@mailwoman/corpus 9.4.0 → 10.1.0

package/out/fr/recipes/lieudit.js

@@ -0,0 +1,231 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `fr-lieudit` recipe: FR lieu-dit (hamlet/place) `dependent_locality` coverage.
+ *
+ * Streams every BAN `adresses-<dept>.csv` dump under `--ban-dir` through `@mailwoman/ban/sdk`'s
+ * `extractBANAddrPoints`, which surfaces a cleaned `lieuDit` per record. Only rows carrying a clean
+ * lieu-dit survive into the pool.
+ *
+ * The recipe maps lieu-dit to `dependent_locality` and commune to `locality`. It renders to match the
+ * formatter's FR `place`-slot convention. House and street go on line 1.
+ * The lieu-dit appears by itself on line 2. Postcode and commune go on line 3.
+ * This follows French postal convention (La Poste's line 5).
+ *
+ * The pool is read in full and Fisher-Yates shuffled with the seeded prng before slicing to
+ * `--count`. With-replacement draws at a large `--count` would produce a large duplicate rate.
+ *
+ * A row's jurisdiction comes from its INSEE commune code, as it does in the `ban` adapter reading
+ * the same files. The BAN fetcher downloads departments 971, 972, 973, 974 and 976 beside the
+ * metropolitan ones, so this recipe emits Guadeloupe, Martinique, French Guiana, Réunion and
+ * Mayotte rows, and labeling them `FR` would mislabel their country and their locale.
+ */
+import { extractBANAddrPoints } from "@mailwoman/ban/sdk/extract";
+import { formatAddress } from "@mailwoman/codex/address/format";
+import { COUNTRY_SURFACE_FORMS } from "@mailwoman/codex/country";
+import { dataRootPath } from "@mailwoman/core/data-root";
+import { stringifyJSON } from "@mailwoman/core/json";
+import { sample, shuffleWith } from "@mailwoman/core/random";
+import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
+import { PathBuilder } from "path-ts";
+import { Globerator } from "spliterator/node/fs";
+import { stableSourceID } from "#adapters/source-id";
+import { composeHouseNumber } from "#adapters/street-line";
+import { countryOfInseeCode } from "#fr/insee-country";
+import { decomposeFrStreet } from "#fr/street-decompose";
+import { defaultRecipeSource } from "#recipes/sources";
+import { alignRow } from "#utils";
+/**
+ * Matches the `ban` adapter's Tier-B election for BAN data.
+ */
+const DEFAULT_LICENSE = "Licence Ouverte 2.0";
+/**
+ * Enumerate `adresses-<dept>.csv[.gz]` files in `banDir`, one path per département.
+ *
+ * Excludes the `merged` and `france` aggregates.
+ * They duplicate the per-département rows.
+ *
+ * When both a `.csv` and a `.csv.gz` exist for the same department, the uncompressed `.csv` wins.
+ */
+async function departementFiles(banDir) {
+    const directory = PathBuilder.from(banDir);
+    const byDept = new Map();
+    for (const name of await Globerator.from("*", { cwd: directory, absolute: false }).toSorted()) {
+        const m = /^adresses-(.+?)\.csv(\.gz)?$/.exec(name);
+        if (!m)
+            continue;
+        const dept = m[1];
+        if (dept === "merged" || dept === "france")
+            continue;
+        const existing = byDept.get(dept);
+        if (!existing || (existing.endsWith(".gz") && !name.endsWith(".gz"))) {
+            byDept.set(dept, name);
+        }
+    }
+    return [...byDept.keys()].toSorted().map((dept) => directory(byDept.get(dept)));
+}
+/**
+ * Stream every département file, keeping only rows with a clean `lieuDit`
+ * (junk/dup filtering lives in `ban/sdk`).
+ */
+async function readLieuDitPool(banDir) {
+    const files = await departementFiles(banDir);
+    if (!files.length) {
+        throw new Error(`No BAN adresses-<dept>.csv files found in ${banDir} — fetch BAN first (\`mailwoman corpus fetch ban\`).`);
+    }
+    const pool = [];
+    let scanned = 0;
+    for (const path of files) {
+        let deptCount = 0;
+        for await (const rec of extractBANAddrPoints(path)) {
+            scanned++;
+            if (!rec.lieuDit || !rec.city)
+                continue;
+            pool.push({
+                numero: rec.numero,
+                rep: rec.rep,
+                street: rec.street,
+                postcode: rec.postcode,
+                locality: rec.city,
+                dependentLocality: rec.lieuDit,
+                codeInsee: rec.codeInsee,
+            });
+            deptCount++;
+        }
+        console.error(`  ${path}: ${deptCount.toLocaleString()} clean lieu-dit rows`);
+    }
+    console.error(`  scanned ${scanned.toLocaleString()} BAN rows across ${files.length} départements → pool=${pool.length.toLocaleString()}`);
+    return pool;
+}
+/**
+ * Recipe registered with the corpus builder.
+ */
+export const frLieuditRecipe = {
+    name: "fr-lieudit",
+    description: "FR lieu-dit rows: BAN nom_ld → dependent_locality (commune → locality), lieu-dit on its own line",
+    mode: "generate",
+    options: [
+        {
+            flag: "--ban-dir <dir>",
+            description: "BAN adresses-<dept>.csv directory. Default $MAILWOMAN_DATA_ROOT/corpus/sources/ban",
+        },
+        {
+            flag: "--country-fraction <f>",
+            description: "Fraction of rows that append an explicit 'France' surface form + a `country` component. Default 0",
+        },
+    ],
+    async run(opts, write) {
+        const random = makeMulberry32(opts.seed);
+        const source = opts.sourceName ?? defaultRecipeSource("synth-fr-lieudit");
+        const count = opts.count ?? 800_000;
+        const banDir = opts.banDir ?? dataRootPath("corpus", "sources", "ban");
+        const countryFraction = opts.countryFraction ?? 0;
+        if (!(countryFraction >= 0 && countryFraction <= 1)) {
+            throw new Error(`--country-fraction must be in [0, 1], got ${countryFraction}`);
+        }
+        const pool = await readLieuDitPool(banDir);
+        if (!pool.length) {
+            throw new Error(`No clean lieu-dit rows found under ${banDir} — see ban/sdk/extract.ts's cleanLieuDit filter.`);
+        }
+        // The recipe shares one mulberry32 stream between this shuffle and the country-fraction
+        // draw below, so a fresh generator here would move every later draw and the committed rows.
+        shuffleWith(pool, random);
+        const selected = pool.slice(0, Math.min(count, pool.length));
+        let emitted = 0;
+        let skipped = 0;
+        let countryAppended = 0;
+        /**
+         * Rows that drew a country append where codex states no surface form for the jurisdiction.
+         *
+         * Counted so the run reports the gap rather than passing over it.
+         */
+        let countrySurfaceFormAbsent = 0;
+        const byCountry = new Map();
+        for (const t of selected) {
+            const house = composeHouseNumber(t.numero, t.rep ?? "", " ");
+            const country = countryOfInseeCode(t.codeInsee ?? "");
+            const decomposed = decomposeFrStreet(t.street);
+            const components = {
+                house_number: house,
+                dependent_locality: t.dependentLocality,
+                locality: t.locality,
+            };
+            if (decomposed.prefix) {
+                components.street_prefix = decomposed.prefix;
+            }
+            if (decomposed.street) {
+                components.street = decomposed.street;
+            }
+            if (t.postcode) {
+                components.postcode = t.postcode;
+            }
+            // The envelope form is the house and street line, the lieu-dit by itself on
+            // its own line, then the postcode and commune line.
+            // That is La Poste's line 5.
+            let raw = formatAddress(components, country);
+            if (!raw) {
+                skipped++;
+                continue;
+            }
+            // Country-append: ~`countryFraction` of the time, append an explicit country surface
+            // form onto the trailing (postcode+commune) line plus a `country` component.
+            // The model relearns to emit country when present without over-firing it on the country-less rows.
+            // `countryFraction <= 0` (the default) never draws from `random`,
+            // so the byte-stream is unaffected when the flag is unset.
+            //
+            // Codex states surface forms for `FR` alone among the jurisdictions BAN covers, so an overseas
+            // row draws from `random` and then leaves `raw` and `components.country` as they were.
+            // The draw happens either way, which keeps the byte-stream a function of `countryFraction`
+            // rather than of the mixture of jurisdictions in the input files.
+            if (countryFraction > 0 && random() < countryFraction) {
+                const forms = COUNTRY_SURFACE_FORMS[country];
+                if (forms) {
+                    const form = sample(forms, random);
+                    raw = `${raw}, ${form}`;
+                    components.country = form;
+                    countryAppended++;
+                }
+                else {
+                    countrySurfaceFormAbsent++;
+                }
+            }
+            const sourceID = stableSourceID(source, {
+                street: t.street,
+                house_number: house,
+                dependent_locality: t.dependentLocality,
+                locality: t.locality,
+                postcode: t.postcode ?? undefined,
+            });
+            const canonical = {
+                raw,
+                components,
+                country,
+                // Every jurisdiction BAN publishes writes its addresses in French,
+                // so the language is constant and the region carries the jurisdiction.
+                locale: `fr-${country}`,
+                source,
+                source_id: sourceID,
+                corpus_version: "",
+                license: DEFAULT_LICENSE,
+            };
+            const aligned = alignRow(canonical);
+            if (aligned.kind !== "labeled" || !aligned.row) {
+                skipped++;
+                continue;
+            }
+            write(stringifyJSON({ ...aligned.row, synth_method: source, synth_base_id: null }));
+            byCountry.set(country, (byCountry.get(country) ?? 0) + 1);
+            emitted++;
+        }
+        console.error(`  emitted=${emitted} skipped=${skipped} country-appended=${countryAppended} pool=${pool.length}`);
+        const jurisdictions = [...byCountry].toSorted((a, b) => b[1] - a[1]);
+        console.error(`  jurisdictions: ${jurisdictions.map(([code, n]) => `${code}=${n}`).join(" ")}`);
+        if (countrySurfaceFormAbsent > 0) {
+            console.error(`  ${countrySurfaceFormAbsent} rows drew a country append and codex states no surface form for their jurisdiction`);
+        }
+        return { emitted, skipped };
+    },
+};
+//# sourceMappingURL=lieudit.js.map

package/out/fr/recipes/lieudit.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"lieudit.js","sourceRoot":"","sources":["../../../lib/fr/recipes/lieudit.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAEH,OAAO,EAAE,oBAAoB,EAAE,MAAM,4BAA4B,CAAA;AACjE,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAA;AAE/D,OAAO,EAAE,qBAAqB,EAAE,MAAM,0BAA0B,CAAA;AAChE,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM,wBAAwB,CAAA;AAC5D,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACpE,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAEhD,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAA;AAC1D,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAA;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,sBAAsB,CAAA;AAExD,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AAEtD,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAEjC;;GAEG;AACH,MAAM,eAAe,GAAG,qBAAqB,CAAA;AAuB7C;;;;;;;GAOG;AACH,KAAK,UAAU,gBAAgB,CAAC,MAAuB;IACtD,MAAM,SAAS,GAAG,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IAC1C,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAA;IAExC,KAAK,MAAM,IAAI,IAAI,MAAM,UAAU,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,GAAG,EAAE,SAAS,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC;QAC/F,MAAM,CAAC,GAAG,8BAA8B,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAEnD,IAAI,CAAC,CAAC;YAAE,SAAQ;QAEhB,MAAM,IAAI,GAAG,CAAC,CAAC,CAAC,CAAE,CAAA;QAElB,IAAI,IAAI,KAAK,QAAQ,IAAI,IAAI,KAAK,QAAQ;YAAE,SAAQ;QAEpD,MAAM,QAAQ,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QAEjC,IAAI,CAAC,QAAQ,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;YACtE,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;QACvB,CAAC;IACF,CAAC;IAED,OAAO,CAAC,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAE,CAAC,CAAC,CAAA;AACjF,CAAC;AAED;;;GAGG;AACH,KAAK,UAAU,eAAe,CAAC,MAAuB;IACrD,MAAM,KAAK,GAAG,MAAM,gBAAgB,CAAC,MAAM,CAAC,CAAA;IAE5C,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QACnB,MAAM,IAAI,KAAK,CACd,6CAA6C,MAAM,sDAAsD,CACzG,CAAA;IACF,CAAC;IAED,MAAM,IAAI,GAAmB,EAAE,CAAA;IAC/B,IAAI,OAAO,GAAG,CAAC,CAAA;IAEf,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,IAAI,SAAS,GAAG,CAAC,CAAA;QAEjB,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,oBAAoB,CAAC,IAAI,CAAC,EAAE,CAAC;YACpD,OAAO,EAAE,CAAA;YAET,IAAI,CAAC,GAAG,CAAC,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI;gBAAE,SAAQ;YAEvC,IAAI,CAAC,IAAI,CAAC;gBACT,MAAM,EAAE,GAAG,CAAC,MAAM;gBAClB,GAAG,EAAE,GAAG,CAAC,GAAG;gBACZ,MAAM,EAAE,GAAG,CAAC,MAAM;gBAClB,QAAQ,EAAE,GAAG,CAAC,QAAQ;gBACtB,QAAQ,EAAE,GAAG,CAAC,IAAI;gBAClB,iBAAiB,EAAE,GAAG,CAAC,OAAO;gBAC9B,SAAS,EAAE,GAAG,CAAC,SAAS;aACxB,CAAC,CAAA;YAEF,SAAS,EAAE,CAAA;QACZ,CAAC;QAED,OAAO,CAAC,KAAK,CAAC,KAAK,IAAI,KAAK,SAAS,CAAC,cAAc,EAAE,sBAAsB,CAAC,CAAA;IAC9E,CAAC;IAED,OAAO,CAAC,KAAK,CACZ,aAAa,OAAO,CAAC,cAAc,EAAE,oBAAoB,KAAK,CAAC,MAAM,wBAAwB,IAAI,CAAC,MAAM,CAAC,cAAc,EAAE,EAAE,CAC3H,CAAA;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAiB;IAC5C,IAAI,EAAE,YAAY;IAClB,WAAW,EAAE,kGAAkG;IAC/G,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE;QACR;YACC,IAAI,EAAE,iBAAiB;YACvB,WAAW,EAAE,oFAAoF;SACjG;QACD;YACC,IAAI,EAAE,wBAAwB;YAC9B,WAAW,EAAE,mGAAmG;SAChH;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,mBAAmB,CAAC,kBAAkB,CAAC,CAAA;QACzE,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,OAAO,CAAA;QACnC,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,IAAI,YAAY,CAAC,QAAQ,EAAE,SAAS,EAAE,KAAK,CAAC,CAAA;QACtE,MAAM,eAAe,GAAG,IAAI,CAAC,eAAe,IAAI,CAAC,CAAA;QAEjD,IAAI,CAAC,CAAC,eAAe,IAAI,CAAC,IAAI,eAAe,IAAI,CAAC,CAAC,EAAE,CAAC;YACrD,MAAM,IAAI,KAAK,CAAC,6CAA6C,eAAe,EAAE,CAAC,CAAA;QAChF,CAAC;QAED,MAAM,IAAI,GAAG,MAAM,eAAe,CAAC,MAAM,CAAC,CAAA;QAE1C,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CAAC,sCAAsC,MAAM,kDAAkD,CAAC,CAAA;QAChH,CAAC;QAED,wFAAwF;QACxF,4FAA4F;QAC5F,WAAW,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;QAEzB,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC,CAAA;QAE5D,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,eAAe,GAAG,CAAC,CAAA;QACvB;;;;WAIG;QACH,IAAI,wBAAwB,GAAG,CAAC,CAAA;QAChC,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAA;QAE3C,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;YAC1B,MAAM,KAAK,GAAG,kBAAkB,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,GAAG,IAAI,EAAE,EAAE,GAAG,CAAC,CAAA;YAC5D,MAAM,OAAO,GAAG,kBAAkB,CAAC,CAAC,CAAC,SAAS,IAAI,EAAE,CAAC,CAAA;YACrD,MAAM,UAAU,GAAG,iBAAiB,CAAC,CAAC,CAAC,MAAM,CAAC,CAAA;YAE9C,MAAM,UAAU,GAA0C;gBACzD,YAAY,EAAE,KAAK;gBACnB,kBAAkB,EAAE,CAAC,CAAC,iBAAiB;gBACvC,QAAQ,EAAE,CAAC,CAAC,QAAQ;aACpB,CAAA;YAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;gBACvB,UAAU,CAAC,aAAa,GAAG,UAAU,CAAC,MAAM,CAAA;YAC7C,CAAC;YAED,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;gBACvB,UAAU,CAAC,MAAM,GAAG,UAAU,CAAC,MAAM,CAAA;YACtC,CAAC;YAED,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC;gBAChB,UAAU,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAA;YACjC,CAAC;YAED,4EAA4E;YAC5E,oDAAoD;YACpD,6BAA6B;YAC7B,IAAI,GAAG,GAAG,aAAa,CAAC,UAAU,EAAE,OAAO,CAAC,CAAA;YAE5C,IAAI,CAAC,GAAG,EAAE,CAAC;gBACV,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,qFAAqF;YACrF,6EAA6E;YAC7E,mGAAmG;YACnG,kEAAkE;YAClE,2DAA2D;YAC3D,EAAE;YACF,+FAA+F;YAC/F,uFAAuF;YACvF,2FAA2F;YAC3F,kEAAkE;YAClE,IAAI,eAAe,GAAG,CAAC,IAAI,MAAM,EAAE,GAAG,eAAe,EAAE,CAAC;gBACvD,MAAM,KAAK,GACV,qBAAqB,CAAC,OAA6C,CAAC,CAAA;gBAErE,IAAI,KAAK,EAAE,CAAC;oBACX,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,CAAA;oBAClC,GAAG,GAAG,GAAG,GAAG,KAAK,IAAI,EAAE,CAAA;oBACvB,UAAU,CAAC,OAAO,GAAG,IAAI,CAAA;oBAEzB,eAAe,EAAE,CAAA;gBAClB,CAAC;qBAAM,CAAC;oBACP,wBAAwB,EAAE,CAAA;gBAC3B,CAAC;YACF,CAAC;YAED,MAAM,QAAQ,GAAG,cAAc,CAAC,MAAM,EAAE;gBACvC,MAAM,EAAE,CAAC,CAAC,MAAM;gBAChB,YAAY,EAAE,KAAK;gBACnB,kBAAkB,EAAE,CAAC,CAAC,iBAAiB;gBACvC,QAAQ,EAAE,CAAC,CAAC,QAAQ;gBACpB,QAAQ,EAAE,CAAC,CAAC,QAAQ,IAAI,SAAS;aACjC,CAAC,CAAA;YAEF,MAAM,SAAS,GAAiB;gBAC/B,GAAG;gBACH,UAAU;gBACV,OAAO;gBACP,mEAAmE;gBACnE,uEAAuE;gBACvE,MAAM,EAAE,MAAM,OAAO,EAAE;gBACvB,MAAM;gBACN,SAAS,EAAE,QAAQ;gBACnB,cAAc,EAAE,EAAE;gBAClB,OAAO,EAAE,eAAe;aACxB,CAAA;YAED,MAAM,OAAO,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAA;YAEnC,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,CAAC,aAAa,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,MAAM,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,CAAC,CAAA;YAEnF,SAAS,CAAC,GAAG,CAAC,OAAO,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;YAEzD,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,CAAC,KAAK,CAAC,aAAa,OAAO,YAAY,OAAO,qBAAqB,eAAe,SAAS,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;QAEhH,MAAM,aAAa,GAAG,CAAC,GAAG,SAAS,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;QAEpE,OAAO,CAAC,KAAK,CAAC,oBAAoB,aAAa,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,IAAI,IAAI,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAE/F,IAAI,wBAAwB,GAAG,CAAC,EAAE,CAAC;YAClC,OAAO,CAAC,KAAK,CACZ,KAAK,wBAAwB,qFAAqF,CAClH,CAAA;QACF,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}

package/out/fr/recipes/order.d.ts

@@ -0,0 +1,15 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Generate canonical and reversed French addresses from real OpenAddresses tuples. Reversed forms
+ *   vary postcode, locality and street order. Optional sub-modes add ordinal house numbers and
+ *   uppercase localities. `--golden` emits a held-out evaluation set.
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+/**
+ * Recipe registered with the corpus builder.
+ */
+export declare const frOrderRecipe: CorpusRecipe;
+//# sourceMappingURL=order.d.ts.map

package/out/fr/recipes/order.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"order.d.ts","sourceRoot":"","sources":["../../../lib/fr/recipes/order.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAUH,OAAO,EAAgB,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AAwEnE;;GAEG;AACH,eAAO,MAAM,aAAa,EAAE,YA6H3B,CAAA"}

package/out/fr/recipes/order.js

@@ -0,0 +1,171 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Generate canonical and reversed French addresses from real OpenAddresses tuples. Reversed forms
+ *   vary postcode, locality and street order. Optional sub-modes add ordinal house numbers and
+ *   uppercase localities. `--golden` emits a held-out evaluation set.
+ */
+import { dataRootPath } from "@mailwoman/core/data-root";
+import { stringifyJSON } from "@mailwoman/core/json";
+import { isPresent } from "@mailwoman/core/objects";
+import { sample } from "@mailwoman/core/random";
+import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
+import { stableSourceID } from "#adapters/source-id";
+import { readOATuples } from "#recipes/scaffold";
+import { defaultRecipeSource } from "#recipes/sources";
+import { alignRow } from "#utils";
+const SOURCE = { zip: dataRootPath("oa-cache", "fr__countrywide.zip"), csv: "fr/countrywide.csv" };
+/**
+ * French house-number ordinal suffixes.
+ */
+const ORDINAL_SUFFIXES = ["bis", "ter", "quater"];
+/**
+ * Probability of adding an ordinal suffix.
+ */
+const ORDINAL_PROB = 0.12;
+/**
+ * Probability of rendering a locality in uppercase.
+ */
+const ALLCAPS_PROB = 0.1;
+/**
+ * Read up to `limit` distinct tuples with house numbers and postcodes.
+ */
+async function readTuples(limit) {
+    return readOATuples(SOURCE, {
+        limit,
+        requirePostcode: true,
+        dedupIncludesPostcode: true,
+        extra: (fields) => fields,
+    });
+}
+/**
+ * Optionally append an ordinal suffix to a house number.
+ */
+function maybeAddOrdinal(random, house_number) {
+    if (random() >= ORDINAL_PROB)
+        return house_number;
+    const suffix = sample(ORDINAL_SUFFIXES, random);
+    // Include lowercase and uppercase suffix forms.
+    return `${house_number} ${random() < 0.5 ? suffix : suffix.toUpperCase()}`;
+}
+/**
+ * Render a tuple in canonical French order.
+ */
+function renderCanonical(hn, street, postcode, locality) {
+    const raw = `${hn} ${street}, ${postcode} ${locality}`;
+    return { raw, components: { house_number: hn, street, postcode, locality } };
+}
+// Cumulative cutoffs for four reversed layouts.
+const REVERSED_VARIANT_A_CUTOFF = 0.25;
+const REVERSED_VARIANT_B_CUTOFF = 0.5;
+const REVERSED_VARIANT_C_CUTOFF = 0.75;
+/**
+ * Recipe registered with the corpus builder.
+ */
+export const frOrderRecipe = {
+    name: "fr-order",
+    description: "French reversed-order rows (#560): real OA FR tuples rendered canonical + 4 postcode-first variants",
+    mode: "generate",
+    options: [
+        { flag: "--reversed-fraction <p>", description: "Fraction rendered reversed-order. Default 0.5" },
+        { flag: "--golden", description: "Emit the held-out reversed-order eval set" },
+    ],
+    async run(opts, write) {
+        if (opts.count == null)
+            throw new Error("fr-order recipe requires --count <N>");
+        const count = opts.count;
+        // Preserve the legacy random stream for generated rows.
+        const random = makeMulberry32(opts.seed);
+        const source = opts.sourceName ?? defaultRecipeSource("synth-fr-order");
+        const reversedFraction = opts.reversedFraction ?? 0.5;
+        // Read extra rows to account for filtering and deduplication.
+        const poolLimit = Math.max(count * 8, 40_000);
+        const pool = await readTuples(poolLimit);
+        console.error(`  ${SOURCE.csv}: ${pool.length} unique tuples (capped read)`);
+        if (!pool.length) {
+            throw new Error(`No FR tuples found — is ${SOURCE.zip} present?`);
+        }
+        let emitted = 0;
+        let skipped = 0;
+        let guard = 0;
+        const N = pool.length;
+        while (emitted < count && guard++ < count * 10) {
+            const base = pool[Math.floor(random() * N)];
+            const { street, postcode } = base;
+            const locality = random() < ALLCAPS_PROB ? base.locality.toUpperCase() : base.locality;
+            const house_number = maybeAddOrdinal(random, base.house_number);
+            // Choose canonical or reversed order.
+            const isReversed = random() < reversedFraction;
+            let rendered;
+            if (isReversed) {
+                const variantRoll = random();
+                let raw;
+                if (variantRoll < REVERSED_VARIANT_A_CUTOFF) {
+                    // Postcode and locality before number and street.
+                    raw = `${postcode} ${locality}, ${house_number} ${street}`;
+                }
+                else if (variantRoll < REVERSED_VARIANT_B_CUTOFF) {
+                    // Locality, postcode, number and street.
+                    raw = `${locality}, ${postcode}, ${house_number} ${street}`;
+                }
+                else if (variantRoll < REVERSED_VARIANT_C_CUTOFF) {
+                    // Locality, number, street and postcode without commas.
+                    raw = `${locality} ${house_number} ${street} ${postcode}`;
+                }
+                else {
+                    // Postcode, number, street, then locality.
+                    raw = `${postcode}, ${house_number} ${street}, ${locality}`;
+                }
+                rendered = { raw, components: { house_number, street, postcode, locality } };
+            }
+            else {
+                rendered = renderCanonical(house_number, street, postcode, locality);
+            }
+            const { raw, components } = rendered;
+            // Require each component value to appear verbatim in the address.
+            const componentValues = Object.values(components).filter(isPresent);
+            if (!componentValues.every((v) => raw.includes(v))) {
+                skipped++;
+                continue;
+            }
+            // Golden output includes parse truth without corpus metadata.
+            if (opts.golden) {
+                write(stringifyJSON({ raw, components, country: "FR" }));
+                emitted++;
+                continue;
+            }
+            const sourceID = stableSourceID(source, {
+                street: components.street,
+                house_number: components.house_number,
+                locality: components.locality,
+                postcode: components.postcode,
+            });
+            const canonical = {
+                raw,
+                components,
+                country: "FR",
+                locale: "fr-FR",
+                source,
+                source_id: sourceID,
+                corpus_version: "0.5.0",
+                license: "OpenAddresses FR countrywide tuples, rendered canonical + reversed-order — see ingest SOURCE",
+            };
+            const aligned = alignRow(canonical);
+            if (aligned.kind !== "labeled" || !aligned.row) {
+                skipped++;
+                continue;
+            }
+            write(stringifyJSON({
+                ...aligned.row,
+                synth_method: "fr-order",
+                synth_order: isReversed ? "reversed" : "canonical",
+                synth_base_id: null,
+            }));
+            emitted++;
+        }
+        return { emitted, skipped };
+    },
+};
+//# sourceMappingURL=order.js.map

package/out/fr/recipes/order.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"order.js","sourceRoot":"","sources":["../../../lib/fr/recipes/order.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAGH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,SAAS,EAAE,MAAM,yBAAyB,CAAA;AACnD,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAC/C,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,YAAY,EAAqB,MAAM,mBAAmB,CAAA;AACnE,OAAO,EAAE,mBAAmB,EAAE,MAAM,kBAAkB,CAAA;AAEtD,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAEjC,MAAM,MAAM,GAAG,EAAE,GAAG,EAAE,YAAY,CAAC,UAAU,EAAE,qBAAqB,CAAC,EAAE,GAAG,EAAE,oBAAoB,EAAE,CAAA;AAElG;;GAEG;AACH,MAAM,gBAAgB,GAAsB,CAAC,KAAK,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAA;AACpE;;GAEG;AACH,MAAM,YAAY,GAAG,IAAI,CAAA;AACzB;;GAEG;AACH,MAAM,YAAY,GAAG,GAAG,CAAA;AAYxB;;GAEG;AACH,KAAK,UAAU,UAAU,CAAC,KAAa;IACtC,OAAO,YAAY,CAAC,MAAM,EAAE;QAC3B,KAAK;QACL,eAAe,EAAE,IAAI;QACrB,qBAAqB,EAAE,IAAI;QAC3B,KAAK,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM;KACzB,CAAC,CAAA;AACH,CAAC;AAED;;GAEG;AACH,SAAS,eAAe,CAAC,MAAoB,EAAE,YAAoB;IAClE,IAAI,MAAM,EAAE,IAAI,YAAY;QAAE,OAAO,YAAY,CAAA;IACjD,MAAM,MAAM,GAAG,MAAM,CAAC,gBAAgB,EAAE,MAAM,CAAC,CAAA;IAE/C,gDAAgD;IAChD,OAAO,GAAG,YAAY,IAAI,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,WAAW,EAAE,EAAE,CAAA;AAC3E,CAAC;AAED;;GAEG;AACH,SAAS,eAAe,CACvB,EAAU,EACV,MAAc,EACd,QAAgB,EAChB,QAAgB;IAEhB,MAAM,GAAG,GAAG,GAAG,EAAE,IAAI,MAAM,KAAK,QAAQ,IAAI,QAAQ,EAAE,CAAA;IAEtD,OAAO,EAAE,GAAG,EAAE,UAAU,EAAE,EAAE,YAAY,EAAE,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,CAAA;AAC7E,CAAC;AAED,gDAAgD;AAChD,MAAM,yBAAyB,GAAG,IAAI,CAAA;AACtC,MAAM,yBAAyB,GAAG,GAAG,CAAA;AACrC,MAAM,yBAAyB,GAAG,IAAI,CAAA;AAEtC;;GAEG;AACH,MAAM,CAAC,MAAM,aAAa,GAAiB;IAC1C,IAAI,EAAE,UAAU;IAChB,WAAW,EAAE,qGAAqG;IAClH,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE;QACR,EAAE,IAAI,EAAE,yBAAyB,EAAE,WAAW,EAAE,+CAA+C,EAAE;QACjG,EAAE,IAAI,EAAE,UAAU,EAAE,WAAW,EAAE,2CAA2C,EAAE;KAC9E;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,IAAI,CAAC,KAAK,IAAI,IAAI;YAAE,MAAM,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAA;QAC/E,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAA;QACxB,wDAAwD;QACxD,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,mBAAmB,CAAC,gBAAgB,CAAC,CAAA;QACvE,MAAM,gBAAgB,GAAG,IAAI,CAAC,gBAAgB,IAAI,GAAG,CAAA;QAErD,8DAA8D;QAC9D,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,EAAE,MAAM,CAAC,CAAA;QAC7C,MAAM,IAAI,GAAG,MAAM,UAAU,CAAC,SAAS,CAAC,CAAA;QAExC,OAAO,CAAC,KAAK,CAAC,KAAK,MAAM,CAAC,GAAG,KAAK,IAAI,CAAC,MAAM,8BAA8B,CAAC,CAAA;QAE5E,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CAAC,2BAA2B,MAAM,CAAC,GAAG,WAAW,CAAC,CAAA;QAClE,CAAC;QAED,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,KAAK,GAAG,CAAC,CAAA;QACb,MAAM,CAAC,GAAG,IAAI,CAAC,MAAM,CAAA;QAErB,OAAO,OAAO,GAAG,KAAK,IAAI,KAAK,EAAE,GAAG,KAAK,GAAG,EAAE,EAAE,CAAC;YAChD,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAE,CAAA;YAC5C,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,IAAI,CAAA;YACjC,MAAM,QAAQ,GAAG,MAAM,EAAE,GAAG,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAA;YACtF,MAAM,YAAY,GAAG,eAAe,CAAC,MAAM,EAAE,IAAI,CAAC,YAAY,CAAC,CAAA;YAE/D,sCAAsC;YACtC,MAAM,UAAU,GAAG,MAAM,EAAE,GAAG,gBAAgB,CAAA;YAE9C,IAAI,QAA4E,CAAA;YAEhF,IAAI,UAAU,EAAE,CAAC;gBAChB,MAAM,WAAW,GAAG,MAAM,EAAE,CAAA;gBAC5B,IAAI,GAAW,CAAA;gBAEf,IAAI,WAAW,GAAG,yBAAyB,EAAE,CAAC;oBAC7C,kDAAkD;oBAClD,GAAG,GAAG,GAAG,QAAQ,IAAI,QAAQ,KAAK,YAAY,IAAI,MAAM,EAAE,CAAA;gBAC3D,CAAC;qBAAM,IAAI,WAAW,GAAG,yBAAyB,EAAE,CAAC;oBACpD,yCAAyC;oBACzC,GAAG,GAAG,GAAG,QAAQ,KAAK,QAAQ,KAAK,YAAY,IAAI,MAAM,EAAE,CAAA;gBAC5D,CAAC;qBAAM,IAAI,WAAW,GAAG,yBAAyB,EAAE,CAAC;oBACpD,wDAAwD;oBACxD,GAAG,GAAG,GAAG,QAAQ,IAAI,YAAY,IAAI,MAAM,IAAI,QAAQ,EAAE,CAAA;gBAC1D,CAAC;qBAAM,CAAC;oBACP,2CAA2C;oBAC3C,GAAG,GAAG,GAAG,QAAQ,KAAK,YAAY,IAAI,MAAM,KAAK,QAAQ,EAAE,CAAA;gBAC5D,CAAC;gBAED,QAAQ,GAAG,EAAE,GAAG,EAAE,UAAU,EAAE,EAAE,YAAY,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,CAAA;YAC7E,CAAC;iBAAM,CAAC;gBACP,QAAQ,GAAG,eAAe,CAAC,YAAY,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAA;YACrE,CAAC;YAED,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,GAAG,QAAQ,CAAA;YAEpC,kEAAkE;YAClE,MAAM,eAAe,GAAG,MAAM,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;YAEnE,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;gBACpD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,8DAA8D;YAC9D,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,KAAK,CAAC,aAAa,CAAC,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAA;gBAExD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,QAAQ,GAAG,cAAc,CAAC,MAAM,EAAE;gBACvC,MAAM,EAAE,UAAU,CAAC,MAAM;gBACzB,YAAY,EAAE,UAAU,CAAC,YAAY;gBACrC,QAAQ,EAAE,UAAU,CAAC,QAAQ;gBAC7B,QAAQ,EAAE,UAAU,CAAC,QAAQ;aAC7B,CAAC,CAAA;YAEF,MAAM,SAAS,GAAiB;gBAC/B,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM;gBACN,SAAS,EAAE,QAAQ;gBACnB,cAAc,EAAE,OAAO;gBACvB,OAAO,EAAE,8FAA8F;aACvG,CAAA;YAED,MAAM,OAAO,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAA;YAEnC,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,CACJ,aAAa,CAAC;gBACb,GAAG,OAAO,CAAC,GAAG;gBACd,YAAY,EAAE,UAAU;gBACxB,WAAW,EAAE,UAAU,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,WAAW;gBAClD,aAAa,EAAE,IAAI;aACnB,CAAC,CACF,CAAA;YAED,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}

package/out/fr/street-decompose.d.ts

@@ -0,0 +1,29 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Decompose a French street name into Stage 3 components. French convention puts the street type as
+ *   a leading word: "Rue de Rivoli", "Avenue des Champs-Élysées", "Bd Voltaire".
+ *
+ *   The street type becomes street_prefix in our schema. The remaining tokens form the street name.
+ *
+ *   Examples: "Rue de Rivoli" → { prefix: "Rue", street: "de Rivoli" } "Avenue des Champs-Élysées" →
+ *   { prefix: "Avenue", street: "des Champs-Élysées" } "Boulevard Voltaire" → { prefix:
+ *   "Boulevard", street: "Voltaire" }
+ *
+ *   Sources street types from `core/data/libpostal/dictionaries/fr/street_types.txt`.
+ */
+export interface DecomposedFrStreet {
+    prefix: string | null;
+    street: string;
+}
+/**
+ * Decompose a French street name into prefix (leading type word) and street name.
+ *
+ * If the first 1-2 tokens match a known street type
+ * (allowing for multi-word like "ancien chemin"), they become the prefix.
+ * Returns `{ prefix: null, street: original }` if no match.
+ */
+export declare function decomposeFrStreet(fullname: string): DecomposedFrStreet;
+//# sourceMappingURL=street-decompose.d.ts.map

package/out/fr/street-decompose.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"street-decompose.d.ts","sourceRoot":"","sources":["../../lib/fr/street-decompose.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAWH,MAAM,WAAW,kBAAkB;IAClC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,MAAM,EAAE,MAAM,CAAA;CACd;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,QAAQ,EAAE,MAAM,GAAG,kBAAkB,CA4BtE"}

package/out/fr/street-decompose.js

@@ -0,0 +1,52 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Decompose a French street name into Stage 3 components. French convention puts the street type as
+ *   a leading word: "Rue de Rivoli", "Avenue des Champs-Élysées", "Bd Voltaire".
+ *
+ *   The street type becomes street_prefix in our schema. The remaining tokens form the street name.
+ *
+ *   Examples: "Rue de Rivoli" → { prefix: "Rue", street: "de Rivoli" } "Avenue des Champs-Élysées" →
+ *   { prefix: "Avenue", street: "des Champs-Élysées" } "Boulevard Voltaire" → { prefix:
+ *   "Boulevard", street: "Voltaire" }
+ *
+ *   Sources street types from `core/data/libpostal/dictionaries/fr/street_types.txt`.
+ */
+import { loadLibpostalDictionary } from "#adapters/libpostal-dictionary";
+/**
+ * Tokens a BAN street needs before a type/article/name decomposition is attempted.
+ */
+const MIN_TOKENS_FOR_DECOMPOSE = 3;
+const STREET_TYPES_FR = await loadLibpostalDictionary("fr", "street_types.txt");
+/**
+ * Decompose a French street name into prefix (leading type word) and street name.
+ *
+ * If the first 1-2 tokens match a known street type
+ * (allowing for multi-word like "ancien chemin"), they become the prefix.
+ * Returns `{ prefix: null, street: original }` if no match.
+ */
+export function decomposeFrStreet(fullname) {
+    const trimmed = fullname.trim();
+    if (!trimmed)
+        return { prefix: null, street: "" };
+    const tokens = trimmed.split(/\s+/);
+    if (tokens.length < 2)
+        return { prefix: null, street: trimmed };
+    const norm = (s) => s.toLowerCase().replace(/[.,;]$/, "");
+    // Try 2-word prefix first (e.g. "ancien chemin")
+    if (tokens.length >= MIN_TOKENS_FOR_DECOMPOSE) {
+        const twoWord = norm(tokens[0]) + " " + norm(tokens[1]);
+        if (STREET_TYPES_FR.has(twoWord)) {
+            return { prefix: tokens.slice(0, 2).join(" "), street: tokens.slice(2).join(" ") };
+        }
+    }
+    // Then try 1-word prefix
+    const first = norm(tokens[0]);
+    if (STREET_TYPES_FR.has(first)) {
+        return { prefix: tokens[0], street: tokens.slice(1).join(" ") };
+    }
+    return { prefix: null, street: trimmed };
+}
+//# sourceMappingURL=street-decompose.js.map

package/out/fr/street-decompose.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"street-decompose.js","sourceRoot":"","sources":["../../lib/fr/street-decompose.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,OAAO,EAAE,uBAAuB,EAAE,MAAM,gCAAgC,CAAA;AAExE;;GAEG;AACH,MAAM,wBAAwB,GAAG,CAAC,CAAA;AAElC,MAAM,eAAe,GAAG,MAAM,uBAAuB,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;AAO/E;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,QAAgB;IACjD,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,EAAE,CAAA;IAE/B,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,EAAE,CAAA;IAEjD,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;IAEnC,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;IAE/D,MAAM,IAAI,GAAG,CAAC,CAAS,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAA;IAEjE,iDAAiD;IACjD,IAAI,MAAM,CAAC,MAAM,IAAI,wBAAwB,EAAE,CAAC;QAC/C,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,GAAG,GAAG,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,CAAA;QAEzD,IAAI,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YAClC,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;QACnF,CAAC;IACF,CAAC;IAED,yBAAyB;IACzB,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,CAAA;IAE9B,IAAI,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;QAChC,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC,CAAE,EAAE,MAAM,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAA;IACjE,CAAC;IAED,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;AACzC,CAAC"}