@mailwoman/core 2.0.0

package/out/classification/BaseClassifier.d.ts

@@ -0,0 +1,38 @@
+/**
+ * @copyright Sister Software
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @license AGPL-3.0
+ * @abstract
+ * @author Teffen Ellis, et al.
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode } from "@mailwoman/core";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+export interface Classifier {
+    /**
+     * Explore a span, classifying its children.
+     *
+     * API consumers, implement this method to classify the children of a span.
+     *
+     * @see {@link classify} for public classification.
+     */
+    explore?(span: Span, ...args: unknown[]): void;
+    /**
+     * Optional async method to prepare the classifier before classification.
+     */
+    ready?(): Promise<this>;
+    /**
+     * Perform classification on the given context.
+     */
+    classifyTokens(context: TokenContext): void;
+    /**
+     * Perform classification on the given input.
+     */
+    classify(input: Span | string): Span;
+}
+export interface ClassifierOptions {
+    languages?: Alpha2LanguageCode[] | readonly Alpha2LanguageCode[] | Set<Alpha2LanguageCode> | ReadonlySet<Alpha2LanguageCode>;
+}
+export type ClassifierConstructor = new (options?: ClassifierOptions) => Classifier;
+//# sourceMappingURL=BaseClassifier.d.ts.map

package/out/classification/BaseClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"BaseClassifier.d.ts","sourceRoot":"","sources":["../../classification/BaseClassifier.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAA;AACpD,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAEjE,MAAM,WAAW,UAAU;IAC1B;;;;;;OAMG;IACH,OAAO,CAAC,CAAC,IAAI,EAAE,IAAI,EAAE,GAAG,IAAI,EAAE,OAAO,EAAE,GAAG,IAAI,CAAA;IAE9C;;OAEG;IACH,KAAK,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,CAAA;IAEvB;;OAEG;IACH,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI,CAAA;IAE3C;;OAEG;IACH,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI,CAAA;CACpC;AAED,MAAM,WAAW,iBAAiB;IACjC,SAAS,CAAC,EACP,kBAAkB,EAAE,GACpB,SAAS,kBAAkB,EAAE,GAC7B,GAAG,CAAC,kBAAkB,CAAC,GACvB,WAAW,CAAC,kBAAkB,CAAC,CAAA;CAClC;AAED,MAAM,MAAM,qBAAqB,GAAG,KAAK,OAAO,CAAC,EAAE,iBAAiB,KAAK,UAAU,CAAA"}

package/out/classification/BaseClassifier.js

@@ -0,0 +1,12 @@
+/**
+ * @copyright Sister Software
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @license AGPL-3.0
+ * @abstract
+ * @author Teffen Ellis, et al.
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode } from "@mailwoman/core";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+//# sourceMappingURL=BaseClassifier.js.map

package/out/classification/BaseClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"BaseClassifier.js","sourceRoot":"","sources":["../../classification/BaseClassifier.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAA;AACpD,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA"}

package/out/classification/Classification.d.ts

@@ -0,0 +1,79 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { Displayable } from "../resources/debugging.js";
+import type { Alpha3bLanguageCode } from "../resources/languages/index.js";
+import type { LibPostalLanguageCode } from "../resources/libpostal.js";
+/**
+ * Classification recognized by Mailwoman.
+ */
+export declare const Classifications: Set<"adjacent" | "alpha" | "alphanumeric" | "area" | "chain" | "country" | "dependency" | "directional" | "end_token_single_character" | "end_token" | "given_name" | "house_number" | "intersection" | "level_designator" | "level" | "locality" | "middle_initial" | "multistreet" | "numeric" | "ordinal" | "person" | "personal_suffix" | "personal_title" | "place" | "postcode" | "punctuation" | "region" | "road_type" | "start_token" | "stop_word" | "street_name" | "street_prefix" | "street_proper_name" | "street_suffix" | "street" | "surname" | "toponym" | "unit_designator" | "unit" | "unknown" | "venue">;
+export type Classification = typeof Classifications extends Set<infer T> ? T : never;
+/**
+ * Public classification labels.
+ */
+declare const VisibleClassification: Set<"country" | "dependency" | "house_number" | "level_designator" | "level" | "locality" | "postcode" | "region" | "street" | "unit_designator" | "unit" | "venue">;
+/**
+ * Classification labels that are exposed to API consumers.
+ */
+export type VisibleClassification = typeof VisibleClassification extends Set<infer T> ? T : never;
+/**
+ * Type guard for visible classification labels.
+ */
+export declare function isVisibleClassification(match: ClassificationMatch): match is ClassificationMatch & {
+    classification: VisibleClassification;
+};
+export declare function isVisibleClassification(classification: Classification): classification is VisibleClassification;
+export type ClassificationMap = Map<VisibleClassification, string[]>;
+export type ClassificationConfidenceMap = Map<VisibleClassification, number>;
+export type ClassificationConfidenceRecord = Record<VisibleClassification, number | undefined>;
+export type ClassificationRecord = Partial<Record<VisibleClassification, string[]>>;
+/**
+ * A classification match for a span of text.
+ */
+export interface ClassificationMatch {
+    /**
+     * The classification assigned to the text.
+     */
+    readonly classification: Classification;
+    /**
+     * The confidence of the classification.
+     */
+    readonly confidence: number;
+    /**
+     * Languages for which the classification is valid.
+     *
+     * If a language is present in this set, it means that the value which it is associated with is a
+     * valid value in that language.
+     */
+    readonly languages?: Displayable<ReadonlySet<LibPostalLanguageCode | Alpha3bLanguageCode>>;
+    /**
+     * Flags associated with the classification, i.e. clarifying details to enhance the
+     * classification.
+     */
+    readonly flags?: Set<string>;
+}
+export type ClassificationsMatchInput = ClassificationMatch | Classification;
+/**
+ * A map of classifications to matches.
+ */
+export declare class ClassificationsMatchMap extends Map<Classification, ClassificationMatch> {
+    /**
+     * Add a classification to the map if it has a higher confidence than the existing classification.
+     */
+    add(classification: Classification, confidence?: number): void;
+    add(classificationMatch: Partial<ClassificationMatch>): void;
+    /**
+     * Predicate to determine if the map contains a classification that is visible.
+     */
+    hasVisibleClassification(classification?: VisibleClassification): boolean;
+    /**
+     * Serialize the classification map to JSON.
+     */
+    toJSON(): ClassificationMatch[];
+    static from(...input: ClassificationsMatchInput[]): ClassificationsMatchMap;
+}
+export {};
+//# sourceMappingURL=Classification.d.ts.map

package/out/classification/Classification.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"Classification.d.ts","sourceRoot":"","sources":["../../classification/Classification.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AAC5D,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AAC1E,OAAO,KAAK,EAAE,qBAAqB,EAAE,MAAM,2BAA2B,CAAA;AAEtE;;GAEG;AACH,eAAO,MAAM,eAAe,gmBA0CjB,CAAA;AAEX,MAAM,MAAM,cAAc,GAAG,OAAO,eAAe,SAAS,GAAG,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,GAAG,KAAK,CAAA;AAIpF;;GAEG;AACH,QAAA,MAAM,qBAAqB,sKAaW,CAAA;AAEtC;;GAEG;AACH,MAAM,MAAM,qBAAqB,GAAG,OAAO,qBAAqB,SAAS,GAAG,CAAC,MAAM,CAAC,CAAC,GAAG,CAAC,GAAG,KAAK,CAAA;AAEjG;;GAEG;AACH,wBAAgB,uBAAuB,CACtC,KAAK,EAAE,mBAAmB,GACxB,KAAK,IAAI,mBAAmB,GAAG;IAAE,cAAc,EAAE,qBAAqB,CAAA;CAAE,CAAA;AAC3E,wBAAgB,uBAAuB,CAAC,cAAc,EAAE,cAAc,GAAG,cAAc,IAAI,qBAAqB,CAAA;AAOhH,MAAM,MAAM,iBAAiB,GAAG,GAAG,CAAC,qBAAqB,EAAE,MAAM,EAAE,CAAC,CAAA;AACpE,MAAM,MAAM,2BAA2B,GAAG,GAAG,CAAC,qBAAqB,EAAE,MAAM,CAAC,CAAA;AAC5E,MAAM,MAAM,8BAA8B,GAAG,MAAM,CAAC,qBAAqB,EAAE,MAAM,GAAG,SAAS,CAAC,CAAA;AAC9F,MAAM,MAAM,oBAAoB,GAAG,OAAO,CAAC,MAAM,CAAC,qBAAqB,EAAE,MAAM,EAAE,CAAC,CAAC,CAAA;AAEnF;;GAEG;AACH,MAAM,WAAW,mBAAmB;IACnC;;OAEG;IACH,QAAQ,CAAC,cAAc,EAAE,cAAc,CAAA;IAEvC;;OAEG;IACH,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAA;IAE3B;;;;;OAKG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,WAAW,CAAC,WAAW,CAAC,qBAAqB,GAAG,mBAAmB,CAAC,CAAC,CAAA;IAE1F;;;OAGG;IACH,QAAQ,CAAC,KAAK,CAAC,EAAE,GAAG,CAAC,MAAM,CAAC,CAAA;CAC5B;AAED,MAAM,MAAM,yBAAyB,GAAG,mBAAmB,GAAG,cAAc,CAAA;AAE5E;;GAEG;AACH,qBAAa,uBAAwB,SAAQ,GAAG,CAAC,cAAc,EAAE,mBAAmB,CAAC;IACpF;;OAEG;IACI,GAAG,CAAC,cAAc,EAAE,cAAc,EAAE,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI;IAC9D,GAAG,CAAC,mBAAmB,EAAE,OAAO,CAAC,mBAAmB,CAAC,GAAG,IAAI;IA+CnE;;OAEG;IACI,wBAAwB,CAAC,cAAc,CAAC,EAAE,qBAAqB,GAAG,OAAO;IAIhF;;OAEG;IACI,MAAM,IAAI,mBAAmB,EAAE;IAItC,MAAM,CAAC,IAAI,CAAC,GAAG,KAAK,EAAE,yBAAyB,EAAE,GAAG,uBAAuB;CAS3E"}

package/out/classification/Classification.js

@@ -0,0 +1,133 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+/**
+ * Classification recognized by Mailwoman.
+ */
+export const Classifications = new Set([
+    "adjacent",
+    "alpha",
+    "alphanumeric",
+    "area",
+    "chain",
+    "country",
+    "dependency",
+    "directional",
+    "end_token_single_character",
+    "end_token",
+    "given_name",
+    "house_number",
+    "intersection",
+    "level_designator",
+    "level",
+    "locality",
+    "middle_initial",
+    "multistreet",
+    "numeric",
+    "ordinal",
+    "person",
+    "personal_suffix",
+    "personal_title",
+    "place",
+    "postcode",
+    "punctuation",
+    "region",
+    "road_type",
+    "start_token",
+    "stop_word",
+    "street_name",
+    "street_prefix",
+    "street_proper_name",
+    "street_suffix",
+    "street",
+    "surname",
+    "toponym",
+    "unit_designator",
+    "unit",
+    "unknown",
+    "venue",
+]);
+const ClassificationConfidenceBias = new Map([["country", 0.9]]);
+/**
+ * Public classification labels.
+ */
+const VisibleClassification = new Set([
+    "country",
+    "dependency",
+    "house_number",
+    "level_designator",
+    "level",
+    "locality",
+    "postcode",
+    "region",
+    "street",
+    "unit_designator",
+    "unit",
+    "venue",
+]);
+export function isVisibleClassification(input) {
+    const classification = typeof input === "string" ? input : input.classification;
+    return VisibleClassification.has(classification);
+}
+/**
+ * A map of classifications to matches.
+ */
+export class ClassificationsMatchMap extends Map {
+    add(...args) {
+        let match;
+        if (typeof args[0] === "string") {
+            const classification = args[0];
+            const confidence = typeof args[1] === "number" ? args[1] : (ClassificationConfidenceBias.get(classification) ?? 1);
+            match = {
+                classification,
+                confidence,
+            };
+        }
+        else {
+            match = {
+                ...args[0],
+                confidence: args[0].confidence ?? ClassificationConfidenceBias.get(args[0].classification) ?? 1,
+            };
+        }
+        if (!match.classification) {
+            throw new TypeError("Classification must be a non-empty string");
+        }
+        if (typeof match.confidence !== "number") {
+            throw new TypeError("Confidence must be a number");
+        }
+        if (match.confidence < 0) {
+            throw new RangeError(`${match.classification}: Confidence must be greater than or equal to 0: ${match.confidence}`);
+        }
+        if (match.confidence > 1) {
+            throw new RangeError(`${match.classification}: Confidence must be less than or equal to 1: ${match.confidence}`);
+        }
+        const existing = this.get(match.classification);
+        // Ensure that duplicate classifications do not reduce confidence
+        if (existing && existing.confidence >= match.confidence) {
+            return;
+        }
+        this.set(match.classification, match);
+    }
+    /**
+     * Predicate to determine if the map contains a classification that is visible.
+     */
+    hasVisibleClassification(classification) {
+        return classification ? this.has(classification) : Iterator.from(this.keys()).some(isVisibleClassification);
+    }
+    /**
+     * Serialize the classification map to JSON.
+     */
+    toJSON() {
+        return this.values().toArray();
+    }
+    static from(...input) {
+        const map = new ClassificationsMatchMap();
+        for (const entry of input) {
+            map.add(entry);
+        }
+        return map;
+    }
+}
+//# sourceMappingURL=Classification.js.map

package/out/classification/Classification.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"Classification.js","sourceRoot":"","sources":["../../classification/Classification.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAMH;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC;IACtC,UAAU;IACV,OAAO;IACP,cAAc;IACd,MAAM;IACN,OAAO;IACP,SAAS;IACT,YAAY;IACZ,aAAa;IACb,4BAA4B;IAC5B,WAAW;IACX,YAAY;IACZ,cAAc;IACd,cAAc;IACd,kBAAkB;IAClB,OAAO;IACP,UAAU;IACV,gBAAgB;IAChB,aAAa;IACb,SAAS;IACT,SAAS;IACT,QAAQ;IACR,iBAAiB;IACjB,gBAAgB;IAChB,OAAO;IACP,UAAU;IACV,aAAa;IACb,QAAQ;IACR,WAAW;IACX,aAAa;IACb,WAAW;IACX,aAAa;IACb,eAAe;IACf,oBAAoB;IACpB,eAAe;IACf,QAAQ;IACR,SAAS;IACT,SAAS;IACT,iBAAiB;IACjB,MAAM;IACN,SAAS;IACT,OAAO;CACE,CAAC,CAAA;AAIX,MAAM,4BAA4B,GAAG,IAAI,GAAG,CAAyB,CAAC,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC,CAAC,CAAA;AAExF;;GAEG;AACH,MAAM,qBAAqB,GAAG,IAAI,GAAG,CAAC;IACrC,SAAS;IACT,YAAY;IACZ,cAAc;IACd,kBAAkB;IAClB,OAAO;IACP,UAAU;IACV,UAAU;IACV,QAAQ;IACR,QAAQ;IACR,iBAAiB;IACjB,MAAM;IACN,OAAO;CAC6B,CAAC,CAAA;AActC,MAAM,UAAU,uBAAuB,CAAC,KAA2C;IAClF,MAAM,cAAc,GAAG,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,cAAc,CAAA;IAE/E,OAAO,qBAAqB,CAAC,GAAG,CAAC,cAAuC,CAAC,CAAA;AAC1E,CAAC;AAsCD;;GAEG;AACH,MAAM,OAAO,uBAAwB,SAAQ,GAAwC;IAM7E,GAAG,CAAC,GAAG,IAAgE;QAC7E,IAAI,KAAmC,CAAA;QAEvC,IAAI,OAAO,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,EAAE,CAAC;YACjC,MAAM,cAAc,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;YAC9B,MAAM,UAAU,GAAG,OAAO,IAAI,CAAC,CAAC,CAAC,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,4BAA4B,CAAC,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC,CAAA;YAElH,KAAK,GAAG;gBACP,cAAc;gBACd,UAAU;aACV,CAAA;QACF,CAAC;aAAM,CAAC;YACP,KAAK,GAAG;gBACP,GAAG,IAAI,CAAC,CAAC,CAAC;gBACV,UAAU,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,IAAI,4BAA4B,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,cAAe,CAAC,IAAI,CAAC;aAChG,CAAA;QACF,CAAC;QAED,IAAI,CAAC,KAAK,CAAC,cAAc,EAAE,CAAC;YAC3B,MAAM,IAAI,SAAS,CAAC,2CAA2C,CAAC,CAAA;QACjE,CAAC;QAED,IAAI,OAAO,KAAK,CAAC,UAAU,KAAK,QAAQ,EAAE,CAAC;YAC1C,MAAM,IAAI,SAAS,CAAC,6BAA6B,CAAC,CAAA;QACnD,CAAC;QAED,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;YAC1B,MAAM,IAAI,UAAU,CACnB,GAAG,KAAK,CAAC,cAAc,oDAAoD,KAAK,CAAC,UAAU,EAAE,CAC7F,CAAA;QACF,CAAC;QAED,IAAI,KAAK,CAAC,UAAU,GAAG,CAAC,EAAE,CAAC;YAC1B,MAAM,IAAI,UAAU,CAAC,GAAG,KAAK,CAAC,cAAc,iDAAiD,KAAK,CAAC,UAAU,EAAE,CAAC,CAAA;QACjH,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,cAAc,CAAC,CAAA;QAE/C,iEAAiE;QACjE,IAAI,QAAQ,IAAI,QAAQ,CAAC,UAAU,IAAI,KAAK,CAAC,UAAU,EAAE,CAAC;YACzD,OAAM;QACP,CAAC;QAED,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,cAAc,EAAE,KAA4B,CAAC,CAAA;IAC7D,CAAC;IAED;;OAEG;IACI,wBAAwB,CAAC,cAAsC;QACrE,OAAO,cAAc,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAA;IAC5G,CAAC;IAED;;OAEG;IACI,MAAM;QACZ,OAAO,IAAI,CAAC,MAAM,EAAE,CAAC,OAAO,EAAE,CAAA;IAC/B,CAAC;IAED,MAAM,CAAC,IAAI,CAAC,GAAG,KAAkC;QAChD,MAAM,GAAG,GAAG,IAAI,uBAAuB,EAAE,CAAA;QAEzC,KAAK,MAAM,KAAK,IAAI,KAAK,EAAE,CAAC;YAC3B,GAAG,CAAC,GAAG,CAAC,KAA4B,CAAC,CAAA;QACtC,CAAC;QAED,OAAO,GAAG,CAAA;IACX,CAAC;CACD"}

package/out/classification/CompositeClassifier.d.ts

@@ -0,0 +1,16 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span } from "@mailwoman/core/tokenization";
+import type { Classification } from "./Classification.js";
+import { SectionClassifier } from "./SectionClassifier.js";
+import { type ClassifierSchemeConfig } from "./scheme.js";
+export declare abstract class CompositeClassifier extends SectionClassifier {
+    schemes: ClassifierSchemeConfig[];
+    readonly primaryClassification: Classification;
+    constructor(primaryClassification: Classification, schemes: Iterable<ClassifierSchemeConfig>);
+    explore(section: Span): void;
+}
+//# sourceMappingURL=CompositeClassifier.d.ts.map

package/out/classification/CompositeClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompositeClassifier.d.ts","sourceRoot":"","sources":["../../classification/CompositeClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,8BAA8B,CAAA;AACnD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACzD,OAAO,EAAE,iBAAiB,EAAE,MAAM,wBAAwB,CAAA;AAC1D,OAAO,EAAE,KAAK,sBAAsB,EAAuB,MAAM,aAAa,CAAA;AAuB9E,8BAAsB,mBAAoB,SAAQ,iBAAiB;IAC3D,OAAO,EAAE,sBAAsB,EAAE,CAAA;IACxC,SAAgB,qBAAqB,EAAE,cAAc,CAAA;gBAEzC,qBAAqB,EAAE,cAAc,EAAE,OAAO,EAAE,QAAQ,CAAC,sBAAsB,CAAC;IAMrF,OAAO,CAAC,OAAO,EAAE,IAAI,GAAG,IAAI;CAkGnC"}

package/out/classification/CompositeClassifier.js

@@ -0,0 +1,113 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span } from "@mailwoman/core/tokenization";
+import { SectionClassifier } from "./SectionClassifier.js";
+import { phraseMatchesScheme } from "./scheme.js";
+/**
+ * Compute the cartesian product of two or more arrays.
+ *
+ * A cartesian product is the set of all possible combinations of the elements.
+ */
+function cartesian(...args) {
+    return args.reduce((acc, val) => {
+        const res = [];
+        acc.forEach((a) => {
+            val.forEach((b) => {
+                res.push(a.concat([b]));
+            });
+        });
+        return res;
+    }, [[]]);
+}
+export class CompositeClassifier extends SectionClassifier {
+    schemes;
+    primaryClassification;
+    constructor(primaryClassification, schemes) {
+        super();
+        this.primaryClassification = primaryClassification;
+        this.schemes = Array.from(schemes);
+    }
+    explore(section) {
+        const { phrases } = section;
+        // Sort phrases so shorter phrases are matched first.
+        section.phrases.sort((a, b) => a.normalized.length - b.normalized.length);
+        for (const config of this.schemes) {
+            const candidates = config.scheme.map((criteria) => {
+                return Iterator.from(phrases)
+                    .filter((phrase) => phraseMatchesScheme(criteria, phrase))
+                    .toArray();
+            });
+            // no candidates were found for one or more schemes
+            if (candidates.some((c) => c.length === 0)) {
+                continue;
+            }
+            // compute composites (each with candidates of the same length as s.scheme)
+            const compositeSpans = cartesian(...candidates);
+            // remove any overlapping composites
+            const adjacentComposites = compositeSpans.filter((composite) => {
+                for (let i = 0; i < composite.length; i++) {
+                    const current = composite[i];
+                    const firstChild = current.children.first;
+                    const lastChild = current.children.last;
+                    const next = composite[i + 1];
+                    const prev = composite[i - 1];
+                    if (!next && !prev)
+                        continue;
+                    // Enforce adjacency.
+                    if (next && lastChild && !Iterator.from(lastChild.nextSiblings).some((s) => s === next.children.first)) {
+                        return false;
+                    }
+                    if (prev && firstChild && !Iterator.from(firstChild.previousSiblings).some((s) => s === prev.children.last)) {
+                        return false;
+                    }
+                    // Avoid adding tokens to the front of a street classification
+                    // that begins with a street prefix.
+                    // e.g. 'A + Ave B' (ave is both a valid prefix & suffix)
+                    if (next?.is("street") && next.children.first?.is("street_prefix")) {
+                        return false;
+                    }
+                }
+                return true;
+            });
+            if (!adjacentComposites.length)
+                continue;
+            // find phrases which equal the composites
+            let superPhrases = [];
+            adjacentComposites.forEach((composites) => {
+                const start = composites[0].start;
+                const end = composites[composites.length - 1].end;
+                superPhrases = superPhrases.concat(Iterator.from(phrases)
+                    .filter((p) => p.start === start && p.end === end)
+                    .toArray());
+            });
+            // classify each super phrase
+            superPhrases.forEach((superPhrase) => {
+                const classification = config.classification ?? this.primaryClassification;
+                superPhrase.classifications.add({
+                    classification,
+                    confidence: config.confidence,
+                    languages: superPhrase.languages,
+                });
+            });
+            // Optionally classify individual phrases
+            for (const spans of adjacentComposites) {
+                for (const [i, childScheme] of config.scheme.entries()) {
+                    if (!childScheme.classification)
+                        continue;
+                    const span = spans[i];
+                    if (!span)
+                        continue;
+                    span.classifications.add({
+                        classification: childScheme.classification,
+                        confidence: childScheme.confidence,
+                        languages: span.languages,
+                    });
+                }
+            }
+        }
+    }
+}
+//# sourceMappingURL=CompositeClassifier.js.map

package/out/classification/CompositeClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompositeClassifier.js","sourceRoot":"","sources":["../../classification/CompositeClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,8BAA8B,CAAA;AAEnD,OAAO,EAAE,iBAAiB,EAAE,MAAM,wBAAwB,CAAA;AAC1D,OAAO,EAA+B,mBAAmB,EAAE,MAAM,aAAa,CAAA;AAE9E;;;;GAIG;AACH,SAAS,SAAS,CAAI,GAAG,IAAW;IACnC,OAAO,IAAI,CAAC,MAAM,CACjB,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE;QACZ,MAAM,GAAG,GAAU,EAAE,CAAA;QAErB,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE;YACjB,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE;gBACjB,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YACxB,CAAC,CAAC,CAAA;QACH,CAAC,CAAC,CAAA;QACF,OAAO,GAAG,CAAA;IACX,CAAC,EACD,CAAC,EAAE,CAAU,CACb,CAAA;AACF,CAAC;AAED,MAAM,OAAgB,mBAAoB,SAAQ,iBAAiB;IAC3D,OAAO,CAA0B;IACxB,qBAAqB,CAAgB;IAErD,YAAY,qBAAqC,EAAE,OAAyC;QAC3F,KAAK,EAAE,CAAA;QACP,IAAI,CAAC,qBAAqB,GAAG,qBAAqB,CAAA;QAClD,IAAI,CAAC,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IACnC,CAAC;IAEM,OAAO,CAAC,OAAa;QAC3B,MAAM,EAAE,OAAO,EAAE,GAAG,OAAO,CAAA;QAE3B,qDAAqD;QACrD,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;QAEzE,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YACnC,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE;gBACjD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC;qBAC3B,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,mBAAmB,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;qBACzD,OAAO,EAAE,CAAA;YACZ,CAAC,CAAC,CAAA;YAEF,mDAAmD;YACnD,IAAI,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,CAAC,CAAC,EAAE,CAAC;gBAC5C,SAAQ;YACT,CAAC;YAED,2EAA2E;YAC3E,MAAM,cAAc,GAAsB,SAAS,CAAC,GAAG,UAAU,CAAC,CAAA;YAElE,oCAAoC;YACpC,MAAM,kBAAkB,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC,SAAS,EAAE,EAAE;gBAC9D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;oBAC3C,MAAM,OAAO,GAAG,SAAS,CAAC,CAAC,CAAE,CAAA;oBAC7B,MAAM,UAAU,GAAG,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAA;oBACzC,MAAM,SAAS,GAAG,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAA;oBAEvC,MAAM,IAAI,GAAG,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;oBAC7B,MAAM,IAAI,GAAG,SAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;oBAE7B,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI;wBAAE,SAAQ;oBAE5B,qBAAqB;oBACrB,IAAI,IAAI,IAAI,SAAS,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;wBACxG,OAAO,KAAK,CAAA;oBACb,CAAC;oBAED,IAAI,IAAI,IAAI,UAAU,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;wBAC7G,OAAO,KAAK,CAAA;oBACb,CAAC;oBAED,8DAA8D;oBAC9D,oCAAoC;oBACpC,yDAAyD;oBACzD,IAAI,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,EAAE,CAAC,eAAe,CAAC,EAAE,CAAC;wBACpE,OAAO,KAAK,CAAA;oBACb,CAAC;gBACF,CAAC;gBAED,OAAO,IAAI,CAAA;YACZ,CAAC,CAAC,CAAA;YAEF,IAAI,CAAC,kBAAkB,CAAC,MAAM;gBAAE,SAAQ;YAExC,0CAA0C;YAC1C,IAAI,YAAY,GAAW,EAAE,CAAA;YAE7B,kBAAkB,CAAC,OAAO,CAAC,CAAC,UAAU,EAAE,EAAE;gBACzC,MAAM,KAAK,GAAG,UAAU,CAAC,CAAC,CAAE,CAAC,KAAK,CAAA;gBAClC,MAAM,GAAG,GAAG,UAAU,CAAC,UAAU,CAAC,MAAM,GAAG,CAAC,CAAE,CAAC,GAAG,CAAA;gBAElD,YAAY,GAAG,YAAY,CAAC,MAAM,CACjC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC;qBACpB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,KAAK,KAAK,IAAI,CAAC,CAAC,GAAG,KAAK,GAAG,CAAC;qBACjD,OAAO,EAAE,CACX,CAAA;YACF,CAAC,CAAC,CAAA;YAEF,6BAA6B;YAC7B,YAAY,CAAC,OAAO,CAAC,CAAC,WAAW,EAAE,EAAE;gBACpC,MAAM,cAAc,GAAG,MAAM,CAAC,cAAc,IAAI,IAAI,CAAC,qBAAqB,CAAA;gBAE1E,WAAW,CAAC,eAAe,CAAC,GAAG,CAAC;oBAC/B,cAAc;oBACd,UAAU,EAAE,MAAM,CAAC,UAAU;oBAC7B,SAAS,EAAE,WAAW,CAAC,SAAS;iBAChC,CAAC,CAAA;YACH,CAAC,CAAC,CAAA;YAEF,yCAAyC;YACzC,KAAK,MAAM,KAAK,IAAI,kBAAkB,EAAE,CAAC;gBACxC,KAAK,MAAM,CAAC,CAAC,EAAE,WAAW,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,OAAO,EAAE,EAAE,CAAC;oBACxD,IAAI,CAAC,WAAW,CAAC,cAAc;wBAAE,SAAQ;oBAEzC,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAA;oBAErB,IAAI,CAAC,IAAI;wBAAE,SAAQ;oBAEnB,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;wBACxB,cAAc,EAAE,WAAW,CAAC,cAAc;wBAC1C,UAAU,EAAE,WAAW,CAAC,UAAU;wBAClC,SAAS,EAAE,IAAI,CAAC,SAAS;qBACzB,CAAC,CAAA;gBACH,CAAC;YACF,CAAC;QACF,CAAC;IACF,CAAC;CACD"}

package/out/classification/PhraseClassifier.d.ts

@@ -0,0 +1,23 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { LibPostalLanguageCode, LocaleIndex } from "@mailwoman/core/resources";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+import type { Classifier, ClassifierOptions } from "./BaseClassifier.js";
+export declare abstract class PhraseClassifier implements Classifier {
+    index: LocaleIndex<LibPostalLanguageCode>;
+    protected languages?: LibPostalLanguageCode[];
+    constructor({ languages }?: ClassifierOptions);
+    /**
+     * Perform classification on the given span.
+     */
+    abstract explore(span: Span, sectionIndex?: number, phraseIndex?: number): void;
+    /**
+     * Perform classification on the given context.
+     */
+    classifyTokens(context: TokenContext): void;
+    classify(input: Span | string): Span;
+}
+//# sourceMappingURL=PhraseClassifier.d.ts.map

package/out/classification/PhraseClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"PhraseClassifier.d.ts","sourceRoot":"","sources":["../../classification/PhraseClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,qBAAqB,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACnF,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AACjE,OAAO,KAAK,EAAE,UAAU,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAA;AAExE,8BAAsB,gBAAiB,YAAW,UAAU;IACpD,KAAK,EAAG,WAAW,CAAC,qBAAqB,CAAC,CAAA;IACjD,SAAS,CAAC,SAAS,CAAC,EAAE,qBAAqB,EAAE,CAAA;gBAEjC,EAAE,SAAS,EAAE,GAAE,iBAAsB;IAIjD;;OAEG;IACH,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,EAAE,YAAY,CAAC,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,MAAM,GAAG,IAAI;IAE/E;;OAEG;IACI,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI;IAc3C,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI;CAO3C"}

package/out/classification/PhraseClassifier.js

@@ -0,0 +1,32 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+export class PhraseClassifier {
+    index;
+    languages;
+    constructor({ languages } = {}) {
+        this.languages = languages ? Array.from(languages) : undefined;
+    }
+    /**
+     * Perform classification on the given context.
+     */
+    classifyTokens(context) {
+        for (const [sectionIndex, section] of context.sections.entries()) {
+            const { phrases } = section;
+            let phraseIndex = 0;
+            for (const phrase of phrases) {
+                this.explore(phrase, sectionIndex, phraseIndex);
+                phraseIndex++;
+            }
+        }
+    }
+    classify(input) {
+        const span = Span.from(input);
+        this.explore(span);
+        return span;
+    }
+}
+//# sourceMappingURL=PhraseClassifier.js.map

package/out/classification/PhraseClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"PhraseClassifier.js","sourceRoot":"","sources":["../../classification/PhraseClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAGH,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAGjE,MAAM,OAAgB,gBAAgB;IAC9B,KAAK,CAAqC;IACvC,SAAS,CAA0B;IAE7C,YAAY,EAAE,SAAS,KAAwB,EAAE;QAChD,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAC/D,CAAC;IAOD;;OAEG;IACI,cAAc,CAAC,OAAqB;QAC1C,KAAK,MAAM,CAAC,YAAY,EAAE,OAAO,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;YAClE,MAAM,EAAE,OAAO,EAAE,GAAG,OAAO,CAAA;YAE3B,IAAI,WAAW,GAAG,CAAC,CAAA;YAEnB,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;gBAC9B,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,YAAY,EAAE,WAAW,CAAC,CAAA;gBAE/C,WAAW,EAAE,CAAA;YACd,CAAC;QACF,CAAC;IACF,CAAC;IAEM,QAAQ,CAAC,KAAoB;QACnC,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAE7B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;QAElB,OAAO,IAAI,CAAA;IACZ,CAAC;CACD"}

package/out/classification/SectionClassifier.d.ts

@@ -0,0 +1,23 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+import type { Classifier } from "./BaseClassifier.js";
+export interface SectionClassifierUtils {
+    findPhrasesContaining: (child: Span) => Span[];
+}
+export declare abstract class SectionClassifier implements Classifier {
+    /**
+     * Explore a span, classifying its children.
+     */
+    abstract explore(span: Span): void;
+    classifyTokens(context: TokenContext): void;
+    classify(input: Span | string): Span;
+    /**
+     * Find all phrases containing a child span.
+     */
+    static findPhrasesContaining(section: Span, child: Span): IteratorObject<Span, undefined, unknown>;
+}
+//# sourceMappingURL=SectionClassifier.d.ts.map

package/out/classification/SectionClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"SectionClassifier.d.ts","sourceRoot":"","sources":["../../classification/SectionClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AACjE,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAErD,MAAM,WAAW,sBAAsB;IACtC,qBAAqB,EAAE,CAAC,KAAK,EAAE,IAAI,KAAK,IAAI,EAAE,CAAA;CAC9C;AAED,8BAAsB,iBAAkB,YAAW,UAAU;IAC5D;;OAEG;IACH,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;IAElC,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI;IAM3C,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI;IAQpC;;OAEG;IACH,MAAM,CAAC,qBAAqB,CAAC,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI;CAKvD"}

package/out/classification/SectionClassifier.js

@@ -0,0 +1,27 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+export class SectionClassifier {
+    classifyTokens(context) {
+        for (const section of context.sections) {
+            this.explore(section);
+        }
+    }
+    classify(input) {
+        const span = Span.from(input);
+        this.explore(span);
+        return span;
+    }
+    /**
+     * Find all phrases containing a child span.
+     */
+    static findPhrasesContaining(section, child) {
+        return Iterator.from(section.phrases).filter((phrase) => {
+            return Iterator.from(phrase.children).some((phraseChild) => phraseChild === child);
+        });
+    }
+}
+//# sourceMappingURL=SectionClassifier.js.map

package/out/classification/SectionClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"SectionClassifier.js","sourceRoot":"","sources":["../../classification/SectionClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAOjE,MAAM,OAAgB,iBAAiB;IAMtC,cAAc,CAAC,OAAqB;QACnC,KAAK,MAAM,OAAO,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;YACxC,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAA;QACtB,CAAC;IACF,CAAC;IAED,QAAQ,CAAC,KAAoB;QAC5B,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAE7B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;QAElB,OAAO,IAAI,CAAA;IACZ,CAAC;IAED;;OAEG;IACH,MAAM,CAAC,qBAAqB,CAAC,OAAa,EAAE,KAAW;QACtD,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE;YACvD,OAAO,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC,WAAW,KAAK,KAAK,CAAC,CAAA;QACnF,CAAC,CAAC,CAAA;IACH,CAAC;CACD"}

package/out/classification/WordClassifier.d.ts

@@ -0,0 +1,22 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { LibPostalLanguageCode, LocaleIndex } from "@mailwoman/core/resources";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+import type { Classifier, ClassifierOptions } from "./BaseClassifier.js";
+export declare abstract class WordClassifier implements Classifier {
+    /**
+     * A mapping of words to their originating languages.
+     *
+     * Each key is a word, and each value is a set of ISO 639-1 language codes.
+     */
+    index: LocaleIndex<LibPostalLanguageCode>;
+    protected languages?: LibPostalLanguageCode[];
+    constructor({ languages }?: ClassifierOptions);
+    abstract explore(span: Span, sectionIndex?: number, childIndex?: number): void;
+    classifyTokens(context: TokenContext): void;
+    classify(input: Span | string): Span;
+}
+//# sourceMappingURL=WordClassifier.d.ts.map

package/out/classification/WordClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"WordClassifier.d.ts","sourceRoot":"","sources":["../../classification/WordClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,qBAAqB,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACnF,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AACjE,OAAO,KAAK,EAAE,UAAU,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAA;AAExE,8BAAsB,cAAe,YAAW,UAAU;IACzD;;;;OAIG;IACI,KAAK,EAAG,WAAW,CAAC,qBAAqB,CAAC,CAAA;IACjD,SAAS,CAAC,SAAS,CAAC,EAAE,qBAAqB,EAAE,CAAA;gBAEjC,EAAE,SAAS,EAAE,GAAE,iBAAsB;IAIjD,QAAQ,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,EAAE,YAAY,CAAC,EAAE,MAAM,EAAE,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI;IAEvE,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI;IAY3C,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI;CAO3C"}

package/out/classification/WordClassifier.js

@@ -0,0 +1,33 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+export class WordClassifier {
+    /**
+     * A mapping of words to their originating languages.
+     *
+     * Each key is a word, and each value is a set of ISO 639-1 language codes.
+     */
+    index;
+    languages;
+    constructor({ languages } = {}) {
+        this.languages = languages ? Array.from(languages) : undefined;
+    }
+    classifyTokens(context) {
+        for (const [sectionIndex, section] of context.sections.entries()) {
+            let childIndex = 0;
+            for (const child of section.children) {
+                this.explore(child, sectionIndex, childIndex);
+                childIndex++;
+            }
+        }
+    }
+    classify(input) {
+        const span = Span.from(input);
+        this.explore(span);
+        return span;
+    }
+}
+//# sourceMappingURL=WordClassifier.js.map

package/out/classification/WordClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"WordClassifier.js","sourceRoot":"","sources":["../../classification/WordClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAGH,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AAGjE,MAAM,OAAgB,cAAc;IACnC;;;;OAIG;IACI,KAAK,CAAqC;IACvC,SAAS,CAA0B;IAE7C,YAAY,EAAE,SAAS,KAAwB,EAAE;QAChD,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAC/D,CAAC;IAIM,cAAc,CAAC,OAAqB;QAC1C,KAAK,MAAM,CAAC,YAAY,EAAE,OAAO,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,OAAO,EAAE,EAAE,CAAC;YAClE,IAAI,UAAU,GAAG,CAAC,CAAA;YAElB,KAAK,MAAM,KAAK,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;gBACtC,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,YAAY,EAAE,UAAU,CAAC,CAAA;gBAE7C,UAAU,EAAE,CAAA;YACb,CAAC;QACF,CAAC;IACF,CAAC;IAEM,QAAQ,CAAC,KAAoB;QACnC,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAE7B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;QAElB,OAAO,IAAI,CAAA;IACZ,CAAC;CACD"}

package/out/classification/index.d.ts

@@ -0,0 +1,13 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+export * from "./BaseClassifier.js";
+export * from "./Classification.js";
+export * from "./CompositeClassifier.js";
+export * from "./PhraseClassifier.js";
+export * from "./scheme.js";
+export * from "./SectionClassifier.js";
+export * from "./WordClassifier.js";
+//# sourceMappingURL=index.d.ts.map

package/out/classification/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../classification/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,qBAAqB,CAAA;AACnC,cAAc,qBAAqB,CAAA;AACnC,cAAc,0BAA0B,CAAA;AACxC,cAAc,uBAAuB,CAAA;AACrC,cAAc,aAAa,CAAA;AAC3B,cAAc,wBAAwB,CAAA;AACtC,cAAc,qBAAqB,CAAA"}