Fix bool narrowing for numeric literal patterns by hsusul · Pull Request #11570 · microsoft/pyright

hsusul · 2026-07-28T06:15:40Z

Summary

Preserve Literal[False] and Literal[True] when matching the numeric literal patterns 0 and 1.
Remove those bool literals from the fallthrough branch because numeric literal patterns compare with equality.
Keep singleton bool patterns identity-based, so case False does not match Literal[0].

Problem

Numeric literal patterns use equality at runtime. This means case 0 matches both 0 and False, and case 1 matches both 1 and True.

from typing import Literal

def f(value: Literal[0, False]) -> None:
    match value:
        case 0:
            reveal_type(value)  # Literal[0, False]
        case _:
            reveal_type(value)  # Never

Pyright previously narrowed the matched branch to Literal[0] and the fallthrough branch to Literal[False], which is unsound relative to runtime behavior.

Root cause

Literal-pattern narrowing relied on assignability between the pattern literal and subject literal. That does not capture Python's equality relationship between 0/False and 1/True.

Solution

Handle the asymmetric numeric-pattern equality case explicitly:

an int literal pattern 0 or 1 retains the equal bool literal in the positive branch;
the same bool literal is removed from the negative branch;
bool singleton patterns remain unchanged because they use identity rather than equality.

Tests

Added focused coverage for:

case 0 over Literal[0, False];
case 1 over Literal[1, True];
case False retaining the identity-based distinction from Literal[0].

Validation

npm run typecheck
npm run build:cli:dev
npm run check
npm run build in packages/pyright-internal
npx jest -t 'MatchLiteral1' --forceExit --runInBand
npx jest src/tests/typeEvaluator6.test.ts --runInBand
npm test -- --runInBand (62 suites, 2549 tests)
CLI repro on Python 3.10 and 3.13

Compatibility and risk

The change is limited to literal-pattern narrowing for int patterns against bool literal subjects. Tests cover both positive and negative branches and the inverse singleton-bool boundary.

Documentation

No documentation changes are needed; this aligns analyzer behavior with Python's existing structural pattern-matching semantics.

Related issues and prior work

#11026 and #11028 addressed broader literal-pattern narrowing involving non-literal and supertype subjects. This change covers the distinct bool/int literal equality edge case.

rchiodo · 2026-07-28T11:48:51Z

🔒 Automated review in progress — @rchiodo is auto-reviewing this PR.

rchiodo

Approved via Review Center.

github-actions · 2026-07-28T16:20:41Z

Diff from mypy_primer, showing the effect of this PR on open source code:

sympy (https://github.com/sympy/sympy)
+   .../projects/sympy/sympy/solvers/ode/subscheck.py:150:48 - error: Argument of type "int | Unknown | str" cannot be assigned to parameter "order" of type "str" in function "checkodesol"
+     Type "int | Unknown | str" is not assignable to type "str"
+       "int" is not assignable to "str" (reportArgumentType)
-   .../projects/sympy/sympy/solvers/ode/systems.py:1167:37 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1167:37 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "Literal[2]" and "Basic*"
-   .../projects/sympy/sympy/solvers/ode/systems.py:1167:37 - error: Operator "*" not supported for types "Unknown | int | Expr" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1167:37 - error: Operator "*" not supported for types "Unknown | Expr | int" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "int" and "Basic*"
-     Operator "*" not supported for types "int" and "Basic"
-     Operator "*" not supported for types "Expr" and "Basic*"
+     Operator "*" not supported for types "Expr" and "Basic"
+     Operator "*" not supported for types "int" and "Basic" (reportOperatorIssue)
-   .../projects/sympy/sympy/solvers/ode/systems.py:1169:30 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1169:30 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "Literal[2]" and "Basic*"
-   .../projects/sympy/sympy/solvers/ode/systems.py:1169:30 - error: Operator "*" not supported for types "Unknown | int | Expr" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1169:30 - error: Operator "*" not supported for types "Unknown | Expr | int" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "int" and "Basic*"
-     Operator "*" not supported for types "int" and "Basic"
-     Operator "*" not supported for types "Expr" and "Basic*"
+     Operator "*" not supported for types "Expr" and "Basic"
+     Operator "*" not supported for types "int" and "Basic" (reportOperatorIssue)
-   .../projects/sympy/sympy/solvers/ode/systems.py:1169:49 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1169:49 - error: Operator "*" not supported for types "Literal[2]" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "Literal[2]" and "Basic*"
-   .../projects/sympy/sympy/solvers/ode/systems.py:1169:49 - error: Operator "*" not supported for types "Unknown | int | Expr" and "Unknown | Self@Basic | bool | Expr | Basic"
+   .../projects/sympy/sympy/solvers/ode/systems.py:1169:49 - error: Operator "*" not supported for types "Unknown | Expr | int" and "Unknown | Expr | Basic | bool"
-     Operator "*" not supported for types "int" and "Basic*"
-     Operator "*" not supported for types "int" and "Basic"
-     Operator "*" not supported for types "Expr" and "Basic*"
+     Operator "*" not supported for types "Expr" and "Basic"
+     Operator "*" not supported for types "int" and "Basic" (reportOperatorIssue)
+   .../projects/sympy/sympy/stats/crv.py:369:24 - error: Operator "*" not supported for types "Expr" and "Unknown | Basic"
+   .../projects/sympy/sympy/stats/crv.py:379:25 - error: Operator "*" not supported for types "Expr" and "Unknown | Basic"
+   .../projects/sympy/sympy/stats/crv.py:392:29 - error: Operator "-" not supported for types "Unknown | Basic" and "Dummy"
+     Operator "-" not supported for types "Basic" and "Dummy" (reportOperatorIssue)
+   .../projects/sympy/sympy/stats/crv_types.py:1289:16 - error: Operator "-" not supported for types "Unknown | Basic | Any" and "Expr"
+     Operator "-" not supported for types "Basic" and "Expr" (reportOperatorIssue)
+   .../projects/sympy/sympy/stats/crv_types.py:3358:23 - error: Operator "*" not supported for types "Expr" and "Unknown | Basic | Any"
+   .../projects/sympy/sympy/stats/drv.py:266:18 - error: No overloads for "sum" match the provided arguments (reportCallIssue)
+   .../projects/sympy/sympy/stats/drv.py:266:22 - error: Argument of type "Generator[Unknown | Basic, None, None]" cannot be assigned to parameter "iterable" of type "Iterable[_SupportsSumNoDefaultT@sum]" in function "sum"
+     "Generator[Unknown | Basic, None, None]" is not assignable to "Iterable[_SupportsSumNoDefaultT@sum]"
+       Type parameter "_T_co@Iterable" is covariant, but "Unknown | Basic" is not a subtype of "_SupportsSumNoDefaultT@sum"
+         Type "Unknown | Basic" is not assignable to type "_SupportsSumWithNoDefaultGiven"
+           Type "Unknown | Basic" is not assignable to type "_SupportsSumWithNoDefaultGiven"
+             "Basic" is incompatible with protocol "_SupportsSumWithNoDefaultGiven" (reportArgumentType)
-   .../projects/sympy/sympy/stats/drv.py:269:22 - error: Argument of type "Generator[tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic | int | None, None, None]" cannot be assigned to parameter "iterable" of type "Iterable[_SupportsSumNoDefaultT@sum]" in function "sum"
+   .../projects/sympy/sympy/stats/drv.py:269:22 - error: Argument of type "Generator[tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic | int | None, None, None]" cannot be assigned to parameter "iterable" of type "Iterable[_SupportsSumNoDefaultT@sum]" in function "sum"
-     "Generator[tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic | int | None, None, None]" is not assignable to "Iterable[_SupportsSumNoDefaultT@sum]"
+     "Generator[tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic | int | None, None, None]" is not assignable to "Iterable[_SupportsSumNoDefaultT@sum]"
-       Type parameter "_T_co@Iterable" is covariant, but "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic | int | None" is not a subtype of "_SupportsSumNoDefaultT@sum"
+       Type parameter "_T_co@Iterable" is covariant, but "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic | int | None" is not a subtype of "_SupportsSumNoDefaultT@sum"
-         Type "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic | int | None" is not assignable to type "_SupportsSumWithNoDefaultGiven"
+         Type "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic | int | None" is not assignable to type "_SupportsSumWithNoDefaultGiven"
-           Type "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic | int | None" is not assignable to type "_SupportsSumWithNoDefaultGiven"
+           Type "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic | int | None" is not assignable to type "_SupportsSumWithNoDefaultGiven"
-   .../projects/sympy/sympy/stats/drv_types.py:293:16 - error: Operator "*" not supported for types "Expr" and "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Add | Zero | NaN | Piecewise | Basic"
+   .../projects/sympy/sympy/stats/drv_types.py:293:16 - error: Operator "*" not supported for types "Expr" and "tuple[Unknown, ...] | Unknown | Sum | Expr | ZeroMatrix | Zero | NaN | Piecewise | Basic"
-   .../projects/sympy/sympy/stats/frv.py:172:24 - error: Cannot access attribute "lhs" for class "Basic*"
+   .../projects/sympy/sympy/stats/frv.py:172:24 - error: Cannot access attribute "lhs" for class "Basic"
-   .../projects/sympy/sympy/stats/frv.py:172:35 - error: Cannot access attribute "rhs" for class "Basic*"
+   .../projects/sympy/sympy/stats/frv.py:172:35 - error: Cannot access attribute "rhs" for class "Basic"
-     Return type mismatch: base method returns type "bool", override returns type "Unknown | Basic* | Literal[False]"
+     Return type mismatch: base method returns type "bool", override returns type "Unknown | Basic | Literal[False]"
-       Type "Unknown | Basic* | Literal[False]" is not assignable to type "bool"
+       Type "Unknown | Basic | Literal[False]" is not assignable to type "bool"
-         "Basic*" is not assignable to "bool" (reportIncompatibleMethodOverride)
+         "Basic" is not assignable to "bool" (reportIncompatibleMethodOverride)
+   .../projects/sympy/sympy/stats/frv.py:401:34 - error: Operator "*" not supported for types "Unknown | Basic | Literal[0]" and "Expr"
+     Operator "*" not supported for types "Basic" and "Expr" (reportOperatorIssue)
+   .../projects/sympy/sympy/stats/frv.py:411:34 - error: Operator "*" not supported for types "Unknown | Basic | Literal[0]" and "Expr"
+     Operator "*" not supported for types "Basic" and "Expr" (reportOperatorIssue)
-   .../projects/sympy/sympy/stats/joint_rv_types.py:576:27 - error: Argument of type "One | NegativeOne | Zero | Integer | NaN | ComplexInfinity | Rational | Unknown | Number | Expr" cannot be assigned to parameter "stop" of type "SupportsIndex" in function "__new__"
+   .../projects/sympy/sympy/stats/joint_rv_types.py:576:27 - error: Argument of type "One | NegativeOne | Zero | Integer | NaN | ComplexInfinity | Rational | Unknown | Infinity | NegativeInfinity | Float | Number | Expr" cannot be assigned to parameter "stop" of type "SupportsIndex" in function "__new__"
-     Type "One | NegativeOne | Zero | Integer | NaN | ComplexInfinity | Rational | Unknown | Number | Expr" is not assignable to type "SupportsIndex"

... (truncated 527 lines) ...

rchiodo · 2026-07-28T17:06:16Z

The sympy mypy_primer diffs are not caused by this PR — they're pre-existing pyright non-determinism

I built pyright at this PR's base (729aabfa3, == current main) and at the PR head (37d3fc744) and ran both on the full sympy source in an identical environment.

The two builds produce identical diagnostics on sympy:

Base vs PR: 0 added, 0 removed (75,789 diagnostics each), comparing full multi-line messages.
Determinism check (same binary run twice): also 0 diff.

So on a deterministic machine the change has zero effect on sympy — a real, PR-caused diff would have reproduced.

Why that's expected from the code: the new isIntLiteralPatternEqualToBool path only activates when a numeric literal pattern 0/1 is matched against a bool-literal subject (Literal[True]/Literal[False]) in a match/case. In all of sympy:

The only case 0:/case 1: is trigonometric.py's match n % 4:, whose subject is int (not a bool literal) → the new path never triggers.
None of the flagged files (subscheck.py, systems.py, crv.py, crv_types.py, drv.py, drv_types.py, frv.py, joint_rv_types.py) contain any match/case.

What the primer diff actually is: analysis-order / recursion-resolution non-determinism over sympy's recursive Basic/Expr/Self hierarchy. The signatures are the usual tells — none of which a literal-pattern-narrowing change can produce:

Basic* ↔ Basic (the * marks a TypeVar/Self-solved type)
Self@Basic dropping in/out of unions
union member reordering (e.g. Unknown | Self@Basic | bool | Expr | Basic → Unknown | Expr | Basic | bool)
Add / Infinity | NegativeInfinity | Float appearing/disappearing in large unions

Bottom line: expected primer noise, not a new bug.

rchiodo · 2026-07-28T17:07:01Z

Thanks for the PR

Fix bool narrowing for numeric literal patterns

37d3fc7

rchiodo approved these changes Jul 28, 2026

View reviewed changes

rchiodo added the review-auto:approved Automated review: no blocking findings (approval posted). label Jul 28, 2026

rchiodo merged commit 559a737 into microsoft:main Jul 28, 2026
16 checks passed

Navigation Menu

Uh oh!

Fix bool narrowing for numeric literal patterns - #11570

Fix bool narrowing for numeric literal patterns#11570
rchiodo merged 1 commit into
microsoft:mainfrom
hsusul:fix/match-numeric-literal-bool-narrowing

hsusul commented Jul 28, 2026

Uh oh!

rchiodo commented Jul 28, 2026

Uh oh!

rchiodo left a comment

Uh oh!

github-actions Bot commented Jul 28, 2026

Uh oh!

rchiodo commented Jul 28, 2026

Uh oh!

Uh oh!

rchiodo commented Jul 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

hsusul commented Jul 28, 2026

Summary

Problem

Root cause

Solution

Tests

Validation

Compatibility and risk

Documentation

Related issues and prior work

Uh oh!

rchiodo commented Jul 28, 2026

Uh oh!

rchiodo left a comment

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jul 28, 2026

Uh oh!

rchiodo commented Jul 28, 2026

The sympy mypy_primer diffs are not caused by this PR — they're pre-existing pyright non-determinism

Uh oh!

Uh oh!

rchiodo commented Jul 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants