merman-editor-core

Protocol-neutral editor intelligence for Merman.

This crate is an internal Rust reuse layer shared by protocol adapters such as merman-lsp and browser adapters such as merman-wasm. External editors should normally integrate through the LSP server rather than depending on this crate directly.

Responsibilities

• Own document snapshots and source/fence lookup through DocumentWorkspace, DocumentSnapshot, and FenceSnapshot.
• Query parser-backed semantic facts for completion, hover, document symbols, workspace symbols, definition, references, prepare-rename, rename, and semantic tokens.
• Preserve semantic fact provenance with FenceTextIndexSource so callers can tell ParserComplete, ParserCompleteDegradedSpans, ParserRecovered, ParserRecoveredDegradedSpans, and TextScan results apart.
• Keep language behavior protocol-neutral: no LSP Url, Range, Diagnostic, or VS Code ownership policy lives here.

ParserCompleteDegradedSpans and ParserRecoveredDegradedSpans remain parser-backed for identity and outline facts, but callers must treat their spans as unavailable for precise source edits when analysis reports source_mapped_spans=false. TextScan is a bounded fallback, not a maturity signal. New editor behavior should deepen parser-backed semantic facts in merman-core / merman-analysis rather than adding protocol-layer scans.