Skip to main content

Module query_shape

Module query_shape 

Source

Structs§

QueryShape
ShapeWeights

Enums§

QueryKind

Functions§

classify
extract_lexical_tokens
Tokens for a lexical second chance after an exact or semantic lane misses. Natural-language and regex shapes normally have no extract_tokens output, but quoted code phrases still carry useful identifiers. Reuse the same quote parser used for semantic name priors so a query such as "outside <touser>" reminder text can rank files containing either quoted token.
extract_short_nl_lexical_tokens
Lexical tokens for a short natural-language concept routed to Hybrid (e.g. “parse imports”). extract_tokens returns nothing for NL (its words are not code identifiers), but a short two-word concept is frequently a literal code phrase the trigram lane can match. Split on whitespace and keep words of at least 3 chars (the trigram floor).
extract_tokens
looks_like_regex
pre_tier_exempt