Skip duplicated grade_ids during autograding

nbgrader/converters/autograde.py

@@ -10,6 +10,7 @@
 from ..preprocessors import (
     AssignLatePenalties, ClearOutput, DeduplicateIds, OverwriteCells, SaveAutoGrades,
     Execute, LimitOutput, OverwriteKernelspec, CheckCellMetadata)
+from ..postprocessors import CheckDuplicateFlag
 from ..api import Gradebook, MissingEntry
 from .. import utils
 
@@ -195,3 +196,6 @@ def convert_single_notebook(self, notebook_filename: str) -> None:
                 super(Autograde, self).convert_single_notebook(notebook_filename)
         finally:
             self._sanitizing = True
+
+        self.log.info(f"Post-processing {notebook_filename}")
+        CheckDuplicateFlag(notebook_filename)

nbgrader/converters/base.py

@@ -17,6 +17,7 @@
 from ..coursedir import CourseDirectory
 from ..utils import find_all_files, rmtree, remove
 from ..preprocessors.execute import UnresponsiveKernelError
+from ..postprocessors import DuplicateIdError
 from ..nbgraderformat import SchemaTooOldError, SchemaTooNewError
 import typing
 from nbconvert.exporters.exporter import ResourcesDict
@@ -405,6 +406,13 @@ def _handle_failure(gd: typing.Dict[str, str]) -> None:
                         errors.append((gd['assignment_id'], gd['student_id']))
                         _handle_failure(gd)
 
+                    except DuplicateIdError:
+                        self.log.error(
+                            f"Encountered a cell with duplicate id when processing {notebook_filename}. "
+                            "Autograding with skipping cells marked as duplicate."
+                        )
+                        errors.append((gd['assignment_id'], gd['student_id']))
+
                     # Raise unhandled exceptions for the outer try/except
                     except Exception as e:
                         raise e

nbgrader/nbgraderformat/__init__.py

@@ -4,3 +4,9 @@
 from .v4 import reads_v4 as reads, writes_v4 as writes
 
 SCHEMA_VERSION = MetadataValidator.schema_version
+
+# Metadata required by latest schema, along with default values
+SCHEMA_REQUIRED = {"schema_version": 4,
+                   "grade": False,
+                   "locked": False,
+                   "solution": False}

nbgrader/postprocessors/__init__.py

@@ -0,0 +1,6 @@
+from .checkduplicateflag import CheckDuplicateFlag, DuplicateIdError
+
+__all__ = [
+    "CheckDuplicateFlag",
+    "DuplicateIdError"
+]

nbgrader/postprocessors/checkduplicateflag.py

@@ -0,0 +1,27 @@
+import nbformat
+from nbformat.notebooknode import NotebookNode
+
+
+class DuplicateIdError(Exception):
+
+    def __init__(self, message):
+        super(DuplicateIdError, self).__init__(message)
+
+
+class CheckDuplicateFlag:
+
+    def __init__(self, notebook_filename):
+        with open(notebook_filename, encoding="utf-8") as f:
+            nb = nbformat.read(f, as_version=nbformat.NO_CONVERT)
+        self.postprocess(nb)
+
+    def postprocess(self, nb: NotebookNode):
+        for cell in nb.cells:
+            self.postprocess_cell(cell)
+
+    @staticmethod
+    def postprocess_cell(cell: NotebookNode):
+        if "nbgrader" in cell.metadata and "duplicate" in cell.metadata.nbgrader:
+            del cell.metadata.nbgrader["duplicate"]
+            msg = "Detected cells with same ids"
+            raise DuplicateIdError(msg)

nbgrader/preprocessors/deduplicateids.py

@@ -1,5 +1,6 @@
 from .. import utils
 from . import NbGraderPreprocessor
+from ..nbgraderformat import SCHEMA_REQUIRED
 from nbconvert.exporters.exporter import ResourcesDict
 from nbformat.notebooknode import NotebookNode
 from typing import Tuple
@@ -27,7 +28,9 @@ def preprocess_cell(self,
         grade_id = cell.metadata.nbgrader['grade_id']
         if grade_id in self.grade_ids:
             self.log.warning("Cell with id '%s' exists multiple times!", grade_id)
-            cell.metadata.nbgrader = {}
+            # Replace problematic metadata and leave message
+            cell.source = "# THIS CELL CONTAINED A DUPLICATE ID DURING AUTOGRADING\n" + cell.source
+            cell.metadata.nbgrader = SCHEMA_REQUIRED | {"duplicate": True}
         else:
             self.grade_ids.add(grade_id)
 

nbgrader/tests/preprocessors/test_deduplicateids.py

@@ -3,6 +3,7 @@
 from nbformat.v4 import new_notebook
 
 from ...preprocessors import DeduplicateIds
+from ...nbgraderformat import SCHEMA_REQUIRED
 from .base import BaseTestPreprocessor
 from .. import (
     create_grade_cell, create_solution_cell, create_locked_cell)
@@ -14,6 +15,10 @@ def preprocessor():
     return pp
 
 
+EXPECTED_DUPLICATE_METADATA = SCHEMA_REQUIRED  # | {"duplicate": True}  # doesn't work in python 3.8
+EXPECTED_DUPLICATE_METADATA["duplicate"] = True
+
+
 class TestDeduplicateIds(BaseTestPreprocessor):
 
     def test_duplicate_grade_cell(self, preprocessor):
@@ -26,7 +31,7 @@ def test_duplicate_grade_cell(self, preprocessor):
         nb, resources = preprocessor.preprocess(nb, {})
 
         assert nb.cells[0].metadata.nbgrader != {}
-        assert nb.cells[1].metadata.nbgrader == {}
+        assert nb.cells[1].metadata.nbgrader == EXPECTED_DUPLICATE_METADATA
 
     def test_duplicate_solution_cell(self, preprocessor):
         cell1 = create_solution_cell("hello", "code", "foo")
@@ -38,7 +43,7 @@ def test_duplicate_solution_cell(self, preprocessor):
         nb, resources = preprocessor.preprocess(nb, {})
 
         assert nb.cells[0].metadata.nbgrader != {}
-        assert nb.cells[1].metadata.nbgrader == {}
+        assert nb.cells[1].metadata.nbgrader == EXPECTED_DUPLICATE_METADATA
 
     def test_duplicate_locked_cell(self, preprocessor):
         cell1 = create_locked_cell("hello", "code", "foo")
@@ -50,4 +55,4 @@ def test_duplicate_locked_cell(self, preprocessor):
         nb, resources = preprocessor.preprocess(nb, {})
 
         assert nb.cells[0].metadata.nbgrader != {}
-        assert nb.cells[1].metadata.nbgrader == {}
+        assert nb.cells[1].metadata.nbgrader == EXPECTED_DUPLICATE_METADATA