Ensure rule_ids are valid. (#6738)

packages/python-packages/apiview-gpt/output.json

@@ -3,8 +3,8 @@
   "violations": [
     {
       "rule_ids": [
-        "python-client-naming",
-        "python-client-separate-sync-async"
+        "python_design.html#python-client-naming",
+        "python_design.html#python-client-separate-sync-async"
       ],
       "line_no": 2,
       "bad_code": "class azure.contoso.ContosoAsyncManager:",
@@ -13,7 +13,7 @@
     },
     {
       "rule_ids": [
-        "python-client-options-naming"
+        "python_design.html#python-client-options-naming"
       ],
       "line_no": 8,
       "bad_code": "options: dict",
@@ -64,16 +64,6 @@
       "bad_code": "red = \"red\"",
       "suggestion": "RED = \"red\"",
       "comment": "Enum names should be in UPPERCASE."
-    },
-    {
-      "rule_ids": [
-        "python_design.html#python-client-naming",
-        "python_design.html#python-models-dict-result"
-      ],
-      "line_no": 30,
-      "bad_code": "class azure.contoso.WidgetResult:",
-      "suggestion": "Use a simple Mapping (e.g. dict) instead of creating a WidgetResult class if the WidgetResult class is not used as an input parameter for other APIs.",
-      "comment": "Service client types should have a Client suffix. Consider using a simple Mapping instead of creating a separate class for operation results."
     }
   ]
 }

packages/python-packages/apiview-gpt/src/__init__.py

@@ -12,7 +12,7 @@ def console_entry_point():
     print("Running apiview-gpt version {}".format(__version__))
     reviewer = GptReviewer()
     # FIXME: Make this a proper CLI
-    input_filename = "test.txt"
+    input_filename = "test2.txt"
     file_path = os.path.join(_PACKAGE_ROOT, input_filename)
     with open(file_path, "r") as f:
         apiview_text = f.read()

packages/python-packages/apiview-gpt/src/_gpt_reviewer.py

@@ -44,8 +44,8 @@ def __init__(self):
         self.chain = LLMChain(llm=self.llm, prompt=self.prompt_template)
 
     def get_response(self, apiview, language):
-        general_guidelines, language_guidelines = self.retrieve_guidelines(language)
-        all_guidelines = general_guidelines + language_guidelines
+        apiview = self.unescape(apiview)
+        all_guidelines = self.retrieve_guidelines(language)
 
         # TODO: Make this not hard-coded!
         guidelines = self.select_guidelines(all_guidelines, [
@@ -76,8 +76,46 @@ def get_response(self, apiview, language):
                 # FIXME see: https://github.com/Azure/azure-sdk-tools/issues/6571
                 if len(output.violations) > 0:
                     final_results.status = "Error"
+        self.process_rule_ids(final_results, all_guidelines)
         return final_results
 
+    """ Ensure that each rule ID matches with an actual guideline ID. 
+        This ensures that the links that appear in APIView should never be broken (404).
+    """
+    def process_rule_ids(self, results, guidelines):
+        # create an index for easy lookup
+        index = { x["id"]: x for x in guidelines }
+        for violation in results.violations:
+            to_remove = []
+            to_add = []
+            for rule_id in violation.rule_ids:
+                try:
+                    index[rule_id]
+                    continue
+                except KeyError:
+                    # see if any guideline ID ends with the rule_id. If so, update it and preserve in the index
+                    matched = False
+                    for guideline in guidelines:
+                        if guideline["id"].endswith(rule_id):
+                            to_remove.append(rule_id)
+                            to_add.append(guideline["id"])
+                            index[rule_id] = guideline["id"]
+                            matched = True
+                            break
+                    if matched:
+                        continue
+                    # no match or partial match found, so remove the rule_id
+                    to_remove.append(rule_id)
+                    print(f"WARNING: Rule ID {rule_id} not found. Possible hallucination.")
+            # update the rule_ids arrays with the new values. Don't modify the array while iterating over it!
+            for rule_id in to_remove:
+                violation.rule_ids.remove(rule_id)
+            for rule_id in to_add:
+                violation.rule_ids.append(rule_id)
+
+    def unescape(self, text: str) -> str:
+        return str(bytes(text, "utf-8").decode("unicode_escape"))
+
     def process_violations(self, violations: List[Violation], section: Section) -> List[Violation]:
         if not violations:
             return violations
@@ -146,4 +184,4 @@ def retrieve_guidelines(self, language):
             with open(os.path.join(language_guidelines_path, filename), "r") as f:
                 items = json.loads(f.read())
                 language_guidelines.extend(items)
-        return general_guidelines, language_guidelines
+        return general_guidelines + language_guidelines

packages/python-packages/apiview-gpt/test2.txt

@@ -0,0 +1 @@
+# Package is parsed using apiview-stub-generator(version:0.3.7), Python version: 3.10.12\n\nclass azure.contoso.ContosoAsyncManager:\n\n    async def __init__(\n            self,\n            endpoint,\n            credential,\n            options: dict\n        )\n\n    async def get_widget_async(self, options: dict) -> WidgetResult\n\nclass azure.contoso.ContosoManager:\n\n    def __init__(\n    self,\n            endpoint,\n            credential,\n            options: dict\n        )\n\n    def get_widget(self, options: dict) -> WidgetResult\n\nclass azure.contoso.WidgetColor(Enum):\n\n    blue = "blue"\n    green = "green"\n    red = "red"\n\nclass azure.contoso.WidgetResult:\n\n    def __init__(\n            self,\n            name: str,\n            color: WidgetColor\n        )\n\n