Refactor existing endpoints and create endpoints

.gitignore

@@ -30,3 +30,5 @@ index.json
 
 # Logging files
 *.log
+
+.DS_Store

.pre-commit-config.yaml

@@ -17,7 +17,9 @@ repos:
     rev: v0.7.2
     hooks:
       - id: ruff
+        args: [--config, pyproject.toml]
       - id: ruff-format
+        args: [--config, pyproject.toml]
   - repo: local
     hooks:
     -   id: validate-schema

pyproject.toml

@@ -5,8 +5,16 @@ readme = "README.md"
 description = ""
 authors = ["ai-validatie-team <[email protected]>"]
 repository = "https://github.com/MinBZK/task-registry"
-keywords = ["AI", "Validation", "Instrument", "Task", "Registry"]
+keywords = ["AI", "Validation", "Instrument", "Requirement", "Measure", "Task", "Registry"]
 license = "EUPL-1.2"
+classifiers = [
+    "Development Status :: Alpha",
+    "Framework :: FastAPI",
+    "Topic :: Software Development :: Libraries :: Python Modules",
+    "Programming Language :: Python :: 3",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Typing :: Typed"
+]
 packages = [
     { include = "task_registry" }
 ]
@@ -48,7 +56,7 @@ types-pyyaml = "^6.0.12.20240724"
 # Ruff settings: https://docs.astral.sh/ruff/configuration/
 [tool.ruff]
 line-length = 120
-target-version = "py311"
+target-version = "py312"
 src = ["task_registry", "tests", "script"]
 include = ['script/validate']
 

task_registry/api/main.py

@@ -1,7 +1,8 @@
 from fastapi import APIRouter
-from task_registry.api.routes import health, instruments, urns
+from task_registry.api.routes import health, instruments, measures, requirements
 
 api_router = APIRouter()
 api_router.include_router(health.router, prefix="/health", tags=["health"])
 api_router.include_router(instruments.router, prefix="/instruments", tags=["instruments"])
-api_router.include_router(urns.router, prefix="/urns", tags=["urns"])
+api_router.include_router(measures.router, prefix="/measures", tags=["measures"])
+api_router.include_router(requirements.router, prefix="/requirements", tags=["requirements"])

task_registry/api/routes/instruments.py

@@ -1,8 +1,9 @@
 import logging
 
-from fastapi import APIRouter
+from fastapi import APIRouter, HTTPException
 from fastapi.responses import JSONResponse
-from task_registry.lifespan import CACHED_DATA
+from task_registry.data import Index, TaskType
+from task_registry.lifespan import CACHED_REGISTRY
 
 router = APIRouter()
 
@@ -11,9 +12,29 @@
 
 @router.get(
     "/",
-    summary="Overview of all the Instruments in the Task Registry",
-    description="This endpoint returns a JSON with all the Instruments in the Task Registry.",
+    summary="Overview of all the instruments in the task registry.",
+    description="This endpoint returns a JSON with all the instruments in the task registry.",
     responses={200: {"description": "JSON with all the instruments."}},
 )
-async def get_root() -> JSONResponse:
-    return JSONResponse(content=CACHED_DATA["index"])
+async def get_instruments() -> Index:
+    return CACHED_REGISTRY.get_tasks_index(TaskType.INSTRUMENTS)
+
+
+# Optional parameter 'version' is included, but not used. In a new ticket
+# versioning of instruments should be handled.
+@router.get(
+    "/urn/{urn}",
+    summary="Get the contents of the specific instrument which has given URN.",
+    description="This endpoint returns a JSON with the contents of a specific instrument identified by URN"
+    " and version.",
+    responses={
+        200: {"description": "JSON with the specific contents of the instrument."},
+        400: {"description": "The URN does not exist or is not valid."},
+    },
+)
+async def get_instrument(urn: str, version: str = "latest") -> JSONResponse:
+    try:
+        content = CACHED_REGISTRY.get_task(urn, TaskType.INSTRUMENTS)
+        return JSONResponse(content=content)
+    except KeyError as err:
+        raise HTTPException(status_code=400, detail=f"invalid urn: {urn}") from err

task_registry/api/routes/measures.py

@@ -0,0 +1,40 @@
+import logging
+
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import JSONResponse
+from task_registry.data import Index, TaskType
+from task_registry.lifespan import CACHED_REGISTRY
+
+router = APIRouter()
+
+logger = logging.getLogger(__name__)
+
+
+@router.get(
+    "/",
+    summary="Overview of all the measures in the task registry",
+    description="This endpoint returns a JSON with all the measures in the task registry.",
+    responses={200: {"description": "JSON with all the measures."}},
+)
+async def get_measures() -> Index:
+    return CACHED_REGISTRY.get_tasks_index(TaskType.MEASURES)
+
+
+# Optional parameter 'version' is included, but not used. In a new ticket
+# versioning of measures should be handled.
+@router.get(
+    "/urn/{urn}",
+    summary="Get the contents of the specific measure by URN",
+    description="This endpoint returns a JSON with the contents of a specific measure identified by URN"
+    " and version.",
+    responses={
+        200: {"description": "JSON with the specific contents of the measure."},
+        400: {"description": "The URN does not exist or is not valid."},
+    },
+)
+async def get_measure(urn: str, version: str = "latest") -> JSONResponse:
+    try:
+        content = CACHED_REGISTRY.get_task(urn, TaskType.MEASURES)
+        return JSONResponse(content=content)
+    except KeyError as err:
+        raise HTTPException(status_code=400, detail=f"invalid urn: {urn}") from err

task_registry/api/routes/requirements.py

@@ -0,0 +1,40 @@
+import logging
+
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import JSONResponse
+from task_registry.data import Index, TaskType
+from task_registry.lifespan import CACHED_REGISTRY
+
+router = APIRouter()
+
+logger = logging.getLogger(__name__)
+
+
+@router.get(
+    "/",
+    summary="Overview of all the requirements in the task registry.",
+    description="This endpoint returns a JSON with all the requirements in the task registry.",
+    responses={200: {"description": "JSON with all the requirements."}},
+)
+async def get_requirements() -> Index:
+    return CACHED_REGISTRY.get_tasks_index(TaskType.REQUIREMENTS)
+
+
+# Optional parameter 'version' is included, but not used. In a new ticket
+# versioning of requirements should be handled.
+@router.get(
+    "/urn/{urn}",
+    summary="Get the contents of the specific instrument which has given URN.",
+    description="This endpoint returns a JSON with the contents of a specific instrument identified by URN"
+    " and version.",
+    responses={
+        200: {"description": "JSON with the specific contents of the instrument."},
+        400: {"description": "The URN does not exist or is not valid."},
+    },
+)
+async def get_requirement(urn: str, version: str = "latest") -> JSONResponse:
+    try:
+        content = CACHED_REGISTRY.get_task(urn, TaskType.REQUIREMENTS)
+        return JSONResponse(content=content)
+    except KeyError as err:
+        raise HTTPException(status_code=400, detail=f"invalid urn: {urn}") from err

task_registry/core/config.py

@@ -10,9 +10,14 @@
 # Self type is not available in Python 3.10 so create our own with TypeVar
 SelfSettings = TypeVar("SelfSettings", bound="Settings")
 
-PROJECT_NAME: str = "TR"
-PROJECT_DESCRIPTION: str = "Task Registry"
+PROJECT_NAME: str = "Task Registry"
+PROJECT_SUMMARY: str = """
+API service for the task registry. This API can be used to retrieve
+measures, requirements and instruments in this registry.
+"""
 VERSION: str = "0.1.0"  # replace in CI/CD pipeline
+LICENSE_NAME: str = "EUPL-1.2 license"
+LICENSE_URL: str = "https://eupl.eu/1.2/en/"
 
 
 class Settings(BaseSettings):

task_registry/data.py

@@ -1,67 +1,115 @@
 import logging
 import os
-from pathlib import Path
+from enum import StrEnum
 from typing import Any
 
 import yaml
+from pydantic import BaseModel, Field
 
 logger = logging.getLogger(__name__)
 
 
-def get_file_size(file_path: str) -> int:
-    return os.path.getsize(file_path)  # pragma: no cover
+class TaskType(StrEnum):
+    INSTRUMENTS = "instruments"
+    REQUIREMENTS = "requirements"
+    MEASURES = "measures"
 
 
-def create_urn_mappper(entries: list[dict[str, Any]]) -> dict[str, Any]:
-    urn_mapper: dict[str, Any] = {}
-    for instrument in entries:
-        path = Path(instrument["path"])
-        try:
-            with open(str(path)) as f:
-                urn_mapper[instrument["urn"]] = yaml.safe_load(f)
-        except FileNotFoundError:
-            logger.exception(f"Instrument file with path {path} not found.")  # pragma: no cover
+class Link(BaseModel):
+    self: str
 
-        except yaml.YAMLError:
-            logger.exception(f"Instrument file with path {path} could not be parsed.")  # pragma: no cover
 
-    return urn_mapper
+class FileInfo(BaseModel):
+    type: str
+    size: int
+    name: str
+    path: str
+    urn: str
+    download_url: str
+    links: Link
+
+
+class Index(BaseModel):
+    type: str = Field(examples=["dir"])
+    size: int = Field(examples=[0])
+    name: str = Field(examples=["task_collection_name"])
+    path: str = Field(examples=["task_collection_path"])
+    download_url: str = Field(examples=["https://task-registry.apps.digilab.network/task_collection"])
+    links: Link = Field(examples=[{"self": "https://task-registry.apps.digilab.network"}])
+    entries: list[FileInfo] = Field(
+        examples=[
+            {
+                "type": "file",
+                "size": 1024,
+                "name": "task_name.yaml",
+                "path": "task_collection_path/task_name.yaml",
+                "urn": "urn:nl:aivt:tr:xx:xx",
+                "download_url": "https://task-registry.apps.digilab.network/task_collection/urn/urn:nl:aivt:tr:aiia:1.0",
+                "links": {
+                    "self": "https://task-registry.apps.digilab.network/task_collection/urn/urn:nl:aivt:tr:aiia:1.0"
+                },
+            }
+        ]
+    )
+
+
+class CachedRegistry:
+    def __init__(self) -> None:
+        self.index_cache: dict[TaskType, Index] = {}
+        self.tasks_cache: dict[tuple[str, TaskType], Any] = {}
+
+    def add_tasks(self, tasks: TaskType) -> None:
+        index = generate_index(tasks)
+        self.index_cache[tasks] = index
+
+        for task in index.entries:
+            try:
+                with open(task.path) as f:
+                    self.tasks_cache[(task.urn, tasks)] = yaml.safe_load(f)
+            except FileNotFoundError:
+                logger.exception(f"Task file with path {task.path} not found.")  # pragma: no cover
+            except yaml.YAMLError:
+                logger.exception(f"Task file with path {task.path} could not be parsed.")  # pragma: no cover
+
+    def get_tasks_index(self, tasks: TaskType) -> Index:
+        return self.index_cache[tasks]
+
+    def get_task(self, urn: str, tasks: TaskType) -> dict[str, Any]:
+        return self.tasks_cache[(urn, tasks)]
 
 
 def generate_index(
+    tasks: TaskType,
     base_url: str = "https://task-registry.apps.digilab.network",
-    directory: str = "instruments",
-) -> dict[str, Any]:
-    index: dict[str, Any] = {
-        "type": "dir",
-        "size": 0,
-        "name": directory,
-        "path": directory,
-        "download_url": f"{base_url}/instruments",
-        "_links": {
-            "self": f"{base_url}/instruments",
-        },
-        "entries": [],
-    }
-
-    for root, _, files in os.walk(directory):
+) -> Index:
+    tasks_url = f"{base_url}/{tasks}"
+    entries: list[FileInfo] = []
+
+    for root, _, files in os.walk(tasks):
         for file in files:
             if file.endswith(".yaml"):
                 file_path = os.path.join(root, file)
                 relative_path = file_path.replace("\\", "/")
                 with open(file_path) as f:
-                    instrument = yaml.safe_load(f)
-                    file_info = {
-                        "type": "file",
-                        "size": get_file_size(file_path),
-                        "name": file,
-                        "path": relative_path,
-                        "urn": instrument["urn"],
-                        "download_url": f"{base_url}/urns/?urn={instrument['urn']}",
-                        "_links": {
-                            "self": f"{base_url}/urns/?urn={instrument['urn']}",
-                        },
-                    }
-                    index["entries"].append(file_info)
-
-    return index
+                    task = yaml.safe_load(f)
+                    task_url = f"{tasks_url}/urn/{task['urn']}"
+                    file_info = FileInfo(
+                        type="file",
+                        size=os.path.getsize(file_path),
+                        name=file,
+                        path=relative_path,
+                        urn=task["urn"],
+                        download_url=task_url,
+                        links=Link(self=task_url),
+                    )
+                    entries.append(file_info)
+
+    return Index(
+        type="dir",
+        size=0,
+        name=tasks.value,
+        path=tasks.value,
+        download_url=tasks_url,
+        links=Link(self=tasks_url),
+        entries=entries,
+    )