cannot find {task_group.archive_path=} in "
                                 "pip-freeze data. Exit."
                             )
 
                         logger.info(
                             f"Now copy wheel file into {task_group.path}."
                         )
-                        new_wheel_path = (
+                        new_archive_path = (
                             Path(task_group.path)
-                            / Path(task_group.wheel_path).name
+                            / Path(task_group.archive_path).name
                         ).as_posix()
-                        shutil.copy(task_group.wheel_path, new_wheel_path)
-                        logger.info(f"Copied wheel file to {new_wheel_path}.")
+                        shutil.copy(task_group.archive_path, new_archive_path)
+                        logger.info(
+                            f"Copied wheel file to {new_archive_path}."
+                        )
 
-                        task_group.wheel_path = new_wheel_path
-                        new_pip_freeze = task_group.pip_freeze.replace(
-                            task_group.wheel_path,
-                            new_wheel_path,
+                        task_group.archive_path = new_archive_path
+                        new_pip_freeze = task_group.env_info.replace(
+                            task_group.archive_path,
+                            new_archive_path,
                         )
-                        task_group.pip_freeze = new_pip_freeze
+                        task_group.env_info = new_pip_freeze
                         task_group = add_commit_refresh(obj=task_group, db=db)
                         logger.info(
-                            "Updated `wheel_path` and `pip_freeze` "
+                            "Updated `archive_path` and `env_info` "
                             "task-group attributes."
                         )
 
                 # Fail if `pip_freeze` includes "github.com", see
                 # https://github.com/fractal-analytics-platform/fractal-server/issues/2142
                 for forbidden_string in FORBIDDEN_DEPENDENCY_STRINGS:
-                    if forbidden_string in task_group.pip_freeze:
+                    if forbidden_string in task_group.env_info:
                         raise ValueError(
                             "Deactivation and reactivation of task packages "
                             f"with direct {forbidden_string} dependencies "

fractal_server/tasks/v2/local/deactivate_pixi.py

@@ -0,0 +1,110 @@
+import logging
+import shutil
+from pathlib import Path
+from tempfile import TemporaryDirectory
+
+from ..utils_background import add_commit_refresh
+from ..utils_background import fail_and_cleanup
+from ..utils_pixi import SOURCE_DIR_NAME
+from fractal_server.app.db import get_sync_db
+from fractal_server.app.models.v2 import TaskGroupActivityV2
+from fractal_server.app.models.v2 import TaskGroupV2
+from fractal_server.app.schemas.v2.task_group import TaskGroupActivityStatusV2
+from fractal_server.logger import reset_logger_handlers
+from fractal_server.logger import set_logger
+from fractal_server.tasks.utils import get_log_path
+from fractal_server.tasks.v2.utils_background import get_current_log
+from fractal_server.utils import get_timestamp
+
+
+def deactivate_local_pixi(
+    *,
+    task_group_activity_id: int,
+    task_group_id: int,
+) -> None:
+    """
+    Deactivate a task group venv.
+
+    This function is run as a background task, therefore exceptions must be
+    handled.
+
+    Arguments:
+        task_group_id:
+        task_group_activity_id:
+    """
+
+    LOGGER_NAME = f"{__name__}.ID{task_group_activity_id}"
+
+    with TemporaryDirectory() as tmpdir:
+        log_file_path = get_log_path(Path(tmpdir))
+        logger = set_logger(
+            logger_name=LOGGER_NAME,
+            log_file_path=log_file_path,
+        )
+
+        with next(get_sync_db()) as db:
+
+            # Get main objects from db
+            activity = db.get(TaskGroupActivityV2, task_group_activity_id)
+            task_group = db.get(TaskGroupV2, task_group_id)
+            if activity is None or task_group is None:
+                # Use `logging` directly
+                logging.error(
+                    "Cannot find database rows with "
+                    f"{task_group_id=} and {task_group_activity_id=}:\n"
+                    f"{task_group=}\n{activity=}. Exit."
+                )
+                return
+
+            # Log some info
+            logger.debug("START")
+
+            for key, value in task_group.model_dump().items():
+                logger.debug(f"task_group.{key}: {value}")
+
+            source_dir = Path(task_group.path, SOURCE_DIR_NAME)
+            # Check that the (local) task_group venv_path does exist
+            if not source_dir.exists():
+                error_msg = f"'{source_dir.as_posix()}' does not exist."
+                logger.error(error_msg)
+                fail_and_cleanup(
+                    task_group=task_group,
+                    task_group_activity=activity,
+                    logger_name=LOGGER_NAME,
+                    log_file_path=log_file_path,
+                    exception=FileNotFoundError(error_msg),
+                    db=db,
+                )
+                return
+
+            try:
+
+                activity.status = TaskGroupActivityStatusV2.ONGOING
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                # Actually mark the task group as non-active
+                logger.info("Now setting `active=False`.")
+                task_group.active = False
+                task_group = add_commit_refresh(obj=task_group, db=db)
+
+                # Proceed with deactivation
+                logger.info(f"Now removing '{source_dir.as_posix()}'.")
+                shutil.rmtree(source_dir)
+                logger.info(f"All good, '{source_dir.as_posix()}' removed.")
+                activity.status = TaskGroupActivityStatusV2.OK
+                activity.log = get_current_log(log_file_path)
+                activity.timestamp_ended = get_timestamp()
+                activity = add_commit_refresh(obj=activity, db=db)
+
+                reset_logger_handlers(logger)
+
+            except Exception as e:
+                fail_and_cleanup(
+                    task_group=task_group,
+                    task_group_activity=activity,
+                    logger_name=LOGGER_NAME,
+                    log_file_path=log_file_path,
+                    exception=e,
+                    db=db,
+                )
+        return

fractal_server/tasks/v2/local/reactivate.py

@@ -95,7 +95,7 @@ def reactivate_local(
                     ),
                 )
                 with open(f"{tmpdir}/pip_freeze.txt", "w") as f:
-                    f.write(task_group.pip_freeze)
+                    f.write(task_group.env_info)
                 replacements.append(
                     ("__PIP_FREEZE_FILE__", f"{tmpdir}/pip_freeze.txt")
                 )

fractal_server/tasks/v2/ssh/__init__.py

@@ -1,3 +1,4 @@
 from .collect import collect_ssh  # noqa
+from .collect_pixi import collect_ssh_pixi  # noqa
 from .deactivate import deactivate_ssh  # noqa
 from .reactivate import reactivate_ssh  # noqa

fractal_server/tasks/v2/ssh/_utils.py

@@ -69,16 +69,16 @@ def _copy_wheel_file_ssh(
     *, task_group: TaskGroupV2, fractal_ssh: FractalSSH, logger_name: str
 ) -> str:
     """
-    Handle the situation where `task_group.wheel_path` is not part of
-    `task_group.path`, by copying `wheel_path` into `path`.
+    Handle the situation where `task_group.archive_path` is not part of
+    `task_group.path`, by copying `archive_path` into `path`.
 
     Returns:
-        The new `wheel_path`.
+        The new `archive_path`.
     """
     logger = get_logger(logger_name=logger_name)
-    source = task_group.wheel_path
+    source = task_group.archive_path
     dest = (
-        Path(task_group.path) / Path(task_group.wheel_path).name
+        Path(task_group.path) / Path(task_group.archive_path).name
     ).as_posix()
     cmd = f"cp {source} {dest}"
     logger.debug(f"[_copy_wheel_file] START {source=} {dest=}")

fractal_server/tasks/v2/ssh/collect.py

@@ -4,15 +4,15 @@ from pathlib import Path
 from tempfile import TemporaryDirectory
 
 from ....ssh._fabric import SingleUseFractalSSH
-from ..utils_background import _prepare_tasks_metadata
 from ..utils_background import fail_and_cleanup
+from ..utils_background import prepare_tasks_metadata
 from ..utils_database import create_db_tasks_and_update_task_group_sync
 from fractal_server.app.db import get_sync_db
 from fractal_server.app.models.v2 import TaskGroupActivityV2
 from fractal_server.app.models.v2 import TaskGroupV2
+from fractal_server.app.schemas.v2 import FractalUploadedFile
 from fractal_server.app.schemas.v2 import TaskGroupActivityActionV2
 from fractal_server.app.schemas.v2 import TaskGroupActivityStatusV2
-from fractal_server.app.schemas.v2 import WheelFile
 from fractal_server.app.schemas.v2.manifest import ManifestV2
 from fractal_server.logger import reset_logger_handlers
 from fractal_server.logger import set_logger
@@ -38,7 +38,7 @@ def collect_ssh(
     task_group_activity_id: int,
     ssh_config: SSHConfig,
     tasks_base_dir: str,
-    wheel_file: WheelFile | None = None,
+    wheel_file: FractalUploadedFile | None = None,
 ) -> None:
     """
     Collect a task package over SSH
@@ -135,25 +135,26 @@ def collect_ssh(
                     fractal_ssh.mkdir(folder=script_dir_remote, parents=True)
 
                     # Write wheel file locally and send it to remote path,
-                    # and set task_group.wheel_path
+                    # and set task_group.archive_path
                     if wheel_file is not None:
                         wheel_filename = wheel_file.filename
-                        wheel_path = (
+                        archive_path = (
                             Path(task_group.path) / wheel_filename
                         ).as_posix()
-                        tmp_wheel_path = (
+                        tmp_archive_path = (
                             Path(tmpdir) / wheel_filename
                         ).as_posix()
                         logger.info(
-                            f"Write wheel-file contents into {tmp_wheel_path}"
+                            "Write wheel-file contents into "
+                            f"{tmp_archive_path}"
                         )
-                        with open(tmp_wheel_path, "wb") as f:
+                        with open(tmp_archive_path, "wb") as f:
                             f.write(wheel_file.contents)
                         fractal_ssh.send_file(
-                            local=tmp_wheel_path,
-                            remote=wheel_path,
+                            local=tmp_archive_path,
+                            remote=archive_path,
                         )
-                        task_group.wheel_path = wheel_path
+                        task_group.archive_path = archive_path
                         task_group = add_commit_refresh(obj=task_group, db=db)
 
                     replacements = get_collection_replacements(
@@ -263,7 +264,7 @@ def collect_ssh(
                     logger.info("Manifest is a valid ManifestV2")
 
                     logger.info("_prepare_tasks_metadata - start")
-                    task_list = _prepare_tasks_metadata(
+                    task_list = prepare_tasks_metadata(
                         package_manifest=pkg_manifest,
                         package_version=task_group.version,
                         package_root=Path(package_root_remote),
@@ -283,15 +284,15 @@ def collect_ssh(
 
                     # Update task_group data
                     logger.info(
-                        "Add pip_freeze, venv_size and venv_file_number "
+                        "Add env_info, venv_size and venv_file_number "
                         "to TaskGroupV2 - start"
                     )
-                    task_group.pip_freeze = pip_freeze_stdout
+                    task_group.env_info = pip_freeze_stdout
                     task_group.venv_size_in_kB = int(venv_size)
                     task_group.venv_file_number = int(venv_file_number)
                     task_group = add_commit_refresh(obj=task_group, db=db)
                     logger.info(
-                        "Add pip_freeze, venv_size and venv_file_number "
+                        "Add env_info, venv_size and venv_file_number "
                         "to TaskGroupV2 - end"
                     )