spb/metrika Public

1//2//  KnownVerbs.swift3//  Metrika4//5//  Author:  Simon-Pierre Boucher6//  Contact: contact@spboucher.ai7//  Copyright © 2026 Simon-Pierre Boucher. All rights reserved.8//910/// Registry of built-in ZQL verbs with their minimum abbreviations,11/// Stata-style: `reg` resolves to `regress`, `su` to `summarize`.12public struct ZQVerbTable: Sendable {13    /// canonical verb → minimum abbreviation length14    public let verbs: [String: Int]15    /// Verbs whose first argument is a raw file path rather than a varlist.16    public let fileVerbs: Set<String>17    /// Verbs taking a `newvar = expression` assignment.18    public let assignmentVerbs: Set<String>19    /// Prefix verbs that wrap another command after a colon.20    public let prefixVerbs: Set<String>21    /// Compound verbs followed by a sub-verb (`graph scatter …`).22    public let compoundVerbs: Set<String>2324    public static let builtin = ZQVerbTable(25        verbs: [26            "use": 3,27            "sysuse": 6,28            "save": 4,29            "import": 3,30            "export": 3,31            "summarize": 2,32            "tabulate": 3,33            "correlate": 3,34            "describe": 1,35            "list": 1,36            "margins": 4,37            "generate": 3,38            "replace": 7,39            "drop": 4,40            "keep": 4,41            "regress": 3,42            "logit": 5,43            "probit": 6,44            "poisson": 7,45            "ivregress": 9,46            "xtreg": 5,47            "xtset": 5,48            "areg": 4,49            "boost": 5,50            "elasticnet": 7,51            "lasso": 5,52            "bayes": 5,53            "bootstrap": 9,54            "permute": 7,55            "jackknife": 9,56            "graph": 2,57            "kdensity": 4,58            "histogram": 4,59            "scatter": 7,60            "display": 2,61            "log": 3,62            "set": 3,63            "clear": 5,64            "count": 3,65            "sort": 4,66            "merge": 5,67            "append": 6,68            "reshape": 7,69            "collapse": 8,70            "egen": 4,71            "predict": 7,72            "test": 4,73            "help": 4,74        ],75        fileVerbs: ["use", "sysuse", "save", "import", "export", "log"],76        assignmentVerbs: ["generate", "replace", "egen"],77        prefixVerbs: ["bayes", "bootstrap", "permute", "jackknife"],78        compoundVerbs: ["graph", "ivregress"]79    )8081    public init(82        verbs: [String: Int],83        fileVerbs: Set<String>,84        assignmentVerbs: Set<String>,85        prefixVerbs: Set<String>,86        compoundVerbs: Set<String>87    ) {88        self.verbs = verbs89        self.fileVerbs = fileVerbs90        self.assignmentVerbs = assignmentVerbs91        self.prefixVerbs = prefixVerbs92        self.compoundVerbs = compoundVerbs93    }9495    /// Resolves a typed verb (possibly abbreviated) to its canonical form.96    /// Returns nil when no verb matches unambiguously.97    public func resolve(_ typed: String) -> String? {98        if let minLength = verbs[typed], typed.count >= minLength { return typed }99        if verbs.keys.contains(typed) { return typed }100101        let candidates = verbs.filter { verb, minLength in102            typed.count >= minLength && verb.hasPrefix(typed)103        }104        return candidates.count == 1 ? candidates.first!.key : nil105    }106107    /// Closest known verb by edit distance, for error suggestions.108    /// Only returns matches within distance 2 to avoid absurd hints.109    public func closest(to typed: String) -> String? {110        let scored = verbs.keys111            .map { (verb: $0, distance: levenshteinDistance(typed, $0)) }112            .sorted { $0.distance < $1.distance }113        guard let best = scored.first, best.distance <= 2 else { return nil }114        return best.verb115    }116}117