feat(yaml): support YAML 1.1

Author: dbohdan

README.md

@@ -14,11 +14,11 @@ Remarshal can also convert all supported formats to Python code.
 
 ## Known limitations and quirks
 
-### YAML 1.2 only
+### YAML versions
 
-Remarshal works with YAML 1.2.
-The last version that read and wrote YAML 1.1 was 0.17.1.
-Install it if you need YAML 1.1.
+Remarshal works with YAML 1.2 by default.
+You can choose the input and/or output format `yaml-1.1` to work with YAML 1.1; `yaml-1.2` lets you emphasize you mean 1.2.
+Conversion from YAML 1.1 to YAML 1.2 and from YAML 1.2 to YAML 1.1 is supported.
 
 ### Lossless by default; lossy must be enabled
 
@@ -133,9 +133,11 @@ uv tool install https://github.com/remarshal-project/remarshal
 ## Usage
 
 ```none
-usage: remarshal [-h] [-v] [-f {cbor,json,msgpack,toml,yaml}] [-i <input>]
-                 [--indent <n>] [-k] [--max-values <n>] [--multiline <n>]
-                 [-o <output>] [-s] [-t {cbor,json,msgpack,python,toml,yaml}]
+usage: remarshal [-h] [-v]
+                 [-f {cbor,json,msgpack,toml,yaml,yaml-1.1,yaml-1.2}]
+                 [-i <input>] [--indent <n>] [-k] [--max-values <n>]
+                 [--multiline <n>] [-o <output>] [-s]
+                 [-t {cbor,json,msgpack,python,toml,yaml,yaml-1.1,yaml-1.2}]
                  [--unwrap <key>] [--verbose] [--width <n>] [--wrap <key>]
                  [--yaml-style {,',",|,>}] [--yaml-style-newline {,',",|,>}]
                  [input] [output]
@@ -149,7 +151,8 @@ positional arguments:
 options:
   -h, --help            show this help message and exit
   -v, --version         show program's version number and exit
-  -f, --from, --if, --input-format {cbor,json,msgpack,toml,yaml}
+  -f, --from, --if, --input-format
+{cbor,json,msgpack,toml,yaml,yaml-1.1,yaml-1.2}
                         input format
   -i, --input <input>   input file
   --indent <n>          JSON and YAML indentation
@@ -164,7 +167,8 @@ options:
                         output file
   -s, --sort-keys       sort JSON, Python, and TOML keys instead of preserving
                         key order
-  -t, --to, --of, --output-format {cbor,json,msgpack,python,toml,yaml}
+  -t, --to, --of, --output-format
+{cbor,json,msgpack,python,toml,yaml,yaml-1.1,yaml-1.2}
                         output format
   --unwrap <key>        only output the data stored under the given key
   --verbose             print debug information when an error occurs

completions/remarshal.bash

@@ -5,7 +5,7 @@ _remarshal() {
     cur=${COMP_WORDS[COMP_CWORD]}
     prev=${COMP_WORDS[COMP_CWORD - 1]}
 
-    formats='cbor json msgpack toml yaml'
+    formats='cbor json msgpack toml yaml yaml-1.1 yaml-1.2'
     input_formats=$formats
     output_formats="$formats python"
 

completions/remarshal.fish

@@ -3,15 +3,15 @@ set --local yaml_styles '"\'" "\\"" "|" ">"'
 complete -c remarshal -s h -l help -d "Show help message and exit"
 complete -c remarshal -s v -l version -d "Show program's version number and exit"
 
-complete -c remarshal -s f -l from -l if -l input-format -x -a "cbor json msgpack toml yaml" -d "Input format"
+complete -c remarshal -s f -l from -l if -l input-format -x -a "cbor json msgpack toml yaml yaml-1.1 yaml-1.2" -d "Input format"
 complete -c remarshal -s i -l input -r -d "Input file"
 complete -c remarshal -l indent -x -d "JSON and YAML indentation"
 complete -c remarshal -s k -l stringify -d "Turn special values into strings"
 complete -c remarshal -l max-values -x -d "Maximum number of values in input data"
 complete -c remarshal -l multiline -x -d "Minimum items for multiline TOML array"
 complete -c remarshal -s o -l output -r -d "Output file"
 complete -c remarshal -s s -l sort-keys -d "Sort JSON, Python, and TOML keys"
-complete -c remarshal -s t -l to -l of -l output-format -x -a "cbor json msgpack python toml yaml" -d "Output format"
+complete -c remarshal -s t -l to -l of -l output-format -x -a "cbor json msgpack python toml yaml yaml-1.1 yaml-1.2" -d "Output format"
 complete -c remarshal -l unwrap -x -d "Only output data under given key"
 complete -c remarshal -l verbose -d "Print debug information on error"
 complete -c remarshal -l width -x -d "Python and YAML line width"

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [project]
 name = "remarshal"
-version = "1.1.0"
+version = "1.2.0"
 description = "Convert between CBOR, JSON, MessagePack, TOML, and YAML"
 authors = [{ name = "D. Bohdan", email = "dbohdan@dbohdan.com" }]
 license = { text = "MIT" }

src/remarshal/main.py

@@ -67,6 +67,7 @@ class Defaults:
 
 Document = Union[bool, bytes, datetime.datetime, Mapping, None, Sequence, str]
 YAMLStyle = Literal["", "'", '"', "|", ">"]
+YAMLVersion = Union[tuple[int, int], None]
 
 
 @dataclass(frozen=True)
@@ -111,6 +112,7 @@ class YAMLOptions(FormatOptions):
     indent: int = Defaults.YAML_INDENT
     style: YAMLStyle = Defaults.YAML_STYLE
     style_newline: YAMLStyle | None = None
+    version: YAMLVersion = (1, 2)
     width: int = Defaults.WIDTH
 
 
@@ -124,6 +126,7 @@ class YAMLOptions(FormatOptions):
     "Document",
     "TooManyValuesError",
     "YAMLStyle",
+    "YAMLVersion",
     # Format dataclasses.
     "FormatOptions",
     "CBOROptions",
@@ -142,16 +145,36 @@ class YAMLOptions(FormatOptions):
 ]
 
 
-INPUT_FORMATS = ["cbor", "json", "msgpack", "toml", "yaml"]
-OUTPUT_FORMATS = ["cbor", "json", "msgpack", "python", "toml", "yaml"]
-OUTPUT_FORMATS_ARGV0 = ["cbor", "json", "msgpack", "py", "toml", "yaml"]
+INPUT_FORMATS = ["cbor", "json", "msgpack", "toml", "yaml", "yaml-1.1", "yaml-1.2"]
+OUTPUT_FORMATS = [
+    "cbor",
+    "json",
+    "msgpack",
+    "python",
+    "toml",
+    "yaml",
+    "yaml-1.1",
+    "yaml-1.2",
+]
+OUTPUT_FORMATS_ARGV0 = [
+    "cbor",
+    "json",
+    "msgpack",
+    "py",
+    "toml",
+    "yaml",
+    "yaml-1.1",
+    "yaml-1.2",
+]
 OPTIONS_CLASSES = {
     "cbor": CBOROptions,
     "json": JSONOptions,
     "msgpack": MsgPackOptions,
     "python": PythonOptions,
     "toml": TOMLOptions,
     "yaml": YAMLOptions,
+    "yaml-1.1": YAMLOptions,
+    "yaml-1.2": YAMLOptions,
 }
 UTF_8 = "utf-8"
 
@@ -436,6 +459,17 @@ def output_width(value: str) -> int:
     return args
 
 
+def _yaml_version(format: str) -> YAMLVersion:
+    match format:
+        case "yaml-1.1":
+            return (1, 1)
+
+        case "yaml-1.2":
+            return (1, 2)
+
+    return None
+
+
 # === Parser/serializer wrappers ===
 
 
@@ -532,9 +566,11 @@ def _decode_toml(input_data: bytes) -> Document:
         raise ValueError(msg)
 
 
-def _decode_yaml(input_data: bytes) -> Document:
+def _decode_yaml(input_data: bytes, version: YAMLVersion) -> Document:
     try:
         yaml = ruamel.yaml.YAML(pure=True, typ="safe")
+        yaml.version = version
+
         doc = yaml.load(input_data)
 
         return cast("Document", doc)
@@ -545,19 +581,25 @@ def _decode_yaml(input_data: bytes) -> Document:
 
 
 def decode(input_format: str, input_data: bytes) -> Document:
-    decoder = {
-        "cbor": _decode_cbor,
-        "json": _decode_json,
-        "msgpack": _decode_msgpack,
-        "toml": _decode_toml,
-        "yaml": _decode_yaml,
-    }
-
-    if input_format not in decoder:
-        msg = f"Unknown input format: {input_format}"
-        raise ValueError(msg)
+    match input_format:
+        case "cbor":
+            return _decode_cbor(input_data)
 
-    return decoder[input_format](input_data)
+        case "json":
+            return _decode_json(input_data)
+
+        case "msgpack":
+            return _decode_msgpack(input_data)
+
+        case "toml":
+            return _decode_toml(input_data)
+
+        case "yaml" | "yaml-1.1" | "yaml-1.2":
+            return _decode_yaml(input_data, version=_yaml_version(input_format))
+
+        case _:
+            msg = f"Unknown input format: {input_format}"
+            raise ValueError(msg)
 
 
 class TooManyValuesError(BaseException):
@@ -776,13 +818,14 @@ def _encode_yaml(
     indent: int | None,
     style: YAMLStyle,
     style_newline: YAMLStyle | None,
+    version: YAMLVersion,
     width: int,
 ) -> str:
     yaml = ruamel.yaml.YAML(pure=True)
     yaml.default_flow_style = False
-
     yaml.default_style = style  # type: ignore
     yaml.indent = indent
+    yaml.version = version
     yaml.width = width
 
     def represent_none(self, data):
@@ -848,11 +891,12 @@ def format_options(
                 stringify=stringify,
             )
 
-        case "yaml":
+        case "yaml" | "yaml-1.1" | "yaml-1.2":
             return YAMLOptions(
                 indent=Defaults.YAML_INDENT if indent is None else indent,
                 style=yaml_style,
                 style_newline=yaml_style_newline,
+                version=_yaml_version(output_format),
                 width=width,
             )
 
@@ -925,7 +969,7 @@ def encode(
                 stringify=options.stringify,
             ).encode(UTF_8)
 
-        case "yaml":
+        case "yaml" | "yaml-1.1" | "yaml-1.2":
             if not isinstance(options, YAMLOptions):
                 msg = "expected 'options' argument to have class 'YAMLOptions'"
                 raise TypeError(msg)
@@ -935,6 +979,7 @@ def encode(
                 indent=options.indent,
                 style=options.style,
                 style_newline=options.style_newline,
+                version=options.version,
                 width=options.width,
             ).encode(UTF_8)
 

tests/test_remarshal.py

@@ -458,6 +458,16 @@ def test_binary_to_toml(self, convert_and_read) -> None:
     def test_binary_to_yaml(self, convert_and_read) -> None:
         convert_and_read("bin.msgpack", "msgpack", "yaml")
 
+    def test_yaml_1_1_features(self, convert_and_read) -> None:
+        output = convert_and_read("yaml-1.1-features.yaml", "yaml-1.1", "json")
+        reference = read_file("yaml-1.1-features.json")
+        assert output == reference
+
+    def test_yaml_1_1_features_in_1_2(self, convert_and_read) -> None:
+        output = convert_and_read("yaml-1.1-features.yaml", "yaml-1.2", "json")
+        reference = read_file("yaml-1.1-features-in-1.2.json")
+        assert output == reference
+
     def test_yaml_null(self, convert_and_read) -> None:
         output = convert_and_read("null.json", "json", "yaml")
         reference = read_file("null.yaml")
@@ -789,7 +799,7 @@ def test_yaml_billion_laughs(self, convert_and_read) -> None:
         with pytest.raises(remarshal.TooManyValuesError):
             convert_and_read("lol.yml", "yaml", "json")
 
-    def test_yaml_norway_problem(self, convert_and_read) -> None:
+    def test_yaml_norway_problem_from_default(self, convert_and_read) -> None:
         output = convert_and_read(
             "norway.yaml",
             "yaml",
@@ -799,6 +809,44 @@ def test_yaml_norway_problem(self, convert_and_read) -> None:
         reference = read_file("norway.json")
         assert output == reference
 
+    def test_yaml_norway_problem_from_yaml_1_1(self, convert_and_read) -> None:
+        output = convert_and_read(
+            "norway.yaml",
+            "yaml-1.1",
+            "json",
+            indent=None,
+        )
+        reference = read_file("norway-yaml-1.1.json")
+        assert output == reference
+
+    def test_yaml_norway_problem_from_yaml_1_2(self, convert_and_read) -> None:
+        output = convert_and_read(
+            "norway.yaml",
+            "yaml-1.2",
+            "json",
+            indent=None,
+        )
+        reference = read_file("norway.json")
+        assert output == reference
+
+    def test_yaml_norway_problem_to_yaml_1_1(self, convert_and_read) -> None:
+        output = convert_and_read(
+            "norway.json",
+            "json",
+            "yaml-1.1",
+        )
+        reference = read_file("norway-yaml-1.1.yaml")
+        assert output == reference
+
+    def test_yaml_norway_problem_to_yaml_1_2(self, convert_and_read) -> None:
+        output = convert_and_read(
+            "norway.json",
+            "json",
+            "yaml-1.2",
+        )
+        reference = read_file("norway-yaml-1.2.yaml")
+        assert output == reference
+
     def test_toml2cbor_date(self, convert_and_read) -> None:
         output = convert_and_read("date.toml", "toml", "cbor")
         reference = read_file("date.cbor")
@@ -826,10 +874,12 @@ def test_toml2msgpack_date(self, convert_and_read) -> None:
     def test_toml2toml_date(self, convert_and_read) -> None:
         output = convert_and_read("date.toml", "toml", "toml")
         reference = read_file("date.toml")
+        assert output == reference
 
     def test_toml2yaml_date(self, convert_and_read) -> None:
         output = convert_and_read("date.toml", "toml", "yaml")
         reference = read_file("date.yaml")
+        assert output == reference
 
     def test_toml2cbor_datetime_local(self, convert_and_read) -> None:
         with pytest.raises(ValueError):