schema: make the schema compatible with 2.0

* application profile 2.0 changes some small things like contenthash ->
  identifier and fileurl to source to provide the naming for link only
  records.

invenio_moodle/schemas.py

@@ -13,6 +13,8 @@
 from marshmallow import Schema, ValidationError, validates_schema
 from marshmallow.fields import Constant, Dict, List, Nested, Number, String
 
+from .utils import extract_moodle_records
+
 
 class ClassificationValuesSchema(Schema):
     """Moodle classification-values schema."""
@@ -79,12 +81,14 @@ class FileSchema(Schema):
 
     abstract = String(required=True)
     classification = List(Nested(ClassificationSchema), required=True)
-    contenthash = String(required=True)
+    contenthash = String()  # application profile 1.0
+    identifier = String()  # application profile 2.0
     context = String(required=True)
     courses = List(Nested(CourseSchema), required=True)
     filecreationtime = String(required=True)
     filesize = Number(required=True)
-    fileurl = String(required=True)
+    fileurl = String()  # application profile 1.0
+    source = String()  # application profile 2.0
     language = String(required=True)
     license = Nested(LicenseSchema)  # noqa: A003
     mimetype = String(required=True)
@@ -101,6 +105,7 @@ class MoodleCourseSchema(Schema):
     """Moodle moodlecourse schema."""
 
     files = List(Nested(FileSchema))
+    elements = List(Nested(FileSchema))
 
 
 class MoodleSchema(Schema):
@@ -109,19 +114,12 @@ class MoodleSchema(Schema):
     Data coming from moodle should be in this format.
     """
 
-    applicationprofile = String(required=True)
-    moodlecourses = Dict(
-        keys=String(),
-        values=Nested(MoodleCourseSchema, required=True),
-    )
-
     @validates_schema
     def validate_urls_unique(self, data: dict, **__: dict) -> None:
         """Check that each file-URL only appears once."""
         urls_counter = Counter(
-            file_["fileurl"]
-            for _, moodlecourse in data["moodlecourses"].items()
-            for file_ in moodlecourse["files"]
+            file_["fileurl"] if "fileurl" in file_ else file_["source"]
+            for file_ in extract_moodle_records(data)
         )
         duplicated_urls = [url for url, count in urls_counter.items() if count > 1]
         if duplicated_urls:
@@ -157,3 +155,21 @@ def validate_course_jsons_unique_per_courseid(self, data: dict, **__: dict) -> N
             course_ids = ", ".join(course_id for course_id in ambiguous_courseids)
             msg = f"Different course-JSONs with same courseid {course_ids}."
             raise ValidationError(msg)
+
+
+class MoodleSchemaApplicationProfile1(MoodleSchema):
+    """Moodle Schema for application profile 1.0."""
+
+    applicationprofile = String(required=True)
+    moodlecourses = Dict(
+        keys=String(),
+        values=Nested(MoodleCourseSchema, required=True),
+    )
+
+
+class MoodleSchemaApplicationProfile2(MoodleSchema):
+    """Moodle schema for application profile 2.0."""
+
+    applicationprofile = String(required=True)
+    release = Number()
+    elements = List(Nested(FileSchema))

invenio_moodle/services.py

@@ -14,7 +14,7 @@
 from marshmallow import ValidationError
 
 from .records import MoodleAPI, MoodleRESTConfig
-from .schemas import MoodleSchema
+from .schemas import MoodleSchemaApplicationProfile1, MoodleSchemaApplicationProfile2
 from .types import FileCacheInfo
 from .utils import extract_moodle_records, post_processing
 
@@ -47,9 +47,12 @@ def fetch_records(self, _: Identity) -> list[dict]:
         moodle_data = self.api.fetch_records()
 
         try:
-            MoodleSchema().load(moodle_data)
-        except ValidationError as error:
-            raise RuntimeError(str(error)) from error
+            MoodleSchemaApplicationProfile1().load(moodle_data)
+        except ValidationError:
+            try:
+                MoodleSchemaApplicationProfile2().load(moodle_data)
+            except ValidationError as error:
+                raise RuntimeError(str(error)) from error
 
         moodle_records = extract_moodle_records(moodle_data)
         post_processing(moodle_records)

invenio_moodle/utils.py

@@ -20,11 +20,32 @@ def is_course_root(sourceid: str) -> bool:
 
 def extract_moodle_records(moodle_data: dict) -> list[dict]:
     """Create moodle file jsons."""
-    return [
-        file_json
-        for _, moodle_course in moodle_data["moodlecourses"].items()
-        for file_json in moodle_course["files"]
-    ]
+
+    # application profile 2.0 uses elements and a flat structure to serve the metadata.
+    # {
+    #   "elements": [
+    #     {METADATA goes here}
+    #   ]
+    # }
+
+    if "elements" in moodle_data:
+        return moodle_data["elements"]
+
+    # application profile 1.0 uses a nested structure with
+    # {
+    #   "moodlecourses":{
+    #     "1": {
+    #        "files OR elements":[
+    #          {METADATA goes here}
+    #      ]
+    # }
+    elements = []
+    for _, moodle_course in moodle_data["moodlecourses"].items():
+        if "files" in moodle_course:
+            elements.extend(file_json for file_json in moodle_course["files"])
+        if "elements" in moodle_course:
+            elements.extend(file_json for file_json in moodle_course["elements"])
+    return elements
 
 
 def remove_moodle_only_course(moodle_records: dict) -> None: