@mailwoman/corpus 4.14.0 → 4.15.0

package/out/src/index.d.ts

@@ -13,6 +13,7 @@ export * from "./golden.js";
 export * from "./parquet.js";
 export * from "./runner.js";
 export * from "./split.js";
+export * from "./synthesize-anchor-absorption.js";
 export * from "./synthesize-german.js";
 export * from "./synthesize-house-venue.js";
 export * from "./synthesize-intersection.js";

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

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,cAAc,CAAA;AAC5B,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,YAAY,CAAA;AAC1B,cAAc,YAAY,CAAA;AAC1B,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,cAAc,CAAA;AAC5B,cAAc,aAAa,CAAA;AAC3B,cAAc,YAAY,CAAA;AAC1B,cAAc,wBAAwB,CAAA;AACtC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,2BAA2B,CAAA;AACzC,cAAc,wBAAwB,CAAA;AACtC,cAAc,wBAAwB,CAAA;AACtC,cAAc,iBAAiB,CAAA;AAC/B,cAAc,eAAe,CAAA;AAC7B,cAAc,YAAY,CAAA"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,cAAc,CAAA;AAC5B,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,YAAY,CAAA;AAC1B,cAAc,YAAY,CAAA;AAC1B,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,cAAc,CAAA;AAC5B,cAAc,aAAa,CAAA;AAC3B,cAAc,YAAY,CAAA;AAC1B,cAAc,mCAAmC,CAAA;AACjD,cAAc,wBAAwB,CAAA;AACtC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,2BAA2B,CAAA;AACzC,cAAc,wBAAwB,CAAA;AACtC,cAAc,wBAAwB,CAAA;AACtC,cAAc,iBAAiB,CAAA;AAC/B,cAAc,eAAe,CAAA;AAC7B,cAAc,YAAY,CAAA"}

package/out/src/index.js

@@ -13,6 +13,7 @@ export * from "./golden.js";
 export * from "./parquet.js";
 export * from "./runner.js";
 export * from "./split.js";
+export * from "./synthesize-anchor-absorption.js";
 export * from "./synthesize-german.js";
 export * from "./synthesize-house-venue.js";
 export * from "./synthesize-intersection.js";

package/out/src/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,cAAc,CAAA;AAC5B,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,YAAY,CAAA;AAC1B,cAAc,YAAY,CAAA;AAC1B,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,cAAc,CAAA;AAC5B,cAAc,aAAa,CAAA;AAC3B,cAAc,YAAY,CAAA;AAC1B,cAAc,wBAAwB,CAAA;AACtC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,2BAA2B,CAAA;AACzC,cAAc,wBAAwB,CAAA;AACtC,cAAc,wBAAwB,CAAA;AACtC,cAAc,iBAAiB,CAAA;AAC/B,cAAc,eAAe,CAAA;AAC7B,cAAc,YAAY,CAAA"}
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/index.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,cAAc,cAAc,CAAA;AAC5B,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,YAAY,CAAA;AAC1B,cAAc,YAAY,CAAA;AAC1B,cAAc,aAAa,CAAA;AAC3B,cAAc,aAAa,CAAA;AAC3B,cAAc,cAAc,CAAA;AAC5B,cAAc,aAAa,CAAA;AAC3B,cAAc,YAAY,CAAA;AAC1B,cAAc,mCAAmC,CAAA;AACjD,cAAc,wBAAwB,CAAA;AACtC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,2BAA2B,CAAA;AACzC,cAAc,wBAAwB,CAAA;AACtC,cAAc,wBAAwB,CAAA;AACtC,cAAc,iBAAiB,CAAA;AAC/B,cAAc,eAAe,CAAA;AAC7B,cAAc,YAAY,CAAA"}

package/out/src/synthesize-anchor-absorption.d.ts

@@ -0,0 +1,54 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Anchor-absorption counter-augmentation (#220/#723, Probe A1). Teaches the model the CONTEXT-
+ *   DEPENDENT leading-5-digit disambiguation that the killed #723 override was faking, AND that the
+ *   `anchor_paint_mode=shaped` WHERE fix alone over-corrected on (Probe A0: it flipped the default to
+ *   house_number, recovering CASE-H but ERODING CASE-P — postcode F1 99.3→86.5 on leading-postcode
+ *   rows like "05764 Finel Hollow Road, VT").
+ *
+ *   THE DISCRIMINATOR the model must learn (from the CASE-H vs CASE-P contrast, NOT a flipped default):
+ *     - a leading 5-digit WITH a trailing postcode + street context  → it is the HOUSE NUMBER (CASE-H)
+ *     - a leading 5-digit with NO trailing postcode (US-rural / DE)   → it IS the POSTCODE   (CASE-P)
+ *   Both leading tokens are real ZIPs (so the painted anchor fires on both); only the surrounding
+ *   context separates them. The model attends to the trailing token to decide the leading one.
+ *
+ *   Slice mix (the A0 learning sets a HEAVY CASE-P floor so the default doesn't flip — DeepSeek's
+ *   ≥35% CASE-P; here CASE-P total = 35%):
+ *     H-adversarial   30%  US street, leading real-ZIP house# + TRAILING postcode → house_number
+ *     P-us-rural      20%  US rural, leading postcode, NO trailing → postcode   (the A0-erosion fix)
+ *     P-de            15%  German leading postcode "{pc} {city}, {street} {hn}" → postcode
+ *     anchor-fp       10%  leading 5-digit that is NOT a real ZIP + trailing postcode → house_number
+ *     locale-ambig    15%  minimal context; the local token (street-type vs none) decides
+ *     standard        10%  normal small house# + trailing postcode → house_number (baseline)
+ *
+ *   Real ZIPs for the leading-5-digit are sampled from the postcode anchor lookup at build time
+ *   (passed via opts.realZips) so the shaped anchor fires on them exactly as at inference; fake ZIPs
+ *   (anchor-fp) are 5-digit strings deliberately absent from the lookup.
+ */
+import type { ComponentTag } from "@mailwoman/core/types";
+export interface AnchorAbsorptionBaseTuple {
+    locality: string;
+    region: string;
+    postcode: string;
+}
+export type AnchorAbsorptionTemplate = "h-adversarial" | "h-no-trailing-locality" | "p-us-rural" | "p-de" | "anchor-fp" | "locale-ambig" | "standard";
+export interface AnchorAbsorptionSynthesisOpts {
+    random?: () => number;
+    forceTemplate?: AnchorAbsorptionTemplate;
+    /** Real US ZIPs (in the anchor lookup) to use as the LEADING 5-digit house number — so the painted
+     * anchor fires on it, the CASE-H/anchor-fp trigger. Builder loads these from pilot-anchor-lookup.json. */
+    realZips?: ReadonlyArray<string>;
+}
+export interface SynthesizedAnchorAbsorptionRow {
+    raw: string;
+    components: Partial<Record<ComponentTag, string>>;
+    locale: string;
+    template: AnchorAbsorptionTemplate;
+}
+/** Build one anchor-absorption counter-augmentation row. */
+export declare function synthesizeAnchorAbsorptionRow(opts?: AnchorAbsorptionSynthesisOpts): SynthesizedAnchorAbsorptionRow;
+export declare const ALL_TEMPLATES: ReadonlyArray<AnchorAbsorptionTemplate>;
+//# sourceMappingURL=synthesize-anchor-absorption.d.ts.map

package/out/src/synthesize-anchor-absorption.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"synthesize-anchor-absorption.d.ts","sourceRoot":"","sources":["../../src/synthesize-anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAEzD,MAAM,WAAW,yBAAyB;IACzC,QAAQ,EAAE,MAAM,CAAA;IAChB,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,MAAM,CAAA;CAChB;AAED,MAAM,MAAM,wBAAwB,GACjC,eAAe,GACf,wBAAwB,GACxB,YAAY,GACZ,MAAM,GACN,WAAW,GACX,cAAc,GACd,UAAU,CAAA;AAEb,MAAM,WAAW,6BAA6B;IAC7C,MAAM,CAAC,EAAE,MAAM,MAAM,CAAA;IACrB,aAAa,CAAC,EAAE,wBAAwB,CAAA;IACxC;8GAC0G;IAC1G,QAAQ,CAAC,EAAE,aAAa,CAAC,MAAM,CAAC,CAAA;CAChC;AAED,MAAM,WAAW,8BAA8B;IAC9C,GAAG,EAAE,MAAM,CAAA;IACX,UAAU,EAAE,OAAO,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;IACjD,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,wBAAwB,CAAA;CAClC;AA6CD,4DAA4D;AAC5D,wBAAgB,6BAA6B,CAC5C,IAAI,GAAE,6BAAkC,GACtC,8BAA8B,CA4FhC;AAaD,eAAO,MAAM,aAAa,EAAE,aAAa,CAAC,wBAAwB,CAQjE,CAAA"}

package/out/src/synthesize-anchor-absorption.js

@@ -0,0 +1,185 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Anchor-absorption counter-augmentation (#220/#723, Probe A1). Teaches the model the CONTEXT-
+ *   DEPENDENT leading-5-digit disambiguation that the killed #723 override was faking, AND that the
+ *   `anchor_paint_mode=shaped` WHERE fix alone over-corrected on (Probe A0: it flipped the default to
+ *   house_number, recovering CASE-H but ERODING CASE-P — postcode F1 99.3→86.5 on leading-postcode
+ *   rows like "05764 Finel Hollow Road, VT").
+ *
+ *   THE DISCRIMINATOR the model must learn (from the CASE-H vs CASE-P contrast, NOT a flipped default):
+ *     - a leading 5-digit WITH a trailing postcode + street context  → it is the HOUSE NUMBER (CASE-H)
+ *     - a leading 5-digit with NO trailing postcode (US-rural / DE)   → it IS the POSTCODE   (CASE-P)
+ *   Both leading tokens are real ZIPs (so the painted anchor fires on both); only the surrounding
+ *   context separates them. The model attends to the trailing token to decide the leading one.
+ *
+ *   Slice mix (the A0 learning sets a HEAVY CASE-P floor so the default doesn't flip — DeepSeek's
+ *   ≥35% CASE-P; here CASE-P total = 35%):
+ *     H-adversarial   30%  US street, leading real-ZIP house# + TRAILING postcode → house_number
+ *     P-us-rural      20%  US rural, leading postcode, NO trailing → postcode   (the A0-erosion fix)
+ *     P-de            15%  German leading postcode "{pc} {city}, {street} {hn}" → postcode
+ *     anchor-fp       10%  leading 5-digit that is NOT a real ZIP + trailing postcode → house_number
+ *     locale-ambig    15%  minimal context; the local token (street-type vs none) decides
+ *     standard        10%  normal small house# + trailing postcode → house_number (baseline)
+ *
+ *   Real ZIPs for the leading-5-digit are sampled from the postcode anchor lookup at build time
+ *   (passed via opts.realZips) so the shaped anchor fires on them exactly as at inference; fake ZIPs
+ *   (anchor-fp) are 5-digit strings deliberately absent from the lookup.
+ */
+function pick(arr, random) {
+    return arr[Math.floor(random() * arr.length)];
+}
+/** A realistic US house number: mostly 1–4 digits (the common range), 25% a real 5-digit ZIP (the HARD
+ * case — a 5-digit leading number that is still a house number when a locality is present). */
+function houseNum(random, realZips) {
+    if (random() < 0.25)
+        return pick(realZips, random);
+    return String(1 + Math.floor(random() * 9999));
+}
+// Curated, provenance-light reference vocab (real US street/city/state + DE). Surface forms must appear
+// in `raw` for alignRow; these are plain ASCII tokens that align cleanly.
+const STREET_NAMES = [
+    "Main", "Oak", "Elm", "Maple", "Cedar", "Pine", "Washington", "Lincoln", "Park", "Hill",
+    "Finel Hollow", "Mt Tabor", "Swasey", "Camperdown", "Mellville", "Rhone", "Westpark", "Crescent Meadow",
+];
+const STREET_TYPES = ["St", "Ave", "Rd", "Dr", "Ln", "Blvd", "Ct", "Way", "Road", "Drive"];
+const US_TUPLES = [
+    { locality: "Springfield", region: "IL", postcode: "62701" },
+    { locality: "Portland", region: "OR", postcode: "97215" },
+    { locality: "Houston", region: "TX", postcode: "77598" },
+    { locality: "Dallas", region: "TX", postcode: "75229" },
+    { locality: "Austin", region: "TX", postcode: "78748" },
+    { locality: "Albuquerque", region: "NM", postcode: "87102" },
+    { locality: "Rochester", region: "NY", postcode: "14606" },
+    { locality: "Sacramento", region: "CA", postcode: "95823" },
+];
+// US rural states where the leading-postcode "{ZIP} {Street}, {STATE}" form (no trailing ZIP) is real —
+// the VT/rural format the #723 override broke and Probe A0 eroded.
+const RURAL_REGIONS = ["VT", "ND", "SD", "NH", "ME", "MT", "WY"];
+// DE leading-postcode tuples: "{postcode} {locality}, {street} {house}".
+const DE_TUPLES = [
+    { postcode: "10115", locality: "Berlin", street: "Hauptstraße" },
+    { postcode: "80331", locality: "München", street: "Sendlinger Straße" },
+    { postcode: "20095", locality: "Hamburg", street: "Mönckebergstraße" },
+    { postcode: "50667", locality: "Köln", street: "Hohe Straße" },
+    { postcode: "01067", locality: "Dresden", street: "Prager Straße" },
+];
+const HOUSE_NUMS = ["5", "12", "27", "100", "212", "1450", "8"];
+// Fake 5-digit strings that are NOT real US ZIPs (so the anchor lookup MISSES them) — for anchor-fp.
+const FAKE_ZIPS = ["00000", "99998", "99997", "00001", "99996"];
+/** Build one anchor-absorption counter-augmentation row. */
+export function synthesizeAnchorAbsorptionRow(opts = {}) {
+    const random = opts.random ?? Math.random;
+    const realZips = opts.realZips && opts.realZips.length ? opts.realZips : US_TUPLES.map((t) => t.postcode);
+    const template = opts.forceTemplate ?? pick(ALL_TEMPLATES, random);
+    const street = `${pick(STREET_NAMES, random)} ${pick(STREET_TYPES, random)}`;
+    if (template === "h-adversarial") {
+        // US street, leading real-ZIP house number, WITH a trailing postcode → the leading is house_number.
+        const zip = pick(realZips, random);
+        const t = pick(US_TUPLES, random);
+        const raw = `${zip} ${street}, ${t.locality}, ${t.region} ${t.postcode}`;
+        return {
+            raw,
+            components: { house_number: zip, street, locality: t.locality, region: t.region, postcode: t.postcode },
+            locale: "en-US",
+            template,
+        };
+    }
+    if (template === "p-us-rural") {
+        // US rural, leading postcode, NO trailing postcode → the leading IS the postcode (the A0-erosion fix).
+        const zip = pick(realZips, random);
+        const region = pick(RURAL_REGIONS, random);
+        const raw = `${zip} ${street}, ${region}`;
+        return {
+            raw,
+            components: { postcode: zip, street, region },
+            locale: "en-US",
+            template,
+        };
+    }
+    if (template === "p-de") {
+        // German leading postcode "{pc} {city}, {street} {hn}" → the leading is the postcode.
+        const d = pick(DE_TUPLES, random);
+        const hn = pick(HOUSE_NUMS, random);
+        const raw = `${d.postcode} ${d.locality}, ${d.street} ${hn}`;
+        return {
+            raw,
+            components: { postcode: d.postcode, locality: d.locality, street: d.street, house_number: hn },
+            locale: "de-DE",
+            template,
+        };
+    }
+    if (template === "anchor-fp") {
+        // Leading 5-digit that is NOT a real ZIP (anchor MISSES) + trailing postcode → still house_number.
+        const fake = pick(FAKE_ZIPS, random);
+        const t = pick(US_TUPLES, random);
+        const raw = `${fake} ${street}, ${t.locality}, ${t.region} ${t.postcode}`;
+        return {
+            raw,
+            components: { house_number: fake, street, locality: t.locality, region: t.region, postcode: t.postcode },
+            locale: "en-US",
+            template,
+        };
+    }
+    if (template === "locale-ambig") {
+        // Minimal context — the LOCAL token decides. Half: "{realZip} {street}" (street-type → house#);
+        // half: "{realZip} {locality}" (no street, leading postcode → postcode). No trailing, no region.
+        const zip = pick(realZips, random);
+        if (random() < 0.5) {
+            return { raw: `${zip} ${street}`, components: { house_number: zip, street }, locale: "en-US", template };
+        }
+        const t = pick(US_TUPLES, random);
+        return { raw: `${zip} ${t.locality}`, components: { postcode: zip, locality: t.locality }, locale: "en-US", template };
+    }
+    if (template === "h-no-trailing-locality") {
+        // The A3 fix (#220): the common US format "{house#} {street}, {locality}, {STATE}" with NO trailing
+        // postcode → the leading number is the HOUSE NUMBER. The CONTRAST to p-us-rural (same no-trailing,
+        // state-bearing shape) is the LOCALITY: present here, absent there. The A2 shard lacked this slice,
+        // so p-us-rural's "leading-number + STATE → postcode" rule over-generalized to 98 golden house#
+        // rows ("36 Oxbow Dr, Bradford, VT" → postcode). The house# spans 1-4 digits AND real 5-digit ZIPs
+        // (the hard case: 5-digit + locality is STILL a house number, distinct from p-us-rural's no-locality).
+        const hn = houseNum(random, realZips);
+        const t = pick(US_TUPLES, random);
+        const region = random() < 0.5 ? pick(RURAL_REGIONS, random) : t.region;
+        const raw = `${hn} ${street}, ${t.locality}, ${region}`;
+        return {
+            raw,
+            components: { house_number: hn, street, locality: t.locality, region },
+            locale: "en-US",
+            template,
+        };
+    }
+    // standard: normal house number + trailing postcode → house_number (baseline, keeps the common case).
+    const hn = houseNum(random, realZips);
+    const t = pick(US_TUPLES, random);
+    const raw = `${hn} ${street}, ${t.locality}, ${t.region} ${t.postcode}`;
+    return {
+        raw,
+        components: { house_number: hn, street, locality: t.locality, region: t.region, postcode: t.postcode },
+        locale: "en-US",
+        template,
+    };
+}
+// Weighted template bag — the slice mix. Expanded to a flat array so `pick` draws at the target
+// frequencies (matches the boundary-stress ALL_TEMPLATES idiom).
+//
+// A3 (#220, after the per-row diagnostic on the A2 probe): A1/A2 both held SLICE-H (100) + postcode
+// (~98) but cost house_number (95.8->92.8), and the A2 mix-rebalance did NOT move it — so it was never a
+// CASE-P-quantity problem. The row-by-row v192-vs-A2 diff (hn-regression-diff.ts) pinned it: 132/132 house#
+// regressions were house#->POSTCODE on "{house#} {street}, {locality}, {STATE}" no-trailing rows — the
+// p-us-rural rule ("leading-number + STATE + no-trailing -> postcode") OVER-GENERALIZED because the shard
+// had NO counter-slice for the common locality-bearing house# case. A3 ADDS h-no-trailing-locality (15%)
+// to teach the LOCALITY discriminator (present -> house#, absent + 5-digit -> postcode = p-us-rural) and
+// trims p-us-rural 16->13. Goal: house_number recovers WITHOUT re-eroding postcode/SLICE-H. CASE-P = 26%.
+export const ALL_TEMPLATES = [
+    ...Array(25).fill("h-adversarial"),
+    ...Array(15).fill("h-no-trailing-locality"),
+    ...Array(13).fill("p-us-rural"),
+    ...Array(13).fill("p-de"),
+    ...Array(8).fill("anchor-fp"),
+    ...Array(14).fill("locale-ambig"),
+    ...Array(12).fill("standard"),
+];
+//# sourceMappingURL=synthesize-anchor-absorption.js.map

package/out/src/synthesize-anchor-absorption.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"synthesize-anchor-absorption.js","sourceRoot":"","sources":["../../src/synthesize-anchor-absorption.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAkCH,SAAS,IAAI,CAAI,GAAqB,EAAE,MAAoB;IAC3D,OAAO,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,GAAG,CAAC,MAAM,CAAC,CAAE,CAAA;AAC/C,CAAC;AAED;+FAC+F;AAC/F,SAAS,QAAQ,CAAC,MAAoB,EAAE,QAA+B;IACtE,IAAI,MAAM,EAAE,GAAG,IAAI;QAAE,OAAO,IAAI,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;IAClD,OAAO,MAAM,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAA;AAC/C,CAAC;AAED,wGAAwG;AACxG,0EAA0E;AAC1E,MAAM,YAAY,GAAG;IACpB,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,YAAY,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM;IACvF,cAAc,EAAE,UAAU,EAAE,QAAQ,EAAE,YAAY,EAAE,WAAW,EAAE,OAAO,EAAE,UAAU,EAAE,iBAAiB;CACvG,CAAA;AACD,MAAM,YAAY,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,CAAC,CAAA;AAC1F,MAAM,SAAS,GAA6C;IAC3D,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IAC5D,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IACzD,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IACxD,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IACvD,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IACvD,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IAC5D,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;IAC1D,EAAE,QAAQ,EAAE,YAAY,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE;CAC3D,CAAA;AACD,wGAAwG;AACxG,mEAAmE;AACnE,MAAM,aAAa,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAA;AAChE,yEAAyE;AACzE,MAAM,SAAS,GAAG;IACjB,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,aAAa,EAAE;IAChE,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,mBAAmB,EAAE;IACvE,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,kBAAkB,EAAE;IACtE,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,aAAa,EAAE;IAC9D,EAAE,QAAQ,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,eAAe,EAAE;CACnE,CAAA;AACD,MAAM,UAAU,GAAG,CAAC,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,CAAC,CAAA;AAC/D,qGAAqG;AACrG,MAAM,SAAS,GAAG,CAAC,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,CAAC,CAAA;AAE/D,4DAA4D;AAC5D,MAAM,UAAU,6BAA6B,CAC5C,OAAsC,EAAE;IAExC,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,CAAA;IACzC,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAA;IACzG,MAAM,QAAQ,GAAG,IAAI,CAAC,aAAa,IAAI,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAA;IAClE,MAAM,MAAM,GAAG,GAAG,IAAI,CAAC,YAAY,EAAE,MAAM,CAAC,IAAI,IAAI,CAAC,YAAY,EAAE,MAAM,CAAC,EAAE,CAAA;IAE5E,IAAI,QAAQ,KAAK,eAAe,EAAE,CAAC;QAClC,oGAAoG;QACpG,MAAM,GAAG,GAAG,IAAI,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;QAClC,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACjC,MAAM,GAAG,GAAG,GAAG,GAAG,IAAI,MAAM,KAAK,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAA;QACxE,OAAO;YACN,GAAG;YACH,UAAU,EAAE,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE;YACvG,MAAM,EAAE,OAAO;YACf,QAAQ;SACR,CAAA;IACF,CAAC;IACD,IAAI,QAAQ,KAAK,YAAY,EAAE,CAAC;QAC/B,uGAAuG;QACvG,MAAM,GAAG,GAAG,IAAI,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;QAClC,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAA;QAC1C,MAAM,GAAG,GAAG,GAAG,GAAG,IAAI,MAAM,KAAK,MAAM,EAAE,CAAA;QACzC,OAAO;YACN,GAAG;YACH,UAAU,EAAE,EAAE,QAAQ,EAAE,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE;YAC7C,MAAM,EAAE,OAAO;YACf,QAAQ;SACR,CAAA;IACF,CAAC;IACD,IAAI,QAAQ,KAAK,MAAM,EAAE,CAAC;QACzB,sFAAsF;QACtF,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACjC,MAAM,EAAE,GAAG,IAAI,CAAC,UAAU,EAAE,MAAM,CAAC,CAAA;QACnC,MAAM,GAAG,GAAG,GAAG,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,EAAE,EAAE,CAAA;QAC5D,OAAO;YACN,GAAG;YACH,UAAU,EAAE,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,YAAY,EAAE,EAAE,EAAE;YAC9F,MAAM,EAAE,OAAO;YACf,QAAQ;SACR,CAAA;IACF,CAAC;IACD,IAAI,QAAQ,KAAK,WAAW,EAAE,CAAC;QAC9B,mGAAmG;QACnG,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACpC,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACjC,MAAM,GAAG,GAAG,GAAG,IAAI,IAAI,MAAM,KAAK,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAA;QACzE,OAAO;YACN,GAAG;YACH,UAAU,EAAE,EAAE,YAAY,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE;YACxG,MAAM,EAAE,OAAO;YACf,QAAQ;SACR,CAAA;IACF,CAAC;IACD,IAAI,QAAQ,KAAK,cAAc,EAAE,CAAC;QACjC,gGAAgG;QAChG,iGAAiG;QACjG,MAAM,GAAG,GAAG,IAAI,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;QAClC,IAAI,MAAM,EAAE,GAAG,GAAG,EAAE,CAAC;YACpB,OAAO,EAAE,GAAG,EAAE,GAAG,GAAG,IAAI,MAAM,EAAE,EAAE,UAAU,EAAE,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAA;QACzG,CAAC;QACD,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACjC,OAAO,EAAE,GAAG,EAAE,GAAG,GAAG,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,UAAU,EAAE,EAAE,QAAQ,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAA;IACvH,CAAC;IACD,IAAI,QAAQ,KAAK,wBAAwB,EAAE,CAAC;QAC3C,oGAAoG;QACpG,mGAAmG;QACnG,oGAAoG;QACpG,gGAAgG;QAChG,mGAAmG;QACnG,uGAAuG;QACvG,MAAM,EAAE,GAAG,QAAQ,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAA;QACrC,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;QACjC,MAAM,MAAM,GAAG,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAA;QACtE,MAAM,GAAG,GAAG,GAAG,EAAE,IAAI,MAAM,KAAK,CAAC,CAAC,QAAQ,KAAK,MAAM,EAAE,CAAA;QACvD,OAAO;YACN,GAAG;YACH,UAAU,EAAE,EAAE,YAAY,EAAE,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE;YACtE,MAAM,EAAE,OAAO;YACf,QAAQ;SACR,CAAA;IACF,CAAC;IACD,sGAAsG;IACtG,MAAM,EAAE,GAAG,QAAQ,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAA;IACrC,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;IACjC,MAAM,GAAG,GAAG,GAAG,EAAE,IAAI,MAAM,KAAK,CAAC,CAAC,QAAQ,KAAK,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAA;IACvE,OAAO;QACN,GAAG;QACH,UAAU,EAAE,EAAE,YAAY,EAAE,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,CAAC,CAAC,QAAQ,EAAE;QACtG,MAAM,EAAE,OAAO;QACf,QAAQ;KACR,CAAA;AACF,CAAC;AAED,gGAAgG;AAChG,iEAAiE;AACjE,EAAE;AACF,oGAAoG;AACpG,yGAAyG;AACzG,4GAA4G;AAC5G,uGAAuG;AACvG,0GAA0G;AAC1G,yGAAyG;AACzG,yGAAyG;AACzG,0GAA0G;AAC1G,MAAM,CAAC,MAAM,aAAa,GAA4C;IACrE,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC;IAC5D,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,wBAAwB,CAAC;IACrE,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC;IACzD,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC;IACnD,GAAG,KAAK,CAA2B,CAAC,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC;IACvD,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,cAAc,CAAC;IAC3D,GAAG,KAAK,CAA2B,EAAE,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC;CACvD,CAAA"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mailwoman/corpus",
-  "version": "4.14.0",
+  "version": "4.15.0",
   "description": "Mailwoman corpus pipeline: BIO-labeled dataset builder for the neural classifier.",
   "license": "AGPL-3.0-only",
   "repository": {
@@ -17,9 +17,9 @@
   },
   "dependencies": {
     "@dsnp/parquetjs": "1.8.7",
-    "@mailwoman/codex": "4.14.0",
-    "@mailwoman/core": "4.14.0",
-    "@mailwoman/formatter": "4.14.0",
+    "@mailwoman/codex": "4.15.0",
+    "@mailwoman/core": "4.15.0",
+    "@mailwoman/formatter": "4.15.0",
     "csv-parse": "^5.6.0",
     "fast-glob": "^3.3.3",
     "fastest-levenshtein": "^1.0.16",