feat: don't erase comment-only yaml documents

src/yamlfix/adapters.py

@@ -224,6 +224,13 @@ def patch_sequence_style(key_node: Node, value_node: Node) -> None:
                 if not sequence_node.value:
                     return
 
+                # if this key_node is the `yamlfix_document_fix_id` node,
+                # the sequence_node is the top-level list, which has to be forced
+                # into block-mode, so the key_node can be removed afterwards
+                force_block_style = force_block_style or self._seq_is_in_top_level_node(
+                    key_node
+                )
+
                 # if this sequence contains non-scalar nodes (i.e. dicts, lists, etc.),
                 # force block-style
                 force_block_style = (
@@ -252,6 +259,9 @@ def patch_sequence_style(key_node: Node, value_node: Node) -> None:
 
         self.patch_functions.append(patch_sequence_style)
 
+    def _seq_is_in_top_level_node(self, key_node: Node) -> bool:
+        return str(key_node.value) == f"yamlfix_{self.config.document_fix_id}"
+
     @staticmethod
     def _seq_contains_non_scalar_nodes(seq_node: Node) -> bool:
         return any(not isinstance(node, ScalarNode) for node in seq_node.value)
@@ -342,9 +352,11 @@ def fix(self, source_code: str) -> str:
         log.debug("Running source code fixers...")
 
         fixers = [
+            self._fix_comment_only_files,
             self._fix_truthy_strings,
             self._fix_jinja_variables,
             self._ruamel_yaml_fixer,
+            self._restore_comment_only_files,
             self._restore_truthy_strings,
             self._restore_jinja_variables,
             self._restore_double_exclamations,
@@ -693,3 +705,33 @@ def _restore_jinja_variables(source_code: str) -> str:
             fixed_source_lines.append(line)
 
         return "\n".join(fixed_source_lines)
+
+    def _fix_comment_only_files(self, source_code: str) -> str:
+        """Add a mapping key with an id to the start of the document\
+            to preserve comments."""
+        fixed_source_lines = []
+        yamlfix_document_id_line = f"yamlfix_{self.config.document_fix_id}:"
+
+        # Add the document id line after each document start
+        has_start_indicator = False
+        for line in source_code.splitlines():
+            fixed_source_lines.append(line)
+            if line.startswith("---"):
+                has_start_indicator = True
+                fixed_source_lines.append(yamlfix_document_id_line)
+
+        # if the document has no start indicator, the document id as the first line
+        if not has_start_indicator:
+            fixed_source_lines.insert(0, yamlfix_document_id_line)
+
+        return "\n".join(fixed_source_lines)
+
+    def _restore_comment_only_files(self, source_code: str) -> str:
+        """Remove the document start id from the document again."""
+        fixed_source_lines = []
+
+        for line in source_code.splitlines():
+            if self.config.document_fix_id not in line:
+                fixed_source_lines.append(line)
+
+        return "\n".join(fixed_source_lines)

src/yamlfix/model.py

@@ -1,5 +1,6 @@
 """Define program entities like configuration value entities."""
 
+import uuid
 from typing import Optional
 
 from maison.schema import ConfigSchema
@@ -12,6 +13,7 @@ class YamlfixConfig(ConfigSchema):
     comments_min_spaces_from_content: int = 2
     comments_require_starting_space: bool = True
     config_path: Optional[str] = None
+    document_fix_id: str = uuid.uuid4().hex
     explicit_start: bool = True
     flow_style_sequence: Optional[bool] = True
     indent_mapping: int = 2

tests/unit/test_services.py

@@ -184,6 +184,21 @@ def test_fix_code_preserves_comments(self) -> None:
 
         assert result == source
 
+    def test_fix_code_preserves_comments_without_start_indication(self) -> None:
+        """Don't delete comments without yaml explicit start indictor."""
+        source = dedent(
+            """\
+            # Keep comments!
+            program: yamlfix
+            """
+        )
+        config = YamlfixConfig()
+        config.explicit_start = False
+
+        result = fix_code(source, config)
+
+        assert result == source
+
     def test_fix_code_preserves_comment_only_file(self) -> None:
         """Don't delete comments even if the file is only comments."""
         source = dedent(
@@ -196,7 +211,24 @@ def test_fix_code_preserves_comment_only_file(self) -> None:
         result = fix_code(source)
 
         assert result == source
-
+
+    def test_fix_code_preserves_comment_only_files_without_start_indication(
+        self,
+    ) -> None:
+        """Don't delete comments even if the file is only comments, without\
+            start indication."""
+        source = dedent(
+            """\
+            # Keep comments!
+            """
+        )
+        config = YamlfixConfig()
+        config.explicit_start = False
+
+        result = fix_code(source, config)
+
+        assert result == source
+
     def test_fix_code_respects_parent_lists_with_comments(self) -> None:
         """Do not indent lists at the first level even if there is a comment."""
         source = dedent(