@mailwoman/query-shape 4.0.0

package/out/character-class.d.ts

@@ -0,0 +1,23 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { CharacterClass, SpanRange, TokenCharacterClass, TokenClass } from "./types.js";
+/** Codepoint-level character class. */
+export type CodepointClass = TokenCharacterClass | "whitespace" | "connector" | "other";
+/** Classify a single Unicode codepoint. */
+export declare function classifyCodepoint(cp: number): CodepointClass;
+/**
+ * Classify a token by walking its codepoints and folding to the dominant class. Mixed alphanumeric
+ * (e.g. `"221B"`, `"10118-1234"`) returns `"mixed"`. Pure-punct tokens return `"punct"`.
+ */
+export declare function classifyToken(text: string): TokenCharacterClass;
+/** Fold per-token classes into the whole-input character class. */
+export declare function foldInputClass(tokens: ReadonlyArray<TokenClass>): CharacterClass;
+/**
+ * Walk a string and emit token spans (whitespace-and-punctuation-separated). Internal helper —
+ * callers receive `TokenClass[]` from `computeQueryShape`.
+ */
+export declare function tokenizeForClass(text: string): SpanRange[];
+//# sourceMappingURL=character-class.d.ts.map

package/out/character-class.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"character-class.d.ts","sourceRoot":"","sources":["../character-class.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,cAAc,EAAE,SAAS,EAAE,mBAAmB,EAAE,UAAU,EAAE,MAAM,YAAY,CAAA;AAE5F,uCAAuC;AACvC,MAAM,MAAM,cAAc,GAAG,mBAAmB,GAAG,YAAY,GAAG,WAAW,GAAG,OAAO,CAAA;AAuFvF,2CAA2C;AAC3C,wBAAgB,iBAAiB,CAAC,EAAE,EAAE,MAAM,GAAG,cAAc,CAY5D;AAED;;;GAGG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,mBAAmB,CA8C/D;AAED,mEAAmE;AACnE,wBAAgB,cAAc,CAAC,MAAM,EAAE,aAAa,CAAC,UAAU,CAAC,GAAG,cAAc,CAyChF;AAED;;;GAGG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS,EAAE,CAmD1D"}

package/out/character-class.js

@@ -0,0 +1,262 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+const CJK_RANGES = [
+    [0x3040, 0x30ff], // Hiragana + Katakana
+    [0x31f0, 0x31ff], // Katakana phonetic extensions
+    [0x3400, 0x4dbf], // CJK Unified Ideographs Extension A
+    [0x4e00, 0x9fff], // CJK Unified Ideographs
+    [0xa000, 0xa4cf], // Yi
+    [0xac00, 0xd7af], // Hangul Syllables
+    [0xf900, 0xfaff], // CJK Compatibility Ideographs
+    [0xff00, 0xffef], // Halfwidth + Fullwidth forms
+    [0x20000, 0x2a6df], // CJK Unified Ideographs Extension B
+];
+const CYRILLIC_RANGES = [
+    [0x0400, 0x04ff],
+    [0x0500, 0x052f], // Cyrillic Supplement
+    [0x2de0, 0x2dff], // Cyrillic Extended-A
+    [0xa640, 0xa69f], // Cyrillic Extended-B
+];
+const ARABIC_RANGES = [
+    [0x0600, 0x06ff],
+    [0x0750, 0x077f], // Arabic Supplement
+    [0x08a0, 0x08ff], // Arabic Extended-A
+    [0xfb50, 0xfdff], // Arabic Presentation Forms-A
+    [0xfe70, 0xfeff], // Arabic Presentation Forms-B
+];
+function inRange(cp, ranges) {
+    for (const [lo, hi] of ranges) {
+        if (cp >= lo && cp <= hi)
+            return true;
+    }
+    return false;
+}
+const PUNCT_CODEPOINTS = new Set([
+    0x21, // !
+    0x22, // "
+    0x23, // #
+    0x25, // %
+    0x26, // &
+    0x28, // (
+    0x29, // )
+    0x2a, // *
+    0x2b, // +
+    0x2c, // ,
+    0x2e, // .
+    0x2f, // /
+    0x3a, // :
+    0x3b, // ;
+    0x3c, // <
+    0x3d, // =
+    0x3e, // >
+    0x3f, // ?
+    0x40, // @
+    0x5b, // [
+    0x5c, // \
+    0x5d, // ]
+    0x5e, // ^
+    0x60, // `
+    0x7b, // {
+    0x7c, // |
+    0x7d, // }
+    0x7e, // ~
+    0x00a1, // ¡
+    0x00bf, // ¿
+    0x201c, // “
+    0x201d, // ”
+    0x2013, // –
+    0x2014, // —
+    0x3001, // 、 (CJK comma)
+    0x3002, // 。 (CJK period)
+]);
+/**
+ * "Connector" codepoints join adjacent tokens instead of separating them. Hyphen, apostrophe,
+ * underscore — surface in "10118-1234", "O'Brien", "Saint-Denis", and similar.
+ */
+const CONNECTOR_CODEPOINTS = new Set([
+    0x2d, // -
+    0x27, // '
+    0x5f, // _
+    0x2018, // ‘
+    0x2019, // ’
+]);
+/** Classify a single Unicode codepoint. */
+export function classifyCodepoint(cp) {
+    if (cp >= 0x30 && cp <= 0x39)
+        return "digit";
+    if ((cp >= 0x41 && cp <= 0x5a) || (cp >= 0x61 && cp <= 0x7a))
+        return "alpha";
+    // Latin-1 letters with diacritics + Latin Extended-A/B
+    if ((cp >= 0x00c0 && cp <= 0x024f) || (cp >= 0x1e00 && cp <= 0x1eff))
+        return "alpha";
+    if (cp === 0x20 || cp === 0x09 || cp === 0x0a || cp === 0x0d || cp === 0xa0)
+        return "whitespace";
+    if (CONNECTOR_CODEPOINTS.has(cp))
+        return "connector";
+    if (PUNCT_CODEPOINTS.has(cp))
+        return "punct";
+    if (inRange(cp, CJK_RANGES))
+        return "cjk";
+    if (inRange(cp, CYRILLIC_RANGES))
+        return "cyrillic";
+    if (inRange(cp, ARABIC_RANGES))
+        return "arabic";
+    return "other";
+}
+/**
+ * Classify a token by walking its codepoints and folding to the dominant class. Mixed alphanumeric
+ * (e.g. `"221B"`, `"10118-1234"`) returns `"mixed"`. Pure-punct tokens return `"punct"`.
+ */
+export function classifyToken(text) {
+    let hasDigit = false;
+    let hasAlpha = false;
+    let hasCjk = false;
+    let hasCyrillic = false;
+    let hasArabic = false;
+    let hasPunct = false;
+    for (let i = 0; i < text.length;) {
+        const cp = text.codePointAt(i);
+        i += cp > 0xffff ? 2 : 1;
+        const cls = classifyCodepoint(cp);
+        switch (cls) {
+            case "digit":
+                hasDigit = true;
+                break;
+            case "alpha":
+                hasAlpha = true;
+                break;
+            case "cjk":
+                hasCjk = true;
+                break;
+            case "cyrillic":
+                hasCyrillic = true;
+                break;
+            case "arabic":
+                hasArabic = true;
+                break;
+            case "punct":
+                hasPunct = true;
+                break;
+            case "connector":
+            case "whitespace":
+            case "other":
+                break;
+        }
+    }
+    if (hasCjk)
+        return "cjk";
+    if (hasCyrillic)
+        return "cyrillic";
+    if (hasArabic)
+        return "arabic";
+    if (hasDigit && hasAlpha)
+        return "mixed";
+    if (hasDigit)
+        return "digit";
+    if (hasAlpha)
+        return "alpha";
+    if (hasPunct)
+        return "punct";
+    return "mixed";
+}
+/** Fold per-token classes into the whole-input character class. */
+export function foldInputClass(tokens) {
+    if (tokens.length === 0)
+        return "alpha";
+    let hasDigit = false;
+    let hasAlpha = false;
+    let hasCjk = false;
+    let hasCyrillic = false;
+    let hasArabic = false;
+    let hasMixed = false;
+    for (const t of tokens) {
+        switch (t.class) {
+            case "cjk":
+                hasCjk = true;
+                break;
+            case "cyrillic":
+                hasCyrillic = true;
+                break;
+            case "arabic":
+                hasArabic = true;
+                break;
+            case "digit":
+                hasDigit = true;
+                break;
+            case "alpha":
+                hasAlpha = true;
+                break;
+            case "mixed":
+                hasMixed = true;
+                break;
+        }
+    }
+    if (hasCjk && !hasAlpha && !hasCyrillic && !hasArabic)
+        return "cjk";
+    if (hasCyrillic && !hasAlpha && !hasCjk && !hasArabic)
+        return "cyrillic";
+    if (hasArabic && !hasAlpha && !hasCjk && !hasCyrillic)
+        return "arabic";
+    if (hasCjk || hasCyrillic || hasArabic)
+        return "mixed";
+    if (hasMixed || (hasDigit && hasAlpha))
+        return "alphanumeric";
+    if (hasDigit && !hasAlpha)
+        return "numeric";
+    if (hasAlpha && !hasDigit)
+        return "alpha";
+    return "mixed";
+}
+/**
+ * Walk a string and emit token spans (whitespace-and-punctuation-separated). Internal helper —
+ * callers receive `TokenClass[]` from `computeQueryShape`.
+ */
+export function tokenizeForClass(text) {
+    const tokens = [];
+    let i = 0;
+    const N = text.length;
+    while (i < N) {
+        const cp = text.codePointAt(i);
+        const cls = classifyCodepoint(cp);
+        if (cls === "whitespace" || cls === "punct") {
+            i += cp > 0xffff ? 2 : 1;
+            continue;
+        }
+        // A leading connector (rare — most inputs don't start with `-`/`'`) is consumed as whitespace.
+        if (cls === "connector") {
+            i += cp > 0xffff ? 2 : 1;
+            continue;
+        }
+        // Start a token at i; walk until we hit whitespace, punct, or a script boundary.
+        // Connectors (`-`, `'`, `_`) join across digit/alpha boundaries.
+        const start = i;
+        const startCls = cls;
+        let cur = i;
+        while (cur < N) {
+            const ncp = text.codePointAt(cur);
+            const nstep = ncp > 0xffff ? 2 : 1;
+            const ncls = classifyCodepoint(ncp);
+            if (ncls === "whitespace" || ncls === "punct")
+                break;
+            if (ncls === "connector") {
+                cur += nstep;
+                continue;
+            }
+            // Break tokens across script transitions (digit↔alpha is fine; alpha↔cjk is a boundary).
+            const isLatinPair = (a, b) => (a === "digit" || a === "alpha") && (b === "digit" || b === "alpha");
+            if (ncls !== startCls &&
+                !isLatinPair(startCls, ncls) &&
+                !(startCls === "other" && (ncls === "digit" || ncls === "alpha"))) {
+                break;
+            }
+            cur += nstep;
+        }
+        tokens.push({ start, end: cur, body: text.slice(start, cur) });
+        i = cur;
+    }
+    return tokens;
+}
+//# sourceMappingURL=character-class.js.map

package/out/character-class.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"character-class.js","sourceRoot":"","sources":["../character-class.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAOH,MAAM,UAAU,GAAoC;IACnD,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,sBAAsB;IACxC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,+BAA+B;IACjD,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,qCAAqC;IACvD,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,yBAAyB;IAC3C,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,KAAK;IACvB,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,mBAAmB;IACrC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,+BAA+B;IACjD,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,8BAA8B;IAChD,CAAC,OAAO,EAAE,OAAO,CAAC,EAAE,qCAAqC;CACzD,CAAA;AAED,MAAM,eAAe,GAAoC;IACxD,CAAC,MAAM,EAAE,MAAM,CAAC;IAChB,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,sBAAsB;IACxC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,sBAAsB;IACxC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,sBAAsB;CACxC,CAAA;AAED,MAAM,aAAa,GAAoC;IACtD,CAAC,MAAM,EAAE,MAAM,CAAC;IAChB,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,oBAAoB;IACtC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,oBAAoB;IACtC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,8BAA8B;IAChD,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,8BAA8B;CAChD,CAAA;AAED,SAAS,OAAO,CAAC,EAAU,EAAE,MAAuC;IACnE,KAAK,MAAM,CAAC,EAAE,EAAE,EAAE,CAAC,IAAI,MAAM,EAAE,CAAC;QAC/B,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE;YAAE,OAAO,IAAI,CAAA;IACtC,CAAC;IACD,OAAO,KAAK,CAAA;AACb,CAAC;AAED,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAS;IACxC,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,gBAAgB;IACxB,MAAM,EAAE,iBAAiB;CACzB,CAAC,CAAA;AAEF;;;GAGG;AACH,MAAM,oBAAoB,GAAG,IAAI,GAAG,CAAS;IAC5C,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,IAAI,EAAE,IAAI;IACV,MAAM,EAAE,IAAI;IACZ,MAAM,EAAE,IAAI;CACZ,CAAC,CAAA;AAEF,2CAA2C;AAC3C,MAAM,UAAU,iBAAiB,CAAC,EAAU;IAC3C,IAAI,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI;QAAE,OAAO,OAAO,CAAA;IAC5C,IAAI,CAAC,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI,CAAC;QAAE,OAAO,OAAO,CAAA;IAC5E,uDAAuD;IACvD,IAAI,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC;QAAE,OAAO,OAAO,CAAA;IACpF,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,IAAI;QAAE,OAAO,YAAY,CAAA;IAChG,IAAI,oBAAoB,CAAC,GAAG,CAAC,EAAE,CAAC;QAAE,OAAO,WAAW,CAAA;IACpD,IAAI,gBAAgB,CAAC,GAAG,CAAC,EAAE,CAAC;QAAE,OAAO,OAAO,CAAA;IAC5C,IAAI,OAAO,CAAC,EAAE,EAAE,UAAU,CAAC;QAAE,OAAO,KAAK,CAAA;IACzC,IAAI,OAAO,CAAC,EAAE,EAAE,eAAe,CAAC;QAAE,OAAO,UAAU,CAAA;IACnD,IAAI,OAAO,CAAC,EAAE,EAAE,aAAa,CAAC;QAAE,OAAO,QAAQ,CAAA;IAC/C,OAAO,OAAO,CAAA;AACf,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,aAAa,CAAC,IAAY;IACzC,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,MAAM,GAAG,KAAK,CAAA;IAClB,IAAI,WAAW,GAAG,KAAK,CAAA;IACvB,IAAI,SAAS,GAAG,KAAK,CAAA;IACrB,IAAI,QAAQ,GAAG,KAAK,CAAA;IAEpB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,GAAI,CAAC;QACnC,MAAM,EAAE,GAAG,IAAI,CAAC,WAAW,CAAC,CAAC,CAAE,CAAA;QAC/B,CAAC,IAAI,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;QACxB,MAAM,GAAG,GAAG,iBAAiB,CAAC,EAAE,CAAC,CAAA;QACjC,QAAQ,GAAG,EAAE,CAAC;YACb,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;YACN,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;YACN,KAAK,KAAK;gBACT,MAAM,GAAG,IAAI,CAAA;gBACb,MAAK;YACN,KAAK,UAAU;gBACd,WAAW,GAAG,IAAI,CAAA;gBAClB,MAAK;YACN,KAAK,QAAQ;gBACZ,SAAS,GAAG,IAAI,CAAA;gBAChB,MAAK;YACN,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;YACN,KAAK,WAAW,CAAC;YACjB,KAAK,YAAY,CAAC;YAClB,KAAK,OAAO;gBACX,MAAK;QACP,CAAC;IACF,CAAC;IAED,IAAI,MAAM;QAAE,OAAO,KAAK,CAAA;IACxB,IAAI,WAAW;QAAE,OAAO,UAAU,CAAA;IAClC,IAAI,SAAS;QAAE,OAAO,QAAQ,CAAA;IAC9B,IAAI,QAAQ,IAAI,QAAQ;QAAE,OAAO,OAAO,CAAA;IACxC,IAAI,QAAQ;QAAE,OAAO,OAAO,CAAA;IAC5B,IAAI,QAAQ;QAAE,OAAO,OAAO,CAAA;IAC5B,IAAI,QAAQ;QAAE,OAAO,OAAO,CAAA;IAC5B,OAAO,OAAO,CAAA;AACf,CAAC;AAED,mEAAmE;AACnE,MAAM,UAAU,cAAc,CAAC,MAAiC;IAC/D,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,OAAO,CAAA;IAEvC,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,MAAM,GAAG,KAAK,CAAA;IAClB,IAAI,WAAW,GAAG,KAAK,CAAA;IACvB,IAAI,SAAS,GAAG,KAAK,CAAA;IACrB,IAAI,QAAQ,GAAG,KAAK,CAAA;IAEpB,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;QACxB,QAAQ,CAAC,CAAC,KAAK,EAAE,CAAC;YACjB,KAAK,KAAK;gBACT,MAAM,GAAG,IAAI,CAAA;gBACb,MAAK;YACN,KAAK,UAAU;gBACd,WAAW,GAAG,IAAI,CAAA;gBAClB,MAAK;YACN,KAAK,QAAQ;gBACZ,SAAS,GAAG,IAAI,CAAA;gBAChB,MAAK;YACN,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;YACN,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;YACN,KAAK,OAAO;gBACX,QAAQ,GAAG,IAAI,CAAA;gBACf,MAAK;QACP,CAAC;IACF,CAAC;IAED,IAAI,MAAM,IAAI,CAAC,QAAQ,IAAI,CAAC,WAAW,IAAI,CAAC,SAAS;QAAE,OAAO,KAAK,CAAA;IACnE,IAAI,WAAW,IAAI,CAAC,QAAQ,IAAI,CAAC,MAAM,IAAI,CAAC,SAAS;QAAE,OAAO,UAAU,CAAA;IACxE,IAAI,SAAS,IAAI,CAAC,QAAQ,IAAI,CAAC,MAAM,IAAI,CAAC,WAAW;QAAE,OAAO,QAAQ,CAAA;IACtE,IAAI,MAAM,IAAI,WAAW,IAAI,SAAS;QAAE,OAAO,OAAO,CAAA;IACtD,IAAI,QAAQ,IAAI,CAAC,QAAQ,IAAI,QAAQ,CAAC;QAAE,OAAO,cAAc,CAAA;IAC7D,IAAI,QAAQ,IAAI,CAAC,QAAQ;QAAE,OAAO,SAAS,CAAA;IAC3C,IAAI,QAAQ,IAAI,CAAC,QAAQ;QAAE,OAAO,OAAO,CAAA;IACzC,OAAO,OAAO,CAAA;AACf,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC5C,MAAM,MAAM,GAAgB,EAAE,CAAA;IAC9B,IAAI,CAAC,GAAG,CAAC,CAAA;IACT,MAAM,CAAC,GAAG,IAAI,CAAC,MAAM,CAAA;IAErB,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACd,MAAM,EAAE,GAAG,IAAI,CAAC,WAAW,CAAC,CAAC,CAAE,CAAA;QAC/B,MAAM,GAAG,GAAG,iBAAiB,CAAC,EAAE,CAAC,CAAA;QAEjC,IAAI,GAAG,KAAK,YAAY,IAAI,GAAG,KAAK,OAAO,EAAE,CAAC;YAC7C,CAAC,IAAI,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YACxB,SAAQ;QACT,CAAC;QACD,+FAA+F;QAC/F,IAAI,GAAG,KAAK,WAAW,EAAE,CAAC;YACzB,CAAC,IAAI,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YACxB,SAAQ;QACT,CAAC;QAED,iFAAiF;QACjF,iEAAiE;QACjE,MAAM,KAAK,GAAG,CAAC,CAAA;QACf,MAAM,QAAQ,GAAG,GAAG,CAAA;QACpB,IAAI,GAAG,GAAG,CAAC,CAAA;QACX,OAAO,GAAG,GAAG,CAAC,EAAE,CAAC;YAChB,MAAM,GAAG,GAAG,IAAI,CAAC,WAAW,CAAC,GAAG,CAAE,CAAA;YAClC,MAAM,KAAK,GAAG,GAAG,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAClC,MAAM,IAAI,GAAG,iBAAiB,CAAC,GAAG,CAAC,CAAA;YACnC,IAAI,IAAI,KAAK,YAAY,IAAI,IAAI,KAAK,OAAO;gBAAE,MAAK;YACpD,IAAI,IAAI,KAAK,WAAW,EAAE,CAAC;gBAC1B,GAAG,IAAI,KAAK,CAAA;gBACZ,SAAQ;YACT,CAAC;YACD,yFAAyF;YACzF,MAAM,WAAW,GAAG,CAAC,CAAiB,EAAE,CAAiB,EAAE,EAAE,CAC5D,CAAC,CAAC,KAAK,OAAO,IAAI,CAAC,KAAK,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,OAAO,IAAI,CAAC,KAAK,OAAO,CAAC,CAAA;YACrE,IACC,IAAI,KAAK,QAAQ;gBACjB,CAAC,WAAW,CAAC,QAAQ,EAAE,IAAI,CAAC;gBAC5B,CAAC,CAAC,QAAQ,KAAK,OAAO,IAAI,CAAC,IAAI,KAAK,OAAO,IAAI,IAAI,KAAK,OAAO,CAAC,CAAC,EAChE,CAAC;gBACF,MAAK;YACN,CAAC;YACD,GAAG,IAAI,KAAK,CAAA;QACb,CAAC;QAED,MAAM,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE,CAAC,CAAA;QAC9D,CAAC,GAAG,GAAG,CAAA;IACR,CAAC;IAED,OAAO,MAAM,CAAA;AACd,CAAC"}

package/out/compute.d.ts

@@ -0,0 +1,14 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { ComputeQueryShapeOpts, NormalizedInputLite, QueryShape } from "./types.js";
+/**
+ * Compute a `QueryShape` from a string or normalized input. Microseconds-cheap, pure-function.
+ *
+ * @example Const shape = computeQueryShape("350 5th Ave, New York, NY 10118") //
+ * shape.knownFormats.find((f) => f.format === "us_zip") → defined // shape.segments.length === 4
+ */
+export declare function computeQueryShape(input: string | NormalizedInputLite, opts?: ComputeQueryShapeOpts): QueryShape;
+//# sourceMappingURL=compute.d.ts.map

package/out/compute.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"compute.d.ts","sourceRoot":"","sources":["../compute.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAMH,OAAO,KAAK,EAAE,qBAAqB,EAAE,mBAAmB,EAAE,UAAU,EAAiC,MAAM,YAAY,CAAA;AA0BvH;;;;;GAKG;AACH,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,GAAG,mBAAmB,EAAE,IAAI,CAAC,EAAE,qBAAqB,GAAG,UAAU,CA0B/G"}

package/out/compute.js

@@ -0,0 +1,70 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { classifyToken, foldInputClass, tokenizeForClass } from "./character-class.js";
+import { detectKnownFormats } from "./known-formats.js";
+import { detectRegionAbbreviations } from "./region-abbreviations.js";
+import { segment } from "./segmentation.js";
+function detectWhitespacePattern(text) {
+    let hasTab = false;
+    let hasDouble = false;
+    let hasSingle = false;
+    let prevSpace = false;
+    for (let i = 0; i < text.length; i++) {
+        const ch = text[i];
+        if (ch === "\t") {
+            hasTab = true;
+        }
+        else if (ch === " ") {
+            if (prevSpace)
+                hasDouble = true;
+            else
+                hasSingle = true;
+            prevSpace = true;
+            continue;
+        }
+        prevSpace = false;
+    }
+    if (hasTab && (hasDouble || hasSingle))
+        return "mixed";
+    if (hasTab)
+        return "tab";
+    if (hasDouble)
+        return "double";
+    if (hasSingle)
+        return "single";
+    return "none";
+}
+/**
+ * Compute a `QueryShape` from a string or normalized input. Microseconds-cheap, pure-function.
+ *
+ * @example Const shape = computeQueryShape("350 5th Ave, New York, NY 10118") //
+ * shape.knownFormats.find((f) => f.format === "us_zip") → defined // shape.segments.length === 4
+ */
+export function computeQueryShape(input, opts) {
+    const text = typeof input === "string" ? input : input.normalized;
+    const locale = opts?.locale ?? (typeof input === "string" ? undefined : input.appliedLocale);
+    const tokenSpans = tokenizeForClass(text);
+    const tokenClasses = tokenSpans.map((span) => ({
+        span,
+        class: classifyToken(span.body),
+        length: span.end - span.start,
+    }));
+    const segments = segment(text, locale);
+    const knownFormats = detectKnownFormats(text, tokenClasses);
+    const regionAbbreviations = detectRegionAbbreviations(tokenClasses, segments);
+    const characterClass = foldInputClass(tokenClasses);
+    const whitespacePattern = detectWhitespacePattern(text);
+    return Object.freeze({
+        characterClass,
+        tokenClasses: Object.freeze(tokenClasses),
+        segments: Object.freeze(segments),
+        knownFormats: Object.freeze(knownFormats),
+        regionAbbreviations: Object.freeze(regionAbbreviations),
+        totalLength: text.length,
+        whitespacePattern,
+    });
+}
+//# sourceMappingURL=compute.js.map

package/out/compute.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"compute.js","sourceRoot":"","sources":["../compute.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,gBAAgB,EAAE,MAAM,sBAAsB,CAAA;AACtF,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAA;AACvD,OAAO,EAAE,yBAAyB,EAAE,MAAM,2BAA2B,CAAA;AACrE,OAAO,EAAE,OAAO,EAAE,MAAM,mBAAmB,CAAA;AAG3C,SAAS,uBAAuB,CAAC,IAAY;IAC5C,IAAI,MAAM,GAAG,KAAK,CAAA;IAClB,IAAI,SAAS,GAAG,KAAK,CAAA;IACrB,IAAI,SAAS,GAAG,KAAK,CAAA;IACrB,IAAI,SAAS,GAAG,KAAK,CAAA;IACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;QAClB,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YACjB,MAAM,GAAG,IAAI,CAAA;QACd,CAAC;aAAM,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACvB,IAAI,SAAS;gBAAE,SAAS,GAAG,IAAI,CAAA;;gBAC1B,SAAS,GAAG,IAAI,CAAA;YACrB,SAAS,GAAG,IAAI,CAAA;YAChB,SAAQ;QACT,CAAC;QACD,SAAS,GAAG,KAAK,CAAA;IAClB,CAAC;IACD,IAAI,MAAM,IAAI,CAAC,SAAS,IAAI,SAAS,CAAC;QAAE,OAAO,OAAO,CAAA;IACtD,IAAI,MAAM;QAAE,OAAO,KAAK,CAAA;IACxB,IAAI,SAAS;QAAE,OAAO,QAAQ,CAAA;IAC9B,IAAI,SAAS;QAAE,OAAO,QAAQ,CAAA;IAC9B,OAAO,MAAM,CAAA;AACd,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAAmC,EAAE,IAA4B;IAClG,MAAM,IAAI,GAAG,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,UAAU,CAAA;IACjE,MAAM,MAAM,GAAG,IAAI,EAAE,MAAM,IAAI,CAAC,OAAO,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,aAAa,CAAC,CAAA;IAE5F,MAAM,UAAU,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAA;IACzC,MAAM,YAAY,GAAiB,UAAU,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QAC5D,IAAI;QACJ,KAAK,EAAE,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC;QAC/B,MAAM,EAAE,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,KAAK;KAC7B,CAAC,CAAC,CAAA;IAEH,MAAM,QAAQ,GAAG,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;IACtC,MAAM,YAAY,GAAG,kBAAkB,CAAC,IAAI,EAAE,YAAY,CAAC,CAAA;IAC3D,MAAM,mBAAmB,GAAG,yBAAyB,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;IAC7E,MAAM,cAAc,GAAG,cAAc,CAAC,YAAY,CAAC,CAAA;IACnD,MAAM,iBAAiB,GAAG,uBAAuB,CAAC,IAAI,CAAC,CAAA;IAEvD,OAAO,MAAM,CAAC,MAAM,CAAC;QACpB,cAAc;QACd,YAAY,EAAE,MAAM,CAAC,MAAM,CAAC,YAAY,CAAiB;QACzD,QAAQ,EAAE,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAoB;QACpD,YAAY,EAAE,MAAM,CAAC,MAAM,CAAC,YAAY,CAAwB;QAChE,mBAAmB,EAAE,MAAM,CAAC,MAAM,CAAC,mBAAmB,CAA+B;QACrF,WAAW,EAAE,IAAI,CAAC,MAAM;QACxB,iBAAiB;KACjB,CAAsB,CAAA;AACxB,CAAC"}

package/out/index.d.ts

@@ -0,0 +1,21 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `@mailwoman/query-shape` — pure-function structural priors for the runtime pipeline.
+ *
+ *   Computes a `QueryShape` from an input string: character class, per-token class, punctuation-
+ *   bounded segments, known-format regex hits, and whitespace pattern. Microseconds-cheap, no ML,
+ *   no runtime dependencies.
+ *
+ *   See `docs/articles/plan/reference/QUERY_SHAPE.md` for the design rationale and
+ *   `docs/articles/plan/reference/STAGES.md` for how this fits into the runtime pipeline.
+ */
+export { classifyCodepoint, classifyToken, foldInputClass } from "./character-class.js";
+export { computeQueryShape } from "./compute.js";
+export { detectKnownFormats } from "./known-formats.js";
+export { detectRegionAbbreviations } from "./region-abbreviations.js";
+export { segment } from "./segmentation.js";
+export type { CharacterClass, ComputeQueryShapeOpts, KnownFormat, KnownFormatHit, NormalizedInputLite, QueryShape, RegionAbbreviationHit, Segment, SegmentSeparator, SpanRange, TokenCharacterClass, TokenClass, WhitespacePattern, } from "./types.js";
+//# sourceMappingURL=index.d.ts.map

package/out/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAA;AACvF,OAAO,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAA;AAChD,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAA;AACvD,OAAO,EAAE,yBAAyB,EAAE,MAAM,2BAA2B,CAAA;AACrE,OAAO,EAAE,OAAO,EAAE,MAAM,mBAAmB,CAAA;AAC3C,YAAY,EACX,cAAc,EACd,qBAAqB,EACrB,WAAW,EACX,cAAc,EACd,mBAAmB,EACnB,UAAU,EACV,qBAAqB,EACrB,OAAO,EACP,gBAAgB,EAChB,SAAS,EACT,mBAAmB,EACnB,UAAU,EACV,iBAAiB,GACjB,MAAM,YAAY,CAAA"}

package/out/index.js

@@ -0,0 +1,20 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `@mailwoman/query-shape` — pure-function structural priors for the runtime pipeline.
+ *
+ *   Computes a `QueryShape` from an input string: character class, per-token class, punctuation-
+ *   bounded segments, known-format regex hits, and whitespace pattern. Microseconds-cheap, no ML,
+ *   no runtime dependencies.
+ *
+ *   See `docs/articles/plan/reference/QUERY_SHAPE.md` for the design rationale and
+ *   `docs/articles/plan/reference/STAGES.md` for how this fits into the runtime pipeline.
+ */
+export { classifyCodepoint, classifyToken, foldInputClass } from "./character-class.js";
+export { computeQueryShape } from "./compute.js";
+export { detectKnownFormats } from "./known-formats.js";
+export { detectRegionAbbreviations } from "./region-abbreviations.js";
+export { segment } from "./segmentation.js";
+//# sourceMappingURL=index.js.map

package/out/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAA;AACvF,OAAO,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAA;AAChD,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAA;AACvD,OAAO,EAAE,yBAAyB,EAAE,MAAM,2BAA2B,CAAA;AACrE,OAAO,EAAE,OAAO,EAAE,MAAM,mBAAmB,CAAA"}

package/out/known-formats.d.ts

@@ -0,0 +1,14 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { KnownFormatHit, TokenClass } from "./types.js";
+/**
+ * Detect known-format hits among the tokenized input.
+ *
+ * Strategy: for each token (or adjacent pair), try every pattern. Multiple format hits on the same
+ * span are allowed (US/FR/DE 5-digit ambiguity surfaces all three).
+ */
+export declare function detectKnownFormats(text: string, tokens: ReadonlyArray<TokenClass>): KnownFormatHit[];
+//# sourceMappingURL=known-formats.d.ts.map

package/out/known-formats.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"known-formats.d.ts","sourceRoot":"","sources":["../known-formats.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAe,cAAc,EAAE,UAAU,EAAE,MAAM,YAAY,CAAA;AAsCzE;;;;;GAKG;AACH,wBAAgB,kBAAkB,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,aAAa,CAAC,UAAU,CAAC,GAAG,cAAc,EAAE,CAqCpG"}

package/out/known-formats.js

@@ -0,0 +1,114 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+const PATTERNS = [
+    // Unambiguous single-token patterns first.
+    { format: "us_zip4", pattern: /^\d{5}-\d{4}$/, tokenSpan: 1, confidence: 0.95 },
+    { format: "ca_postcode", pattern: /^[A-Z]\d[A-Z]\d[A-Z]\d$/i, tokenSpan: 1, confidence: 0.95 },
+    { format: "jp_postcode", pattern: /^\d{3}-\d{4}$/, tokenSpan: 1, confidence: 0.95 },
+    // UK postcode is 2 tokens when split on space (e.g. "SW1A 1AA"), 1 token otherwise.
+    { format: "uk_postcode", pattern: /^[A-Z]{1,2}\d[A-Z\d]?\d[A-Z]{2}$/i, tokenSpan: 1, confidence: 0.9 },
+    { format: "uk_postcode", pattern: /^[A-Z]{1,2}\d[A-Z\d]? \d[A-Z]{2}$/i, tokenSpan: 2, confidence: 0.9 },
+    { format: "ca_postcode", pattern: /^[A-Z]\d[A-Z] \d[A-Z]\d$/i, tokenSpan: 2, confidence: 0.9 },
+    // Ambiguous 5-digit (US/FR/DE). Tag as us_zip with reduced confidence; caller disambiguates by
+    // locale prior. Multiple format hits on the same span are possible.
+    { format: "us_zip", pattern: /^\d{5}$/, tokenSpan: 1, confidence: 0.6 },
+    { format: "fr_postcode", pattern: /^\d{5}$/, tokenSpan: 1, confidence: 0.6 },
+    { format: "de_postcode", pattern: /^\d{5}$/, tokenSpan: 1, confidence: 0.6 },
+    // PO Box variants (US + FR). The pattern matches across 2-3 tokens — handled separately.
+];
+const PO_BOX_LEADERS = new Set(["po", "p.o.", "p.o", "box", "bp", "b.p.", "b.p", "casilla", "apartado"]);
+/**
+ * Detect known-format hits among the tokenized input.
+ *
+ * Strategy: for each token (or adjacent pair), try every pattern. Multiple format hits on the same
+ * span are allowed (US/FR/DE 5-digit ambiguity surfaces all three).
+ */
+export function detectKnownFormats(text, tokens) {
+    const hits = [];
+    // Single-token patterns.
+    for (const tok of tokens) {
+        for (const p of PATTERNS) {
+            if (p.tokenSpan !== 1)
+                continue;
+            if (p.pattern.test(tok.span.body)) {
+                hits.push({ format: p.format, span: tok.span, confidence: p.confidence });
+            }
+        }
+    }
+    // Two-token patterns (joined by a single space).
+    for (let i = 0; i + 1 < tokens.length; i++) {
+        const a = tokens[i];
+        const b = tokens[i + 1];
+        if (!a || !b)
+            continue;
+        const joined = `${a.span.body} ${b.span.body}`;
+        for (const p of PATTERNS) {
+            if (p.tokenSpan !== 2)
+                continue;
+            if (p.pattern.test(joined)) {
+                hits.push({
+                    format: p.format,
+                    span: { start: a.span.start, end: b.span.end, body: text.slice(a.span.start, b.span.end) },
+                    confidence: p.confidence,
+                });
+            }
+        }
+    }
+    // PO Box detection — handled separately because the leader can be 1-3 tokens and the number
+    // can be alphanumeric.
+    const poHit = detectPoBox(text, tokens);
+    if (poHit)
+        hits.push(poHit);
+    return hits;
+}
+function detectPoBox(text, tokens) {
+    if (tokens.length === 0)
+        return null;
+    // Find a leader token + optional "Box" + numeric/alphanumeric.
+    for (let i = 0; i < tokens.length; i++) {
+        const leadTok = tokens[i];
+        if (!leadTok)
+            continue;
+        const lead = leadTok.span.body.toLowerCase();
+        if (!PO_BOX_LEADERS.has(lead))
+            continue;
+        // Walk forward up to 3 tokens looking for the box number.
+        let last = i;
+        let foundNumber = false;
+        for (let j = i + 1; j <= Math.min(i + 3, tokens.length - 1); j++) {
+            const tj = tokens[j];
+            if (!tj)
+                break;
+            const tjBody = tj.span.body.toLowerCase();
+            if (PO_BOX_LEADERS.has(tjBody)) {
+                last = j;
+                continue;
+            }
+            // Numeric or alphanumeric token = the box number.
+            if (tj.class === "digit" || tj.class === "mixed") {
+                last = j;
+                foundNumber = true;
+                break;
+            }
+            break;
+        }
+        if (foundNumber) {
+            const startTok = tokens[i];
+            const endTok = tokens[last];
+            if (!startTok || !endTok)
+                return null;
+            const start = startTok.span.start;
+            const end = endTok.span.end;
+            return {
+                format: "po_box",
+                span: { start, end, body: text.slice(start, end) },
+                confidence: 0.85,
+            };
+        }
+    }
+    return null;
+}
+//# sourceMappingURL=known-formats.js.map

package/out/known-formats.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"known-formats.js","sourceRoot":"","sources":["../known-formats.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAqBH,MAAM,QAAQ,GAAiC;IAC9C,2CAA2C;IAC3C,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;IAC/E,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,0BAA0B,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;IAC9F,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;IACnF,oFAAoF;IACpF,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,mCAAmC,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IACtG,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,oCAAoC,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IACvG,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,2BAA2B,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IAC9F,+FAA+F;IAC/F,oEAAoE;IACpE,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IACvE,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IAC5E,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS,EAAE,CAAC,EAAE,UAAU,EAAE,GAAG,EAAE;IAC5E,yFAAyF;CACzF,CAAA;AAED,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,EAAE,UAAU,CAAC,CAAC,CAAA;AAExG;;;;;GAKG;AACH,MAAM,UAAU,kBAAkB,CAAC,IAAY,EAAE,MAAiC;IACjF,MAAM,IAAI,GAAqB,EAAE,CAAA;IAEjC,yBAAyB;IACzB,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;QAC1B,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;YAC1B,IAAI,CAAC,CAAC,SAAS,KAAK,CAAC;gBAAE,SAAQ;YAC/B,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;gBACnC,IAAI,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,UAAU,EAAE,CAAC,CAAC,UAAU,EAAE,CAAC,CAAA;YAC1E,CAAC;QACF,CAAC;IACF,CAAC;IAED,iDAAiD;IACjD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC5C,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;QACnB,MAAM,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;QACvB,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC;YAAE,SAAQ;QACtB,MAAM,MAAM,GAAG,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,IAAI,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA;QAC9C,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;YAC1B,IAAI,CAAC,CAAC,SAAS,KAAK,CAAC;gBAAE,SAAQ;YAC/B,IAAI,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC5B,IAAI,CAAC,IAAI,CAAC;oBACT,MAAM,EAAE,CAAC,CAAC,MAAM;oBAChB,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE;oBAC1F,UAAU,EAAE,CAAC,CAAC,UAAU;iBACxB,CAAC,CAAA;YACH,CAAC;QACF,CAAC;IACF,CAAC;IAED,4FAA4F;IAC5F,uBAAuB;IACvB,MAAM,KAAK,GAAG,WAAW,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;IACvC,IAAI,KAAK;QAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;IAE3B,OAAO,IAAI,CAAA;AACZ,CAAC;AAED,SAAS,WAAW,CAAC,IAAY,EAAE,MAAiC;IACnE,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IACpC,+DAA+D;IAC/D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACxC,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;QACzB,IAAI,CAAC,OAAO;YAAE,SAAQ;QACtB,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,CAAA;QAC5C,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC;YAAE,SAAQ;QAEvC,0DAA0D;QAC1D,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,WAAW,GAAG,KAAK,CAAA;QACvB,KAAK,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YAClE,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;YACpB,IAAI,CAAC,EAAE;gBAAE,MAAK;YACd,MAAM,MAAM,GAAG,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,CAAA;YACzC,IAAI,cAAc,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;gBAChC,IAAI,GAAG,CAAC,CAAA;gBACR,SAAQ;YACT,CAAC;YACD,kDAAkD;YAClD,IAAI,EAAE,CAAC,KAAK,KAAK,OAAO,IAAI,EAAE,CAAC,KAAK,KAAK,OAAO,EAAE,CAAC;gBAClD,IAAI,GAAG,CAAC,CAAA;gBACR,WAAW,GAAG,IAAI,CAAA;gBAClB,MAAK;YACN,CAAC;YACD,MAAK;QACN,CAAC;QACD,IAAI,WAAW,EAAE,CAAC;YACjB,MAAM,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;YAC1B,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,CAAA;YAC3B,IAAI,CAAC,QAAQ,IAAI,CAAC,MAAM;gBAAE,OAAO,IAAI,CAAA;YACrC,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAA;YACjC,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,GAAG,CAAA;YAC3B,OAAO;gBACN,MAAM,EAAE,QAAQ;gBAChB,IAAI,EAAE,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,EAAE;gBAClD,UAAU,EAAE,IAAI;aAChB,CAAA;QACF,CAAC;IACF,CAAC;IACD,OAAO,IAAI,CAAA;AACZ,CAAC"}

package/out/region-abbreviations.d.ts

@@ -0,0 +1,17 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Detect unambiguous region abbreviations (e.g., "DC", "NY", "CA") for the locality soft prior.
+ *   Only fires after a comma-space boundary in en-us — the canonical "City, ST ZIP" pattern.
+ */
+import type { RegionAbbreviationHit, Segment, TokenClass } from "./types.js";
+/**
+ * Find region abbreviation hits. A hit is a 2-letter all-uppercase token that appears after a
+ * comma-separated segment boundary — the canonical "City, ST" or "City, ST ZIP" tail pattern.
+ *
+ * Returns empty array for non-Western locales or inputs without comma segmentation.
+ */
+export declare function detectRegionAbbreviations(tokens: ReadonlyArray<TokenClass>, segments: ReadonlyArray<Segment>): RegionAbbreviationHit[];
+//# sourceMappingURL=region-abbreviations.d.ts.map

package/out/region-abbreviations.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"region-abbreviations.d.ts","sourceRoot":"","sources":["../region-abbreviations.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,qBAAqB,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,YAAY,CAAA;AAI5E;;;;;GAKG;AACH,wBAAgB,yBAAyB,CACxC,MAAM,EAAE,aAAa,CAAC,UAAU,CAAC,EACjC,QAAQ,EAAE,aAAa,CAAC,OAAO,CAAC,GAC9B,qBAAqB,EAAE,CAkBzB"}

package/out/region-abbreviations.js

@@ -0,0 +1,35 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Detect unambiguous region abbreviations (e.g., "DC", "NY", "CA") for the locality soft prior.
+ *   Only fires after a comma-space boundary in en-us — the canonical "City, ST ZIP" pattern.
+ */
+const REGION_ABBREV_RE = /^[A-Z]{2}$/;
+/**
+ * Find region abbreviation hits. A hit is a 2-letter all-uppercase token that appears after a
+ * comma-separated segment boundary — the canonical "City, ST" or "City, ST ZIP" tail pattern.
+ *
+ * Returns empty array for non-Western locales or inputs without comma segmentation.
+ */
+export function detectRegionAbbreviations(tokens, segments) {
+    if (segments.length < 2)
+        return [];
+    const hits = [];
+    for (const seg of segments) {
+        if (seg.separator !== "comma")
+            continue;
+        for (const tok of tokens) {
+            if (tok.span.start < seg.span.start || tok.span.end > seg.span.end)
+                continue;
+            if (tok.class !== "alpha")
+                continue;
+            if (!REGION_ABBREV_RE.test(tok.span.body))
+                continue;
+            hits.push({ start: tok.span.start, span: tok.span.body });
+        }
+    }
+    return hits;
+}
+//# sourceMappingURL=region-abbreviations.js.map

package/out/region-abbreviations.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"region-abbreviations.js","sourceRoot":"","sources":["../region-abbreviations.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,MAAM,gBAAgB,GAAG,YAAY,CAAA;AAErC;;;;;GAKG;AACH,MAAM,UAAU,yBAAyB,CACxC,MAAiC,EACjC,QAAgC;IAEhC,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,EAAE,CAAA;IAElC,MAAM,IAAI,GAA4B,EAAE,CAAA;IAExC,KAAK,MAAM,GAAG,IAAI,QAAQ,EAAE,CAAC;QAC5B,IAAI,GAAG,CAAC,SAAS,KAAK,OAAO;YAAE,SAAQ;QAEvC,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;YAC1B,IAAI,GAAG,CAAC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,IAAI,CAAC,KAAK,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC,IAAI,CAAC,GAAG;gBAAE,SAAQ;YAC5E,IAAI,GAAG,CAAC,KAAK,KAAK,OAAO;gBAAE,SAAQ;YACnC,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC;gBAAE,SAAQ;YAEnD,IAAI,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAA;QAC1D,CAAC;IACF,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC"}

package/out/segmentation.d.ts

@@ -0,0 +1,12 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import type { Segment } from "./types.js";
+/**
+ * Comma / newline / tab separate segments. Locale-aware grammar reserved for future (JP whitespace,
+ * KR honorifics). Default rules apply when no locale-specific override exists.
+ */
+export declare function segment(text: string, _locale?: string): Segment[];
+//# sourceMappingURL=segmentation.d.ts.map

package/out/segmentation.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"segmentation.d.ts","sourceRoot":"","sources":["../segmentation.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,KAAK,EAAE,OAAO,EAAoB,MAAM,YAAY,CAAA;AAE3D;;;GAGG;AACH,wBAAgB,OAAO,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,CAAC,EAAE,MAAM,GAAG,OAAO,EAAE,CA8CjE"}

package/out/segmentation.js

@@ -0,0 +1,57 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+/**
+ * Comma / newline / tab separate segments. Locale-aware grammar reserved for future (JP whitespace,
+ * KR honorifics). Default rules apply when no locale-specific override exists.
+ */
+export function segment(text, _locale) {
+    const segments = [];
+    if (text.length === 0)
+        return segments;
+    let start = 0;
+    let lastSeparator = null;
+    let index = 0;
+    const flush = (end, separator) => {
+        // Trim leading + trailing whitespace from each segment but record the original span.
+        const raw = text.slice(start, end);
+        const leftPad = raw.match(/^\s*/)[0].length;
+        const rightPad = raw.match(/\s*$/)[0].length;
+        const innerStart = start + leftPad;
+        const innerEnd = end - rightPad;
+        if (innerEnd > innerStart) {
+            segments.push({
+                span: { start: innerStart, end: innerEnd, body: text.slice(innerStart, innerEnd) },
+                body: text.slice(innerStart, innerEnd),
+                index,
+                separator: lastSeparator,
+            });
+            index += 1;
+        }
+        lastSeparator = separator;
+    };
+    for (let i = 0; i < text.length; i++) {
+        const ch = text[i];
+        if (ch === ",") {
+            flush(i, "comma");
+            start = i + 1;
+        }
+        else if (ch === "\n") {
+            flush(i, "newline");
+            start = i + 1;
+        }
+        else if (ch === "\t") {
+            flush(i, "tab");
+            start = i + 1;
+        }
+        else if (ch === ";") {
+            flush(i, "comma"); // semicolon treated as comma-equivalent
+            start = i + 1;
+        }
+    }
+    flush(text.length, null);
+    return segments;
+}
+//# sourceMappingURL=segmentation.js.map

package/out/segmentation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"segmentation.js","sourceRoot":"","sources":["../segmentation.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAIH;;;GAGG;AACH,MAAM,UAAU,OAAO,CAAC,IAAY,EAAE,OAAgB;IACrD,MAAM,QAAQ,GAAc,EAAE,CAAA;IAC9B,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,QAAQ,CAAA;IAEtC,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,IAAI,aAAa,GAAqB,IAAI,CAAA;IAC1C,IAAI,KAAK,GAAG,CAAC,CAAA;IAEb,MAAM,KAAK,GAAG,CAAC,GAAW,EAAE,SAA2B,EAAE,EAAE;QAC1D,qFAAqF;QACrF,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,EAAE,GAAG,CAAC,CAAA;QAClC,MAAM,OAAO,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,CAAE,CAAC,CAAC,CAAC,CAAC,MAAM,CAAA;QAC5C,MAAM,QAAQ,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,CAAE,CAAC,CAAC,CAAC,CAAC,MAAM,CAAA;QAC7C,MAAM,UAAU,GAAG,KAAK,GAAG,OAAO,CAAA;QAClC,MAAM,QAAQ,GAAG,GAAG,GAAG,QAAQ,CAAA;QAC/B,IAAI,QAAQ,GAAG,UAAU,EAAE,CAAC;YAC3B,QAAQ,CAAC,IAAI,CAAC;gBACb,IAAI,EAAE,EAAE,KAAK,EAAE,UAAU,EAAE,GAAG,EAAE,QAAQ,EAAE,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,EAAE;gBAClF,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC;gBACtC,KAAK;gBACL,SAAS,EAAE,aAAa;aACxB,CAAC,CAAA;YACF,KAAK,IAAI,CAAC,CAAA;QACX,CAAC;QACD,aAAa,GAAG,SAAS,CAAA;IAC1B,CAAC,CAAA;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;QAClB,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YAChB,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,CAAA;YACjB,KAAK,GAAG,CAAC,GAAG,CAAC,CAAA;QACd,CAAC;aAAM,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YACxB,KAAK,CAAC,CAAC,EAAE,SAAS,CAAC,CAAA;YACnB,KAAK,GAAG,CAAC,GAAG,CAAC,CAAA;QACd,CAAC;aAAM,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YACxB,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAA;YACf,KAAK,GAAG,CAAC,GAAG,CAAC,CAAA;QACd,CAAC;aAAM,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACvB,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,CAAA,CAAC,wCAAwC;YAC1D,KAAK,GAAG,CAAC,GAAG,CAAC,CAAA;QACd,CAAC;IACF,CAAC;IACD,KAAK,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;IAExB,OAAO,QAAQ,CAAA;AAChB,CAAC"}

package/out/types.d.ts

@@ -0,0 +1,87 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+/**
+ * Minimal character-range descriptor used internally. Compatible with `@mailwoman/core`'s `Span`
+ * class by shape — consumers holding a `Span` can pass it where `SpanRange` is expected.
+ */
+export interface SpanRange {
+    start: number;
+    end: number;
+    body: string;
+}
+/** Per-token character classification. */
+export type TokenCharacterClass = "digit" | "alpha" | "mixed" | "punct" | "cjk" | "cyrillic" | "arabic";
+/** Whole-input character class — folded from `TokenCharacterClass`. */
+export type CharacterClass = "numeric" | "alpha" | "alphanumeric" | "cjk" | "cyrillic" | "arabic" | "mixed";
+/** Known-format identifier. The set is intentionally small + universal. */
+export type KnownFormat = "us_zip" | "us_zip4" | "uk_postcode" | "fr_postcode" | "ca_postcode" | "de_postcode" | "jp_postcode" | "po_box";
+/** Punctuation grammar separator between consecutive segments. */
+export type SegmentSeparator = "comma" | "newline" | "tab" | "whitespace" | "japanese-style" | null;
+/** Whitespace pattern of the whole input. */
+export type WhitespacePattern = "single" | "double" | "tab" | "mixed" | "none";
+export interface TokenClass {
+    span: SpanRange;
+    class: TokenCharacterClass;
+    length: number;
+}
+export interface Segment {
+    span: SpanRange;
+    body: string;
+    /** Position in the segment list, 0-indexed. */
+    index: number;
+    /** The separator that preceded this segment, or `null` for the first segment. */
+    separator: SegmentSeparator;
+}
+export interface KnownFormatHit {
+    format: KnownFormat;
+    span: SpanRange;
+    /** 0..1. Ambiguous patterns (`fr_postcode`/`de_postcode` overlap with `us_zip`) score lower. */
+    confidence: number;
+}
+/**
+ * A detected region abbreviation (e.g., "DC", "NY", "CA"). Used by the locality soft prior to bias
+ * preceding place-name tokens toward `B-locality`.
+ */
+export interface RegionAbbreviationHit {
+    /** Character offset into the normalized input. */
+    start: number;
+    /** The abbreviation text (e.g., "DC", "NY"). */
+    span: string;
+}
+/**
+ * Structural snapshot of an input string, computed once at the boundary between Stage 1 and Stage 2
+ * of the runtime pipeline. Microseconds-cheap. Consumed by stages 2, 2.5, 3 (optional), and 6 as
+ * additional context.
+ *
+ * Bitter-lesson-safe: recognizes universal structural patterns (character class, punctuation,
+ * postcode shape) rather than place-specific knowledge.
+ */
+export interface QueryShape {
+    characterClass: CharacterClass;
+    tokenClasses: TokenClass[];
+    segments: Segment[];
+    knownFormats: KnownFormatHit[];
+    /**
+     * Region abbreviation hits detected in the input. The locality soft prior uses these to bias
+     * preceding place-name tokens toward `B-locality` / `I-locality` during Viterbi decoding.
+     */
+    regionAbbreviations: RegionAbbreviationHit[];
+    totalLength: number;
+    whitespacePattern: WhitespacePattern;
+}
+/**
+ * Minimal shape that satisfies `computeQueryShape`'s input contract. The full `NormalizedInput`
+ * from `@mailwoman/normalize` is structurally compatible — no import required.
+ */
+export interface NormalizedInputLite {
+    normalized: string;
+    appliedLocale?: string;
+}
+export interface ComputeQueryShapeOpts {
+    /** Locale hint for segmentation grammar (default: comma-based Western). */
+    locale?: string;
+}
+//# sourceMappingURL=types.d.ts.map

package/out/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH;;;GAGG;AACH,MAAM,WAAW,SAAS;IACzB,KAAK,EAAE,MAAM,CAAA;IACb,GAAG,EAAE,MAAM,CAAA;IACX,IAAI,EAAE,MAAM,CAAA;CACZ;AAED,0CAA0C;AAC1C,MAAM,MAAM,mBAAmB,GAAG,OAAO,GAAG,OAAO,GAAG,OAAO,GAAG,OAAO,GAAG,KAAK,GAAG,UAAU,GAAG,QAAQ,CAAA;AAEvG,uEAAuE;AACvE,MAAM,MAAM,cAAc,GAAG,SAAS,GAAG,OAAO,GAAG,cAAc,GAAG,KAAK,GAAG,UAAU,GAAG,QAAQ,GAAG,OAAO,CAAA;AAE3G,2EAA2E;AAC3E,MAAM,MAAM,WAAW,GACpB,QAAQ,GACR,SAAS,GACT,aAAa,GACb,aAAa,GACb,aAAa,GACb,aAAa,GACb,aAAa,GACb,QAAQ,CAAA;AAEX,kEAAkE;AAClE,MAAM,MAAM,gBAAgB,GAAG,OAAO,GAAG,SAAS,GAAG,KAAK,GAAG,YAAY,GAAG,gBAAgB,GAAG,IAAI,CAAA;AAEnG,6CAA6C;AAC7C,MAAM,MAAM,iBAAiB,GAAG,QAAQ,GAAG,QAAQ,GAAG,KAAK,GAAG,OAAO,GAAG,MAAM,CAAA;AAE9E,MAAM,WAAW,UAAU;IAC1B,IAAI,EAAE,SAAS,CAAA;IACf,KAAK,EAAE,mBAAmB,CAAA;IAC1B,MAAM,EAAE,MAAM,CAAA;CACd;AAED,MAAM,WAAW,OAAO;IACvB,IAAI,EAAE,SAAS,CAAA;IACf,IAAI,EAAE,MAAM,CAAA;IACZ,+CAA+C;IAC/C,KAAK,EAAE,MAAM,CAAA;IACb,iFAAiF;IACjF,SAAS,EAAE,gBAAgB,CAAA;CAC3B;AAED,MAAM,WAAW,cAAc;IAC9B,MAAM,EAAE,WAAW,CAAA;IACnB,IAAI,EAAE,SAAS,CAAA;IACf,gGAAgG;IAChG,UAAU,EAAE,MAAM,CAAA;CAClB;AAED;;;GAGG;AACH,MAAM,WAAW,qBAAqB;IACrC,kDAAkD;IAClD,KAAK,EAAE,MAAM,CAAA;IACb,gDAAgD;IAChD,IAAI,EAAE,MAAM,CAAA;CACZ;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,UAAU;IAC1B,cAAc,EAAE,cAAc,CAAA;IAC9B,YAAY,EAAE,UAAU,EAAE,CAAA;IAC1B,QAAQ,EAAE,OAAO,EAAE,CAAA;IACnB,YAAY,EAAE,cAAc,EAAE,CAAA;IAC9B;;;OAGG;IACH,mBAAmB,EAAE,qBAAqB,EAAE,CAAA;IAC5C,WAAW,EAAE,MAAM,CAAA;IACnB,iBAAiB,EAAE,iBAAiB,CAAA;CACpC;AAED;;;GAGG;AACH,MAAM,WAAW,mBAAmB;IACnC,UAAU,EAAE,MAAM,CAAA;IAClB,aAAa,CAAC,EAAE,MAAM,CAAA;CACtB;AAED,MAAM,WAAW,qBAAqB;IACrC,2EAA2E;IAC3E,MAAM,CAAC,EAAE,MAAM,CAAA;CACf"}

package/out/types.js

@@ -0,0 +1,7 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+export {};
+//# sourceMappingURL=types.js.map

package/out/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;GAIG"}

package/package.json

@@ -0,0 +1,25 @@
+{
+  "name": "@mailwoman/query-shape",
+  "version": "4.0.0",
+  "description": "Cheap structural priors for the runtime pipeline: character class, segmentation, known-format detection. Pure functions, no ML.",
+  "license": "AGPL-3.0-only",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/sister-software/mailwoman.git",
+    "directory": "query-shape"
+  },
+  "type": "module",
+  "exports": {
+    "./package.json": "./package.json",
+    ".": "./out/index.js"
+  },
+  "files": [
+    "out/**/*.js",
+    "out/**/*.js.map",
+    "out/**/*.d.ts",
+    "out/**/*.d.ts.map"
+  ],
+  "publishConfig": {
+    "access": "public"
+  }
+}