DIG-1669: Automate incrementing schema version when schemas get updated

.github/workflows/update-schema.yaml

@@ -1,13 +1,13 @@
 name: Docs
 
 on:
-  push:
+  pull_request:
+    branches:
+      - develop
     paths:
-      - 'chord_metadata_service/mohpackets/schemas/**'
-      - 'chord_metadata_service/mohpackets/models.py'
-      - 'chord_metadata_service/mohpackets/docs/schema.json'
+      - "chord_metadata_service/mohpackets/schemas/**"
+      - "chord_metadata_service/mohpackets/models.py"
 
-
 jobs:
   generate-moh-schema:
     name: Update MoH schema
@@ -36,37 +36,35 @@ jobs:
           echo "CACHE_DURATION = 0" >> config/settings/base.py
           echo "AGGREGATE_COUNT_THRESHOLD = 5" >> config/settings/base.py
           python manage.py export_openapi_schema --api chord_metadata_service.mohpackets.apis.core.api | python -m json.tool > chord_metadata_service/mohpackets/docs/schema.json
-
-      - name: Commit new schema.json
-        uses: stefanzweifel/git-auto-commit-action@v5
-        id: auto-commit-action
-        with:
-          commit_message: Update schema.json
-          file_pattern: 'chord_metadata_service/mohpackets/docs/schema.json'
-
-      - name: Update schema.json with new SHA
-        if: steps.auto-commit-action.outputs.changes_detected == 'true'
-        run: |
-          REPO_NAME=${{ github.repository }}
-          SHA=${{ steps.auto-commit-action.outputs.commit_hash }}
-          SCHEMA_PATH=chord_metadata_service/mohpackets/docs/schema.json
-          sed -i 's|"description": "This is the RESTful API for the MoH Service."|"description": "This is the RESTful API for the MoH Service. Based on https://raw.githubusercontent.com/'"$REPO_NAME"'/'"$SHA"'/'"$SCHEMA_PATH"'"|' $SCHEMA_PATH
 
       - name: Install widdershins
         run: npm install -g widdershins
 
       - name: Convert schema to OpenAPI documentation
         run: |
           npx widdershins ./chord_metadata_service/mohpackets/docs/schema.json -o ./chord_metadata_service/mohpackets/docs/schema.md -u ./chord_metadata_service/mohpackets/docs/widdershins/templates/openapi3 -c true --omitHeader true
-      
+
       - name: Install PyYAML
         run: pip install PyYAML
 
       - name: Convert schema.json to schema.yml
         run: python -c 'import json, yaml; json.load(open("chord_metadata_service/mohpackets/docs/schema.json")); print(yaml.dump(json.load(open("chord_metadata_service/mohpackets/docs/schema.json"))))' > chord_metadata_service/mohpackets/docs/schema.yml
 
-      - name: Update schemas with SHA
+      - name: Update schemas
+        uses: stefanzweifel/git-auto-commit-action@v5
+        with:
+          commit_message: Update schemas
+          file_pattern: "chord_metadata_service/mohpackets/docs/schema.json chord_metadata_service/mohpackets/docs/schema.md chord_metadata_service/mohpackets/docs/schema.yml"
+
+      - name: Get SHA commit of schema.json
+        id: get-sha
+        run: echo "SHA=$(git log -1 --format=%H chord_metadata_service/mohpackets/docs/schema.json)" >> $GITHUB_ENV
+
+      - name: Write SHA to schema_version.txt
+        run: echo ${{ env.SHA }} > chord_metadata_service/mohpackets/docs/schema_version.txt
+
+      - name: Commit updated schema_version.txt
         uses: stefanzweifel/git-auto-commit-action@v5
         with:
-          commit_message: Update schemas with SHA
-          file_pattern: 'chord_metadata_service/mohpackets/docs/schema.json chord_metadata_service/mohpackets/docs/schema.md chord_metadata_service/mohpackets/docs/schema.yml '
+          commit_message: Update schema version
+          file_pattern: "chord_metadata_service/mohpackets/docs/schema_version.txt"

chord_metadata_service/mohpackets/apis/core.py

@@ -21,7 +21,7 @@
     explorer_router as explorer_router,
 )
 from chord_metadata_service.mohpackets.apis.ingestion import router as ingest_router
-from chord_metadata_service.mohpackets.utils import get_schema_url
+from chord_metadata_service.mohpackets.utils import get_schema_version
 
 from django.views.decorators.cache import cache_page
 from ninja.decorators import decorate_view
@@ -192,14 +192,12 @@ def authenticate(self, request, service_token):
 @api.get("/service-info")
 @decorate_view(cache_page(settings.CACHE_DURATION))
 def service_info(request):
-    schema_url = get_schema_url()
-
     return JsonResponse(
         {
             "name": "katsu",
             "description": "A CanDIG clinical data service",
             "version": settings.KATSU_VERSION,
-            "schema_url": schema_url,
+            "schema_version": get_schema_version(),
         },
         status=200,
         safe=False,

chord_metadata_service/mohpackets/docs/schema_version.txt

@@ -0,0 +1 @@
+440

chord_metadata_service/mohpackets/utils.py

@@ -1,39 +1,17 @@
-import json
 import logging
-import re
 from enum import Enum
 
 logger = logging.getLogger(__name__)
 
 
-def get_schema_url():
+def get_schema_version():
     """
-    Retrieve the SHA URL from schema.json.
-    It reads the first 6 lines of the JSON file and extracts the URL from the "description".
+    Reads the version number from the schema file.
     """
-    url_pattern = r"https://[^\s]+"  # get everything after https
-
-    schema_url = "None"
-
-    try:
-        with open(
-            "chord_metadata_service/mohpackets/docs/schema.json", "r"
-        ) as json_file:
-            # Read the first 6 lines of the JSON file only
-            # The line we are looking for is on the 6th
-            first_6_lines = [next(json_file) for _ in range(6)]
-            first_6_lines.extend(["}", "}"])  # make valid JSON
-            # Concatenate the lines to form a JSON string
-            json_str = "".join(first_6_lines)
-            data = json.loads(json_str)
-            desc_str = data["info"]["description"]
-            schema_url = re.search(url_pattern, desc_str).group()
-    except Exception as e:
-        logger.debug(
-            f"An error occurred while fetching the schema URL. Details: {str(e)}"
-        )
-
-    return schema_url
+    file_path = "chord_metadata_service/mohpackets/docs/schema_version.txt"
+    with open(file_path, "r") as file:
+        version = file.read().strip()
+    return int(version)
 
 
 def list_to_enum(enum_name, value_list):