@mailwoman/classifiers 2.0.0

package/out/WhosOnFirstClassifier.js

@@ -0,0 +1,148 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { PhraseClassifier, Span, TextNormalizer, WOFPlacenameCache, } from "@mailwoman/core";
+import { resourceDictionaryPathBuilder } from "@mailwoman/core/utils";
+const wofPlacetypeDictionary = resourceDictionaryPathBuilder("whosonfirst");
+const wofInternalPlacetypeDictionary = resourceDictionaryPathBuilder("internal", "whosonfirst");
+const WOFNormalizer = new TextNormalizer({
+    lowercase: true,
+    removeHyphen: true,
+    removeAccents: true,
+    minLength: 2,
+});
+// Note: These should be defined from most granular to least granular.
+const placetypeConfigMap = new Map([
+    [
+        "locality",
+        {
+            files: ["name:*_x_preferred.txt"],
+            classifications: ["area", "locality"],
+        },
+    ],
+    [
+        "region",
+        {
+            files: ["abrv:*_x_preferred.txt", "name:*_x_preferred.txt"],
+            classifications: ["area", "region"],
+        },
+    ],
+    [
+        "country",
+        {
+            files: ["name:*_x_preferred.txt", "wof:country.txt", "wof:country_alpha3.txt"],
+            classifications: ["area", "country"],
+        },
+    ],
+]);
+const tokenBlacklist = new Set([
+    // Cardinal directions
+    "north",
+    "south",
+    "east",
+    "west",
+    // Generic placetype names
+    "town",
+    "street",
+    "city",
+    "king",
+    // Stop words
+    "at",
+    "rue",
+    // Ordinal numerics
+    "one",
+    "two",
+    "three",
+    "four",
+    "five",
+    "six",
+    "seven",
+    "eight",
+    "nine",
+    "ten",
+    "cafe",
+    "small",
+    "grand",
+]);
+const localityBlacklist = new Set([
+    // ---
+    "avenue",
+    "lane",
+    "terrace",
+    "street",
+    "road",
+    "crescent",
+    "furlong",
+    "broadway",
+]);
+export class WhosOnFirstClassifier extends PhraseClassifier {
+    placetypeToCacheMap = new Map();
+    async ready() {
+        for (const [placetype, config] of placetypeConfigMap) {
+            const placenameCache = new WOFPlacenameCache({
+                patterns: config.files,
+                normalizer: WOFNormalizer,
+                blacklist: tokenBlacklist,
+                dataDirectory: wofPlacetypeDictionary(placetype),
+                internalDataDirectory: wofInternalPlacetypeDictionary(placetype),
+            });
+            await placenameCache.ready();
+            this.placetypeToCacheMap.set(placetype, placenameCache);
+            // Placetype specific modifications
+            if (placetype === "locality") {
+                const localityTokens = this.placetypeToCacheMap.get("locality");
+                // Remove locality names that sound like streets.
+                for (const [token] of localityTokens) {
+                    const split = token.split(/\s/);
+                    const lastWord = split[split.length - 1];
+                    if (lastWord && localityBlacklist.has(lastWord)) {
+                        localityTokens.delete(token);
+                    }
+                }
+            }
+        }
+        return this;
+    }
+    explore(span) {
+        if (span.is("stop_word") || span.children.first?.is("stop_word")) {
+            return;
+        }
+        const lastChild = span.children.last || span;
+        const { nextSibling } = lastChild;
+        if (nextSibling && (nextSibling.is("street_suffix") || nextSibling.is("place"))) {
+            return;
+        }
+        const firstChild = span.children.first || span;
+        const { previousSibling } = firstChild;
+        let confidence;
+        if (previousSibling) {
+            if (previousSibling.is("intersection"))
+                return;
+            if (previousSibling.is("stop_word")) {
+                confidence = 0.5;
+            }
+        }
+        const normalizedPlacename = WOFNormalizer.normalize(span.normalized);
+        if (!normalizedPlacename)
+            return;
+        for (const [placetype, placetypeConfig] of placetypeConfigMap) {
+            const placetypeTokens = this.placetypeToCacheMap.get(placetype);
+            if (!placetypeTokens)
+                continue;
+            const languages = placetypeTokens.get(normalizedPlacename);
+            if (!languages)
+                continue;
+            // Finally, we add the classifications.
+            for (const classification of placetypeConfig.classifications) {
+                span.classifications.add({
+                    classification,
+                    confidence,
+                    languages,
+                });
+            }
+        }
+    }
+}
+//# sourceMappingURL=WhosOnFirstClassifier.js.map

package/out/WhosOnFirstClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"WhosOnFirstClassifier.js","sourceRoot":"","sources":["../WhosOnFirstClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAEN,gBAAgB,EAChB,IAAI,EACJ,cAAc,EAEd,iBAAiB,GACjB,MAAM,iBAAiB,CAAA;AAExB,OAAO,EAAE,6BAA6B,EAAE,MAAM,uBAAuB,CAAA;AACrE,MAAM,sBAAsB,GAAG,6BAA6B,CAAC,aAAa,CAAC,CAAA;AAC3E,MAAM,8BAA8B,GAAG,6BAA6B,CAAC,UAAU,EAAE,aAAa,CAAC,CAAA;AAE/F,MAAM,aAAa,GAAG,IAAI,cAAc,CAAC;IACxC,SAAS,EAAE,IAAI;IACf,YAAY,EAAE,IAAI;IAClB,aAAa,EAAE,IAAI;IACnB,SAAS,EAAE,CAAC;CACZ,CAAC,CAAA;AAOF,sEAAsE;AACtE,MAAM,kBAAkB,GAAG,IAAI,GAAG,CAAmD;IACpF;QACC,UAAU;QACV;YACC,KAAK,EAAE,CAAC,wBAAwB,CAAC;YACjC,eAAe,EAAE,CAAC,MAAM,EAAE,UAAU,CAAC;SACrC;KACD;IACD;QACC,QAAQ;QACR;YACC,KAAK,EAAE,CAAC,wBAAwB,EAAE,wBAAwB,CAAC;YAC3D,eAAe,EAAE,CAAC,MAAM,EAAE,QAAQ,CAAC;SACnC;KACD;IAED;QACC,SAAS;QACT;YACC,KAAK,EAAE,CAAC,wBAAwB,EAAE,iBAAiB,EAAE,wBAAwB,CAAC;YAC9E,eAAe,EAAE,CAAC,MAAM,EAAE,SAAS,CAAC;SACpC;KACD;CACD,CAAC,CAAA;AAEF,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC;IAC9B,sBAAsB;IACtB,OAAO;IACP,OAAO;IACP,MAAM;IACN,MAAM;IAEN,0BAA0B;IAC1B,MAAM;IACN,QAAQ;IACR,MAAM;IACN,MAAM;IACN,aAAa;IACb,IAAI;IACJ,KAAK;IAEL,mBAAmB;IACnB,KAAK;IACL,KAAK;IACL,OAAO;IACP,MAAM;IACN,MAAM;IACN,KAAK;IACL,OAAO;IACP,OAAO;IACP,MAAM;IACN,KAAK;IAEL,MAAM;IACN,OAAO;IACP,OAAO;CACE,CAAC,CAAA;AAEX,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC;IACjC,MAAM;IACN,QAAQ;IACR,MAAM;IACN,SAAS;IACT,QAAQ;IACR,MAAM;IACN,UAAU;IACV,SAAS;IACT,UAAU;CACV,CAAC,CAAA;AAEF,MAAM,OAAO,qBAAsB,SAAQ,gBAAgB;IACnD,mBAAmB,GAAG,IAAI,GAAG,EAA2C,CAAA;IAE/E,KAAK,CAAC,KAAK;QACV,KAAK,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,IAAI,kBAAkB,EAAE,CAAC;YACtD,MAAM,cAAc,GAAG,IAAI,iBAAiB,CAAC;gBAC5C,QAAQ,EAAE,MAAM,CAAC,KAAK;gBACtB,UAAU,EAAE,aAAa;gBACzB,SAAS,EAAE,cAAc;gBACzB,aAAa,EAAE,sBAAsB,CAAC,SAAS,CAAC;gBAChD,qBAAqB,EAAE,8BAA8B,CAAC,SAAS,CAAC;aAChE,CAAC,CAAA;YAEF,MAAM,cAAc,CAAC,KAAK,EAAE,CAAA;YAE5B,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,SAAS,EAAE,cAAc,CAAC,CAAA;YAEvD,mCAAmC;YAEnC,IAAI,SAAS,KAAK,UAAU,EAAE,CAAC;gBAC9B,MAAM,cAAc,GAAG,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,UAAU,CAAE,CAAA;gBAChE,iDAAiD;gBAEjD,KAAK,MAAM,CAAC,KAAK,CAAC,IAAI,cAAc,EAAE,CAAC;oBACtC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;oBAC/B,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;oBAExC,IAAI,QAAQ,IAAI,iBAAiB,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;wBACjD,cAAc,CAAC,MAAM,CAAC,KAAK,CAAC,CAAA;oBAC7B,CAAC;gBACF,CAAC;YACF,CAAC;QACF,CAAC;QAED,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU;QACxB,IAAI,IAAI,CAAC,EAAE,CAAC,WAAW,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,EAAE,CAAC,WAAW,CAAC,EAAE,CAAC;YAClE,OAAM;QACP,CAAC;QAED,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,IAAI,IAAI,IAAI,CAAA;QAC5C,MAAM,EAAE,WAAW,EAAE,GAAG,SAAS,CAAA;QAEjC,IAAI,WAAW,IAAI,CAAC,WAAW,CAAC,EAAE,CAAC,eAAe,CAAC,IAAI,WAAW,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;YACjF,OAAM;QACP,CAAC;QAED,MAAM,UAAU,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,IAAI,IAAI,CAAA;QAC9C,MAAM,EAAE,eAAe,EAAE,GAAG,UAAU,CAAA;QAEtC,IAAI,UAA8B,CAAA;QAElC,IAAI,eAAe,EAAE,CAAC;YACrB,IAAI,eAAe,CAAC,EAAE,CAAC,cAAc,CAAC;gBAAE,OAAM;YAE9C,IAAI,eAAe,CAAC,EAAE,CAAC,WAAW,CAAC,EAAE,CAAC;gBACrC,UAAU,GAAG,GAAG,CAAA;YACjB,CAAC;QACF,CAAC;QAED,MAAM,mBAAmB,GAAG,aAAa,CAAC,SAAS,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QAEpE,IAAI,CAAC,mBAAmB;YAAE,OAAM;QAEhC,KAAK,MAAM,CAAC,SAAS,EAAE,eAAe,CAAC,IAAI,kBAAkB,EAAE,CAAC;YAC/D,MAAM,eAAe,GAAG,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;YAE/D,IAAI,CAAC,eAAe;gBAAE,SAAQ;YAE9B,MAAM,SAAS,GAAG,eAAe,CAAC,GAAG,CAAC,mBAAmB,CAAC,CAAA;YAE1D,IAAI,CAAC,SAAS;gBAAE,SAAQ;YAExB,uCAAuC;YACvC,KAAK,MAAM,cAAc,IAAI,eAAe,CAAC,eAAe,EAAE,CAAC;gBAC9D,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;oBACxB,cAAc;oBACd,UAAU;oBACV,SAAS;iBACT,CAAC,CAAA;YACH,CAAC;QACF,CAAC;IACF,CAAC;CACD"}

package/out/adapter.d.ts

@@ -0,0 +1,77 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Adapter that wraps Mailwoman's legacy mutation-based rule classifiers (`classifyTokens(context):
+ *   void`) into the proposal-emitting contract declared in `@mailwoman/core/types` (#6).
+ *
+ *   Phase 0 contract scaffolding: the adapter lets a single rule classifier be exposed as a
+ *   `ProposalClassifier`. The exhaustive one-wrapper-per-classifier sweep + solver rewire is a
+ *   follow-up; see Phase 0 task 3 success criteria in plan #8.
+ */
+import { type Classification, type Classifier as LegacyClassifier, type ClassifierConstructor as LegacyClassifierConstructor, type ClassifierOptions as LegacyClassifierOptions } from "@mailwoman/core/classification";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+import { type ComponentTag, type ProposalClassifier } from "@mailwoman/core/types";
+/**
+ * Options describing how to wrap a legacy classifier in the proposal interface.
+ */
+export interface WrapLegacyClassifierOptions {
+    /** Stable identifier for this classifier; used as `source_id`. */
+    id: string;
+    /** Legacy classifier instance or constructor. */
+    classifier: LegacyClassifier | LegacyClassifierConstructor;
+    /**
+     * Constructor options forwarded when `classifier` is a constructor. Ignored when it's already an
+     * instance.
+     */
+    classifierOptions?: LegacyClassifierOptions;
+    /**
+     * `ComponentTag`s this classifier may emit. Proposals carrying a tag not in this set are dropped
+     * (with a console warning in development).
+     */
+    emits: readonly ComponentTag[];
+    /**
+     * Legacy classification tags this classifier is known to produce. The wrapper traverses the span
+     * graph for spans bearing one of these tags. Defaults to "any tag with a `ComponentTag`
+     * mapping."
+     */
+    legacyTags?: readonly Classification[];
+    /**
+     * Locales this classifier is active for. `["*"]` (locale-agnostic) by default — matches the
+     * pre-refactor behavior.
+     */
+    locales?: readonly (string | "*")[];
+    /** Default penalty applied to emitted proposals. */
+    penalty?: number;
+    /**
+     * Override the default mapping for a legacy → component pair. Returning `null` drops the
+     * proposal.
+     */
+    mapTag?: (legacy: Classification) => ComponentTag | null;
+}
+/**
+ * Wrap a single legacy rule classifier as a `ProposalClassifier`.
+ *
+ * Mechanics:
+ *
+ * 1. The legacy classifier is instantiated (if a constructor) and its `ready()` step is awaited.
+ * 2. On `classify(section)`, a fresh local `TokenContext` is built around the section's body text so
+ *    legacy mutations don't bleed into the caller's graph.
+ * 3. After `classifyTokens(localContext)` runs, the wrapper walks all spans (sections, words, phrases)
+ *    and collects spans bearing any of the wrapper's `legacyTags`.
+ * 4. Each such (span, classification) pair becomes a `ClassificationProposal`. The span is re-anchored
+ *    to the caller's section so character offsets are correct relative to the original input.
+ *
+ * Note on isolation: building a fresh `TokenContext` is deliberately coarse — it ignores any prior
+ * classifications from earlier classifiers in the pipeline. Composite classifiers that depend on
+ * upstream tags will need a different bridge; that lives in a higher-level orchestrator, not in
+ * this adapter.
+ */
+export declare function wrapLegacyClassifier(options: WrapLegacyClassifierOptions): ProposalClassifier;
+/**
+ * Walk every span attached to a `TokenContext` — sections, words, and phrases — in a single flat
+ * pass. Order is not guaranteed.
+ */
+export declare function iterateSpans(context: TokenContext): Generator<Span>;
+//# sourceMappingURL=adapter.d.ts.map

package/out/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EACN,KAAK,cAAc,EACnB,KAAK,UAAU,IAAI,gBAAgB,EACnC,KAAK,qBAAqB,IAAI,2BAA2B,EACzD,KAAK,iBAAiB,IAAI,uBAAuB,EACjD,MAAM,gCAAgC,CAAA;AACvC,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AACjE,OAAO,EAGN,KAAK,YAAY,EAEjB,KAAK,kBAAkB,EAEvB,MAAM,uBAAuB,CAAA;AAE9B;;GAEG;AACH,MAAM,WAAW,2BAA2B;IAC3C,kEAAkE;IAClE,EAAE,EAAE,MAAM,CAAA;IAEV,iDAAiD;IACjD,UAAU,EAAE,gBAAgB,GAAG,2BAA2B,CAAA;IAE1D;;;OAGG;IACH,iBAAiB,CAAC,EAAE,uBAAuB,CAAA;IAE3C;;;OAGG;IACH,KAAK,EAAE,SAAS,YAAY,EAAE,CAAA;IAE9B;;;;OAIG;IACH,UAAU,CAAC,EAAE,SAAS,cAAc,EAAE,CAAA;IAEtC;;;OAGG;IACH,OAAO,CAAC,EAAE,SAAS,CAAC,MAAM,GAAG,GAAG,CAAC,EAAE,CAAA;IAEnC,oDAAoD;IACpD,OAAO,CAAC,EAAE,MAAM,CAAA;IAEhB;;;OAGG;IACH,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,cAAc,KAAK,YAAY,GAAG,IAAI,CAAA;CACxD;AAED;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,oBAAoB,CAAC,OAAO,EAAE,2BAA2B,GAAG,kBAAkB,CA6E7F;AAED;;;GAGG;AACH,wBAAiB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,SAAS,CAAC,IAAI,CAAC,CAgBpE"}

package/out/adapter.js

@@ -0,0 +1,125 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Adapter that wraps Mailwoman's legacy mutation-based rule classifiers (`classifyTokens(context):
+ *   void`) into the proposal-emitting contract declared in `@mailwoman/core/types` (#6).
+ *
+ *   Phase 0 contract scaffolding: the adapter lets a single rule classifier be exposed as a
+ *   `ProposalClassifier`. The exhaustive one-wrapper-per-classifier sweep + solver rewire is a
+ *   follow-up; see Phase 0 task 3 success criteria in plan #8.
+ */
+import {} from "@mailwoman/core/classification";
+import { Span, TokenContext } from "@mailwoman/core/tokenization";
+import { legacyClassificationToComponentTag, } from "@mailwoman/core/types";
+/**
+ * Wrap a single legacy rule classifier as a `ProposalClassifier`.
+ *
+ * Mechanics:
+ *
+ * 1. The legacy classifier is instantiated (if a constructor) and its `ready()` step is awaited.
+ * 2. On `classify(section)`, a fresh local `TokenContext` is built around the section's body text so
+ *    legacy mutations don't bleed into the caller's graph.
+ * 3. After `classifyTokens(localContext)` runs, the wrapper walks all spans (sections, words, phrases)
+ *    and collects spans bearing any of the wrapper's `legacyTags`.
+ * 4. Each such (span, classification) pair becomes a `ClassificationProposal`. The span is re-anchored
+ *    to the caller's section so character offsets are correct relative to the original input.
+ *
+ * Note on isolation: building a fresh `TokenContext` is deliberately coarse — it ignores any prior
+ * classifications from earlier classifiers in the pipeline. Composite classifiers that depend on
+ * upstream tags will need a different bridge; that lives in a higher-level orchestrator, not in
+ * this adapter.
+ */
+export function wrapLegacyClassifier(options) {
+    const { id, classifier: classifierOrCtor, classifierOptions, emits, legacyTags, locales = ["*"], penalty = 0, mapTag, } = options;
+    const instance = typeof classifierOrCtor === "function" ? new classifierOrCtor(classifierOptions) : classifierOrCtor;
+    const tagFilter = mapTag ??
+        ((legacy) => {
+            const mapped = legacyClassificationToComponentTag(legacy);
+            return mapped && emits.includes(mapped) ? mapped : null;
+        });
+    const expectedLegacy = legacyTags ? new Set(legacyTags) : null;
+    let readyPromise = null;
+    function ensureReady() {
+        if (!instance.ready)
+            return Promise.resolve();
+        readyPromise ??= instance.ready().then(() => undefined);
+        return readyPromise;
+    }
+    async function runOnSection(section) {
+        await ensureReady();
+        const localContext = new TokenContext(section.body);
+        instance.classifyTokens(localContext);
+        const sectionOffset = section.start;
+        const proposals = [];
+        for (const span of iterateSpans(localContext)) {
+            for (const [legacy, match] of span.classifications) {
+                if (expectedLegacy && !expectedLegacy.has(legacy))
+                    continue;
+                const component = tagFilter(legacy);
+                if (!component)
+                    continue;
+                if (!emits.includes(component))
+                    continue;
+                proposals.push({
+                    span: rebaseSpan(span, sectionOffset),
+                    component,
+                    confidence: match.confidence,
+                    source: "rule",
+                    source_id: id,
+                    penalty,
+                    metadata: {
+                        legacyClassification: legacy,
+                        ...(match.languages ? { languages: Array.from(match.languages) } : {}),
+                        ...(match.flags ? { flags: Array.from(match.flags) } : {}),
+                    },
+                });
+            }
+        }
+        return proposals;
+    }
+    const wrapped = {
+        id,
+        emits,
+        locales,
+        ready: () => ensureReady(),
+        classify: (section, _context) => runOnSection(section),
+    };
+    return wrapped;
+}
+/**
+ * Walk every span attached to a `TokenContext` — sections, words, and phrases — in a single flat
+ * pass. Order is not guaranteed.
+ */
+export function* iterateSpans(context) {
+    const visited = new Set();
+    const queue = [];
+    if (context.span)
+        queue.push(context.span);
+    for (const section of context.sections)
+        queue.push(section);
+    while (queue.length > 0) {
+        const span = queue.pop();
+        if (visited.has(span.id))
+            continue;
+        visited.add(span.id);
+        yield span;
+        for (const child of span.children)
+            queue.push(child);
+        for (const phrase of span.phrases)
+            queue.push(phrase);
+    }
+}
+/**
+ * Produce a Span whose character offsets are anchored to the caller's original input (not the local
+ * context's). The returned span is a thin clone for use in proposals; the source span continues to
+ * live in the local TokenContext.
+ */
+function rebaseSpan(span, sectionOffset) {
+    if (sectionOffset === 0)
+        return span;
+    const rebased = Span.from(span.body, { start: span.start + sectionOffset });
+    return rebased;
+}
+//# sourceMappingURL=adapter.js.map

package/out/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,EAKN,MAAM,gCAAgC,CAAA;AACvC,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,8BAA8B,CAAA;AACjE,OAAO,EAIN,kCAAkC,GAGlC,MAAM,uBAAuB,CAAA;AA+C9B;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,oBAAoB,CAAC,OAAoC;IACxE,MAAM,EACL,EAAE,EACF,UAAU,EAAE,gBAAgB,EAC5B,iBAAiB,EACjB,KAAK,EACL,UAAU,EACV,OAAO,GAAG,CAAC,GAAG,CAAC,EACf,OAAO,GAAG,CAAC,EACX,MAAM,GACN,GAAG,OAAO,CAAA;IAEX,MAAM,QAAQ,GACb,OAAO,gBAAgB,KAAK,UAAU,CAAC,CAAC,CAAC,IAAI,gBAAgB,CAAC,iBAAiB,CAAC,CAAC,CAAC,CAAC,gBAAgB,CAAA;IAEpG,MAAM,SAAS,GACd,MAAM;QACN,CAAC,CAAC,MAAM,EAAE,EAAE;YACX,MAAM,MAAM,GAAG,kCAAkC,CAAC,MAAM,CAAC,CAAA;YACzD,OAAO,MAAM,IAAI,KAAK,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAA;QACxD,CAAC,CAAC,CAAA;IAEH,MAAM,cAAc,GAAG,UAAU,CAAC,CAAC,CAAC,IAAI,GAAG,CAAiB,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAE9E,IAAI,YAAY,GAAyB,IAAI,CAAA;IAE7C,SAAS,WAAW;QACnB,IAAI,CAAC,QAAQ,CAAC,KAAK;YAAE,OAAO,OAAO,CAAC,OAAO,EAAE,CAAA;QAC7C,YAAY,KAAK,QAAQ,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAA;QACvD,OAAO,YAAY,CAAA;IACpB,CAAC;IAED,KAAK,UAAU,YAAY,CAAC,OAAgB;QAC3C,MAAM,WAAW,EAAE,CAAA;QAEnB,MAAM,YAAY,GAAG,IAAI,YAAY,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;QACnD,QAAQ,CAAC,cAAc,CAAC,YAAY,CAAC,CAAA;QAErC,MAAM,aAAa,GAAG,OAAO,CAAC,KAAK,CAAA;QACnC,MAAM,SAAS,GAA6B,EAAE,CAAA;QAE9C,KAAK,MAAM,IAAI,IAAI,YAAY,CAAC,YAAY,CAAC,EAAE,CAAC;YAC/C,KAAK,MAAM,CAAC,MAAM,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC;gBACpD,IAAI,cAAc,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,MAAM,CAAC;oBAAE,SAAQ;gBAE3D,MAAM,SAAS,GAAG,SAAS,CAAC,MAAM,CAAC,CAAA;gBACnC,IAAI,CAAC,SAAS;oBAAE,SAAQ;gBACxB,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,CAAC;oBAAE,SAAQ;gBAExC,SAAS,CAAC,IAAI,CAAC;oBACd,IAAI,EAAE,UAAU,CAAC,IAAI,EAAE,aAAa,CAAC;oBACrC,SAAS;oBACT,UAAU,EAAE,KAAK,CAAC,UAAU;oBAC5B,MAAM,EAAE,MAAM;oBACd,SAAS,EAAE,EAAE;oBACb,OAAO;oBACP,QAAQ,EAAE;wBACT,oBAAoB,EAAE,MAAM;wBAC5B,GAAG,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;wBACtE,GAAG,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;qBAC1D;iBACD,CAAC,CAAA;YACH,CAAC;QACF,CAAC;QAED,OAAO,SAAS,CAAA;IACjB,CAAC;IAED,MAAM,OAAO,GAAuB;QACnC,EAAE;QACF,KAAK;QACL,OAAO;QACP,KAAK,EAAE,GAAG,EAAE,CAAC,WAAW,EAAE;QAC1B,QAAQ,EAAE,CAAC,OAAO,EAAE,QAA2B,EAAE,EAAE,CAAC,YAAY,CAAC,OAAO,CAAC;KACzE,CAAA;IAED,OAAO,OAAO,CAAA;AACf,CAAC;AAED;;;GAGG;AACH,MAAM,SAAS,CAAC,CAAC,YAAY,CAAC,OAAqB;IAClD,MAAM,OAAO,GAAG,IAAI,GAAG,EAAU,CAAA;IACjC,MAAM,KAAK,GAAW,EAAE,CAAA;IAExB,IAAI,OAAO,CAAC,IAAI;QAAE,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;IAC1C,KAAK,MAAM,OAAO,IAAI,OAAO,CAAC,QAAQ;QAAE,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAE3D,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,EAAG,CAAA;QACzB,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YAAE,SAAQ;QAClC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;QACpB,MAAM,IAAI,CAAA;QAEV,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,QAAQ;YAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACpD,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,OAAO;YAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IACtD,CAAC;AACF,CAAC;AAED;;;;GAIG;AACH,SAAS,UAAU,CAAC,IAAU,EAAE,aAAqB;IACpD,IAAI,aAAa,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IAEpC,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,GAAG,aAAa,EAAE,CAAC,CAAA;IAC3E,OAAO,OAAO,CAAA;AACf,CAAC"}

package/out/composite/index.d.ts

@@ -0,0 +1,12 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+export * from "./intersection.js";
+export * from "./person.js";
+export * from "./street.js";
+export * from "./street_name.js";
+export * from "./subdivision.js";
+export * from "./venue.js";
+//# sourceMappingURL=index.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../composite/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,mBAAmB,CAAA;AACjC,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,kBAAkB,CAAA;AAChC,cAAc,kBAAkB,CAAA;AAChC,cAAc,YAAY,CAAA"}

package/out/composite/index.js

@@ -0,0 +1,12 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+export * from "./intersection.js";
+export * from "./person.js";
+export * from "./street.js";
+export * from "./street_name.js";
+export * from "./subdivision.js";
+export * from "./venue.js";
+//# sourceMappingURL=index.js.map

package/out/composite/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../composite/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,mBAAmB,CAAA;AACjC,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,kBAAkB,CAAA;AAChC,cAAc,kBAAkB,CAAA;AAChC,cAAc,YAAY,CAAA"}

package/out/composite/intersection.d.ts

@@ -0,0 +1,10 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { CompositeClassifier } from "@mailwoman/core";
+export declare class CompositeIntersectionClassifier extends CompositeClassifier {
+    constructor();
+}
+//# sourceMappingURL=intersection.d.ts.map

package/out/composite/intersection.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"intersection.d.ts","sourceRoot":"","sources":["../../composite/intersection.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAA+B,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AA0GlF,qBAAa,+BAAgC,SAAQ,mBAAmB;;CAIvE"}

package/out/composite/intersection.js

@@ -0,0 +1,115 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { CompositeClassifier } from "@mailwoman/core";
+const configs = [
+    {
+        // SW 6th & Pine
+        scheme: [
+            {
+                is: ["directional"],
+                not: ["intersection", "street_suffix"],
+            },
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection", "street_suffix"],
+            },
+            {
+                is: ["intersection"],
+                not: ["street", "street_suffix"],
+            },
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection"],
+            },
+        ],
+    },
+    {
+        // Foo St and Bar St
+        scheme: [
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection", "street_suffix"],
+                confidence: 0.81,
+                classification: "street",
+            },
+            {
+                is: ["intersection"],
+                not: ["street", "street_suffix"],
+            },
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection"],
+                confidence: 0.82,
+                classification: "street",
+            },
+        ],
+    },
+    {
+        // Foo and Bar St
+        scheme: [
+            {
+                is: ["alpha"],
+                not: ["intersection", "street", "street_suffix"],
+                confidence: 0.53,
+                classification: "street",
+            },
+            {
+                is: ["intersection"],
+                not: ["street", "street_suffix"],
+            },
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection"],
+            },
+        ],
+    },
+    {
+        // Foo St and Bar
+        scheme: [
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection"],
+            },
+            {
+                is: ["intersection"],
+                not: ["street", "street_suffix"],
+            },
+            {
+                is: ["alpha", "numeric", "ordinal"],
+                not: ["intersection", "street"],
+                confidence: 0.56,
+                classification: "street",
+            },
+        ],
+    },
+    {
+        // Foo and Bar
+        scheme: [
+            {
+                is: ["alpha"],
+                not: ["intersection", "street", "street_suffix"],
+                confidence: 0.57,
+                classification: "street",
+            },
+            {
+                is: ["intersection"],
+                not: ["street", "street_suffix"],
+            },
+            {
+                is: ["alpha"],
+                not: ["intersection", "street", "street_suffix"],
+                confidence: 0.58,
+                classification: "street",
+            },
+        ],
+    },
+];
+export class CompositeIntersectionClassifier extends CompositeClassifier {
+    constructor() {
+        super("multistreet", configs);
+    }
+}
+//# sourceMappingURL=intersection.js.map

package/out/composite/intersection.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"intersection.js","sourceRoot":"","sources":["../../composite/intersection.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAA+B,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAElF,MAAM,OAAO,GAA6B;IACzC;QACC,gBAAgB;QAChB,MAAM,EAAE;YACP;gBACC,EAAE,EAAE,CAAC,aAAa,CAAC;gBACnB,GAAG,EAAE,CAAC,cAAc,EAAE,eAAe,CAAC;aACtC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,EAAE,eAAe,CAAC;aACtC;YACD;gBACC,EAAE,EAAE,CAAC,cAAc,CAAC;gBACpB,GAAG,EAAE,CAAC,QAAQ,EAAE,eAAe,CAAC;aAChC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,CAAC;aACrB;SACD;KACD;IACD;QACC,oBAAoB;QACpB,MAAM,EAAE;YACP;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,EAAE,eAAe,CAAC;gBACtC,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;YACD;gBACC,EAAE,EAAE,CAAC,cAAc,CAAC;gBACpB,GAAG,EAAE,CAAC,QAAQ,EAAE,eAAe,CAAC;aAChC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,CAAC;gBACrB,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;SACD;KACD;IACD;QACC,iBAAiB;QACjB,MAAM,EAAE;YACP;gBACC,EAAE,EAAE,CAAC,OAAO,CAAC;gBACb,GAAG,EAAE,CAAC,cAAc,EAAE,QAAQ,EAAE,eAAe,CAAC;gBAChD,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;YACD;gBACC,EAAE,EAAE,CAAC,cAAc,CAAC;gBACpB,GAAG,EAAE,CAAC,QAAQ,EAAE,eAAe,CAAC;aAChC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,CAAC;aACrB;SACD;KACD;IACD;QACC,iBAAiB;QACjB,MAAM,EAAE;YACP;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,CAAC;aACrB;YACD;gBACC,EAAE,EAAE,CAAC,cAAc,CAAC;gBACpB,GAAG,EAAE,CAAC,QAAQ,EAAE,eAAe,CAAC;aAChC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,EAAE,SAAS,EAAE,SAAS,CAAC;gBACnC,GAAG,EAAE,CAAC,cAAc,EAAE,QAAQ,CAAC;gBAC/B,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;SACD;KACD;IACD;QACC,cAAc;QACd,MAAM,EAAE;YACP;gBACC,EAAE,EAAE,CAAC,OAAO,CAAC;gBACb,GAAG,EAAE,CAAC,cAAc,EAAE,QAAQ,EAAE,eAAe,CAAC;gBAChD,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;YACD;gBACC,EAAE,EAAE,CAAC,cAAc,CAAC;gBACpB,GAAG,EAAE,CAAC,QAAQ,EAAE,eAAe,CAAC;aAChC;YACD;gBACC,EAAE,EAAE,CAAC,OAAO,CAAC;gBACb,GAAG,EAAE,CAAC,cAAc,EAAE,QAAQ,EAAE,eAAe,CAAC;gBAChD,UAAU,EAAE,IAAI;gBAChB,cAAc,EAAE,QAAQ;aACxB;SACD;KACD;CACD,CAAA;AAED,MAAM,OAAO,+BAAgC,SAAQ,mBAAmB;IACvE;QACC,KAAK,CAAC,aAAa,EAAE,OAAO,CAAC,CAAA;IAC9B,CAAC;CACD"}

package/out/composite/person.d.ts

@@ -0,0 +1,10 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { CompositeClassifier } from "@mailwoman/core";
+export declare class CompositePersonClassifier extends CompositeClassifier {
+    constructor();
+}
+//# sourceMappingURL=person.d.ts.map

package/out/composite/person.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"person.d.ts","sourceRoot":"","sources":["../../composite/person.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAA+B,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAuGlF,qBAAa,yBAA0B,SAAQ,mBAAmB;;CAIjE"}

package/out/composite/person.js

@@ -0,0 +1,112 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { CompositeClassifier } from "@mailwoman/core";
+const configs = [
+    {
+        // Anne Marie
+        classification: "given_name",
+        confidence: 0.25,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["given_name"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+    {
+        // Georges Bizet
+        confidence: 0.5,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["surname"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+    {
+        // Rose de Lima
+        confidence: 0.5,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["stop_word"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["surname"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+    {
+        // Raul Leite Magalhães (first name, middle name, family name)
+        // Donald W. Reynolds
+        confidence: 0.5,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["given_name", "surname", "middle_initial"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["surname"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+    {
+        // Unknown surname
+        confidence: 0.1,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["alpha"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+    {
+        // Unknown surname
+        confidence: 0.1,
+        scheme: [
+            {
+                is: ["given_name"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["stop_word"],
+                not: ["street", "intersection"],
+            },
+            {
+                is: ["alpha"],
+                not: ["street", "street_prefix", "stop_word"],
+            },
+        ],
+    },
+];
+export class CompositePersonClassifier extends CompositeClassifier {
+    constructor() {
+        super("person", configs);
+    }
+}
+//# sourceMappingURL=person.js.map