dsp-tools 17.0.0.post29__py3-none-any.whl → 18.0.0.post3__py3-none-any.whl

dsp_tools/cli/args.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 from dataclasses import dataclass
+from dataclasses import field
 from enum import Enum
 from pathlib import Path
 
@@ -32,3 +33,15 @@ class ValidationSeverity(Enum):
     ERROR = 3
     WARNING = 2
     INFO = 1
+
+
+@dataclass
+class PathDependencies:
+    required_files: list[Path] = field(default_factory=list)
+    required_directories: list[Path] = field(default_factory=list)
+
+
+@dataclass
+class NetworkRequirements:
+    api_url: str
+    always_requires_docker: bool = False

dsp_tools/cli/call_action.py

@@ -1,41 +1,24 @@
 import argparse
-import subprocess
-from pathlib import Path
 
-import requests
 from loguru import logger
 
-from dsp_tools.cli.args import ServerCredentials
-from dsp_tools.cli.args import ValidationSeverity
-from dsp_tools.commands.excel2json.lists.make_lists import excel2lists
-from dsp_tools.commands.excel2json.old_lists import old_excel2lists
-from dsp_tools.commands.excel2json.old_lists import validate_lists_section_with_schema
-from dsp_tools.commands.excel2json.project import excel2json
-from dsp_tools.commands.excel2json.project import old_excel2json
-from dsp_tools.commands.excel2json.properties import excel2properties
-from dsp_tools.commands.excel2json.resources import excel2resources
-from dsp_tools.commands.id2iri import id2iri
-from dsp_tools.commands.ingest_xmlupload.create_resources.upload_xml import ingest_xmlupload
-from dsp_tools.commands.ingest_xmlupload.ingest_files.ingest_files import ingest_files
-from dsp_tools.commands.ingest_xmlupload.upload_files.upload_files import upload_files
-from dsp_tools.commands.project.create.project_create_all import create_project
-from dsp_tools.commands.project.create.project_create_lists import create_only_lists
-from dsp_tools.commands.project.create.project_validate import validate_project
-from dsp_tools.commands.project.get.get import get_project
-from dsp_tools.commands.resume_xmlupload.resume_xmlupload import resume_xmlupload
-from dsp_tools.commands.start_stack import StackConfiguration
-from dsp_tools.commands.start_stack import StackHandler
-from dsp_tools.commands.validate_data.validate_data import validate_data
-from dsp_tools.commands.xmlupload.upload_config import UploadConfig
-from dsp_tools.commands.xmlupload.xmlupload import xmlupload
-from dsp_tools.error.exceptions import DockerNotReachableError
-from dsp_tools.error.exceptions import DspApiNotReachableError
-from dsp_tools.error.exceptions import InputError
-from dsp_tools.error.exceptions import UserDirectoryNotFoundError
-from dsp_tools.error.exceptions import UserFilepathNotFoundError
-from dsp_tools.utils.xml_parsing.parse_clean_validate_xml import parse_and_validate_xml_file
-
-LOCALHOST_API = "http://0.0.0.0:3333"
+from dsp_tools.cli.call_action_files_only import call_excel2json
+from dsp_tools.cli.call_action_files_only import call_excel2lists
+from dsp_tools.cli.call_action_files_only import call_excel2properties
+from dsp_tools.cli.call_action_files_only import call_excel2resources
+from dsp_tools.cli.call_action_files_only import call_id2iri
+from dsp_tools.cli.call_action_files_only import call_old_excel2json
+from dsp_tools.cli.call_action_files_only import call_old_excel2lists
+from dsp_tools.cli.call_action_with_network import call_create
+from dsp_tools.cli.call_action_with_network import call_get
+from dsp_tools.cli.call_action_with_network import call_ingest_files
+from dsp_tools.cli.call_action_with_network import call_ingest_xmlupload
+from dsp_tools.cli.call_action_with_network import call_resume_xmlupload
+from dsp_tools.cli.call_action_with_network import call_start_stack
+from dsp_tools.cli.call_action_with_network import call_stop_stack
+from dsp_tools.cli.call_action_with_network import call_upload_files
+from dsp_tools.cli.call_action_with_network import call_validate_data
+from dsp_tools.cli.call_action_with_network import call_xmlupload
 
 
 def call_requested_action(args: argparse.Namespace) -> bool:  # noqa: PLR0912 (too many branches)
@@ -58,321 +41,42 @@ def call_requested_action(args: argparse.Namespace) -> bool:  # noqa: PLR0912 (t
     match args.action:
         # commands with Docker / API interactions
         case "start-stack":
-            result = _call_start_stack(args)
+            result = call_start_stack(args)
         case "stop-stack":
-            result = _call_stop_stack()
+            result = call_stop_stack()
         case "create":
-            result = _call_create(args)
+            result = call_create(args)
         case "get":
-            result = _call_get(args)
+            result = call_get(args)
         case "validate-data":
-            result = _call_validate_data(args)
+            result = call_validate_data(args)
         case "xmlupload":
-            result = _call_xmlupload(args)
+            result = call_xmlupload(args)
         case "resume-xmlupload":
-            result = _call_resume_xmlupload(args)
+            result = call_resume_xmlupload(args)
         case "upload-files":
-            result = _call_upload_files(args)
+            result = call_upload_files(args)
         case "ingest-files":
-            result = _call_ingest_files(args)
+            result = call_ingest_files(args)
         case "ingest-xmlupload":
-            result = _call_ingest_xmlupload(args)
+            result = call_ingest_xmlupload(args)
         # commands that do not require docker
         case "excel2json":
-            result = _call_excel2json(args)
+            result = call_excel2json(args)
         case "old-excel2json":
-            result = _call_old_excel2json(args)
+            result = call_old_excel2json(args)
         case "excel2lists":
-            result = _call_excel2lists(args)
+            result = call_excel2lists(args)
         case "old-excel2lists":
-            result = _call_old_excel2lists(args)
+            result = call_old_excel2lists(args)
         case "excel2resources":
-            result = _call_excel2resources(args)
+            result = call_excel2resources(args)
         case "excel2properties":
-            result = _call_excel2properties(args)
+            result = call_excel2properties(args)
         case "id2iri":
-            result = _call_id2iri(args)
+            result = call_id2iri(args)
         case _:
             print(f"ERROR: Unknown action '{args.action}'")
             logger.error(f"Unknown action '{args.action}'")
             result = False
     return result
-
-
-def _call_stop_stack() -> bool:
-    _check_docker_health()
-    stack_handler = StackHandler(StackConfiguration())
-    return stack_handler.stop_stack()
-
-
-def _call_start_stack(args: argparse.Namespace) -> bool:
-    _check_docker_health()
-    stack_handler = StackHandler(
-        StackConfiguration(
-            max_file_size=args.max_file_size,
-            enforce_docker_system_prune=args.prune,
-            suppress_docker_system_prune=args.no_prune,
-            latest_dev_version=args.latest,
-            upload_test_data=args.with_test_data,
-            custom_host=args.custom_host,
-        )
-    )
-    return stack_handler.start_stack()
-
-
-def _call_id2iri(args: argparse.Namespace) -> bool:
-    _check_filepath_exists(Path(args.xmlfile))
-    _check_filepath_exists(Path(args.mapping))
-    return id2iri(
-        xml_file=args.xmlfile,
-        json_file=args.mapping,
-        remove_resource_if_id_in_mapping=args.remove_resources,
-    )
-
-
-def _call_excel2properties(args: argparse.Namespace) -> bool:
-    _check_filepath_exists(Path(args.excelfile))
-    _, _, success = excel2properties(
-        excelfile=args.excelfile,
-        path_to_output_file=args.properties_section,
-    )
-    return success
-
-
-def _call_excel2resources(args: argparse.Namespace) -> bool:
-    _check_filepath_exists(Path(args.excelfile))
-    _, _, success = excel2resources(
-        excelfile=args.excelfile,
-        path_to_output_file=args.resources_section,
-    )
-    return success
-
-
-def _call_old_excel2lists(args: argparse.Namespace) -> bool:
-    _check_directory_exists(Path(args.excelfolder))
-    _, success = old_excel2lists(
-        excelfolder=args.excelfolder,
-        path_to_output_file=args.lists_section,
-        verbose=args.verbose,
-    )
-    return success
-
-
-def _call_excel2lists(args: argparse.Namespace) -> bool:
-    _check_directory_exists(Path(args.excelfolder))
-    _, success = excel2lists(
-        excelfolder=args.excelfolder,
-        path_to_output_file=args.lists_section,
-    )
-    return success
-
-
-def _call_excel2json(args: argparse.Namespace) -> bool:
-    _check_directory_exists(Path(args.excelfolder))
-    return excel2json(
-        data_model_files=args.excelfolder,
-        path_to_output_file=args.project_definition,
-    )
-
-
-def _call_old_excel2json(args: argparse.Namespace) -> bool:
-    _check_directory_exists(Path(args.excelfolder))
-    return old_excel2json(
-        data_model_files=args.excelfolder,
-        path_to_output_file=args.project_definition,
-    )
-
-
-def _call_upload_files(args: argparse.Namespace) -> bool:
-    _check_health_with_docker_on_localhost(args.server)
-    xml_path = Path(args.xml_file)
-    _check_filepath_exists(xml_path)
-    return upload_files(
-        xml_file=xml_path,
-        creds=_get_creds(args),
-        imgdir=Path(args.imgdir),
-    )
-
-
-def _call_ingest_files(args: argparse.Namespace) -> bool:
-    _check_health_with_docker_on_localhost(args.server)
-    return ingest_files(creds=_get_creds(args), shortcode=args.shortcode)
-
-
-def _call_ingest_xmlupload(args: argparse.Namespace) -> bool:
-    _check_health_with_docker(args.server)
-    xml_path = Path(args.xml_file)
-    _check_filepath_exists(xml_path)
-    interrupt_after = args.interrupt_after if args.interrupt_after > 0 else None
-    return ingest_xmlupload(
-        xml_file=xml_path,
-        creds=_get_creds(args),
-        interrupt_after=interrupt_after,
-        skip_validation=args.skip_validation,
-        skip_ontology_validation=args.skip_ontology_validation,
-        id2iri_replacement_file=args.id2iri_replacement_with_file,
-        do_not_request_resource_metadata_from_db=args.do_not_request_resource_metadata_from_db,
-    )
-
-
-def _call_xmlupload(args: argparse.Namespace) -> bool:
-    _check_health_with_docker(args.server)
-    xml_path = Path(args.xmlfile)
-    _check_filepath_exists(xml_path)
-    id_2_iri_file = args.id2iri_replacement_with_file
-    if id_2_iri_file:
-        _check_filepath_exists(Path(id_2_iri_file))
-    if args.validate_only:
-        success = parse_and_validate_xml_file(xml_path)
-        print("The XML file is syntactically correct.")
-        return success
-    else:
-        interrupt_after = args.interrupt_after if args.interrupt_after > 0 else None
-        match args.validation_severity:
-            case "info":
-                severity = ValidationSeverity.INFO
-            case "warning":
-                severity = ValidationSeverity.WARNING
-            case "error":
-                severity = ValidationSeverity.ERROR
-            case _:
-                raise InputError(
-                    f"The entered validation severity '{args.validation_severity}' "
-                    f"is not part of the allowed values: info, warning, error."
-                )
-        return xmlupload(
-            input_file=xml_path,
-            creds=_get_creds(args),
-            imgdir=args.imgdir,
-            config=UploadConfig(
-                interrupt_after=interrupt_after,
-                skip_iiif_validation=args.no_iiif_uri_validation,
-                skip_validation=args.skip_validation,
-                ignore_duplicate_files_warning=args.ignore_duplicate_files_warning,
-                validation_severity=severity,
-                skip_ontology_validation=args.skip_ontology_validation,
-                do_not_request_resource_metadata_from_db=args.do_not_request_resource_metadata_from_db,
-                id2iri_replacement_file=id_2_iri_file,
-            ),
-        )
-
-
-def _call_validate_data(args: argparse.Namespace) -> bool:
-    _check_health_with_docker(args.server)
-    xml_path = Path(args.xmlfile)
-    _check_filepath_exists(xml_path)
-    return validate_data(
-        filepath=xml_path,
-        creds=_get_creds(args),
-        save_graphs=args.save_graphs,
-        ignore_duplicate_files_warning=args.ignore_duplicate_files_warning,
-        skip_ontology_validation=args.skip_ontology_validation,
-        id2iri_replacement_file=args.id2iri_replacement_with_file,
-        do_not_request_resource_metadata_from_db=args.do_not_request_resource_metadata_from_db,
-    )
-
-
-def _call_resume_xmlupload(args: argparse.Namespace) -> bool:
-    # this does not need docker if not on localhost, as does not need to validate
-    _check_health_with_docker_on_localhost(args.server)
-    return resume_xmlupload(
-        creds=_get_creds(args),
-        skip_first_resource=args.skip_first_resource,
-    )
-
-
-def _call_get(args: argparse.Namespace) -> bool:
-    _check_health_with_docker_on_localhost(args.server)
-    _check_directory_exists(Path(args.project_definition).parent)
-    return get_project(
-        project_identifier=args.project,
-        outfile_path=args.project_definition,
-        creds=_get_creds(args),
-        verbose=args.verbose,
-    )
-
-
-def _call_create(args: argparse.Namespace) -> bool:
-    _check_health_with_docker_on_localhost(args.server)
-    _check_filepath_exists(Path(args.project_definition))
-    success = False
-    match args.lists_only, args.validate_only:
-        case True, True:
-            success = validate_lists_section_with_schema(args.project_definition)
-            print("'Lists' section of the JSON project file is syntactically correct and passed validation.")
-        case True, False:
-            _, success = create_only_lists(
-                project_file_as_path_or_parsed=args.project_definition,
-                creds=_get_creds(args),
-            )
-        case False, True:
-            success = validate_project(args.project_definition)
-            print("JSON project file is syntactically correct and passed validation.")
-        case False, False:
-            success = create_project(
-                project_file_as_path_or_parsed=args.project_definition,
-                creds=_get_creds(args),
-                verbose=args.verbose,
-            )
-    return success
-
-
-def _get_creds(args: argparse.Namespace) -> ServerCredentials:
-    return ServerCredentials(
-        server=args.server,
-        user=args.user,
-        password=args.password,
-        dsp_ingest_url=args.dsp_ingest_url,
-    )
-
-
-def _check_health_with_docker_on_localhost(api_url: str) -> None:
-    if api_url == LOCALHOST_API:
-        _check_docker_health()
-    _check_api_health(api_url)
-
-
-def _check_health_with_docker(api_url: str) -> None:
-    # validate always needs docker running
-    _check_docker_health()
-    _check_api_health(api_url)
-
-
-def _check_docker_health() -> None:
-    if subprocess.run("docker stats --no-stream".split(), check=False, capture_output=True).returncode != 0:
-        raise DockerNotReachableError()
-
-
-def _check_api_health(api_url: str) -> None:
-    health_url = f"{api_url}/health"
-    msg = (
-        "The DSP-API could not be reached. Please check if your stack is healthy "
-        "or start a stack with 'dsp-tools start-stack' if none is running."
-    )
-    try:
-        response = requests.get(health_url, timeout=2)
-        if not response.ok:
-            if api_url != LOCALHOST_API:
-                msg = (
-                    f"The DSP-API could not be reached (returned status {response.status_code}). "
-                    f"Please contact the DaSCH engineering team for help."
-                )
-            logger.error(msg)
-            raise DspApiNotReachableError(msg)
-        logger.debug(f"DSP API health check passed: {health_url}")
-    except requests.exceptions.RequestException as e:
-        logger.error(e)
-        if api_url != LOCALHOST_API:
-            msg = "The DSP-API responded with a request exception. Please contact the DaSCH engineering team for help."
-        logger.error(msg)
-        raise DspApiNotReachableError(msg) from None
-
-
-def _check_filepath_exists(file_path: Path) -> None:
-    if not file_path.exists():
-        raise UserFilepathNotFoundError(file_path)
-
-
-def _check_directory_exists(dir_path: Path) -> None:
-    if not dir_path.is_dir():
-        raise UserDirectoryNotFoundError(dir_path)

dsp_tools/cli/call_action_files_only.py

@@ -0,0 +1,74 @@
+import argparse
+from pathlib import Path
+
+from dsp_tools.cli.args import PathDependencies
+from dsp_tools.cli.utils import check_path_dependencies
+from dsp_tools.commands.excel2json.lists.make_lists import excel2lists
+from dsp_tools.commands.excel2json.old_lists import old_excel2lists
+from dsp_tools.commands.excel2json.project import excel2json
+from dsp_tools.commands.excel2json.project import old_excel2json
+from dsp_tools.commands.excel2json.properties import excel2properties
+from dsp_tools.commands.excel2json.resources import excel2resources
+from dsp_tools.commands.id2iri import id2iri
+
+
+def call_id2iri(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies([Path(args.xmlfile), Path(args.mapping)]))
+    return id2iri(
+        xml_file=args.xmlfile,
+        json_file=args.mapping,
+        remove_resource_if_id_in_mapping=args.remove_resources,
+    )
+
+
+def call_excel2properties(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies([Path(args.excelfile)]))
+    _, _, success = excel2properties(
+        excelfile=args.excelfile,
+        path_to_output_file=args.properties_section,
+    )
+    return success
+
+
+def call_excel2resources(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies([Path(args.excelfile)]))
+    _, _, success = excel2resources(
+        excelfile=args.excelfile,
+        path_to_output_file=args.resources_section,
+    )
+    return success
+
+
+def call_old_excel2lists(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies(required_directories=[Path(args.excelfolder)]))
+    _, success = old_excel2lists(
+        excelfolder=args.excelfolder,
+        path_to_output_file=args.lists_section,
+        verbose=args.verbose,
+    )
+    return success
+
+
+def call_excel2lists(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies(required_directories=[Path(args.excelfolder)]))
+    _, success = excel2lists(
+        excelfolder=args.excelfolder,
+        path_to_output_file=args.lists_section,
+    )
+    return success
+
+
+def call_excel2json(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies(required_directories=[Path(args.excelfolder)]))
+    return excel2json(
+        data_model_files=args.excelfolder,
+        path_to_output_file=args.project_definition,
+    )
+
+
+def call_old_excel2json(args: argparse.Namespace) -> bool:
+    check_path_dependencies(PathDependencies(required_directories=[Path(args.excelfolder)]))
+    return old_excel2json(
+        data_model_files=args.excelfolder,
+        path_to_output_file=args.project_definition,
+    )