Full Hoogle-style type search #23

New issue

Open

opened 2026-01-05 21:17:05 +00:00 by mprihoda · 0 comments

mprihoda commented

2026-01-05 21:17:05 +00:00

(Migrated from github.com)

Problem

Simplified type search (#17) handles basic cases, but advanced queries need:

Polymorphic matching: `(A => B) => List[A] => List[B]` finds `map`
Fuzzy matching: `F[A] => A` finds `unsafeRunSync`, `getOrElse`, etc.
Curried function support
Higher-kinded type matching

Proposed Solution

Full Hoogle-style search engine for Scala/Java.

Enhanced `search_by_type`

{
  "name": "search_by_type",
  "description": "Find methods by type signature (advanced)",
  "input": {
    "signature": "Type signature query",
    "coordinates": "optional, limit to specific library",
    "projectPath": "optional, search project deps"
  }
}

Example queries:

search_by_type("(A => B) => List[A] => List[B]")
→ List.map, Vector.map, Array.map, ...

search_by_type("List[A] => NonEmptyList[A]")
→ NonEmptyList.fromList (returns Option), NonEmptyList.fromListUnsafe

search_by_type("F[Either[E, A]] => EitherT[F, E, A]")
→ EitherT.apply, EitherT.fromEither

Technical Approach

Type Signature Language

signature := type '=>' signature | type
type := name '[' type (',' type)* ']' | name | typevar
typevar := uppercase letter (A, B, F, etc.)

Matching Algorithm

Parse query into type AST
Match against method signatures from TASTy index
Unify type variables
Rank by specificity and popularity

Index Structure

Per-artifact index of all method signatures
Normalized for matching (strip package prefixes)
Precomputed for common libraries

Complexity

This is a significant engineering effort:

Type unification algorithm
Index storage and querying
Cross-artifact search
Performance optimization

Consider: Start with #17 (simplified) and evolve based on usage patterns.

Value for Agent

"How do I transform X to Y?" answered for any types
Discovers non-obvious conversions
Finds correct method even with complex signatures

Acceptance Criteria

Parse type signature queries with type variables
Match polymorphic signatures
Handle higher-kinded types (F[_])
Rank results by relevance
Sub-second query time for indexed libraries

## Problem Simplified type search (#17) handles basic cases, but advanced queries need: - Polymorphic matching: \`(A => B) => List[A] => List[B]\` finds \`map\` - Fuzzy matching: \`F[A] => A\` finds \`unsafeRunSync\`, \`getOrElse\`, etc. - Curried function support - Higher-kinded type matching ## Proposed Solution Full Hoogle-style search engine for Scala/Java. ### Enhanced \`search_by_type\` ```json { "name": "search_by_type", "description": "Find methods by type signature (advanced)", "input": { "signature": "Type signature query", "coordinates": "optional, limit to specific library", "projectPath": "optional, search project deps" } } ``` **Example queries:** ``` search_by_type("(A => B) => List[A] => List[B]") → List.map, Vector.map, Array.map, ... search_by_type("List[A] => NonEmptyList[A]") → NonEmptyList.fromList (returns Option), NonEmptyList.fromListUnsafe search_by_type("F[Either[E, A]] => EitherT[F, E, A]") → EitherT.apply, EitherT.fromEither ``` ## Technical Approach ### Type Signature Language ``` signature := type '=>' signature | type type := name '[' type (',' type)* ']' | name | typevar typevar := uppercase letter (A, B, F, etc.) ``` ### Matching Algorithm 1. Parse query into type AST 2. Match against method signatures from TASTy index 3. Unify type variables 4. Rank by specificity and popularity ### Index Structure - Per-artifact index of all method signatures - Normalized for matching (strip package prefixes) - Precomputed for common libraries ## Complexity This is a significant engineering effort: - Type unification algorithm - Index storage and querying - Cross-artifact search - Performance optimization Consider: Start with #17 (simplified) and evolve based on usage patterns. ## Value for Agent - "How do I transform X to Y?" answered for any types - Discovers non-obvious conversions - Finds correct method even with complex signatures ## Acceptance Criteria - [ ] Parse type signature queries with type variables - [ ] Match polymorphic signatures - [ ] Handle higher-kinded types (F[_]) - [ ] Rank results by relevance - [ ] Sub-second query time for indexed libraries