refactor(export): request for Study export function (#1669)

Co-authored-by: LAIDI Takfarinas (Externe) <[email protected]>
Co-authored-by: TLAIDI <[email protected]>

antarest/study/common/studystorage.py

@@ -230,27 +230,6 @@ def export_output(self, metadata: T, output_id: str, target: Path) -> None:
         """
         raise NotImplementedError()
 
-    @abstractmethod
-    def export_study_flat(
-        self,
-        metadata: T,
-        dst_path: Path,
-        outputs: bool = True,
-        output_list_filter: Optional[List[str]] = None,
-        denormalize: bool = True,
-    ) -> None:
-        """
-        Export study to destination
-
-        Args:
-            metadata: study.
-            dst_path: destination path.
-            outputs: list of outputs to keep.
-            output_list_filter: list of outputs to keep (None indicate all outputs).
-            denormalize: denormalize the study (replace matrix links by real matrices).
-        """
-        raise NotImplementedError()
-
     @abstractmethod
     def get_synthesis(self, metadata: T, params: Optional[RequestParameters] = None) -> FileStudyTreeConfigDTO:
         """
@@ -274,3 +253,6 @@ def archive_study_output(self, study: T, output_id: str) -> bool:
     @abstractmethod
     def unarchive_study_output(self, study: T, output_id: str, keep_src_zip: bool) -> bool:
         raise NotImplementedError()
+
+    def unarchive(self, study: T) -> None:
+        raise NotImplementedError()

antarest/study/service.py

@@ -2,6 +2,7 @@
 import io
 import json
 import logging
+import shutil
 import os
 from datetime import datetime, timedelta
 from http import HTTPStatus
@@ -106,6 +107,7 @@
     remove_from_cache,
     study_matcher,
 )
+from antarest.study.storage.abstract_storage_service import export_study_flat
 from antarest.study.storage.variantstudy.model.command.icommand import ICommand
 from antarest.study.storage.variantstudy.model.command.replace_matrix import ReplaceMatrix
 from antarest.study.storage.variantstudy.model.command.update_comments import UpdateComments
@@ -918,7 +920,17 @@ def export_study(
         def export_task(notifier: TaskUpdateNotifier) -> TaskResult:
             try:
                 target_study = self.get_study(uuid)
-                self.storage_service.get_storage(target_study).export_study(target_study, export_path, outputs)
+                storage = self.storage_service.get_storage(target_study)
+                if isinstance(target_study, RawStudy):
+                    if target_study.archived:
+                        storage.unarchive(target_study)
+                    try:
+                        storage.export_study(target_study, export_path, outputs)
+                    finally:
+                        if target_study.archived:
+                            shutil.rmtree(target_study.path)
+                else:
+                    storage.export_study(target_study, export_path, outputs)
                 self.file_transfer_manager.set_ready(export_id)
                 return TaskResult(success=True, message=f"Study {uuid} successfully exported")
             except Exception as e:
@@ -1020,9 +1032,28 @@ def export_study_flat(
         study = self.get_study(uuid)
         assert_permission(params.user, study, StudyPermissionType.READ)
         self._assert_study_unarchived(study)
-
-        return self.storage_service.get_storage(study).export_study_flat(
-            study, dest, len(output_list or []) > 0, output_list
+        path_study = Path(study.path)
+        if isinstance(study, RawStudy):
+            if study.archived:
+                self.storage_service.get_storage(study).unarchive(study)
+            try:
+                return export_study_flat(
+                    path_study=path_study,
+                    dest=dest,
+                    outputs=len(output_list or []) > 0,
+                    output_list_filter=output_list,
+                )
+            finally:
+                if study.archived:
+                    shutil.rmtree(study.path)
+        snapshot_path = path_study / "snapshot"
+        output_src_path = path_study / "output"
+        export_study_flat(
+            path_study=snapshot_path,
+            dest=dest,
+            outputs=len(output_list or []) > 0,
+            output_list_filter=output_list,
+            output_src_path=output_src_path,
         )
 
     def delete_study(self, uuid: str, children: bool, params: RequestParameters) -> None:

antarest/study/storage/abstract_storage_service.py

@@ -5,6 +5,10 @@
 from pathlib import Path
 from typing import IO, List, Optional, Union
 from uuid import uuid4
+import time
+from zipfile import ZipFile
+import os
+
 
 from antarest.core.config import Config
 from antarest.core.exceptions import BadOutputError, StudyOutputNotFoundError
@@ -32,6 +36,60 @@
 logger = logging.getLogger(__name__)
 
 
+def export_study_flat(
+    path_study: Path,
+    dest: Path,
+    outputs: bool = True,
+    output_list_filter: Optional[List[str]] = None,
+    output_src_path: Optional[Path] = None,
+) -> None:
+    """
+    Export study to destination
+
+    Args:
+        path_study: Study source path
+        dest: Destination path.
+        outputs: List of outputs to keep.
+        output_list_filter: List of outputs to keep (None indicate all outputs).
+        output_src_path: Denormalize the study (replace matrix links by real matrices).
+
+    """
+    start_time = time.time()
+    output_src_path = output_src_path or path_study / "output"
+    output_dest_path = dest / "output"
+    ignore_patterns = (
+        lambda directory, contents: ["output"]
+        if str(directory) == str(path_study)
+        else []
+    )
+
+    shutil.copytree(src=path_study, dst=dest, ignore=ignore_patterns)
+    if outputs and output_src_path.is_dir():
+        if output_dest_path.exists():
+            shutil.rmtree(output_dest_path)
+        if output_list_filter is not None:
+            os.mkdir(output_dest_path)
+            for output in output_list_filter:
+                zip_path = output_src_path / f"{output}.zip"
+                if zip_path.exists():
+                    with ZipFile(zip_path) as zf:
+                        zf.extractall(output_dest_path / output)
+                else:
+                    shutil.copytree(
+                        src=output_src_path / output,
+                        dst=output_dest_path / output,
+                    )
+        else:
+            shutil.copytree(
+                src=output_src_path,
+                dst=output_dest_path,
+            )
+
+    stop_time = time.time()
+    duration = "{:.3f}".format(stop_time - start_time)
+    logger.info(f"Study {path_study} exported (flat mode) in {duration}s")
+
+
 class AbstractStorageService(IStudyStorageService[T], ABC):
     def __init__(
         self,
@@ -230,7 +288,16 @@ def export_study(self, metadata: T, target: Path, outputs: bool = True) -> Path:
         with tempfile.TemporaryDirectory(dir=self.config.storage.tmp_dir) as tmpdir:
             logger.info(f"Exporting study {metadata.id} to temporary path {tmpdir}")
             tmp_study_path = Path(tmpdir) / "tmp_copy"
-            self.export_study_flat(metadata, tmp_study_path, outputs)
+            if not isinstance(metadata, RawStudy):
+                snapshot_path = path_study / "snapshot"
+                output_src_path = path_study / "output"
+                export_study_flat(
+                    path_study=snapshot_path,
+                    dest=tmp_study_path,
+                    outputs=outputs,
+                    output_src_path=output_src_path,
+                )
+            export_study_flat(path_study, tmp_study_path, outputs)
             stopwatch = StopWatch()
             zip_dir(tmp_study_path, target)
             stopwatch.log_elapsed(lambda x: logger.info(f"Study {path_study} exported (zipped mode) in {x}s"))

antarest/study/storage/rawstudy/raw_study_service.py

@@ -23,7 +23,6 @@
 from antarest.study.storage.rawstudy.model.filesystem.lazy_node import LazyNode
 from antarest.study.storage.utils import (
     create_new_empty_study,
-    export_study_flat,
     fix_study_root,
     get_default_workspace_path,
     is_managed,
@@ -323,31 +322,6 @@ def import_study(self, metadata: RawStudy, stream: IO[bytes]) -> Study:
         metadata.path = str(path_study)
         return metadata
 
-    def export_study_flat(
-        self,
-        metadata: RawStudy,
-        dst_path: Path,
-        outputs: bool = True,
-        output_list_filter: Optional[List[str]] = None,
-        denormalize: bool = True,
-    ) -> None:
-        path_study = Path(metadata.path)
-
-        if metadata.archived:
-            self.unarchive(metadata)
-        try:
-            export_study_flat(
-                path_study,
-                dst_path,
-                self.study_factory,
-                outputs,
-                output_list_filter,
-                denormalize,
-            )
-        finally:
-            if metadata.archived:
-                shutil.rmtree(metadata.path)
-
     def check_errors(
         self,
         metadata: RawStudy,

antarest/study/storage/utils.py

@@ -326,50 +326,3 @@ def get_start_date(
         first_week_size=first_week_size,
         level=level,
     )
-
-
-def export_study_flat(
-    path_study: Path,
-    dest: Path,
-    study_factory: StudyFactory,
-    outputs: bool = True,
-    output_list_filter: Optional[List[str]] = None,
-    denormalize: bool = True,
-    output_src_path: Optional[Path] = None,
-) -> None:
-    start_time = time.time()
-
-    output_src_path = output_src_path or path_study / "output"
-    output_dest_path = dest / "output"
-    ignore_patterns = lambda directory, contents: ["output"] if str(directory) == str(path_study) else []
-
-    shutil.copytree(src=path_study, dst=dest, ignore=ignore_patterns)
-
-    if outputs and output_src_path.exists():
-        if output_list_filter is None:
-            # Retrieve all directories or ZIP files without duplicates
-            output_list_filter = list(
-                {f.with_suffix("").name for f in output_src_path.iterdir() if f.is_dir() or f.suffix == ".zip"}
-            )
-        # Copy each folder or uncompress each ZIP file to the destination dir.
-        shutil.rmtree(output_dest_path, ignore_errors=True)
-        output_dest_path.mkdir()
-        for output in output_list_filter:
-            zip_path = output_src_path / f"{output}.zip"
-            if zip_path.exists():
-                with ZipFile(zip_path) as zf:
-                    zf.extractall(output_dest_path / output)
-            else:
-                shutil.copytree(
-                    src=output_src_path / output,
-                    dst=output_dest_path / output,
-                )
-
-    stop_time = time.time()
-    duration = "{:.3f}".format(stop_time - start_time)
-    logger.info(f"Study {path_study} exported (flat mode) in {duration}s")
-    study = study_factory.create_from_fs(dest, "", use_cache=False)
-    if denormalize:
-        study.tree.denormalize()
-        duration = "{:.3f}".format(time.time() - stop_time)
-        logger.info(f"Study {path_study} denormalized in {duration}s")

antarest/study/storage/variantstudy/variant_study_service.py

@@ -34,14 +34,13 @@
 from antarest.core.utils.utils import assert_this, suppress_exception
 from antarest.matrixstore.service import MatrixService
 from antarest.study.model import RawStudy, Study, StudyAdditionalData, StudyMetadataDTO, StudySimResultDTO
-from antarest.study.storage.abstract_storage_service import AbstractStorageService
+from antarest.study.storage.abstract_storage_service import AbstractStorageService, export_study_flat
 from antarest.study.storage.patch_service import PatchService
 from antarest.study.storage.rawstudy.model.filesystem.config.model import FileStudyTreeConfig, FileStudyTreeConfigDTO
 from antarest.study.storage.rawstudy.model.filesystem.factory import FileStudy, StudyFactory
 from antarest.study.storage.rawstudy.raw_study_service import RawStudyService
 from antarest.study.storage.utils import (
     assert_permission,
-    export_study_flat,
     get_default_workspace_path,
     is_managed,
     remove_from_cache,
@@ -719,22 +718,30 @@ def _generate(
             last_executed_command_index = None
 
         if last_executed_command_index is None:
-            # Copy parent study to destination
             if isinstance(parent_study, VariantStudy):
                 self._safe_generation(parent_study)
-                self.export_study_flat(
-                    metadata=parent_study,
-                    dst_path=dst_path,
+                path_study = Path(parent_study.path)
+                snapshot_path = path_study / SNAPSHOT_RELATIVE_PATH
+                output_src_path = path_study / "output"
+                export_study_flat(
+                    snapshot_path,
+                    dst_path,
                     outputs=False,
-                    denormalize=False,
+                    output_src_path=output_src_path,
                 )
             else:
-                self.raw_study_service.export_study_flat(
-                    metadata=parent_study,
-                    dst_path=dst_path,
-                    outputs=False,
-                    denormalize=False,
-                )
+                path_study = Path(parent_study.path)
+                if parent_study.archived:
+                    self.raw_study_service.unarchive(parent_study)
+                try:
+                    export_study_flat(
+                        path_study=path_study,
+                        dest=dst_path,
+                        outputs=False,
+                    )
+                finally:
+                    if parent_study.archived:
+                        shutil.rmtree(parent_study.path)
 
         command_start_index = last_executed_command_index + 1 if last_executed_command_index is not None else 0
         logger.info(f"Generating study snapshot from command index {command_start_index}")
@@ -1068,29 +1075,6 @@ def get_study_path(self, metadata: Study) -> Path:
         """
         return Path(metadata.path) / SNAPSHOT_RELATIVE_PATH
 
-    def export_study_flat(
-        self,
-        metadata: VariantStudy,
-        dst_path: Path,
-        outputs: bool = True,
-        output_list_filter: Optional[List[str]] = None,
-        denormalize: bool = True,
-    ) -> None:
-        self._safe_generation(metadata)
-        path_study = Path(metadata.path)
-
-        snapshot_path = path_study / SNAPSHOT_RELATIVE_PATH
-        output_src_path = path_study / "output"
-        export_study_flat(
-            snapshot_path,
-            dst_path,
-            self.study_factory,
-            outputs,
-            output_list_filter,
-            denormalize,
-            output_src_path,
-        )
-
     def get_synthesis(
         self,
         metadata: VariantStudy,

tests/storage/business/test_export.py

@@ -8,6 +8,7 @@
 from antarest.core.config import Config, StorageConfig
 from antarest.study.model import DEFAULT_WORKSPACE_NAME, RawStudy
 from antarest.study.storage.rawstudy.raw_study_service import RawStudyService
+from antarest.study.storage.abstract_storage_service import export_study_flat
 
 
 @pytest.mark.unit_test
@@ -104,17 +105,14 @@ def test_export_flat(tmp_path: Path):
     study_factory.create_from_fs.return_value = study_tree
 
     study = RawStudy(id="id", path=root)
+    path_study = Path(study.path)
 
-    study_service.export_study_flat(study, tmp_path / "copy_with_output", outputs=True)
-
+    export_study_flat(path_study, tmp_path / "copy_with_output", outputs=True)
     copy_with_output_hash = dirhash(tmp_path / "copy_with_output", "md5")
-
     assert root_hash == copy_with_output_hash
 
-    study_service.export_study_flat(study, tmp_path / "copy_without_output", outputs=False)
-
+    export_study_flat(path_study, tmp_path / "copy_without_output", outputs=False)
     copy_without_output_hash = dirhash(tmp_path / "copy_without_output", "md5")
-
     assert root_without_output_hash == copy_without_output_hash