@mailwoman/kind-classifier 10.0.0 → 10.1.0

package/out/rules.js

@@ -3,56 +3,47 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   Rule-based classifiers for each `QueryKind`. Each rule inspects the normalized input + QueryShape
- *   and returns a confidence score in [0, 1], or 0 if the rule doesn't fire.
- *
- *   Bitter-lesson-safe: only universal structural patterns — no place-name dictionaries. ~1 small
- *   regex set per new locale, not 50K dictionary entries.
+ *   Rule-based `QueryKind` scorers. Each scorer reads structural patterns in the normalized input and
+ *   its QueryShape and returns a confidence from 0 to 1.
  */
-import { NAME_PRONE_US_SUFFIXES, US_STREET_SUFFIX_LOOKUP } from "@mailwoman/codex/us/street-suffix";
+import { NAME_PRONE_US_SUFFIXES, US_STREET_SUFFIX_LOOKUP } from "@mailwoman/codex/us/street/suffix";
+import { classifyTokens, foldInputClass } from "@mailwoman/query-shape/character-class";
 import { isPostcodeFormat } from "@mailwoman/query-shape/known-formats";
 /**
- * Longest input still plausible as a bare venue or landmark name. Beyond it the query is carrying an address as well,
- * and belongs to the structured-address scorer.
+ * Maximum length for a bare venue or landmark query.
  */
 const MAX_LANDMARK_LENGTH = 50;
 /**
- * Longest input still plausible as a bare postcode, allowing for a country prefix and separators.
+ * Maximum length for a bare postcode, including an optional country prefix and separators.
  */
 const MAX_POSTCODE_ONLY_LENGTH = 16;
 /**
- * Share of the input the postcode must occupy before `postcode_only` fires. Below it the query is carrying something
- * else too — a locality, a street — and another kind should win.
+ * Minimum fraction of the input that a postcode must cover for `postcode_only`.
  */
 const MIN_POSTCODE_COVERAGE = 0.7;
 /**
- * Longest input still plausible as a bare locality name, including a trailing region code.
- *
- * Exported so `intent-rules.ts`'s `bare_toponym` can share the exact same ceiling. Sharing it is what makes
- * "bare_toponym is a strict refinement of locality_only" a structural property rather than two numbers that happen to
- * agree today.
+ * Maximum length for a bare locality.
+ * The `bare_toponym` intent rule uses the same limit.
  */
 export const MAX_LOCALITY_ONLY_LENGTH = 30;
 /**
- * Word count of a short capitalized phrase — the shape of a venue name like `Empire State Building`. Wider than this
- * and the phrase is more likely a full address line.
+ * Minimum word count for a short venue phrase.
  */
 const VENUE_PHRASE_MIN_WORDS = 2;
 /**
- * Upper bound of the short-phrase venue window; see {@link VENUE_PHRASE_MIN_WORDS}.
+ * Maximum word count for a short venue phrase.
  */
 const VENUE_PHRASE_MAX_WORDS = 4;
 /**
- * Word count above which a single-segment proper-case phrase stops reading as a venue name.
+ * Maximum word count for a longer single-segment venue phrase.
  */
 const LONG_VENUE_PHRASE_MAX_WORDS = 6;
 /**
- * Length at which a single-segment alphanumeric input reads as a full postcode rather than a fragment. Shorter
- * alphanumeric inputs score lower because they are as likely to be a unit number.
+ * Length at which single-segment alphanumeric input scores as a likely structured address.
  */
 const ALPHANUMERIC_POSTCODE_MIN_LENGTH = 15;
 /**
- * Landmark vocabulary — phrases that suggest a vague-location description rather than an address.
+ * Phrases that signal a relative landmark description.
  */
 const LANDMARK_LEADERS = [
     "behind",
@@ -66,29 +57,27 @@ const LANDMARK_LEADERS = [
     "beside",
 ];
 /**
- * Intersection vocabulary — words that signal "where two streets cross" rather than an address.
+ * Patterns that signal a street intersection.
  */
 const INTERSECTION_PATTERNS = [
     /\bcorner of\b/i,
     /\bintersection of\b/i,
     /\bat the corner of\b/i,
-    // "5th and Main", "Broadway & 42nd"
+    // Examples: "5th and Main", "Broadway & 42nd".
     /\b\w+(?:st|nd|rd|th|street|ave|avenue|blvd|boulevard|road|rd|lane|ln)?\s+(?:and|&|@)\s+\w+/i,
 ];
 /**
- * `po_box` rule: high-confidence iff QueryShape detected a po_box format hit. Confidence comes directly from the hit;
- * covers all locale variants (US "PO Box 123", FR "BP 42", etc.).
+ * Scores input in which QueryShape detected a PO Box format.
  */
 export function scorePoBox(_input, shape) {
     const hit = shape.knownFormats.find((f) => f.format === "po_box");
     if (!hit)
         return 0;
-    // Boost slightly above the raw hit confidence so po_box wins ties with structured_address when
-    // both rules fire on the same input.
+    // The bonus makes a PO Box win a tie with a structured address.
     return Math.min(1, hit.confidence + 0.1);
 }
 /**
- * `intersection` rule: text matches one of the conventional intersection phrasings.
+ * Scores street-intersection phrasing.
  */
 export function scoreIntersection(input, _shape) {
     const text = input.normalized;
@@ -99,8 +88,7 @@ export function scoreIntersection(input, _shape) {
     return 0;
 }
 /**
- * `landmark` rule: text begins with a landmark-leader phrase. These inputs are not addresses proper — they describe a
- * location relative to another place.
+ * Scores input that begins with a relative-landmark phrase.
  */
 export function scoreLandmark(input, _shape) {
     const lc = input.normalized.toLowerCase().trim();
@@ -111,59 +99,101 @@ export function scoreLandmark(input, _shape) {
     return 0;
 }
 /**
- * Split on whitespace + commas and drop empty words — the word grammar shared with `intent-rules.ts`. Dropping empties
- * matters: a trailing comma or doubled separator otherwise yields an empty string that inflates the word count and
- * falsifies every-word predicates like the proper-case check below.
+ * Splits text on whitespace and commas and drops empty tokens.
  */
 export function wordsOf(text) {
-    return text.split(/[\s,]+/).filter((word) => word.length > 0);
+    return text.split(/[\s,]+/).filter((word) => word.length);
 }
 /**
- * True when a word is USPS street-suffix vocabulary that unambiguously signals an address — the full Pub-28 table via
- * `@mailwoman/codex`, minus its curated name-prone canonicals (PARK, FIELD, HILL, LAKE, …). Those double as ordinary
- * proper-name heads ("Wrigley Field", "Menlo Park"), and disqualifying on them would reject the very venue and place
- * names the rules below exist to capture. Shared with `intent-rules.ts` so both rule sets read one definition.
+ * Reports whether a word is a USPS street suffix that rarely appears in place names.
  */
 export function isDisqualifyingStreetSuffix(word) {
     const canonical = US_STREET_SUFFIX_LOOKUP.get(word.trim().toLowerCase());
     return canonical !== undefined && !NAME_PRONE_US_SUFFIXES.has(canonical);
 }
 /**
- * `landmark` rule (venue/named-place variant): short capitalized input with no street suffixes, no postcode hits, and
- * no region abbreviations. Captures "Pier 39", "Empire State Building", "Wrigley Field", "Grand Central Terminal".
+ * Removes postcode spans in the final segment and collapses the remaining separators.
+ */
+export function withoutPostcodeSpans(text, shape) {
+    const merged = mergedPostcodeSpans(shape);
+    if (!merged.length)
+        return text;
+    let remainder = text;
+    // The loop removes spans from last to first.
+    // This order keeps the remaining offsets valid.
+    for (let index = merged.length - 1; index >= 0; index--) {
+        const span = merged[index];
+        remainder = remainder.slice(0, span.start) + remainder.slice(span.end);
+    }
+    return remainder
+        .replaceAll(/[\s,]+/gu, " ")
+        .replace(/[\s,]+$/u, "")
+        .trim();
+}
+/**
+ * Reports whether text contains a Unicode letter.
  *
- * Fires at moderate confidence (0.65) — below structured_address (0.9) so addresses always win, but above vague (0.3)
- * so the pipeline can route landmark queries to the venue resolver.
+ * By itself, the `alpha` character class can match punctuation-only input.
+ */
+export function carriesLetter(text) {
+    return /\p{L}/u.test(text);
+}
+/**
+ * Merges overlapping postcode hits in the final segment into sorted, disjoint spans.
+ */
+function mergedPostcodeSpans(shape) {
+    // Postcode hits in earlier segments are often house numbers.
+    const tail = shape.segments?.at(-1)?.span;
+    if (!tail)
+        return [];
+    const merged = [];
+    for (const hit of shape.knownFormats) {
+        if (!isPostcodeFormat(hit.format))
+            continue;
+        if (hit.span.start < tail.start || hit.span.end > tail.end)
+            continue;
+        merged.push({ start: hit.span.start, end: hit.span.end });
+    }
+    if (merged.length < 2)
+        return merged;
+    merged.sort((left, right) => left.start - right.start);
+    const disjoint = [merged[0]];
+    for (const span of merged.slice(1)) {
+        const last = disjoint.at(-1);
+        if (span.start <= last.end) {
+            last.end = Math.max(last.end, span.end);
+            continue;
+        }
+        disjoint.push(span);
+    }
+    return disjoint;
+}
+/**
+ * Scores short capitalized venue names that have no street suffix or recognized format.
  */
 export function scoreVenueLandmark(input, shape) {
     const text = input.normalized.trim();
     const len = text.length;
     if (len === 0 || len > MAX_LANDMARK_LENGTH)
         return 0;
-    // Must have at least one capitalized word.
     if (!/[A-Z]/.test(text))
         return 0;
-    // Reject if any known postcode format hit exists.
     if (shape.knownFormats.length)
         return 0;
-    // Reject if it looks like a multi-segment structured address (City, ST ZIP).
     const segCount = shape.segments?.length ?? 1;
     if (segCount > 2)
         return 0;
-    // Reject if any word is an unambiguous street suffix.
     const words = wordsOf(text);
     for (const w of words) {
         if (isDisqualifyingStreetSuffix(w))
             return 0;
     }
-    // Reject if the first token is a pure number (house-number-leading pattern).
+    // A leading number usually is a house number.
     if (/^\d+\s/.test(text))
         return 0;
-    // Boost if the input has a number NOT at the start (venue-style: "Pier 39", "Terminal 5").
+    // Venue names often contain a number after the first word.
     const hasInternalNumber = /\s\d+/.test(text) && !/^\d/.test(text);
-    // Check if every word starts with uppercase (proper-noun pattern).
     const allProperCase = words.length > 1 && words.every((w) => /^[A-Z]/.test(w));
-    // Boost for short single-segment capitalized phrases (2-4 words).
     const wordCount = words.length;
     if (wordCount >= VENUE_PHRASE_MIN_WORDS && wordCount <= VENUE_PHRASE_MAX_WORDS && segCount === 1) {
         if (hasInternalNumber)
@@ -172,17 +202,13 @@ export function scoreVenueLandmark(input, shape) {
             return 0.88;
         return 0.65;
     }
-    // Longer single-segment capitalized phrases get moderate confidence.
     if (wordCount <= LONG_VENUE_PHRASE_MAX_WORDS && segCount === 1 && allProperCase) {
         return 0.75;
     }
     return 0;
 }
 /**
- * `postcode_only` rule: input is short AND has a postcode format hit covering most of it.
- *
- * The "covering most of it" check is what distinguishes `"10118"` (postcode-only) from `"350 5th Ave 10118"`
- * (structured-address with a postcode in it).
+ * Scores short input in which a postcode covers most of the text.
  */
 export function scorePostcodeOnly(input, shape) {
     const len = input.normalized.length;
@@ -192,61 +218,72 @@ export function scorePostcodeOnly(input, shape) {
     if (!postcodeHit)
         return 0;
     const hitLen = postcodeHit.span.end - postcodeHit.span.start;
-    // At least 70% of the input must be the postcode for the rule to fire confidently.
     if (hitLen / len < MIN_POSTCODE_COVERAGE)
         return 0;
-    // Confidence scales with how much of the input is the postcode and how confident the format hit was.
     return Math.min(1, postcodeHit.confidence * (hitLen / len) + 0.1);
 }
 /**
- * `locality_only` rule: short input, alpha-class, single segment, no format hits.
- *
- * Examples: `"Paris"`, `"NYC NY"`, `"Tokyo"`. Distinguishes from `structured_address` (multiple segments) and `vague`
- * (long or mixed-class).
+ * Scores a place name with at most an administrative tail and an optional postcode.
  */
 export function scoreLocalityOnly(input, shape) {
-    const len = input.normalized.length;
-    if (len === 0 || len > MAX_LOCALITY_ONLY_LENGTH)
+    // Any recognized format other than a postcode rules out a bare locality.
+    let carriesPostcode = false;
+    for (const hit of shape.knownFormats) {
+        if (isPostcodeFormat(hit.format)) {
+            carriesPostcode = true;
+            continue;
+        }
         return 0;
-    if (shape.characterClass !== "alpha")
+    }
+    // Only postcode hits in the final segment can be removed.
+    const removable = carriesPostcode ? mergedPostcodeSpans(shape) : [];
+    if (carriesPostcode && !removable.length)
         return 0;
-    if (shape.knownFormats.length)
+    const withoutPostcode = carriesPostcode ? withoutPostcodeSpans(input.normalized, shape) : input.normalized;
+    const len = withoutPostcode.length;
+    if (len === 0 || len > MAX_LOCALITY_ONLY_LENGTH)
         return 0;
-    // Locality-only inputs typically have 1-3 segments (e.g. "New York" is 1 segment, "Paris, FR" is 2).
-    // We allow up to 2 segments before deciding it's structured.
+    if (!carriesLetter(withoutPostcode))
+        return 0;
+    // The remaining text must be alphabetic.
+    // Its class is recomputed only after a postcode was removed.
+    if (shape.characterClass !== "alpha") {
+        if (!carriesPostcode)
+            return 0;
+        if (foldInputClass(classifyTokens(withoutPostcode)) !== "alpha")
+            return 0;
+    }
+    // A locality and an administrative tail use at most two segments.
     const segCount = shape.segments?.length ?? 1;
     if (segCount > 2)
         return 0;
     return 0.85;
 }
 /**
- * `structured_address` rule: looks like a real multi-component address. Either has multiple segments or is long and
- * mixed-class.
+ * Scores multi-component address shapes.
  */
 export function scoreStructuredAddress(input, shape) {
     const len = input.normalized.length;
     if (len === 0)
         return 0;
     const segCount = shape.segments?.length ?? 1;
-    // Multi-segment input with mixed character class = high confidence structured.
-    if (segCount >= 2 && shape.characterClass === "alphanumeric")
+    // Digits outside the postcode are required, so a locality with a postcode does not score here.
+    if (segCount >= 2 &&
+        shape.characterClass === "alphanumeric" &&
+        /\d/u.test(withoutPostcodeSpans(input.normalized, shape))) {
         return 0.9;
-    // Single-segment but reasonably long and alphanumeric = moderate confidence.
+    }
     if (len >= ALPHANUMERIC_POSTCODE_MIN_LENGTH && shape.characterClass === "alphanumeric")
         return 0.75;
-    // Multi-segment but pure-alpha = moderate (could be a multi-word locality).
+    // Multi-segment input without digits may still be a multi-word locality, so it scores lower.
     if (segCount >= 2)
         return 0.6;
-    // Single-segment, short, alphanumeric (e.g. "10118-1234" with no other content) — weak.
     if (len < ALPHANUMERIC_POSTCODE_MIN_LENGTH && shape.characterClass === "alphanumeric")
         return 0.4;
     return 0;
 }
 /**
- * `vague` rule: nothing else fired with high confidence — input is ambiguous.
- *
- * Returns a moderate baseline so `vague` always shows up as an alternative, even when other rules dominate. The
- * coordinator decides whether to trust vague as the primary kind.
+ * Returns a low constant fallback score for ambiguous input.
  */
 export function scoreVague(_input, _shape) {
     return 0.3;

package/out/rules.js.map

@@ -1 +1 @@
-{"version":3,"file":"rules.js","sourceRoot":"","sources":["../lib/rules.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,sBAAsB,EAAE,uBAAuB,EAAE,MAAM,mCAAmC,CAAA;AAEnG,OAAO,EAAE,gBAAgB,EAAE,MAAM,sCAAsC,CAAA;AACvE;;;GAGG;AACH,MAAM,mBAAmB,GAAG,EAAE,CAAA;AAE9B;;GAEG;AACH,MAAM,wBAAwB,GAAG,EAAE,CAAA;AAEnC;;;GAGG;AACH,MAAM,qBAAqB,GAAG,GAAG,CAAA;AAEjC;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,EAAE,CAAA;AAE1C;;;GAGG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAA;AAEhC;;GAEG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAA;AAEhC;;GAEG;AACH,MAAM,2BAA2B,GAAG,CAAC,CAAA;AAErC;;;GAGG;AACH,MAAM,gCAAgC,GAAG,EAAE,CAAA;AAE3C;;GAEG;AACH,MAAM,gBAAgB,GAAG;IACxB,QAAQ;IACR,MAAM;IACN,aAAa;IACb,UAAU;IACV,SAAS;IACT,QAAQ;IACR,aAAa;IACb,UAAU;IACV,QAAQ;CACR,CAAA;AAED;;GAEG;AACH,MAAM,qBAAqB,GAAG;IAC7B,gBAAgB;IAChB,sBAAsB;IACtB,uBAAuB;IACvB,oCAAoC;IACpC,6FAA6F;CAC7F,CAAA;AAED;;;GAGG;AACH,MAAM,UAAU,UAAU,CAAC,MAA2B,EAAE,KAAqB;IAC5E,MAAM,GAAG,GAAG,KAAK,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAA;IAEjE,IAAI,CAAC,GAAG;QAAE,OAAO,CAAC,CAAA;IAElB,+FAA+F;IAC/F,qCAAqC;IACrC,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,UAAU,GAAG,GAAG,CAAC,CAAA;AACzC,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,MAAsB;IACnF,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAA;IAE7B,KAAK,MAAM,OAAO,IAAI,qBAAqB,EAAE,CAAC;QAC7C,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,IAAI,CAAA;IACpC,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,aAAa,CAAC,KAA0B,EAAE,MAAsB;IAC/E,MAAM,EAAE,GAAG,KAAK,CAAC,UAAU,CAAC,WAAW,EAAE,CAAC,IAAI,EAAE,CAAA;IAEhD,KAAK,MAAM,MAAM,IAAI,gBAAgB,EAAE,CAAC;QACvC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,GAAG,GAAG,CAAC,IAAI,EAAE,KAAK,MAAM;YAAE,OAAO,GAAG,CAAA;IAC7D,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,OAAO,CAAC,IAAY;IACnC,OAAO,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;AAC9D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,2BAA2B,CAAC,IAAY;IACvD,MAAM,SAAS,GAAG,uBAAuB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAA;IAExE,OAAO,SAAS,KAAK,SAAS,IAAI,CAAC,sBAAsB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;AACzE,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,KAA0B,EAAE,KAAqB;IACnF,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,IAAI,EAAE,CAAA;IACpC,MAAM,GAAG,GAAG,IAAI,CAAC,MAAM,CAAA;IAEvB,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,mBAAmB;QAAE,OAAO,CAAC,CAAA;IAEpD,2CAA2C;IAC3C,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,CAAC,CAAA;IAEjC,kDAAkD;IAClD,IAAI,KAAK,CAAC,YAAY,CAAC,MAAM;QAAE,OAAO,CAAC,CAAA;IAEvC,6EAA6E;IAC7E,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,IAAI,QAAQ,GAAG,CAAC;QAAE,OAAO,CAAC,CAAA;IAE1B,sDAAsD;IACtD,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,CAAA;IAE3B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACvB,IAAI,2BAA2B,CAAC,CAAC,CAAC;YAAE,OAAO,CAAC,CAAA;IAC7C,CAAC;IAED,6EAA6E;IAC7E,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,CAAC,CAAA;IAEjC,2FAA2F;IAC3F,MAAM,iBAAiB,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEjE,mEAAmE;IACnE,MAAM,aAAa,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;IAE9E,kEAAkE;IAClE,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,CAAA;IAE9B,IAAI,SAAS,IAAI,sBAAsB,IAAI,SAAS,IAAI,sBAAsB,IAAI,QAAQ,KAAK,CAAC,EAAE,CAAC;QAClG,IAAI,iBAAiB;YAAE,OAAO,IAAI,CAAA;QAElC,IAAI,aAAa;YAAE,OAAO,IAAI,CAAA;QAE9B,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,qEAAqE;IACrE,IAAI,SAAS,IAAI,2BAA2B,IAAI,QAAQ,KAAK,CAAC,IAAI,aAAa,EAAE,CAAC;QACjF,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,KAAqB;IAClF,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAA;IAEnC,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,wBAAwB;QAAE,OAAO,CAAC,CAAA;IACzD,MAAM,WAAW,GAAG,KAAK,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAA;IAE9E,IAAI,CAAC,WAAW;QAAE,OAAO,CAAC,CAAA;IAC1B,MAAM,MAAM,GAAG,WAAW,CAAC,IAAI,CAAC,GAAG,GAAG,WAAW,CAAC,IAAI,CAAC,KAAK,CAAA;IAE5D,mFAAmF;IACnF,IAAI,MAAM,GAAG,GAAG,GAAG,qBAAqB;QAAE,OAAO,CAAC,CAAA;IAElD,qGAAqG;IACrG,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,WAAW,CAAC,UAAU,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,GAAG,GAAG,CAAC,CAAA;AAClE,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,KAAqB;IAClF,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAA;IAEnC,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,wBAAwB;QAAE,OAAO,CAAC,CAAA;IAEzD,IAAI,KAAK,CAAC,cAAc,KAAK,OAAO;QAAE,OAAO,CAAC,CAAA;IAE9C,IAAI,KAAK,CAAC,YAAY,CAAC,MAAM;QAAE,OAAO,CAAC,CAAA;IACvC,qGAAqG;IACrG,6DAA6D;IAC7D,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,IAAI,QAAQ,GAAG,CAAC;QAAE,OAAO,CAAC,CAAA;IAE1B,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,sBAAsB,CAAC,KAA0B,EAAE,KAAqB;IACvF,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAA;IAEnC,IAAI,GAAG,KAAK,CAAC;QAAE,OAAO,CAAC,CAAA;IACvB,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,+EAA+E;IAC/E,IAAI,QAAQ,IAAI,CAAC,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,OAAO,GAAG,CAAA;IAExE,6EAA6E;IAC7E,IAAI,GAAG,IAAI,gCAAgC,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,OAAO,IAAI,CAAA;IAEnG,4EAA4E;IAC5E,IAAI,QAAQ,IAAI,CAAC;QAAE,OAAO,GAAG,CAAA;IAE7B,wFAAwF;IACxF,IAAI,GAAG,GAAG,gCAAgC,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,OAAO,GAAG,CAAA;IAEjG,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,UAAU,CAAC,MAA2B,EAAE,MAAsB;IAC7E,OAAO,GAAG,CAAA;AACX,CAAC"}
+{"version":3,"file":"rules.js","sourceRoot":"","sources":["../lib/rules.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAE,sBAAsB,EAAE,uBAAuB,EAAE,MAAM,mCAAmC,CAAA;AAEnG,OAAO,EAAE,cAAc,EAAE,cAAc,EAAE,MAAM,wCAAwC,CAAA;AACvF,OAAO,EAAE,gBAAgB,EAAE,MAAM,sCAAsC,CAAA;AACvE;;GAEG;AACH,MAAM,mBAAmB,GAAG,EAAE,CAAA;AAE9B;;GAEG;AACH,MAAM,wBAAwB,GAAG,EAAE,CAAA;AAEnC;;GAEG;AACH,MAAM,qBAAqB,GAAG,GAAG,CAAA;AAEjC;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,EAAE,CAAA;AAE1C;;GAEG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAA;AAEhC;;GAEG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAA;AAEhC;;GAEG;AACH,MAAM,2BAA2B,GAAG,CAAC,CAAA;AAErC;;GAEG;AACH,MAAM,gCAAgC,GAAG,EAAE,CAAA;AAE3C;;GAEG;AACH,MAAM,gBAAgB,GAAG;IACxB,QAAQ;IACR,MAAM;IACN,aAAa;IACb,UAAU;IACV,SAAS;IACT,QAAQ;IACR,aAAa;IACb,UAAU;IACV,QAAQ;CACR,CAAA;AAED;;GAEG;AACH,MAAM,qBAAqB,GAAG;IAC7B,gBAAgB;IAChB,sBAAsB;IACtB,uBAAuB;IACvB,+CAA+C;IAC/C,6FAA6F;CAC7F,CAAA;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,MAA2B,EAAE,KAAqB;IAC5E,MAAM,GAAG,GAAG,KAAK,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAA;IAEjE,IAAI,CAAC,GAAG;QAAE,OAAO,CAAC,CAAA;IAElB,gEAAgE;IAChE,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,UAAU,GAAG,GAAG,CAAC,CAAA;AACzC,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,MAAsB;IACnF,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAA;IAE7B,KAAK,MAAM,OAAO,IAAI,qBAAqB,EAAE,CAAC;QAC7C,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,IAAI,CAAA;IACpC,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,aAAa,CAAC,KAA0B,EAAE,MAAsB;IAC/E,MAAM,EAAE,GAAG,KAAK,CAAC,UAAU,CAAC,WAAW,EAAE,CAAC,IAAI,EAAE,CAAA;IAEhD,KAAK,MAAM,MAAM,IAAI,gBAAgB,EAAE,CAAC;QACvC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,GAAG,GAAG,CAAC,IAAI,EAAE,KAAK,MAAM;YAAE,OAAO,GAAG,CAAA;IAC7D,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,OAAO,CAAC,IAAY;IACnC,OAAO,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;AAC1D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,2BAA2B,CAAC,IAAY;IACvD,MAAM,SAAS,GAAG,uBAAuB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAA;IAExE,OAAO,SAAS,KAAK,SAAS,IAAI,CAAC,sBAAsB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;AACzE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAY,EAAE,KAAqB;IACvE,MAAM,MAAM,GAAG,mBAAmB,CAAC,KAAK,CAAC,CAAA;IAEzC,IAAI,CAAC,MAAM,CAAC,MAAM;QAAE,OAAO,IAAI,CAAA;IAE/B,IAAI,SAAS,GAAG,IAAI,CAAA;IAEpB,6CAA6C;IAC7C,gDAAgD;IAChD,KAAK,IAAI,KAAK,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,KAAK,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC;QACzD,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAE,CAAA;QAE3B,SAAS,GAAG,SAAS,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,GAAG,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;IACvE,CAAC;IAED,OAAO,SAAS;SACd,UAAU,CAAC,UAAU,EAAE,GAAG,CAAC;SAC3B,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC;SACvB,IAAI,EAAE,CAAA;AACT,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACzC,OAAO,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AAC3B,CAAC;AAED;;GAEG;AACH,SAAS,mBAAmB,CAAC,KAAqB;IACjD,6DAA6D;IAC7D,MAAM,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,CAAA;IAEzC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAA;IAEpB,MAAM,MAAM,GAA0C,EAAE,CAAA;IAExD,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,YAAY,EAAE,CAAC;QACtC,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,MAAM,CAAC;YAAE,SAAQ;QAE3C,IAAI,GAAG,CAAC,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG;YAAE,SAAQ;QAEpE,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,GAAG,EAAE,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,CAAA;IAC1D,CAAC;IAED,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,MAAM,CAAA;IAEpC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAA;IAEtD,MAAM,QAAQ,GAA0C,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,CAAA;IAEpE,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAC;QACpC,MAAM,IAAI,GAAG,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAE,CAAA;QAE7B,IAAI,IAAI,CAAC,KAAK,IAAI,IAAI,CAAC,GAAG,EAAE,CAAC;YAC5B,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,CAAA;YAEvC,SAAQ;QACT,CAAC;QAED,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACpB,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,KAA0B,EAAE,KAAqB;IACnF,MAAM,IAAI,GAAG,KAAK,CAAC,UAAU,CAAC,IAAI,EAAE,CAAA;IACpC,MAAM,GAAG,GAAG,IAAI,CAAC,MAAM,CAAA;IAEvB,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,mBAAmB;QAAE,OAAO,CAAC,CAAA;IAEpD,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,CAAC,CAAA;IAEjC,IAAI,KAAK,CAAC,YAAY,CAAC,MAAM;QAAE,OAAO,CAAC,CAAA;IAEvC,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,IAAI,QAAQ,GAAG,CAAC;QAAE,OAAO,CAAC,CAAA;IAE1B,MAAM,KAAK,GAAG,OAAO,CAAC,IAAI,CAAC,CAAA;IAE3B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;QACvB,IAAI,2BAA2B,CAAC,CAAC,CAAC;YAAE,OAAO,CAAC,CAAA;IAC7C,CAAC;IAED,8CAA8C;IAC9C,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,CAAC,CAAA;IAEjC,2DAA2D;IAC3D,MAAM,iBAAiB,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEjE,MAAM,aAAa,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;IAE9E,MAAM,SAAS,GAAG,KAAK,CAAC,MAAM,CAAA;IAE9B,IAAI,SAAS,IAAI,sBAAsB,IAAI,SAAS,IAAI,sBAAsB,IAAI,QAAQ,KAAK,CAAC,EAAE,CAAC;QAClG,IAAI,iBAAiB;YAAE,OAAO,IAAI,CAAA;QAElC,IAAI,aAAa;YAAE,OAAO,IAAI,CAAA;QAE9B,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,IAAI,SAAS,IAAI,2BAA2B,IAAI,QAAQ,KAAK,CAAC,IAAI,aAAa,EAAE,CAAC;QACjF,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,KAAqB;IAClF,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAA;IAEnC,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,wBAAwB;QAAE,OAAO,CAAC,CAAA;IACzD,MAAM,WAAW,GAAG,KAAK,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAA;IAE9E,IAAI,CAAC,WAAW;QAAE,OAAO,CAAC,CAAA;IAC1B,MAAM,MAAM,GAAG,WAAW,CAAC,IAAI,CAAC,GAAG,GAAG,WAAW,CAAC,IAAI,CAAC,KAAK,CAAA;IAE5D,IAAI,MAAM,GAAG,GAAG,GAAG,qBAAqB;QAAE,OAAO,CAAC,CAAA;IAElD,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,WAAW,CAAC,UAAU,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,GAAG,GAAG,CAAC,CAAA;AAClE,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAA0B,EAAE,KAAqB;IAClF,yEAAyE;IACzE,IAAI,eAAe,GAAG,KAAK,CAAA;IAE3B,KAAK,MAAM,GAAG,IAAI,KAAK,CAAC,YAAY,EAAE,CAAC;QACtC,IAAI,gBAAgB,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;YAClC,eAAe,GAAG,IAAI,CAAA;YAEtB,SAAQ;QACT,CAAC;QAED,OAAO,CAAC,CAAA;IACT,CAAC;IAED,0DAA0D;IAC1D,MAAM,SAAS,GAAG,eAAe,CAAC,CAAC,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAEnE,IAAI,eAAe,IAAI,CAAC,SAAS,CAAC,MAAM;QAAE,OAAO,CAAC,CAAA;IAElD,MAAM,eAAe,GAAG,eAAe,CAAC,CAAC,CAAC,oBAAoB,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,UAAU,CAAA;IAC1G,MAAM,GAAG,GAAG,eAAe,CAAC,MAAM,CAAA;IAElC,IAAI,GAAG,KAAK,CAAC,IAAI,GAAG,GAAG,wBAAwB;QAAE,OAAO,CAAC,CAAA;IAEzD,IAAI,CAAC,aAAa,CAAC,eAAe,CAAC;QAAE,OAAO,CAAC,CAAA;IAE7C,yCAAyC;IACzC,6DAA6D;IAC7D,IAAI,KAAK,CAAC,cAAc,KAAK,OAAO,EAAE,CAAC;QACtC,IAAI,CAAC,eAAe;YAAE,OAAO,CAAC,CAAA;QAE9B,IAAI,cAAc,CAAC,cAAc,CAAC,eAAe,CAAC,CAAC,KAAK,OAAO;YAAE,OAAO,CAAC,CAAA;IAC1E,CAAC;IAED,kEAAkE;IAClE,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,IAAI,QAAQ,GAAG,CAAC;QAAE,OAAO,CAAC,CAAA;IAE1B,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,sBAAsB,CAAC,KAA0B,EAAE,KAAqB;IACvF,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,CAAA;IAEnC,IAAI,GAAG,KAAK,CAAC;QAAE,OAAO,CAAC,CAAA;IACvB,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,MAAM,IAAI,CAAC,CAAA;IAE5C,+FAA+F;IAC/F,IACC,QAAQ,IAAI,CAAC;QACb,KAAK,CAAC,cAAc,KAAK,cAAc;QACvC,KAAK,CAAC,IAAI,CAAC,oBAAoB,CAAC,KAAK,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC,EACxD,CAAC;QACF,OAAO,GAAG,CAAA;IACX,CAAC;IAED,IAAI,GAAG,IAAI,gCAAgC,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,OAAO,IAAI,CAAA;IAEnG,6FAA6F;IAC7F,IAAI,QAAQ,IAAI,CAAC;QAAE,OAAO,GAAG,CAAA;IAE7B,IAAI,GAAG,GAAG,gCAAgC,IAAI,KAAK,CAAC,cAAc,KAAK,cAAc;QAAE,OAAO,GAAG,CAAA;IAEjG,OAAO,CAAC,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,MAA2B,EAAE,MAAsB;IAC7E,OAAO,GAAG,CAAA;AACX,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mailwoman/kind-classifier",
-  "version": "10.0.0",
+  "version": "10.1.0",
   "description": "Stage 2.5 of the runtime pipeline — categorize inputs by query shape (postcode_only / locality_only / structured_address / intersection / po_box / landmark / vague). Rule-based for v1.",
   "license": "AGPL-3.0-only OR LicenseRef-Commercial",
   "repository": {
@@ -13,15 +13,10 @@
     "out/**/*.js.map",
     "out/**/*.d.ts",
     "out/**/*.d.ts.map",
-    "*.ts",
-    "*.tsx",
-    "**/*.ts",
-    "**/*.tsx",
-    "!*.test.ts",
-    "!*.test.tsx",
+    "lib/**/*.ts",
+    "lib/**/*.tsx",
     "!**/*.test.ts",
-    "!**/*.test.tsx",
-    "!test/**"
+    "!**/*.test.tsx"
   ],
   "type": "module",
   "sideEffects": false,
@@ -39,25 +34,10 @@
       "node": "./out/index.js",
       "default": "./out/index.js"
     },
-    "./classify": {
-      "types": "./out/classify.d.ts",
-      "node": "./out/classify.js",
-      "default": "./out/classify.js"
-    },
-    "./intent-rules": {
-      "types": "./out/intent/rules.d.ts",
-      "node": "./out/intent/rules.js",
-      "default": "./out/intent/rules.js"
-    },
-    "./rules": {
-      "types": "./out/rules.d.ts",
-      "node": "./out/rules.js",
-      "default": "./out/rules.js"
-    },
-    "./poi": {
-      "types": "./out/poi.d.ts",
-      "node": "./out/poi.js",
-      "default": "./out/poi.js"
+    "./*": {
+      "types": "./out/*.d.ts",
+      "node": "./out/*.js",
+      "default": "./out/*.js"
     }
   },
   "publishConfig": {
@@ -69,25 +49,10 @@
         "node": "./out/index.js",
         "default": "./out/index.js"
       },
-      "./classify": {
-        "types": "./out/classify.d.ts",
-        "node": "./out/classify.js",
-        "default": "./out/classify.js"
-      },
-      "./intent-rules": {
-        "types": "./out/intent/rules.d.ts",
-        "node": "./out/intent/rules.js",
-        "default": "./out/intent/rules.js"
-      },
-      "./rules": {
-        "types": "./out/rules.d.ts",
-        "node": "./out/rules.js",
-        "default": "./out/rules.js"
-      },
-      "./poi": {
-        "types": "./out/poi.d.ts",
-        "node": "./out/poi.js",
-        "default": "./out/poi.js"
+      "./*": {
+        "types": "./out/*.d.ts",
+        "node": "./out/*.js",
+        "default": "./out/*.js"
       }
     },
     "imports": {
@@ -99,8 +64,8 @@
     }
   },
   "dependencies": {
-    "@mailwoman/codex": "10.0.0",
-    "@mailwoman/core": "10.0.0",
-    "@mailwoman/query-shape": "10.0.0"
+    "@mailwoman/codex": "10.1.0",
+    "@mailwoman/core": "10.1.0",
+    "@mailwoman/query-shape": "10.1.0"
   }
 }