From 063d80a037041d7892eaa2e59156d1198531b432 Mon Sep 17 00:00:00 2001
From: Daniel <139119540+hf-ddernbach@users.noreply.github.com>
Date: Mon, 18 Sep 2023 14:24:09 +0200
Subject: [PATCH] Added error handling for ambiguous outcome_codes -> A**
 (#127)

cf. https://github.com/Hochfrequenz/ebdtable2graph/issues/38
---
 src/ebdtable2graph/graph_conversion.py       | 18 ++++++++++++++++++
 src/ebdtable2graph/models/errors/__init__.py | 12 +++++++++++-
 2 files changed, 29 insertions(+), 1 deletion(-)

diff --git a/src/ebdtable2graph/graph_conversion.py b/src/ebdtable2graph/graph_conversion.py
index 9ee7e3f..57cfba9 100644
--- a/src/ebdtable2graph/graph_conversion.py
+++ b/src/ebdtable2graph/graph_conversion.py
@@ -23,6 +23,7 @@
 from ebdtable2graph.models.errors import (
     EbdCrossReferenceNotSupportedError,
     EndeInWrongColumnError,
+    OutcomeCodeAmbiguousError,
     OutcomeNodeCreationError,
 )
 
@@ -89,6 +90,8 @@ def get_all_edges(table: EbdTable) -> List[EbdGraphEdge]:
     first_node_after_start = _get_key_and_node_with_lowest_step_number(table)[1]
     result: List[EbdGraphEdge] = [EbdGraphEdge(source=nodes["Start"], target=first_node_after_start, note=None)]
 
+    outcome_nodes_duplicates: dict[str, OutcomeNode] = {}  # map to check for duplicate outcome nodes
+
     for row in table.rows:
         decision_node = _convert_row_to_decision_node(row)
         for sub_row in row.sub_rows:
@@ -100,6 +103,7 @@ def get_all_edges(table: EbdTable) -> List[EbdGraphEdge]:
                 )
             else:
                 outcome_node: Optional[OutcomeNode] = _convert_sub_row_to_outcome_node(sub_row)
+
                 if outcome_node is None:
                     if all(sr.result_code is None for sr in row.sub_rows) and any(
                         sr.note is not None and sr.note.startswith("EBD ") for sr in row.sub_rows
@@ -110,6 +114,20 @@ def get_all_edges(table: EbdTable) -> List[EbdGraphEdge]:
                     ):
                         raise EndeInWrongColumnError(row=row)
                     raise OutcomeNodeCreationError(decision_node=decision_node, sub_row=sub_row)
+
+                # check for ambiguous outcome nodes, i.e. A** with different notes
+                is_ambiguous_outcome_node = (
+                    outcome_node.result_code in outcome_nodes_duplicates
+                    and outcome_nodes_duplicates[outcome_node.result_code].note != outcome_node.note
+                )
+
+                if not is_ambiguous_outcome_node:
+                    outcome_nodes_duplicates[outcome_node.result_code] = outcome_node
+                else:
+                    raise OutcomeCodeAmbiguousError(
+                        outcome_node1=outcome_nodes_duplicates[outcome_node.result_code], outcome_node2=outcome_node
+                    )
+
                 edge = _yes_no_edge(
                     sub_row.check_result.result,
                     source=decision_node,
diff --git a/src/ebdtable2graph/models/errors/__init__.py b/src/ebdtable2graph/models/errors/__init__.py
index ce4de07..6100425 100644
--- a/src/ebdtable2graph/models/errors/__init__.py
+++ b/src/ebdtable2graph/models/errors/__init__.py
@@ -4,7 +4,7 @@
 """
 from typing import Optional
 
-from ebdtable2graph.models import DecisionNode, EbdTableRow, EbdTableSubRow
+from ebdtable2graph.models import DecisionNode, EbdTableRow, EbdTableSubRow, OutcomeNode
 
 
 class NotExactlyTwoOutgoingEdgesError(NotImplementedError):
@@ -109,3 +109,13 @@ def __init__(self, decision_node: DecisionNode, sub_row: EbdTableSubRow):
         super().__init__(f"Cannot create outcome node from sub row {sub_row} for DecisionNode {decision_node}.")
         self.sub_row = sub_row
         self.decision_node = decision_node
+
+
+class OutcomeCodeAmbiguousError(ValueError):
+    """
+    Raised when the result nodes are ambiguous. This can be the case for "A**" results.
+    """
+
+    def __init__(self, outcome_node1: OutcomeNode, outcome_node2: OutcomeNode):
+        super().__init__(f"Ambiguous result codes:  for [{outcome_node1, outcome_node2}].")
+        self.outcome_nodes = [outcome_node1, outcome_node2]