Standardize type comments to form # type: (value) (#2097)

CHANGES.md

@@ -17,6 +17,7 @@
 - Fix bug where module docstrings would be treated as normal strings if preceeded by
   comments (#4764)
 - Fix bug where python 3.12 generics syntax split line happens weirdly (#4777)
+- Standardize type comments to form `# type: <value>` (#4645)
 
 ### Configuration
 

docs/the_black_code_style/future_style.md

@@ -29,6 +29,8 @@ Currently, the following features are included in the preview style:
 - `fix_fmt_skip_in_one_liners`: Fix `# fmt: skip` behaviour on one-liner declarations,
   such as `def foo(): return "mock" # fmt: skip`, where previously the declaration would
   have been incorrectly collapsed.
+- `standardize_type_comments`: Format type comments which have zero or more spaces
+  between `#` and `type:` or between `type:` and value to `# type: (value)`
 - `wrap_comprehension_in`: Wrap the `in` clause of list and dictionary comprehensions
   across lines if it would otherwise exceed the maximum line length.
 - `remove_parens_around_except_types`: Remove parentheses around multiple exception

src/black/comments.py

@@ -11,6 +11,7 @@
     WHITESPACE,
     container_of,
     first_leaf_of,
+    is_type_comment_string,
     make_simple_prefix,
     preceding_leaf,
     syms,
@@ -50,7 +51,7 @@ class ProtoComment:
     leading_whitespace: str  # leading whitespace before the comment, if any
 
 
-def generate_comments(leaf: LN) -> Iterator[Leaf]:
+def generate_comments(leaf: LN, mode: Mode) -> Iterator[Leaf]:
     """Clean the prefix of the `leaf` and generate comments from it, if any.
 
     Comments in lib2to3 are shoved into the whitespace prefix.  This happens
@@ -70,15 +71,17 @@ def generate_comments(leaf: LN) -> Iterator[Leaf]:
     are emitted with a fake STANDALONE_COMMENT token identifier.
     """
     total_consumed = 0
-    for pc in list_comments(leaf.prefix, is_endmarker=leaf.type == token.ENDMARKER):
+    for pc in list_comments(
+        leaf.prefix, is_endmarker=leaf.type == token.ENDMARKER, mode=mode
+    ):
         total_consumed = pc.consumed
         prefix = make_simple_prefix(pc.newlines, pc.form_feed)
         yield Leaf(pc.type, pc.value, prefix=prefix)
     normalize_trailing_prefix(leaf, total_consumed)
 
 
 @lru_cache(maxsize=4096)
-def list_comments(prefix: str, *, is_endmarker: bool) -> list[ProtoComment]:
+def list_comments(prefix: str, *, is_endmarker: bool, mode: Mode) -> list[ProtoComment]:
     """Return a list of :class:`ProtoComment` objects parsed from the given `prefix`."""
     result: list[ProtoComment] = []
     if not prefix or "#" not in prefix:
@@ -109,7 +112,7 @@ def list_comments(prefix: str, *, is_endmarker: bool) -> list[ProtoComment]:
             comment_type = token.COMMENT  # simple trailing comment
         else:
             comment_type = STANDALONE_COMMENT
-        comment = make_comment(line)
+        comment = make_comment(line, mode=mode)
         result.append(
             ProtoComment(
                 type=comment_type,
@@ -140,7 +143,7 @@ def normalize_trailing_prefix(leaf: LN, total_consumed: int) -> None:
     leaf.prefix = ""
 
 
-def make_comment(content: str) -> str:
+def make_comment(content: str, mode: Mode) -> str:
     """Return a consistently formatted comment from the given `content` string.
 
     All comments (except for "##", "#!", "#:", '#'") should have a single
@@ -157,9 +160,18 @@ def make_comment(content: str) -> str:
     if (
         content
         and content[0] == "\N{NO-BREAK SPACE}"
-        and not content.lstrip().startswith("type:")
+        and not is_type_comment_string("# " + content.lstrip(), mode=mode)
     ):
         content = " " + content[1:]  # Replace NBSP by a simple space
+    if (
+        Preview.standardize_type_comments in mode
+        and content
+        and "\N{NO-BREAK SPACE}" not in content
+        and is_type_comment_string("#" + content, mode=mode)
+    ):
+        type_part, value_part = content.split(":", 1)
+        content = type_part.strip() + ": " + value_part.strip()
+
     if content and content[0] not in COMMENT_EXCEPTIONS:
         content = " " + content
     return "#" + content
@@ -183,7 +195,7 @@ def convert_one_fmt_off_pair(
     """
     for leaf in node.leaves():
         previous_consumed = 0
-        for comment in list_comments(leaf.prefix, is_endmarker=False):
+        for comment in list_comments(leaf.prefix, is_endmarker=False, mode=mode):
             is_fmt_off = comment.value in FMT_OFF
             is_fmt_skip = _contains_fmt_skip_comment(comment.value, mode)
             if (not is_fmt_off and not is_fmt_skip) or (
@@ -273,13 +285,13 @@ def generate_ignored_nodes(
         return
     container: Optional[LN] = container_of(leaf)
     while container is not None and container.type != token.ENDMARKER:
-        if is_fmt_on(container):
+        if is_fmt_on(container, mode=mode):
             return
 
         # fix for fmt: on in children
-        if children_contains_fmt_on(container):
+        if children_contains_fmt_on(container, mode=mode):
             for index, child in enumerate(container.children):
-                if isinstance(child, Leaf) and is_fmt_on(child):
+                if isinstance(child, Leaf) and is_fmt_on(child, mode=mode):
                     if child.type in CLOSING_BRACKETS:
                         # This means `# fmt: on` is placed at a different bracket level
                         # than `# fmt: off`. This is an invalid use, but as a courtesy,
@@ -290,12 +302,14 @@ def generate_ignored_nodes(
                 if (
                     child.type == token.INDENT
                     and index < len(container.children) - 1
-                    and children_contains_fmt_on(container.children[index + 1])
+                    and children_contains_fmt_on(
+                        container.children[index + 1], mode=mode
+                    )
                 ):
                     # This means `# fmt: on` is placed right after an indentation
                     # level, and we shouldn't swallow the previous INDENT token.
                     return
-                if children_contains_fmt_on(child):
+                if children_contains_fmt_on(child, mode=mode):
                     return
                 yield child
         else:
@@ -316,7 +330,7 @@ def _generate_ignored_nodes_from_fmt_skip(
     ignored_nodes: list[LN] = []
     # Need to properly format the leaf prefix to compare it to comment.value,
     # which is also formatted
-    comments = list_comments(leaf.prefix, is_endmarker=False)
+    comments = list_comments(leaf.prefix, is_endmarker=False, mode=mode)
     if not comments or comment.value != comments[0].value:
         return
     if prev_sibling is not None:
@@ -392,24 +406,24 @@ def _generate_ignored_nodes_from_fmt_skip(
         yield from iter(ignored_nodes)
 
 
-def is_fmt_on(container: LN) -> bool:
+def is_fmt_on(container: LN, mode: Mode) -> bool:
     """Determine whether formatting is switched on within a container.
     Determined by whether the last `# fmt:` comment is `on` or `off`.
     """
     fmt_on = False
-    for comment in list_comments(container.prefix, is_endmarker=False):
+    for comment in list_comments(container.prefix, is_endmarker=False, mode=mode):
         if comment.value in FMT_ON:
             fmt_on = True
         elif comment.value in FMT_OFF:
             fmt_on = False
     return fmt_on
 
 
-def children_contains_fmt_on(container: LN) -> bool:
+def children_contains_fmt_on(container: LN, mode: Mode) -> bool:
     """Determine if children have formatting switched on."""
     for child in container.children:
         leaf = first_leaf_of(child)
-        if leaf is not None and is_fmt_on(leaf):
+        if leaf is not None and is_fmt_on(leaf, mode=mode):
             return True
 
     return False

src/black/linegen.py

@@ -140,7 +140,7 @@ def visit_default(self, node: LN) -> Iterator[Line]:
         """Default `visit_*()` implementation. Recurses to children of `node`."""
         if isinstance(node, Leaf):
             any_open_brackets = self.current_line.bracket_tracker.any_open_brackets()
-            for comment in generate_comments(node):
+            for comment in generate_comments(node, mode=self.mode):
                 if any_open_brackets:
                     # any comment within brackets is subject to splitting
                     self.current_line.append(comment)
@@ -1420,7 +1420,7 @@ def normalize_invisible_parens(  # noqa: C901
     Standardizes on visible parentheses for single-element tuples, and keeps
     existing visible parentheses for other tuples and generator expressions.
     """
-    for pc in list_comments(node.prefix, is_endmarker=False):
+    for pc in list_comments(node.prefix, is_endmarker=False, mode=mode):
         if pc.value in FMT_OFF:
             # This `node` has a prefix with `# fmt: off`, don't mess with parens.
             return
@@ -1748,7 +1748,7 @@ def maybe_make_parens_invisible_in_atom(
         if (
             # If the prefix of `middle` includes a type comment with
             # ignore annotation, then we do not remove the parentheses
-            not is_type_ignore_comment_string(middle.prefix.strip())
+            not is_type_ignore_comment_string(middle.prefix.strip(), mode=mode)
         ):
             first.value = ""
             last.value = ""

src/black/lines.py

@@ -286,9 +286,9 @@ def contains_uncollapsable_type_comments(self) -> bool:
         comment_seen = False
         for leaf_id, comments in self.comments.items():
             for comment in comments:
-                if is_type_comment(comment):
+                if is_type_comment(comment, mode=self.mode):
                     if comment_seen or (
-                        not is_type_ignore_comment(comment)
+                        not is_type_ignore_comment(comment, mode=self.mode)
                         and leaf_id not in ignored_ids
                     ):
                         return True
@@ -325,7 +325,7 @@ def contains_unsplittable_type_ignore(self) -> bool:
             # line.
             for node in self.leaves[-2:]:
                 for comment in self.comments.get(id(node), []):
-                    if is_type_ignore_comment(comment):
+                    if is_type_ignore_comment(comment, mode=self.mode):
                         return True
 
         return False
@@ -400,7 +400,7 @@ def append_comment(self, comment: Leaf) -> bool:
             and not last_leaf.value
             and last_leaf.parent
             and len(list(last_leaf.parent.leaves())) <= 3
-            and not is_type_comment(comment)
+            and not is_type_comment(comment, mode=self.mode)
         ):
             # Comments on an optional parens wrapping a single leaf should belong to
             # the wrapped node except if it's a type comment. Pinning the comment like

src/black/mode.py

@@ -231,6 +231,7 @@ class Preview(Enum):
     multiline_string_handling = auto()
     always_one_newline_after_import = auto()
     fix_fmt_skip_in_one_liners = auto()
+    standardize_type_comments = auto()
     wrap_comprehension_in = auto()
     # Remove parentheses around multiple exception types in except and
     # except* without as. See PEP 758 for details.
@@ -319,3 +320,18 @@ def get_cache_key(self) -> str:
             features_and_magics,
         ]
         return ".".join(parts)
+
+    def __hash__(self) -> int:
+        return hash((
+            frozenset(self.target_versions),
+            self.line_length,
+            self.string_normalization,
+            self.is_pyi,
+            self.is_ipynb,
+            self.skip_source_first_line,
+            self.magic_trailing_comma,
+            frozenset(self.python_cell_magics),
+            self.preview,
+            self.unstable,
+            frozenset(self.enabled_features),
+        ))

src/black/nodes.py

@@ -14,7 +14,7 @@
 from mypy_extensions import mypyc_attr
 
 from black.cache import CACHE_DIR
-from black.mode import Mode
+from black.mode import Mode, Preview
 from black.strings import get_string_prefix, has_triple_quotes
 from blib2to3 import pygram
 from blib2to3.pgen2 import token
@@ -931,27 +931,44 @@ def is_async_stmt_or_funcdef(leaf: Leaf) -> bool:
     )
 
 
-def is_type_comment(leaf: Leaf) -> bool:
+def is_type_comment(leaf: Leaf, mode: Mode) -> bool:
     """Return True if the given leaf is a type comment. This function should only
     be used for general type comments (excluding ignore annotations, which should
     use `is_type_ignore_comment`). Note that general type comments are no longer
     used in modern version of Python, this function may be deprecated in the future."""
     t = leaf.type
     v = leaf.value
-    return t in {token.COMMENT, STANDALONE_COMMENT} and v.startswith("# type:")
+    return t in {token.COMMENT, STANDALONE_COMMENT} and is_type_comment_string(v, mode)
 
 
-def is_type_ignore_comment(leaf: Leaf) -> bool:
+def is_type_comment_string(value: str, mode: Mode) -> bool:
+    if Preview.standardize_type_comments in mode:
+        is_valid = value.startswith("#") and value[1:].lstrip().startswith("type:")
+    else:
+        is_valid = value.startswith("# type:")
+    return is_valid
+
+
+def is_type_ignore_comment(leaf: Leaf, mode: Mode) -> bool:
     """Return True if the given leaf is a type comment with ignore annotation."""
     t = leaf.type
     v = leaf.value
-    return t in {token.COMMENT, STANDALONE_COMMENT} and is_type_ignore_comment_string(v)
+    return t in {token.COMMENT, STANDALONE_COMMENT} and is_type_ignore_comment_string(
+        v, mode
+    )
 
 
-def is_type_ignore_comment_string(value: str) -> bool:
+def is_type_ignore_comment_string(value: str, mode: Mode) -> bool:
     """Return True if the given string match with type comment with
     ignore annotation."""
-    return value.startswith("# type: ignore")
+    if Preview.standardize_type_comments in mode:
+        is_valid = is_type_comment_string(value, mode) and value.split(":", 1)[
+            1
+        ].lstrip().startswith("ignore")
+    else:
+        is_valid = value.startswith("# type: ignore")
+
+    return is_valid
 
 
 def wrap_in_parentheses(parent: Node, child: LN, *, visible: bool = True) -> None:

src/black/resources/black.schema.json

@@ -86,6 +86,7 @@
           "multiline_string_handling",
           "always_one_newline_after_import",
           "fix_fmt_skip_in_one_liners",
+          "standardize_type_comments",
           "wrap_comprehension_in",
           "remove_parens_around_except_types",
           "normalize_cr_newlines",

tests/data/cases/preview_standardize_type_comments.py

@@ -0,0 +1,23 @@
+# flags: --preview
+def foo(
+    a, #type:int
+    b, #type: str
+    c, # type: List[int]
+    d, #  type:  Dict[int, str]
+    e, #       type:     ignore
+    f, # type : ignore
+    g, #  type : ignore
+):
+    pass
+
+# output
+def foo(
+    a,  # type: int
+    b,  # type: str
+    c,  # type: List[int]
+    d,  # type: Dict[int, str]
+    e,  # type: ignore
+    f,  # type : ignore
+    g,  #  type : ignore
+):
+    pass