Skip to content

no-dynamic-string-replacement: bare-identifier replacement isn't resolved to its literal initializer (FP on send_otlp_span.cjs R #75595

Closed

Description

@github-actions

Summary

no-dynamic-string-replacement's isDynamic() helper (eslint-factory/src/rules/no-dynamic-string-replacement.ts:137-164) treats every bare Identifier replacement argument as dynamic:

case "Identifier":
  return expression.name !== "undefined";

It never resolves the identifier back to its declaration the way the rule's own sibling helpers already do. isKnownFunction (lines 81-135) and isDefinitelyNonFunction (lines 166-197) both call resolveVariable/getVariableInitializer to trace a single-definition const binding back to its initializer before deciding — isDynamic has no equivalent step for its Identifier case, so a const bound to a plain string literal is indistinguishable from a genuinely runtime-computed value.

Live grounding (1 site, actions/setup/js/send_otlp_span.cjs)

// send_otlp_span.cjs:1062
const REDACTED = "[REDACTED]";
...
// send_otlp_span.cjs:1141
.replace((bhttps/redacted)?:\/\/[^\s"'`<>]+/gi, REDACTED)

REDACTED is a single-definition module-scope const whose initializer is the literal "[REDACTED]" (no $ characters at all, so there is zero actual replacement-pattern risk). The pattern argument is a regex literal (isRegexLiteral is true), so the call passes the rule's method/argument-count/pattern gates, then isDynamic(replacement) returns true purely because REDACTED is an Identifier, not "undefined" — a false positive on a provably-safe, non-varying constant.

Because isDefinitelyNonFunction does resolve the same identifier back to a Literal initializer and returns true, and hasStaticReplacementToken has no Identifier case (defaults to false), the rule also offers a "wrap in a replacer function" suggestion here — busywork that doesn't change behavior, since the value is a fixed, $-free string.

Suggested fix

Give isDynamic's Identifier branch the same one-hop (or resolving-guarded recursive) initializer resolution already used by isKnownFunction/isDefinitelyNonFunction: if the identifier resolves to a single-definition const whose initializer is itself non-dynamic per isDynamic (e.g. a Literal/static TemplateLiteral), treat it as non-dynamic instead of unconditionally dynamic.

Acceptance criteria

  • const REDACTED = "[REDACTED]"; s.replace(/x/g, REDACTED); is not flagged.
  • Reassigned bindings remain flagged: let value = "ok"; value = input; s.replace(/x/g, value); (existing invalid test case at no-dynamic-string-replacement.test.ts:64-67) must still fire.
  • A const whose initializer does contain $ (e.g. const R = "$1";) still fires, since the risk is in the literal's content, not just its mutability.
  • A const initialized from a non-literal expression (e.g. const x = foo();) remains flagged as dynamic.
  • New test cases added to no-dynamic-string-replacement.test.ts covering the traceable-literal-constant shape (valid) alongside the existing reassigned-binding shape (invalid).

Generated by 🤖 ESLint Refiner · claude · agent · ~122 AIC · ⌖ ~1 AIC · ⊞ 4.2K · ◷
Claim finished

  • expires on Nov 13, 2026, 7:34 PM UTC-08:00

Activity

  1. github-actions commented on Nov 7, 2026

    @github-actions
    ContributorAuthor

    🍪 Issue Monster selected this for Copilot

    I've identified this issue as a good candidate for automated resolution and requested assignment to the Copilot coding agent.

    If assignment succeeds, Copilot will analyze the issue and create a pull request with the fix.

    Om nom nom! 🍪

    🍪 Om nom nom by Issue Monster · pi · gpt50 · ~1 AIC · ⌖ ~1 AIC · ⊞ 12.5K · ◷

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Labels

cookieIssue Monster Loves Cookies!eslint

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions