bug: fix description parsing for back-ticked columns

scripts/python/idc_index_data_manager.py

@@ -222,21 +222,21 @@ def _extract_column_name(column_def: str) -> str | None:
         column_def = column_def.strip().rstrip(",").strip()
 
         # Look for the last AS clause (to handle nested AS in CAST expressions)
-        # Use a regex that finds the rightmost AS followed by a word
-        as_matches = list(re.finditer(r"\bAS\b\s+(\w+)", column_def, re.IGNORECASE))
+        # Use a regex that finds the rightmost AS followed by a word (optionally backtick-quoted)
+        as_matches = list(re.finditer(r"\bAS\b\s+`?(\w+)`?", column_def, re.IGNORECASE))
         if as_matches:
             # Return the last match (rightmost AS clause)
             return as_matches[-1].group(1)
 
         # If no AS clause, try to get the column name
         # Remove function calls and get the last word before comma
-        # Handle cases like: column_name, or just column_name
+        # Handle cases like: column_name, or `column_name` (backtick-quoted reserved words)
         parts = column_def.split()
         if parts:
             # Get the last word that looks like an identifier
             for original_part in reversed(parts):
-                # Remove trailing punctuation
-                part = original_part.rstrip(",").strip()
+                # Remove trailing punctuation and backtick quoting
+                part = original_part.rstrip(",").strip().strip("`")
                 # Check if it's a valid identifier (word characters only)
                 if re.match(r"^\w+$", part):
                     return part

tests/test_column_description_parser.py

@@ -125,6 +125,21 @@ def test_extract_column_name_with_as(self):
         )
         assert IDCIndexDataManager._extract_column_name("column AS alias,") == "alias"
 
+    def test_extract_column_name_backtick_quoted(self):
+        """Test extracting column name from backtick-quoted reserved words (e.g. Rows, Columns)."""
+        assert IDCIndexDataManager._extract_column_name("`Rows`,") == "Rows"
+        assert IDCIndexDataManager._extract_column_name("`Columns`,") == "Columns"
+        assert (
+            IDCIndexDataManager._extract_column_name("ANY_VALUE(`Rows`) AS `Rows`,")
+            == "Rows"
+        )
+        assert (
+            IDCIndexDataManager._extract_column_name(
+                "ANY_VALUE(`Columns`) AS `Columns`,"
+            )
+            == "Columns"
+        )
+
     def test_extract_column_name_complex(self):
         """Test extracting column name from complex expressions."""
         assert (
@@ -151,6 +166,24 @@ def test_complex_multiline_select(self):
         assert "series_size_MB" in descriptions
         assert descriptions["series_size_MB"] == "total size of the series in megabytes"
 
+    def test_backtick_quoted_columns(self):
+        """Test parsing descriptions for backtick-quoted reserved-word columns (Rows, Columns)."""
+        sql_query = """
+SELECT
+  # description:
+  # number of pixel rows per image slice
+  ANY_VALUE(`Rows`) AS `Rows`,
+  # description:
+  # number of pixel columns per image slice
+  ANY_VALUE(`Columns`) AS `Columns`,
+FROM table
+"""
+        descriptions = IDCIndexDataManager.parse_column_descriptions(sql_query)
+        assert "Rows" in descriptions
+        assert descriptions["Rows"] == "number of pixel rows per image slice"
+        assert "Columns" in descriptions
+        assert descriptions["Columns"] == "number of pixel columns per image slice"
+
     def test_no_descriptions(self):
         """Test SQL query with no descriptions."""
         sql_query = """