@mailwoman/corpus 10.0.0 → 10.1.0

package/out/no/recipes/fragment.js

@@ -2,55 +2,27 @@
  * @copyright Sister Software
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
- *
- *   `no-fragment` — the Norwegian house-number-licence change (Track B, 2026-07-16). The mirror of
- *   `fr-fragment`, which earned +50pp on the same defect shape in French.
- *
- *   WHY THIS EXISTS AND `no-street-led` DOES NOT SUFFICE. Board 3 (the NO digit board) measured that
- *   `synth-no-street-led`'s three forms — all carrying postcode+city — are ALREADY at 0.940-0.968 on
- *   a model with zero Norwegian rows. The headroom is in the forms that slice never emits:
- *
- *     bare-street-hn   "Hallingrudveien 32"     0.693   — no postcode competing, still fails 31%
- *     slash-hn         "Øvrabø 124/1"           0.650   — cadastral gnr/bnr, ONE component
- *
- *   THE MECHANISM the board exposed: `Hallingrudveien 32` -> locality + postcode, while
- *   `Hallingrudveien 32, 3370 Vikersund` parses perfectly. The street loses its street reading and
- *   the digit loses its anchor TOGETHER. That is Track A's bare-street LICENCE in Norwegian — the
- *   model will not read a street without its postcode/locality partner — not a digit-ownership prior.
- *   fr-fragment fixed exactly this in French by teaching the street WITHOUT its partners.
- *
- *   THE COUNTER-DISTRIBUTION IS THE POINT (fr-fragment's lesson, and board 2's bare-locality guard).
- *   Teaching bare `{street} {number}` alone lets the model satisfy every row by flipping its default
- *   from "bare toponym -> locality" to "bare toponym -> street", trading one broken prior for
- *   another. Two counter-classes hold the line:
- *     - bare LOCALITIES (no street) so "bare -> street" is not free.
- *     - bare POSTCODES so the model does not learn to stop emitting postcode to win the digit — the
- *       board 3 bare-pc negative class (1.000) must HOLD.
- *
- *   SLASH HAZARD, pinned deliberately: NO `124/1` is ONE house_number (cadastral gnr/bnr). AU
- *   `12/345` is TWO (unit + house_number). This slice teaches the Norwegian reading; a future AU
- *   intra-word-split slice (B5) must not generalize over it. The two are locale-restricted by design.
- *
- *   SPLIT: `--exclude-surfaces` REQUIRED (throws otherwise) — the digit board's reserved surface
- *   list. Diacritic-KEEPING normalizer, matching the board (see the norm docstring).
  */
 import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
 import { TextSpliterator } from "spliterator";
-import { alignAndWrite, foldNOSurface, readTuples, sliceSourceID } from "#recipes/scaffold";
+import { alignAndWrite, foldNOSurface, readTuples, requireRegister, recipeSourceID, } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
+import { SurfaceOrigin } from "#types";
 /**
- * Title-case a Kartverket ALL-CAPS locality (HELLVIK -> Hellvik); #690, all-caps is OOD.
+ * Resolved once, because `source_id` includes the source as its prefix and a pair
+ * that disagreed would name a source no row of this output declares.
  */
+const NO_FRAGMENT_SOURCE = defaultRecipeSource("synth-no-fragment");
 const titleNO = (value) => value
     .split(/\s+/)
     .map((w) => (w ? w[0].toUpperCase() + w.slice(1).toLowerCase() : w))
     .join(" ");
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Norwegian fragment recipe.
  */
 export const noFragmentRecipe = {
     name: "no-fragment",
-    description: "NO street fragments — the house-number-licence change (Track B): '«st» «n»' / bare «st» with NO postcode partner, + bare-locality & bare-postcode counters",
+    description: "NO street fragments — the house-number-license change (Track B): '«st» «n»' / bare «st» with NO postcode partner, + bare-locality & bare-postcode counters",
     mode: "tuples",
     options: [
         {
@@ -59,7 +31,7 @@ export const noFragmentRecipe = {
         },
         {
             flag: "--bare-street-prob <n>",
-            description: "Share of street rows emitted as a BARE street, no number (default 0.30 — the pure licence signal)",
+            description: "Share of street rows emitted as a bare street without a house number, the pure license signal (default 0.30)",
         },
         {
             flag: "--counter-prob <n>",
@@ -79,14 +51,13 @@ export const noFragmentRecipe = {
         const excludePath = opts.excludeSurfaces;
         if (!excludePath) {
             throw new Error("no-fragment: --exclude-surfaces is REQUIRED. Pass the NO digit board's reserved list " +
-                "(mailwoman/eval-harness/fixtures/no-digits.surfaces.txt) or this slice trains on its own eval set. " +
+                "(mailwoman/eval-harness/fixtures/no-digits.surfaces.txt) or this recipe trains on its own eval set. " +
                 "Source-disjoint by street SURFACE is the split discipline; there is no safe default.");
         }
         const excluded = new Set();
         for await (const line of TextSpliterator.fromAsync(excludePath)) {
-            const trimmed = line.trim();
-            if (trimmed && !trimmed.startsWith("#")) {
-                excluded.add(trimmed);
+            if (!line.startsWith("#")) {
+                excluded.add(line);
             }
         }
         if (!excluded.size)
@@ -95,8 +66,6 @@ export const noFragmentRecipe = {
         const counterProb = opts.counterProb ?? 0.3;
         const longNumberBoost = Math.max(1, Math.floor(opts.longNumberBoost ?? 1));
         const longNumberMinDigits = opts.longNumberMinDigits ?? 3;
-        // Harvested from the tuples — every NO row carries its locality and postcode, so the two
-        // counter-classes need no second source.
         const localities = new Set();
         const postcodes = new Set();
         let read = 0;
@@ -105,26 +74,32 @@ export const noFragmentRecipe = {
         let contaminated = 0;
         let emitSeq = 0;
         const emit = (raw, components, klass) => {
-            // emitSeq keeps every emit distinct — knob 3 emits N copies of one long-number row, and
-            // (components, read) alone would collide their source_id and let downstream dedup drop the boost.
-            const source_id = sliceSourceID("synth-no-fragment", { ...components, k: klass, v: `${read}:${emitSeq++}` });
+            const source_id = recipeSourceID(NO_FRAGMENT_SOURCE, { ...components, k: klass, v: `${read}:${emitSeq++}` });
             const canonical = {
                 raw,
                 components,
                 country: "NO",
                 locale: "nb-NO",
-                source: "synth-no-fragment",
+                source: NO_FRAGMENT_SOURCE,
                 source_id,
                 corpus_version: "0.11.0",
                 license: "Synthetic — no-fragment; (street, number, postcode, city) from OpenAddresses NO / Kartverket",
             };
-            if (alignAndWrite(write, canonical, "no-fragment")) {
+            if (alignAndWrite(write, canonical, "no-fragment", NO_FRAGMENT_PROVENANCE)) {
                 emitted++;
             }
             else {
                 skipped++;
             }
         };
+        const NO_FRAGMENT_PROVENANCE = {
+            register: requireRegister(opts, "no-fragment"),
+            surface: SurfaceOrigin.Composed,
+            // `emit` is a closure over the run rather than over one tuple, so this cannot name a per-row record.
+            // The tuples file has no `sourceID` either.
+            // Both inputs are required to populate this field (#2359).
+            baseSourceID: null,
+        };
         for await (const tuple of readTuples(opts.input)) {
             read++;
             const street = String(tuple.street ?? "").trim();
@@ -141,14 +116,10 @@ export const noFragmentRecipe = {
                 skipped++;
                 continue;
             }
-            // THE SPLIT. A surface on the digit board never enters training.
             if (excluded.has(foldNOSurface(street))) {
                 contaminated++;
                 continue;
             }
-            // COUNTER-DISTRIBUTION — drawn from the harvested pools, not this row's street. Half bare
-            // localities (so "bare -> street" is not free), half bare postcodes (so the model does not
-            // stop emitting postcode to win the digit — board 3's bare-pc must hold).
             if (random() < counterProb) {
                 if (random() < 0.5 && localities.size) {
                     const loc = [...localities][Math.floor(random() * localities.size)];
@@ -160,16 +131,11 @@ export const noFragmentRecipe = {
                 }
                 continue;
             }
-            // THE SIGNAL. A street with NO postcode/locality partner. Either bare, or street+number —
-            // both are the forms board 3 measured as the headroom (bare-street-hn 0.693, slash-hn 0.650).
             if (!number || random() < bareStreetProb) {
                 emit(street, { street }, "bare-street");
             }
             else {
                 const klass = number.includes("/") ? "slash-hn" : "street-hn";
-                // knob 3: the failing class is street + LONG number (Leppdalsvegen 1285 -> postcode). The
-                // digit count, not the slash, is what tips the length prior toward postcode. Oversample
-                // those rows to fight the prior with volume and teach the street/number boundary directly.
                 const digits = (number.match(/\d/g) ?? []).length;
                 const copies = digits >= longNumberMinDigits ? longNumberBoost : 1;
                 for (let c = 0; c < copies; c++) {

package/out/no/recipes/fragment.js.map

@@ -1 +1 @@
-{"version":3,"file":"fragment.js","sourceRoot":"","sources":["../../../lib/no/recipes/fragment.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACpE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,UAAU,EAAqB,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAE9G;;GAEG;AACH,MAAM,OAAO,GAAG,CAAC,KAAa,EAAU,EAAE,CACzC,KAAK;KACH,KAAK,CAAC,KAAK,CAAC;KACZ,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACpE,IAAI,CAAC,GAAG,CAAC,CAAA;AAEZ;;;GAGG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC7C,IAAI,EAAE,aAAa;IACnB,WAAW,EACV,4JAA4J;IAC7J,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE;QACR;YACC,IAAI,EAAE,2BAA2B;YACjC,WAAW,EACV,gHAAgH;SACjH;QACD;YACC,IAAI,EAAE,wBAAwB;YAC9B,WAAW,EAAE,mGAAmG;SAChH;QACD;YACC,IAAI,EAAE,oBAAoB;YAC1B,WAAW,EAAE,iGAAiG;SAC9G;QACD;YACC,IAAI,EAAE,yBAAyB;YAC/B,WAAW,EACV,kMAAkM;SACnM;QACD;YACC,IAAI,EAAE,8BAA8B;YACpC,WAAW,EAAE,+EAA+E;SAC5F;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,WAAW,GAAG,IAAI,CAAC,eAAe,CAAA;QAExC,IAAI,CAAC,WAAW,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CACd,uFAAuF;gBACtF,qGAAqG;gBACrG,sFAAsF,CACvF,CAAA;QACF,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;QAElC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAAC;YACjE,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAA;YAE3B,IAAI,OAAO,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBACzC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;YACtB,CAAC;QACF,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,oCAAoC,WAAW,sBAAsB,CAAC,CAAA;QAE1G,MAAM,cAAc,GAAG,IAAI,CAAC,QAAQ,IAAI,GAAG,CAAA;QAC3C,MAAM,WAAW,GAAG,IAAI,CAAC,WAAW,IAAI,GAAG,CAAA;QAC3C,MAAM,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,eAAe,IAAI,CAAC,CAAC,CAAC,CAAA;QAC1E,MAAM,mBAAmB,GAAG,IAAI,CAAC,mBAAmB,IAAI,CAAC,CAAA;QAEzD,yFAAyF;QACzF,yCAAyC;QACzC,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU,CAAA;QACpC,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAA;QAEnC,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,YAAY,GAAG,CAAC,CAAA;QACpB,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,MAAM,IAAI,GAAG,CAAC,GAAW,EAAE,UAAkC,EAAE,KAAa,EAAQ,EAAE;YACrF,wFAAwF;YACxF,kGAAkG;YAClG,MAAM,SAAS,GAAG,aAAa,CAAC,mBAAmB,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,IAAI,IAAI,OAAO,EAAE,EAAE,EAAE,CAAC,CAAA;YAE5G,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,mBAAmB;gBAC3B,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EAAE,8FAA8F;aACvG,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,aAAa,CAAC,EAAE,CAAC;gBACpD,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC,CAAA;QAED,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAM,CAAC,EAAE,CAAC;YACnD,IAAI,EAAE,CAAA;YACN,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAChD,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAA;YAC7D,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAChD,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAEpD,IAAI,QAAQ,EAAE,CAAC;gBACd,UAAU,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;YACzB,CAAC;YAED,IAAI,QAAQ,EAAE,CAAC;gBACd,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,CAAC,MAAM,EAAE,CAAC;gBACb,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,iEAAiE;YACjE,IAAI,QAAQ,CAAC,GAAG,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzC,YAAY,EAAE,CAAA;gBAEd,SAAQ;YACT,CAAC;YAED,0FAA0F;YAC1F,2FAA2F;YAC3F,0EAA0E;YAC1E,IAAI,MAAM,EAAE,GAAG,WAAW,EAAE,CAAC;gBAC5B,IAAI,MAAM,EAAE,GAAG,GAAG,IAAI,UAAU,CAAC,IAAI,EAAE,CAAC;oBACvC,MAAM,GAAG,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,UAAU,CAAC,IAAI,CAAC,CAAE,CAAA;oBAEpE,IAAI,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,GAAG,EAAE,EAAE,uBAAuB,CAAC,CAAA;gBACtD,CAAC;qBAAM,IAAI,SAAS,CAAC,IAAI,EAAE,CAAC;oBAC3B,MAAM,EAAE,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,CAAE,CAAA;oBAEjE,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,uBAAuB,CAAC,CAAA;gBACpD,CAAC;gBAED,SAAQ;YACT,CAAC;YAED,0FAA0F;YAC1F,8FAA8F;YAC9F,IAAI,CAAC,MAAM,IAAI,MAAM,EAAE,GAAG,cAAc,EAAE,CAAC;gBAC1C,IAAI,CAAC,MAAM,EAAE,EAAE,MAAM,EAAE,EAAE,aAAa,CAAC,CAAA;YACxC,CAAC;iBAAM,CAAC;gBACP,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,CAAA;gBAC7D,0FAA0F;gBAC1F,wFAAwF;gBACxF,2FAA2F;gBAC3F,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAA;gBACjD,MAAM,MAAM,GAAG,MAAM,IAAI,mBAAmB,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,CAAA;gBAElE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;oBACjC,IAAI,CAAC,GAAG,MAAM,IAAI,MAAM,EAAE,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,EAAE,KAAK,CAAC,CAAA;gBACrE,CAAC;YACF,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,CAAA;IAChD,CAAC;CACD,CAAA"}
+{"version":3,"file":"fragment.js","sourceRoot":"","sources":["../../../lib/no/recipes/fragment.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACpE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EACN,aAAa,EACb,aAAa,EACb,UAAU,EACV,eAAe,EAEf,cAAc,GACd,MAAM,mBAAmB,CAAA;AAC1B,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AACtD,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEtC;;;GAGG;AACH,MAAM,kBAAkB,GAAG,mBAAmB,CAAC,mBAAmB,CAAC,CAAA;AAEnE,MAAM,OAAO,GAAG,CAAC,KAAa,EAAU,EAAE,CACzC,KAAK;KACH,KAAK,CAAC,KAAK,CAAC;KACZ,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAE,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;KACpE,IAAI,CAAC,GAAG,CAAC,CAAA;AAEZ;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC7C,IAAI,EAAE,aAAa;IACnB,WAAW,EACV,4JAA4J;IAC7J,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE;QACR;YACC,IAAI,EAAE,2BAA2B;YACjC,WAAW,EACV,gHAAgH;SACjH;QACD;YACC,IAAI,EAAE,wBAAwB;YAC9B,WAAW,EACV,8GAA8G;SAC/G;QACD;YACC,IAAI,EAAE,oBAAoB;YAC1B,WAAW,EAAE,iGAAiG;SAC9G;QACD;YACC,IAAI,EAAE,yBAAyB;YAC/B,WAAW,EACV,kMAAkM;SACnM;QACD;YACC,IAAI,EAAE,8BAA8B;YACpC,WAAW,EAAE,+EAA+E;SAC5F;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,WAAW,GAAG,IAAI,CAAC,eAAe,CAAA;QAExC,IAAI,CAAC,WAAW,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CACd,uFAAuF;gBACtF,sGAAsG;gBACtG,sFAAsF,CACvF,CAAA;QACF,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;QAElC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAAC;YACjE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBAC3B,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;YACnB,CAAC;QACF,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,oCAAoC,WAAW,sBAAsB,CAAC,CAAA;QAE1G,MAAM,cAAc,GAAG,IAAI,CAAC,QAAQ,IAAI,GAAG,CAAA;QAC3C,MAAM,WAAW,GAAG,IAAI,CAAC,WAAW,IAAI,GAAG,CAAA;QAC3C,MAAM,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,eAAe,IAAI,CAAC,CAAC,CAAC,CAAA;QAC1E,MAAM,mBAAmB,GAAG,IAAI,CAAC,mBAAmB,IAAI,CAAC,CAAA;QAEzD,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU,CAAA;QACpC,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAA;QAEnC,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,YAAY,GAAG,CAAC,CAAA;QACpB,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,MAAM,IAAI,GAAG,CAAC,GAAW,EAAE,UAAkC,EAAE,KAAa,EAAQ,EAAE;YACrF,MAAM,SAAS,GAAG,cAAc,CAAC,kBAAkB,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,IAAI,IAAI,OAAO,EAAE,EAAE,EAAE,CAAC,CAAA;YAE5G,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,kBAAkB;gBAC1B,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EAAE,8FAA8F;aACvG,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,aAAa,EAAE,sBAAsB,CAAC,EAAE,CAAC;gBAC5E,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC,CAAA;QAED,MAAM,sBAAsB,GAAG;YAC9B,QAAQ,EAAE,eAAe,CAAC,IAAI,EAAE,aAAa,CAAC;YAC9C,OAAO,EAAE,aAAa,CAAC,QAAQ;YAC/B,qGAAqG;YACrG,4CAA4C;YAC5C,2DAA2D;YAC3D,YAAY,EAAE,IAAI;SAClB,CAAA;QAED,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAM,CAAC,EAAE,CAAC;YACnD,IAAI,EAAE,CAAA;YACN,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAChD,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAA;YAC7D,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAChD,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAEpD,IAAI,QAAQ,EAAE,CAAC;gBACd,UAAU,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;YACzB,CAAC;YAED,IAAI,QAAQ,EAAE,CAAC;gBACd,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,CAAC,MAAM,EAAE,CAAC;gBACb,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,QAAQ,CAAC,GAAG,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzC,YAAY,EAAE,CAAA;gBAEd,SAAQ;YACT,CAAC;YAED,IAAI,MAAM,EAAE,GAAG,WAAW,EAAE,CAAC;gBAC5B,IAAI,MAAM,EAAE,GAAG,GAAG,IAAI,UAAU,CAAC,IAAI,EAAE,CAAC;oBACvC,MAAM,GAAG,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,UAAU,CAAC,IAAI,CAAC,CAAE,CAAA;oBAEpE,IAAI,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,GAAG,EAAE,EAAE,uBAAuB,CAAC,CAAA;gBACtD,CAAC;qBAAM,IAAI,SAAS,CAAC,IAAI,EAAE,CAAC;oBAC3B,MAAM,EAAE,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,SAAS,CAAC,IAAI,CAAC,CAAE,CAAA;oBAEjE,IAAI,CAAC,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,EAAE,uBAAuB,CAAC,CAAA;gBACpD,CAAC;gBAED,SAAQ;YACT,CAAC;YAED,IAAI,CAAC,MAAM,IAAI,MAAM,EAAE,GAAG,cAAc,EAAE,CAAC;gBAC1C,IAAI,CAAC,MAAM,EAAE,EAAE,MAAM,EAAE,EAAE,aAAa,CAAC,CAAA;YACxC,CAAC;iBAAM,CAAC;gBACP,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW,CAAA;gBAE7D,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAA;gBACjD,MAAM,MAAM,GAAG,MAAM,IAAI,mBAAmB,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,CAAA;gBAElE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;oBACjC,IAAI,CAAC,GAAG,MAAM,IAAI,MAAM,EAAE,EAAE,EAAE,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,EAAE,KAAK,CAAC,CAAA;gBACrE,CAAC;YACF,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,CAAA;IAChD,CAAC;CACD,CAAA"}

package/out/no/recipes/street/led.d.ts

@@ -3,22 +3,19 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `no-street-led` — the Norwegian street-led slice, third orthography of the #901
- *   leading-name-boundary family. The post-#920 NO row-read measured the residual as
- *   ORDER-SENSITIVE decode: street-led forms ("Tangavegen 40, 5620 Tørvikbygd") carry a 30%
- *   residual vs pc-first's 7% — the diacritic street head (…vegen/…veien with ø/å/æ) is the same
- *   leading-name-before-comma boundary the FR bare-street and SI village forms exercise, in a
- *   third orthography. All three real orders cycle per tuple (balanced polarity — the v1.9.9
- *   lesson: a one-order gradient loses at convergence):
+ * `no-street-led`: the Norwegian street-led recipe.
  *
- *   1. canonical  "«st» «n», «pc» «city»"   (the 30% residual class — the lesson)
- *   2. city-first "«city», «pc», «st» «n»"
- *   3. pc-first   "«pc» «city», «st» «n»"   (the 7% floor — the anchor the others converge to)
+ * Street-led forms ("Tangavegen 40, 5620 Tørvikbygd") have a larger residual than the pc-first form.
+ * A diacritic street head (…vegen/…veien with ø/å/æ) creates a leading-name-before-comma boundary.
+ * The recipe cycles each tuple through all three real orders to balance polarity.
+ *
+ * 1. canonical  "«st» «n», «pc» «city»"   (the larger residual class)
+ * 2. city-first "«city», «pc», «st» «n»"
+ * 3. pc-first   "«pc» «city», «st» «n»"   (the floor the others converge to)
  */
 import { type CorpusRecipe } from "#recipes/scaffold";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Recipe registered with the corpus builder.
  */
 export declare const noStreetLedRecipe: CorpusRecipe;
 //# sourceMappingURL=led.d.ts.map

package/out/no/recipes/street/led.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"led.d.ts","sourceRoot":"","sources":["../../../../lib/no/recipes/street/led.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAKH,OAAO,EAA4C,KAAK,YAAY,EAAiB,MAAM,mBAAmB,CAAA;AAE9G;;;GAGG;AACH,eAAO,MAAM,iBAAiB,EAAE,YA2G/B,CAAA"}
+{"version":3,"file":"led.d.ts","sourceRoot":"","sources":["../../../../lib/no/recipes/street/led.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAKH,OAAO,EAKN,KAAK,YAAY,EAEjB,MAAM,mBAAmB,CAAA;AAI1B;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,YAwH/B,CAAA"}

package/out/no/recipes/street/led.js

@@ -3,24 +3,23 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `no-street-led` — the Norwegian street-led slice, third orthography of the #901
- *   leading-name-boundary family. The post-#920 NO row-read measured the residual as
- *   ORDER-SENSITIVE decode: street-led forms ("Tangavegen 40, 5620 Tørvikbygd") carry a 30%
- *   residual vs pc-first's 7% — the diacritic street head (…vegen/…veien with ø/å/æ) is the same
- *   leading-name-before-comma boundary the FR bare-street and SI village forms exercise, in a
- *   third orthography. All three real orders cycle per tuple (balanced polarity — the v1.9.9
- *   lesson: a one-order gradient loses at convergence):
+ * `no-street-led`: the Norwegian street-led recipe.
  *
- *   1. canonical  "«st» «n», «pc» «city»"   (the 30% residual class — the lesson)
- *   2. city-first "«city», «pc», «st» «n»"
- *   3. pc-first   "«pc» «city», «st» «n»"   (the 7% floor — the anchor the others converge to)
+ * Street-led forms ("Tangavegen 40, 5620 Tørvikbygd") have a larger residual than the pc-first form.
+ * A diacritic street head (…vegen/…veien with ø/å/æ) creates a leading-name-before-comma boundary.
+ * The recipe cycles each tuple through all three real orders to balance polarity.
+ *
+ * 1. canonical  "«st» «n», «pc» «city»"   (the larger residual class)
+ * 2. city-first "«city», «pc», «st» «n»"
+ * 3. pc-first   "«pc» «city», «st» «n»"   (the floor the others converge to)
  */
 import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
 import { TextSpliterator } from "spliterator";
-import { alignAndWrite, foldNOSurface, readTuples, sliceSourceID } from "#recipes/scaffold";
+import { alignAndWrite, foldNOSurface, readTuples, requireRegister, recipeSourceID, } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
+import { SurfaceOrigin } from "#types";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * Recipe registered with the corpus builder.
  */
 export const noStreetLedRecipe = {
     name: "no-street-led",
@@ -35,21 +34,19 @@ export const noStreetLedRecipe = {
     ],
     async run(opts, write) {
         makeMulberry32(opts.seed);
-        // THE SPLIT (ported from fr-fragment, #727 T2). Without it this slice trains on all 10,697
-        // NO surfaces, 1,952 of which the digit board reserves — so a Norway retrain would grade
-        // memorization of `Hallingrudveien` while claiming to measure the boundary form. There is no
-        // safe default: source-disjoint by street SURFACE is the discipline, so the flag throws.
+        // The split excludes every surface the digit board reserves, so a retrain
+        // grades the boundary form on unseen streets.
+        // There is no safe default, so the flag throws.
         const excludePath = opts.excludeSurfaces;
         if (!excludePath) {
             throw new Error("no-street-led: --exclude-surfaces is REQUIRED. Pass the NO digit board's reserved list " +
-                "(mailwoman/eval-harness/fixtures/no-digits.surfaces.txt) or this slice trains on its own eval set. " +
+                "(mailwoman/eval-harness/fixtures/no-digits.surfaces.txt) or this recipe trains on its own eval set. " +
                 "Source-disjoint by street SURFACE is the split discipline; there is no safe default.");
         }
         const excluded = new Set();
         for await (const line of TextSpliterator.fromAsync(excludePath)) {
-            const trimmed = line.trim();
-            if (trimmed && !trimmed.startsWith("#")) {
-                excluded.add(trimmed);
+            if (!line.startsWith("#")) {
+                excluded.add(line);
             }
         }
         if (!excluded.size)
@@ -58,6 +55,10 @@ export const noStreetLedRecipe = {
         let emitted = 0;
         let skipped = 0;
         let contaminated = 0;
+        const NO_STREET_LED_PROVENANCE = {
+            register: requireRegister(opts, "no-street-led"),
+            surface: SurfaceOrigin.Composed,
+        };
         for await (const t of readTuples(opts.input)) {
             read++;
             const street = String(t.street ?? "").trim();
@@ -90,18 +91,27 @@ export const noStreetLedRecipe = {
             else {
                 raw = `${postcode} ${city}, ${street} ${number}`;
             }
-            const source_id = sliceSourceID("synth-no-street-led", { ...components, o: String(order), v: String(read) });
+            // One resolution for both, because `source_id` includes the source as its prefix
+            // and a pair that disagreed would name a source no row of this output declares.
+            const source = defaultRecipeSource("synth-no-street-led");
+            const source_id = recipeSourceID(source, { ...components, o: String(order), v: String(read) });
             const canonical = {
                 raw,
                 components,
                 country: "NO",
                 locale: "nb-NO",
-                source: "synth-no-street-led",
+                source,
                 source_id,
                 corpus_version: "0.10.0",
                 license: "Synthetic — no-street-led; (street, number, postcode, city) from OpenAddresses NO (per-source attribution in the model card)",
             };
-            if (alignAndWrite(write, canonical, "no-street-led")) {
+            // Per row, because the id names this tuple's record.
+            // Tuple files created before `sourceID` have no identifier to forward.
+            // A `null` base ID preserves that absence.
+            if (alignAndWrite(write, canonical, "no-street-led", {
+                ...NO_STREET_LED_PROVENANCE,
+                baseSourceID: t.sourceID ?? null,
+            })) {
                 emitted++;
             }
             else {

package/out/no/recipes/street/led.js.map

@@ -1 +1 @@
-{"version":3,"file":"led.js","sourceRoot":"","sources":["../../../../lib/no/recipes/street/led.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACpE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,UAAU,EAAqB,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAE9G;;;GAGG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAiB;IAC9C,IAAI,EAAE,eAAe;IACrB,WAAW,EAAE,6FAA6F;IAC1G,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE;QACR;YACC,IAAI,EAAE,2BAA2B;YACjC,WAAW,EACV,uDAAuD;gBACvD,2FAA2F;SAC5F;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAEzB,2FAA2F;QAC3F,yFAAyF;QACzF,6FAA6F;QAC7F,yFAAyF;QACzF,MAAM,WAAW,GAAG,IAAI,CAAC,eAAe,CAAA;QAExC,IAAI,CAAC,WAAW,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CACd,yFAAyF;gBACxF,qGAAqG;gBACrG,sFAAsF,CACvF,CAAA;QACF,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;QAElC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAAC;YACjE,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAA;YAE3B,IAAI,OAAO,IAAI,CAAC,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBACzC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;YACtB,CAAC;QACF,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,WAAW,sBAAsB,CAAC,CAAA;QAE5G,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,YAAY,GAAG,CAAC,CAAA;QAEpB,IAAI,KAAK,EAAE,MAAM,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,KAAM,CAAC,EAAE,CAAC;YAC/C,IAAI,EAAE,CAAA;YACN,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAEhD,IAAI,CAAC,MAAM,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAC9C,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,sDAAsD;YACtD,IAAI,QAAQ,CAAC,GAAG,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzC,YAAY,EAAE,CAAA;gBAEd,SAAQ;YACT,CAAC;YAED,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAA;YACtB,IAAI,GAAW,CAAA;YAEf,MAAM,UAAU,GAA2B;gBAC1C,MAAM;gBACN,YAAY,EAAE,MAAM;gBACpB,QAAQ;gBACR,QAAQ,EAAE,IAAI;aACd,CAAA;YAED,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACjB,GAAG,GAAG,GAAG,MAAM,IAAI,MAAM,KAAK,QAAQ,IAAI,IAAI,EAAE,CAAA;YACjD,CAAC;iBAAM,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACxB,GAAG,GAAG,GAAG,IAAI,KAAK,QAAQ,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YAClD,CAAC;iBAAM,CAAC;gBACP,GAAG,GAAG,GAAG,QAAQ,IAAI,IAAI,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YACjD,CAAC;YAED,MAAM,SAAS,GAAG,aAAa,CAAC,qBAAqB,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAE5G,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM,EAAE,qBAAqB;gBAC7B,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EACN,8HAA8H;aAC/H,CAAA;YAED,IAAI,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,eAAe,CAAC,EAAE,CAAC;gBACtD,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,CAAA;IAChD,CAAC;CACD,CAAA"}
+{"version":3,"file":"led.js","sourceRoot":"","sources":["../../../../lib/no/recipes/street/led.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;GAcG;AAEH,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACpE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EACN,aAAa,EACb,aAAa,EACb,UAAU,EACV,eAAe,EAEf,cAAc,GACd,MAAM,mBAAmB,CAAA;AAC1B,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AACtD,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEtC;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAiB;IAC9C,IAAI,EAAE,eAAe;IACrB,WAAW,EAAE,6FAA6F;IAC1G,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE;QACR;YACC,IAAI,EAAE,2BAA2B;YACjC,WAAW,EACV,uDAAuD;gBACvD,2FAA2F;SAC5F;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAEzB,0EAA0E;QAC1E,8CAA8C;QAC9C,gDAAgD;QAChD,MAAM,WAAW,GAAG,IAAI,CAAC,eAAe,CAAA;QAExC,IAAI,CAAC,WAAW,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CACd,yFAAyF;gBACxF,sGAAsG;gBACtG,sFAAsF,CACvF,CAAA;QACF,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;QAElC,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,eAAe,CAAC,SAAS,CAAC,WAAW,CAAC,EAAE,CAAC;YACjE,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;gBAC3B,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;YACnB,CAAC;QACF,CAAC;QAED,IAAI,CAAC,QAAQ,CAAC,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,WAAW,sBAAsB,CAAC,CAAA;QAE5G,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,YAAY,GAAG,CAAC,CAAA;QAEpB,MAAM,wBAAwB,GAAG;YAChC,QAAQ,EAAE,eAAe,CAAC,IAAI,EAAE,eAAe,CAAC;YAChD,OAAO,EAAE,aAAa,CAAC,QAAQ;SAC/B,CAAA;QAED,IAAI,KAAK,EAAE,MAAM,CAAC,IAAI,UAAU,CAAC,IAAI,CAAC,KAAM,CAAC,EAAE,CAAC;YAC/C,IAAI,EAAE,CAAA;YACN,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC5C,MAAM,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAEhD,IAAI,CAAC,MAAM,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAC9C,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,sDAAsD;YACtD,IAAI,QAAQ,CAAC,GAAG,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;gBACzC,YAAY,EAAE,CAAA;gBAEd,SAAQ;YACT,CAAC;YAED,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAA;YACtB,IAAI,GAAW,CAAA;YAEf,MAAM,UAAU,GAA2B;gBAC1C,MAAM;gBACN,YAAY,EAAE,MAAM;gBACpB,QAAQ;gBACR,QAAQ,EAAE,IAAI;aACd,CAAA;YAED,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACjB,GAAG,GAAG,GAAG,MAAM,IAAI,MAAM,KAAK,QAAQ,IAAI,IAAI,EAAE,CAAA;YACjD,CAAC;iBAAM,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBACxB,GAAG,GAAG,GAAG,IAAI,KAAK,QAAQ,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YAClD,CAAC;iBAAM,CAAC;gBACP,GAAG,GAAG,GAAG,QAAQ,IAAI,IAAI,KAAK,MAAM,IAAI,MAAM,EAAE,CAAA;YACjD,CAAC;YAED,iFAAiF;YACjF,gFAAgF;YAChF,MAAM,MAAM,GAAG,mBAAmB,CAAC,qBAAqB,CAAC,CAAA;YACzD,MAAM,SAAS,GAAG,cAAc,CAAC,MAAM,EAAE,EAAE,GAAG,UAAU,EAAE,CAAC,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAA;YAE9F,MAAM,SAAS,GAAG;gBACjB,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM;gBACN,SAAS;gBACT,cAAc,EAAE,QAAQ;gBACxB,OAAO,EACN,8HAA8H;aAC/H,CAAA;YAED,qDAAqD;YACrD,uEAAuE;YACvE,2CAA2C;YAC3C,IACC,aAAa,CAAC,KAAK,EAAE,SAAS,EAAE,eAAe,EAAE;gBAChD,GAAG,wBAAwB;gBAC3B,YAAY,EAAE,CAAC,CAAC,QAAQ,IAAI,IAAI;aAChC,CAAC,EACD,CAAC;gBACF,OAAO,EAAE,CAAA;YACV,CAAC;iBAAM,CAAC;gBACP,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,YAAY,EAAE,CAAA;IAChD,CAAC;CACD,CAAA"}

package/out/no/recipes/street.d.ts

@@ -0,0 +1,19 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   `no-street` recipe — synthetic no-street counter-example rows: tuples →
+ *   {@link synthesizeNoStreetRow} → aligned LabeledRow. The corpus-side counterweight to the
+ *   `invented-street` source — venue+admin and admin-only rows with explicit absence of any
+ *   street-side component.
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+/**
+ * Recipe registered with the corpus builder.
+ *
+ * The file header describes the parse behavior this recipe exercises.
+ * `description` below shows the generated surface form.
+ */
+export declare const noStreetRecipe: CorpusRecipe;
+//# sourceMappingURL=street.d.ts.map

package/out/no/recipes/street.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"street.d.ts","sourceRoot":"","sources":["../../../lib/no/recipes/street.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAIH,OAAO,EAKN,KAAK,YAAY,EACjB,MAAM,mBAAmB,CAAA;AAmB1B;;;;;GAKG;AACH,eAAO,MAAM,cAAc,EAAE,YAgE5B,CAAA"}

package/out/no/recipes/{street/index.js → street.js}

@@ -3,18 +3,34 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   `no-street` slice recipe — synthetic NO-street counter-example rows: tuples →
+ *   `no-street` recipe — synthetic no-street counter-example rows: tuples →
  *   {@link synthesizeNoStreetRow} → aligned LabeledRow. The corpus-side counterweight to the
- *   synth-street slice that drove v0.6.1's `dependent_locality` regression — venue+admin and
- *   admin-only rows with explicit absence of any street-side component. Ported from
- *   scripts/build-no-street-slice.mjs.
+ *   `invented-street` source — venue+admin and admin-only rows with explicit absence of any
+ *   street-side component.
  */
 import { makeLcg } from "@mailwoman/core/utils";
-import { alignAndWrite, readTuples, sliceSourceID, SYNTHETIC_TUPLE_LICENSE as LICENSE, } from "#recipes/scaffold";
+import { alignAndWrite, readTuples, recipeSourceID, SYNTHETIC_TUPLE_LICENSE as LICENSE, } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
 import { synthesizeNoStreetRow } from "#synthesizers/no-street";
+import { SurfaceOrigin } from "#types";
 /**
- * Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
+ * A venue or admin-only row asserting that no street-side component is present.
+ *
+ * The row uses components from the `--input` tuples.
+ * It teaches the absence of a street component instead of representing a source record.
+ * No register asserts the row.
+ */
+const NO_STREET_PROVENANCE = {
+    register: null,
+    surface: SurfaceOrigin.Invented,
+    // An invented row has no underlying record to name.
+    baseSourceID: null,
+};
+/**
+ * Recipe registered with the corpus builder.
+ *
+ * The file header describes the parse behavior this recipe exercises.
+ * `description` below shows the generated surface form.
  */
 export const noStreetRecipe = {
     name: "no-street",
@@ -23,9 +39,9 @@ export const noStreetRecipe = {
     async run(opts, write) {
         if (!opts.input)
             throw new Error("no-street recipe requires --input <tuples.jsonl>");
-        // Legacy build-no-street-slice.mjs seeded the LCG via makeRandom(opts.seed) (s = seed).
+        // The legacy build script seeded the LCG via makeRandom(opts.seed) (s = seed).
         const random = makeLcg(opts.seed);
-        const source = opts.sourceName ?? "synth-no-street";
+        const source = opts.sourceName ?? defaultRecipeSource("synth-no-street");
         let read = 0;
         let emitted = 0;
         let skipped = 0;
@@ -47,7 +63,7 @@ export const noStreetRecipe = {
                     country: tuple.country,
                     locale: synth.locale,
                     source,
-                    source_id: sliceSourceID(source, {
+                    source_id: recipeSourceID(source, {
                         locality: tuple.locality,
                         region: tuple.region,
                         postcode: tuple.postcode,
@@ -57,7 +73,7 @@ export const noStreetRecipe = {
                     }),
                     corpus_version: "0.4.0",
                     license: LICENSE,
-                }, synth.template);
+                }, synth.template, NO_STREET_PROVENANCE);
                 if (ok) {
                     emitted++;
                 }
@@ -69,4 +85,4 @@ export const noStreetRecipe = {
         return { read, emitted, skipped };
     },
 };
-//# sourceMappingURL=index.js.map
+//# sourceMappingURL=street.js.map

package/out/no/recipes/street.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"street.js","sourceRoot":"","sources":["../../../lib/no/recipes/street.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAA;AAE/C,OAAO,EACN,aAAa,EACb,UAAU,EACV,cAAc,EACd,uBAAuB,IAAI,OAAO,GAElC,MAAM,mBAAmB,CAAA;AAC1B,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AACtD,OAAO,EAAE,qBAAqB,EAA0B,MAAM,yBAAyB,CAAA;AACvF,OAAO,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEtC;;;;;;GAMG;AACH,MAAM,oBAAoB,GAAG;IAC5B,QAAQ,EAAE,IAAI;IACd,OAAO,EAAE,aAAa,CAAC,QAAQ;IAC/B,oDAAoD;IACpD,YAAY,EAAE,IAAI;CAClB,CAAA;AAED;;;;;GAKG;AACH,MAAM,CAAC,MAAM,cAAc,GAAiB;IAC3C,IAAI,EAAE,WAAW;IACjB,WAAW,EAAE,qFAAqF;IAClG,IAAI,EAAE,QAAQ;IACd,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,MAAM,IAAI,KAAK,CAAC,kDAAkD,CAAC,CAAA;QACpF,+EAA+E;QAC/E,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACjC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,mBAAmB,CAAC,iBAAiB,CAAC,CAAA;QACxE,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAClD,IAAI,EAAE,CAAA;YAEN,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBAC3E,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,KAAK,GAAG,qBAAqB,CAAC,KAA0B,EAAE,EAAE,MAAM,EAAE,CAAC,CAAA;gBAE3E,IAAI,CAAC,KAAK,EAAE,CAAC;oBACZ,OAAO,EAAE,CAAA;oBAET,SAAQ;gBACT,CAAC;gBAED,MAAM,EAAE,GAAG,aAAa,CACvB,KAAK,EACL;oBACC,GAAG,EAAE,KAAK,CAAC,GAAG;oBACd,UAAU,EAAE,KAAK,CAAC,UAAU;oBAC5B,OAAO,EAAE,KAAK,CAAC,OAAO;oBACtB,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,MAAM;oBACN,SAAS,EAAE,cAAc,CAAC,MAAM,EAAE;wBACjC,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,MAAM,EAAE,KAAK,CAAC,MAAM;wBACpB,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,OAAO,EAAE,KAAK,CAAC,OAAO;wBACtB,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC;qBACZ,CAAC;oBACF,cAAc,EAAE,OAAO;oBACvB,OAAO,EAAE,OAAO;iBAChB,EACD,KAAK,CAAC,QAAQ,EACd,oBAAoB,CACpB,CAAA;gBAED,IAAI,EAAE,EAAE,CAAC;oBACR,OAAO,EAAE,CAAA;gBACV,CAAC;qBAAM,CAAC;oBACP,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAClC,CAAC;CACD,CAAA"}