import type { CodeChunk } from './types.js';
/**
 * A word/path-boundary regex for `token` — see `CONTINUATION_CHARS` for why
 * this is stricter than plain `\b`. Pass `flags: 'g'` for a reusable
 * global-match regex (`matchAll`).
 */
export declare function wordBoundaryRe(token: string, flags?: string): RegExp;
/**
 * True iff `token` has an internal camelCase/PascalCase case transition or an
 * internal underscore — the shape no ordinary lowercase English word has
 * (`index`, `config`, `platform`), so it CANNOT be mistaken for one no matter
 * how many plain-prose occurrences turn up in the corpus.
 *
 * Deliberately NARROWER than "contains any uppercase letter or underscore
 * anywhere": a bare ALL-CAPS acronym (`API`, `ID`, `URL`, `DB`, `UI`, `HTTP`,
 * `TODO`) is exactly the kind of ordinary, high-frequency word this gate
 * exists to catch — "call the API", "check the ID" are completely mundane
 * prose, and a removed export literally named `ID` or `API` would otherwise
 * false-fire against nearly every doc in a real corpus. A single trailing or
 * leading underscore (`foo_`, `_prefix`) is excluded for the same reason: it
 * doesn't turn an otherwise-ordinary word into an identifier the way an
 * INTERNAL separator does. False-fires are this nudge's worst failure mode
 * (an incorrect "N docs reference X" is actively misleading, whereas a missed
 * reference is merely silent) — this is why the requirement is a genuine
 * shape signal, not merely "has a capital letter somewhere." Exposed for
 * testing.
 */
export declare function isUnambiguousIdentifierShape(token: string): boolean;
/**
 * True iff EVERY word-boundary occurrence of `token` across `docChunks` reads
 * as code/path context — never as ordinary prose describing something
 * unrelated (e.g. a bare directory named `platform` inside "supports every
 * platform", or a bare symbol named `index`/`config` inside "the index is
 * built here"). A single prose hit disqualifies the token: when in doubt,
 * suppress.
 *
 * Skipped entirely for a token with `isUnambiguousIdentifierShape` — a
 * camelCase/PascalCase identifier or an underscored name reads as code no
 * matter what surrounds it in prose, so gating it the same way a bare
 * lowercase word needs to be gated would only produce false suppressions
 * (found dogfooding: this feature's own architecture-doc writeup mentions
 * `createVectorDB` inline, un-backticked, in running prose — a real doc
 * reference that must not be thrown away). The strict corpus-driven check
 * below remains exactly as it was for a bare lowercase word (the review
 * pass's bare-top-level-directory case, `platform`/`runner`, is unaffected
 * by construction: directory names here are always lowercase).
 *
 * Corpus-driven rather than a hardcoded stopword list: a fixed word list
 * needs constant upkeep across languages/domains and still misses whatever
 * wasn't anticipated. Checking what the corpus actually does with the word
 * is self-maintaining. Chunks that don't even contain `token` trivially pass
 * (nothing to disqualify), so callers may pass either the full doc corpus or
 * an already-narrowed "chunks containing this token" set — both produce the
 * same result.
 */
export declare function isDistinctiveToken(token: string, docChunks: CodeChunk[]): boolean;
//# sourceMappingURL=doc-reference-matching.d.ts.map