feat: Generate expr method signatures, docs

tests/expr/test_expr.py

@@ -2,14 +2,20 @@
 
 import operator
 import sys
-from inspect import classify_class_attrs, getmembers
-from typing import Any, Iterator
+from inspect import classify_class_attrs, getmembers, signature
+from typing import TYPE_CHECKING, Any, Callable, Iterable, Iterator, TypeVar, cast
 
 import pytest
 from jsonschema.exceptions import ValidationError
 
 from altair import datum, expr, ExprRef
-from altair.expr import _ConstExpressionType
+from altair.expr import _ExprMeta
+from altair.expr.core import Expression, GetAttrExpression
+
+if TYPE_CHECKING:
+    from inspect import _IntrospectableCallable
+
+T = TypeVar("T")
 
 # This maps vega expression function names to the Python name
 VEGA_REMAP = {"if_": "if"}
@@ -19,20 +25,29 @@ def _is_property(obj: Any, /) -> bool:
     return isinstance(obj, property)
 
 
-def _get_classmethod_names(tp: type[Any], /) -> Iterator[str]:
-    for m in classify_class_attrs(tp):
-        if m.kind == "class method" and m.defining_class is tp:
-            yield m.name
+def _get_property_names(tp: type[Any], /) -> Iterator[str]:
+    for nm, _ in getmembers(tp, _is_property):
+        yield nm
 
 
-def _remap_classmethod_names(tp: type[Any], /) -> Iterator[tuple[str, str]]:
-    for name in _get_classmethod_names(tp):
-        yield VEGA_REMAP.get(name, name), name
+def signature_n_params(
+    obj: _IntrospectableCallable,
+    /,
+    *,
+    exclude: Iterable[str] = frozenset(("cls", "self")),
+) -> int:
+    sig = signature(obj)
+    return len(set(sig.parameters).difference(exclude))
 
 
-def _get_property_names(tp: type[Any], /) -> Iterator[str]:
-    for nm, _ in getmembers(tp, _is_property):
-        yield nm
+def _iter_classmethod_specs(
+    tp: type[T], /
+) -> Iterator[tuple[str, Callable[..., Expression], int]]:
+    for m in classify_class_attrs(tp):
+        if m.kind == "class method" and m.defining_class is tp:
+            name = m.name
+            fn = cast("classmethod[T, ..., Expression]", m.object).__func__
+            yield (VEGA_REMAP.get(name, name), fn.__get__(tp), signature_n_params(fn))
 
 
 def test_unary_operations():
@@ -86,23 +101,26 @@ def test_abs():
     assert repr(z) == "abs(datum.xxx)"
 
 
-@pytest.mark.parametrize(("veganame", "methodname"), _remap_classmethod_names(expr))
-def test_expr_funcs(veganame: str, methodname: str):
-    """Test all functions defined in expr.funcs."""
-    func = getattr(expr, methodname)
-    z = func(datum.xxx)
-    assert repr(z) == f"{veganame}(datum.xxx)"
+@pytest.mark.parametrize(("veganame", "fn", "n_params"), _iter_classmethod_specs(expr))
+def test_expr_methods(
+    veganame: str, fn: Callable[..., Expression], n_params: int
+) -> None:
+    datum_names = [f"col_{n}" for n in range(n_params)]
+    datum_args = ",".join(f"datum.{nm}" for nm in datum_names)
+
+    fn_call = fn(*(GetAttrExpression("datum", nm) for nm in datum_names))
+    assert repr(fn_call) == f"{veganame}({datum_args})"
 
 
-@pytest.mark.parametrize("constname", _get_property_names(_ConstExpressionType))
+@pytest.mark.parametrize("constname", _get_property_names(_ExprMeta))
 def test_expr_consts(constname: str):
     """Test all constants defined in expr.consts."""
     const = getattr(expr, constname)
     z = const * datum.xxx
     assert repr(z) == f"({constname} * datum.xxx)"
 
 
-@pytest.mark.parametrize("constname", _get_property_names(_ConstExpressionType))
+@pytest.mark.parametrize("constname", _get_property_names(_ExprMeta))
 def test_expr_consts_immutable(constname: str):
     """Ensure e.g `alt.expr.PI = 2` is prevented."""
     if sys.version_info >= (3, 11):

tests/vegalite/v5/test_api.py

@@ -557,9 +557,13 @@ def test_when_labels_position_based_on_condition() -> None:
     # `mypy` will flag structural errors here
     cond = when["condition"][0]
     otherwise = when["value"]
-    param_color_py_when = alt.param(
-        expr=alt.expr.if_(cond["test"], cond["value"], otherwise)
-    )
+
+    # TODO: Open an issue on making `OperatorMixin` generic
+    # Something like this would be used as the return type for all `__dunder__` methods:
+    # R = TypeVar("R", Expression, SelectionPredicateComposition)
+    test = cond["test"]
+    assert not isinstance(test, alt.PredicateComposition)
+    param_color_py_when = alt.param(expr=alt.expr.if_(test, cond["value"], otherwise))
     assert param_color_py_expr.expr == param_color_py_when.expr
 
     chart = (
@@ -600,7 +604,9 @@ def test_when_expressions_inside_parameters() -> None:
     cond = when_then_otherwise["condition"][0]
     otherwise = when_then_otherwise["value"]
     expected = alt.expr.if_(alt.datum.b >= 0, 10, -20)
-    actual = alt.expr.if_(cond["test"], cond["value"], otherwise)
+    test = cond["test"]
+    assert not isinstance(test, alt.PredicateComposition)
+    actual = alt.expr.if_(test, cond["value"], otherwise)
     assert expected == actual
 
     text_conditioned = bar.mark_text(

tools/__init__.py

@@ -1,8 +1,15 @@
-from tools import generate_api_docs, generate_schema_wrapper, schemapi, update_init_file
+from tools import (
+    generate_api_docs,
+    generate_schema_wrapper,
+    markup,
+    schemapi,
+    update_init_file,
+)
 
 __all__ = [
     "generate_api_docs",
     "generate_schema_wrapper",
+    "markup",
     "schemapi",
     "update_init_file",
 ]

tools/generate_schema_wrapper.py

@@ -20,13 +20,15 @@
 sys.path.insert(0, str(Path.cwd()))
 
 
+from tools.markup import rst_syntax_for_class
 from tools.schemapi import (  # noqa: F401
     CodeSnippet,
     SchemaInfo,
     arg_invalid_kwds,
     arg_kwds,
     arg_required_kwds,
     codegen,
+    write_expr_module,
 )
 from tools.schemapi.utils import (
     SchemaProperties,
@@ -37,7 +39,6 @@
     import_typing_extensions,
     indent_docstring,
     resolve_references,
-    rst_syntax_for_class,
     ruff_format_py,
     ruff_write_lint_format_str,
     spell_literal,
@@ -61,6 +62,9 @@
 SCHEMA_URL_TEMPLATE: Final = "https://vega.github.io/schema/{library}/{version}.json"
 SCHEMA_FILE = "vega-lite-schema.json"
 THEMES_FILE = "vega-themes.json"
+EXPR_FILE: Path = (
+    Path(__file__).parent / ".." / "altair" / "expr" / "__init__.py"
+).resolve()
 
 CHANNEL_MYPY_IGNORE_STATEMENTS: Final = """\
 # These errors need to be ignored as they come from the overload methods
@@ -1206,6 +1210,7 @@ def main() -> None:
     args = parser.parse_args()
     copy_schemapi_util()
     vegalite_main(args.skip_download)
+    write_expr_module(source_url="static", output=EXPR_FILE)
 
     # The modules below are imported after the generation of the new schema files
     # as these modules import Altair. This allows them to use the new changes

tools/markup.py

@@ -0,0 +1,140 @@
+"""Tools for working with formats like ``.md``, ``.rst``."""
+
+from __future__ import annotations
+
+import re
+from html import unescape
+from typing import TYPE_CHECKING, Any, Iterable, Literal
+
+import mistune.util
+from mistune import InlineParser as _InlineParser
+from mistune import Markdown as _Markdown
+from mistune.renderers.rst import RSTRenderer as _RSTRenderer
+
+if TYPE_CHECKING:
+    import sys
+    from pathlib import Path
+
+    if sys.version_info >= (3, 11):
+        from typing import TypeAlias
+    else:
+        from typing_extensions import TypeAlias
+    from re import Pattern
+
+    from mistune import BaseRenderer, BlockParser, BlockState, InlineState
+
+Token: TypeAlias = "dict[str, Any]"
+
+_RE_LINK: Pattern[str] = re.compile(r"(?<=\[)([^\]]+)(?=\]\([^\)]+\))", re.MULTILINE)
+_RE_SPECIAL: Pattern[str] = re.compile(r"[*_]{2,3}|`", re.MULTILINE)
+_RE_LIQUID_INCLUDE: Pattern[str] = re.compile(r"( \{% include.+%\})")
+
+
+class RSTRenderer(_RSTRenderer):
+    def __init__(self) -> None:
+        super().__init__()
+
+    def inline_html(self, token: Token, state: BlockState) -> str:
+        html = token["raw"]
+        return rf"\ :raw-html:`{html}`\ "
+
+
+class RSTParse(_Markdown):
+    """
+    Minor extension to support partial `ast`_ conversion.
+
+    Only need to convert the docstring tokens to `.rst`.
+
+    .. _ast:
+        https://mistune.lepture.com/en/latest/guide.html#abstract-syntax-tree
+    """
+
+    def __init__(
+        self,
+        renderer: BaseRenderer | Literal["ast"] | None,
+        block: BlockParser | None = None,
+        inline: _InlineParser | None = None,
+        plugins=None,
+    ) -> None:
+        if renderer == "ast":
+            renderer = None
+        super().__init__(renderer, block, inline, plugins)
+
+    def __call__(self, s: str) -> str:
+        s = super().__call__(s)  # pyright: ignore[reportAssignmentType]
+        return unescape(s).replace(r"\ ,", ",").replace(r"\ ", " ")
+
+    def render_tokens(self, tokens: Iterable[Token], /) -> str:
+        """
+        Render ast tokens originating from another parser.
+
+        Parameters
+        ----------
+        tokens
+            All tokens will be rendered into a single `.rst` string
+        """
+        if self.renderer is None:
+            msg = "Unable to render tokens without a renderer."
+            raise TypeError(msg)
+        state = self.block.state_cls()
+        s = self.renderer(self._iter_render(tokens, state), state)
+        return mistune.util.unescape(s)
+
+
+class RSTParseVegaLite(RSTParse):
+    def __init__(
+        self,
+        renderer: RSTRenderer | None = None,
+        block: BlockParser | None = None,
+        inline: _InlineParser | None = None,
+        plugins=None,
+    ) -> None:
+        super().__init__(renderer or RSTRenderer(), block, inline, plugins)
+
+    def __call__(self, s: str) -> str:
+        # remove formatting from links
+        description = "".join(
+            _RE_SPECIAL.sub("", d) if i % 2 else d
+            for i, d in enumerate(_RE_LINK.split(s))
+        )
+
+        description = super().__call__(description)
+        # Some entries in the Vega-Lite schema miss the second occurence of '__'
+        description = description.replace("__Default value: ", "__Default value:__ ")
+        # Links to the vega-lite documentation cannot be relative but instead need to
+        # contain the full URL.
+        description = description.replace(
+            "types#datetime", "https://vega.github.io/vega-lite/docs/datetime.html"
+        )
+        # Fixing ambiguous unicode, RUF001 produces RUF002 in docs
+        description = description.replace("’", "'")  # noqa: RUF001 [RIGHT SINGLE QUOTATION MARK]
+        description = description.replace("–", "-")  # noqa: RUF001 [EN DASH]
+        description = description.replace(" ", " ")  # noqa: RUF001 [NO-BREAK SPACE]
+        return description.strip()
+
+
+class InlineParser(_InlineParser):
+    def __init__(self, hard_wrap: bool = False) -> None:
+        super().__init__(hard_wrap)
+
+    def process_text(self, text: str, state: InlineState) -> None:
+        """
+        Removes `liquid`_ templating markup.
+
+        .. _liquid:
+        https://shopify.github.io/liquid/
+        """
+        state.append_token({"type": "text", "raw": _RE_LIQUID_INCLUDE.sub(r"", text)})
+
+
+def read_ast_tokens(source: Path, /) -> list[Token]:
+    """
+    Read from ``source``, drop ``BlockState``.
+
+    Factored out to provide accurate typing.
+    """
+    return _Markdown(renderer=None, inline=InlineParser()).read(source)[0]
+
+
+def rst_syntax_for_class(class_name: str) -> str:
+    return f":class:`{class_name}`"

tools/schemapi/__init__.py

@@ -9,6 +9,7 @@
 )
 from tools.schemapi.schemapi import SchemaBase, Undefined
 from tools.schemapi.utils import OneOrSeq, SchemaInfo
+from tools.schemapi.vega_expr import write_expr_module
 
 __all__ = [
     "CodeSnippet",
@@ -21,4 +22,5 @@
     "arg_required_kwds",
     "codegen",
     "utils",
+    "write_expr_module",
 ]