//
//  KnownVerbs.swift
//  Metrika
//
//  Author:  Simon-Pierre Boucher
//  Contact: contact@spboucher.ai
//  Copyright © 2026 Simon-Pierre Boucher. All rights reserved.
//

/// Registry of built-in ZQL verbs with their minimum abbreviations,
/// Stata-style: `reg` resolves to `regress`, `su` to `summarize`.
public struct ZQVerbTable: Sendable {
    /// canonical verb → minimum abbreviation length
    public let verbs: [String: Int]
    /// Verbs whose first argument is a raw file path rather than a varlist.
    public let fileVerbs: Set<String>
    /// Verbs taking a `newvar = expression` assignment.
    public let assignmentVerbs: Set<String>
    /// Prefix verbs that wrap another command after a colon.
    public let prefixVerbs: Set<String>
    /// Compound verbs followed by a sub-verb (`graph scatter …`).
    public let compoundVerbs: Set<String>

    public static let builtin = ZQVerbTable(
        verbs: [
            "use": 3,
            "sysuse": 6,
            "save": 4,
            "import": 3,
            "export": 3,
            "summarize": 2,
            "tabulate": 3,
            "correlate": 3,
            "describe": 1,
            "list": 1,
            "margins": 4,
            "generate": 3,
            "replace": 7,
            "drop": 4,
            "keep": 4,
            "regress": 3,
            "logit": 5,
            "probit": 6,
            "poisson": 7,
            "ivregress": 9,
            "xtreg": 5,
            "xtset": 5,
            "areg": 4,
            "boost": 5,
            "elasticnet": 7,
            "lasso": 5,
            "bayes": 5,
            "bootstrap": 9,
            "permute": 7,
            "jackknife": 9,
            "graph": 2,
            "kdensity": 4,
            "histogram": 4,
            "scatter": 7,
            "display": 2,
            "log": 3,
            "set": 3,
            "clear": 5,
            "count": 3,
            "sort": 4,
            "merge": 5,
            "append": 6,
            "reshape": 7,
            "collapse": 8,
            "egen": 4,
            "predict": 7,
            "test": 4,
            "help": 4,
        ],
        fileVerbs: ["use", "sysuse", "save", "import", "export", "log"],
        assignmentVerbs: ["generate", "replace", "egen"],
        prefixVerbs: ["bayes", "bootstrap", "permute", "jackknife"],
        compoundVerbs: ["graph", "ivregress"]
    )

    public init(
        verbs: [String: Int],
        fileVerbs: Set<String>,
        assignmentVerbs: Set<String>,
        prefixVerbs: Set<String>,
        compoundVerbs: Set<String>
    ) {
        self.verbs = verbs
        self.fileVerbs = fileVerbs
        self.assignmentVerbs = assignmentVerbs
        self.prefixVerbs = prefixVerbs
        self.compoundVerbs = compoundVerbs
    }

    /// Resolves a typed verb (possibly abbreviated) to its canonical form.
    /// Returns nil when no verb matches unambiguously.
    public func resolve(_ typed: String) -> String? {
        if let minLength = verbs[typed], typed.count >= minLength { return typed }
        if verbs.keys.contains(typed) { return typed }

        let candidates = verbs.filter { verb, minLength in
            typed.count >= minLength && verb.hasPrefix(typed)
        }
        return candidates.count == 1 ? candidates.first!.key : nil
    }

    /// Closest known verb by edit distance, for error suggestions.
    /// Only returns matches within distance 2 to avoid absurd hints.
    public func closest(to typed: String) -> String? {
        let scored = verbs.keys
            .map { (verb: $0, distance: levenshteinDistance(typed, $0)) }
            .sorted { $0.distance < $1.distance }
        guard let best = scored.first, best.distance <= 2 else { return nil }
        return best.verb
    }
}