@mailwoman/corpus 9.4.0 → 10.1.0

package/out/tools/fetch.d.ts

@@ -0,0 +1,140 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Registers the corpus source fetchers. They download raw files and manifests for the adapters to read.
+ */
+import { fetchBrussels } from "#be/tools/fetch/brussels";
+import { fetchWallonie } from "#be/tools/fetch/wallonie";
+import { fetchCzCuzk } from "#cz/tools/fetch/cuzk";
+import { fetchDKAddresses } from "#dk/tools/fetch/inspire";
+import { fetchADSEE } from "#ee/tools/fetch/ads";
+import { fetchESBizkaia } from "#es/tools/fetch/bizkaia";
+import { fetchESCatastro } from "#es/tools/fetch/catastro";
+import { fetchESGipuzkoa } from "#es/tools/fetch/gipuzkoa";
+import { fetchESNavarra } from "#es/tools/fetch/navarra";
+import { fetchRyhti } from "#fi/tools/fetch/ryhti";
+import { fetchAnnuaireEducationOverseas } from "#fr/tools/fetch/annuaire-education";
+import { fetchBan } from "#fr/tools/fetch/ban";
+import { fetchFinessOverseas } from "#fr/tools/fetch/finess";
+import { fetchITANAC } from "#it/tools/fetch/anac";
+import { fetchHoujinJP } from "#jp/tools/fetch/houjin";
+import { fetchJusoKR } from "#kr/tools/fetch/juso";
+import { fetchLocaldataKR } from "#kr/tools/fetch/localdata";
+import { fetchBDAdresses } from "#lu/tools/fetch/bd-adresses";
+import { fetchVaris } from "#lv/tools/fetch/varis";
+import { fetchNLKadaster } from "#nl/tools/fetch/kadaster";
+import { fetchMatrikkelen } from "#no/tools/fetch/matrikkelen";
+import { fetchEMUiAPL } from "#pl/tools/fetch/emuia";
+import { fetchReteaScolara } from "#ro/tools/fetch/retea-scolara";
+import { fetchACRASG } from "#sg/tools/fetch/acra";
+import { fetchSKInspire } from "#sk/tools/fetch/inspire";
+import { fetchCORDIS } from "#tools/fetch/cordis";
+import { fetchGeonamesDumps } from "#tools/fetch/geonames/dump";
+import { fetchGeonamesPostal } from "#tools/fetch/geonames/postal";
+import { fetchGLEIF } from "#tools/fetch/gleif";
+import { fetchOpenAddresses } from "#tools/fetch/openaddresses";
+import { fetchOurAirports } from "#tools/fetch/ourairports";
+import { fetchVlaanderenAD } from "#tools/fetch/vlaanderen-ad";
+import { fetchWikidataSubVenue } from "#tools/fetch/wikidata-subvenue";
+import { fetchGCISTW } from "#tw/tools/fetch/gcis";
+import { fetchHRSA } from "#us/tools/fetch/hrsa";
+import { fetchIMLSPLS } from "#us/tools/fetch/imls-pls";
+import { fetchNAD } from "#us/tools/fetch/nad";
+import { fetchNPPES } from "#us/tools/fetch/nppes";
+import { fetchStateHISchools } from "#us/tools/fetch/state/hi-schools";
+import { fetchStateSources } from "#us/tools/fetch/state/sources";
+import { fetchTigerFull } from "#us/tools/fetch/tiger-full";
+export * from "#sg/tools/fetch/acra";
+export * from "#ee/tools/fetch/ads";
+export * from "#be/tools/fetch/brussels";
+export * from "#tools/fetch/vlaanderen-ad";
+export * from "#be/tools/fetch/wallonie";
+export * from "#it/tools/fetch/anac";
+export * from "#fr/tools/fetch/finess";
+export * from "#fr/tools/fetch/annuaire-education";
+export * from "#ro/tools/fetch/retea-scolara";
+export * from "#tools/fetch/cordis";
+export * from "#tools/fetch/gleif";
+export * from "#fr/tools/fetch/ban";
+export * from "#lu/tools/fetch/bd-adresses";
+export * from "#lv/tools/fetch/varis";
+export * from "#es/tools/fetch/bizkaia";
+export * from "#es/tools/fetch/catastro";
+export * from "#cz/tools/fetch/cuzk";
+export * from "#es/tools/fetch/gipuzkoa";
+export * from "#es/tools/fetch/navarra";
+export * from "#pl/tools/fetch/emuia";
+export * from "#dk/tools/fetch/inspire";
+export * from "#sk/tools/fetch/inspire";
+export * from "#tw/tools/fetch/gcis";
+export * from "#tools/fetch/geonames/dump";
+export * from "#tools/fetch/geonames/postal";
+export * from "#jp/tools/fetch/houjin";
+export * from "#us/tools/fetch/hrsa";
+export * from "#us/tools/fetch/imls-pls";
+export * from "#kr/tools/fetch/juso";
+export * from "#nl/tools/fetch/kadaster";
+export * from "#kr/tools/fetch/localdata";
+export * from "#no/tools/fetch/matrikkelen";
+export * from "#us/tools/fetch/nad";
+export * from "#us/tools/fetch/nppes";
+export * from "#tools/fetch/openaddresses";
+export * from "#tools/fetch/ourairports";
+export * from "#fi/tools/fetch/ryhti";
+export * from "#us/tools/fetch/state/hi-schools";
+export * from "#us/tools/fetch/state/sources";
+export * from "#us/tools/fetch/tiger-full";
+export * from "#tools/fetch/wikidata-subvenue";
+/**
+ * The fetcher for each source id that `mailwoman corpus fetch` accepts.
+ */
+export declare const FETCH_SOURCES: {
+    readonly "acra-sg": typeof fetchACRASG;
+    readonly "ads-ee": typeof fetchADSEE;
+    readonly ban: typeof fetchBan;
+    readonly brussels: typeof fetchBrussels;
+    readonly cordis: typeof fetchCORDIS;
+    readonly "bd-adresses": typeof fetchBDAdresses;
+    readonly "cz-cuzk": typeof fetchCzCuzk;
+    readonly "dk-inspire": typeof fetchDKAddresses;
+    readonly "emuia-pl": typeof fetchEMUiAPL;
+    readonly "es-bizkaia": typeof fetchESBizkaia;
+    readonly "es-catastro": typeof fetchESCatastro;
+    readonly "es-gipuzkoa": typeof fetchESGipuzkoa;
+    readonly "es-navarra": typeof fetchESNavarra;
+    readonly "fr-annuaire-education-overseas": typeof fetchAnnuaireEducationOverseas;
+    readonly "fr-finess-overseas": typeof fetchFinessOverseas;
+    readonly "gcis-tw": typeof fetchGCISTW;
+    readonly gleif: typeof fetchGLEIF;
+    readonly "houjin-jp": typeof fetchHoujinJP;
+    readonly "it-anac": typeof fetchITANAC;
+    readonly "juso-kr": typeof fetchJusoKR;
+    readonly "localdata-kr": typeof fetchLocaldataKR;
+    readonly "lv-varis": typeof fetchVaris;
+    readonly matrikkelen: typeof fetchMatrikkelen;
+    readonly nad: typeof fetchNAD;
+    readonly "nl-kadaster": typeof fetchNLKadaster;
+    readonly "ro-retea-scolara": typeof fetchReteaScolara;
+    readonly ryhti: typeof fetchRyhti;
+    readonly "sk-inspire": typeof fetchSKInspire;
+    readonly "geonames-dump": typeof fetchGeonamesDumps;
+    readonly "geonames-postal": typeof fetchGeonamesPostal;
+    readonly hrsa: typeof fetchHRSA;
+    readonly "imls-pls": typeof fetchIMLSPLS;
+    readonly nppes: typeof fetchNPPES;
+    readonly openaddresses: typeof fetchOpenAddresses;
+    readonly ourairports: typeof fetchOurAirports;
+    readonly vlaanderen: typeof fetchVlaanderenAD;
+    readonly wallonie: typeof fetchWallonie;
+    readonly "state-sources": typeof fetchStateSources;
+    readonly "state-hi-schools": typeof fetchStateHISchools;
+    readonly "tiger-full": typeof fetchTigerFull;
+    readonly "wikidata-subvenue": typeof fetchWikidataSubVenue;
+};
+/**
+ * A source id in {@link FETCH_SOURCES}.
+ */
+export type FetchSourceID = keyof typeof FETCH_SOURCES;
+//# sourceMappingURL=fetch.d.ts.map

package/out/tools/fetch.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"fetch.d.ts","sourceRoot":"","sources":["../../lib/tools/fetch.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACxD,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAA;AAC1D,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAChD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,8BAA8B,EAAE,MAAM,oCAAoC,CAAA;AACnF,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAA;AAC9C,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAA;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAA;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAA;AAC5D,OAAO,EAAE,eAAe,EAAE,MAAM,6BAA6B,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,gBAAgB,EAAE,MAAM,6BAA6B,CAAA;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AACpD,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAA;AACjD,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,mBAAmB,EAAE,MAAM,8BAA8B,CAAA;AAClE,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAA;AAC/C,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,gBAAgB,EAAE,MAAM,0BAA0B,CAAA;AAC3D,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,qBAAqB,EAAE,MAAM,gCAAgC,CAAA;AACtE,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAA;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AACvD,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAA;AAC9C,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,mBAAmB,EAAE,MAAM,kCAAkC,CAAA;AACtE,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,cAAc,EAAE,MAAM,4BAA4B,CAAA;AAE3D,cAAc,sBAAsB,CAAA;AACpC,cAAc,qBAAqB,CAAA;AACnC,cAAc,0BAA0B,CAAA;AACxC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,wBAAwB,CAAA;AACtC,cAAc,oCAAoC,CAAA;AAClD,cAAc,+BAA+B,CAAA;AAC7C,cAAc,qBAAqB,CAAA;AACnC,cAAc,oBAAoB,CAAA;AAClC,cAAc,qBAAqB,CAAA;AACnC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,uBAAuB,CAAA;AACrC,cAAc,yBAAyB,CAAA;AACvC,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,yBAAyB,CAAA;AACvC,cAAc,uBAAuB,CAAA;AACrC,cAAc,yBAAyB,CAAA;AACvC,cAAc,yBAAyB,CAAA;AACvC,cAAc,sBAAsB,CAAA;AACpC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,wBAAwB,CAAA;AACtC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,2BAA2B,CAAA;AACzC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,uBAAuB,CAAA;AACrC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,0BAA0B,CAAA;AACxC,cAAc,uBAAuB,CAAA;AACrC,cAAc,kCAAkC,CAAA;AAChD,cAAc,+BAA+B,CAAA;AAC7C,cAAc,4BAA4B,CAAA;AAC1C,cAAc,gCAAgC,CAAA;AAE9C;;GAEG;AACH,eAAO,MAAM,aAAa;aACzB,SAAS;aACT,QAAQ;aACR,GAAG;aACH,QAAQ;aACR,MAAM;aACN,aAAa;aACb,SAAS;aACT,YAAY;aACZ,UAAU;aACV,YAAY;aACZ,aAAa;aACb,aAAa;aACb,YAAY;aACZ,gCAAgC;aAChC,oBAAoB;aACpB,SAAS;aACT,KAAK;aACL,WAAW;aACX,SAAS;aACT,SAAS;aACT,cAAc;aACd,UAAU;aACV,WAAW;aACX,GAAG;aACH,aAAa;aACb,kBAAkB;aAClB,KAAK;aACL,YAAY;aACZ,eAAe;aACf,iBAAiB;aACjB,IAAI;aACJ,UAAU;aACV,KAAK;aACL,aAAa;aACb,WAAW;aACX,UAAU;aACV,QAAQ;aACR,eAAe;aACf,kBAAkB;aAClB,YAAY;aACZ,mBAAmB;CACV,CAAA;AAEV;;GAEG;AACH,MAAM,MAAM,aAAa,GAAG,MAAM,OAAO,aAAa,CAAA"}

package/out/tools/fetch.js

@@ -0,0 +1,136 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Registers the corpus source fetchers. They download raw files and manifests for the adapters to read.
+ */
+import { fetchBrussels } from "#be/tools/fetch/brussels";
+import { fetchWallonie } from "#be/tools/fetch/wallonie";
+import { fetchCzCuzk } from "#cz/tools/fetch/cuzk";
+import { fetchDKAddresses } from "#dk/tools/fetch/inspire";
+import { fetchADSEE } from "#ee/tools/fetch/ads";
+import { fetchESBizkaia } from "#es/tools/fetch/bizkaia";
+import { fetchESCatastro } from "#es/tools/fetch/catastro";
+import { fetchESGipuzkoa } from "#es/tools/fetch/gipuzkoa";
+import { fetchESNavarra } from "#es/tools/fetch/navarra";
+import { fetchRyhti } from "#fi/tools/fetch/ryhti";
+import { fetchAnnuaireEducationOverseas } from "#fr/tools/fetch/annuaire-education";
+import { fetchBan } from "#fr/tools/fetch/ban";
+import { fetchFinessOverseas } from "#fr/tools/fetch/finess";
+import { fetchITANAC } from "#it/tools/fetch/anac";
+import { fetchHoujinJP } from "#jp/tools/fetch/houjin";
+import { fetchJusoKR } from "#kr/tools/fetch/juso";
+import { fetchLocaldataKR } from "#kr/tools/fetch/localdata";
+import { fetchBDAdresses } from "#lu/tools/fetch/bd-adresses";
+import { fetchVaris } from "#lv/tools/fetch/varis";
+import { fetchNLKadaster } from "#nl/tools/fetch/kadaster";
+import { fetchMatrikkelen } from "#no/tools/fetch/matrikkelen";
+import { fetchEMUiAPL } from "#pl/tools/fetch/emuia";
+import { fetchReteaScolara } from "#ro/tools/fetch/retea-scolara";
+import { fetchACRASG } from "#sg/tools/fetch/acra";
+import { fetchSKInspire } from "#sk/tools/fetch/inspire";
+import { fetchCORDIS } from "#tools/fetch/cordis";
+import { fetchGeonamesDumps } from "#tools/fetch/geonames/dump";
+import { fetchGeonamesPostal } from "#tools/fetch/geonames/postal";
+import { fetchGLEIF } from "#tools/fetch/gleif";
+import { fetchOpenAddresses } from "#tools/fetch/openaddresses";
+import { fetchOurAirports } from "#tools/fetch/ourairports";
+import { fetchVlaanderenAD } from "#tools/fetch/vlaanderen-ad";
+import { fetchWikidataSubVenue } from "#tools/fetch/wikidata-subvenue";
+import { fetchGCISTW } from "#tw/tools/fetch/gcis";
+import { fetchHRSA } from "#us/tools/fetch/hrsa";
+import { fetchIMLSPLS } from "#us/tools/fetch/imls-pls";
+import { fetchNAD } from "#us/tools/fetch/nad";
+import { fetchNPPES } from "#us/tools/fetch/nppes";
+import { fetchStateHISchools } from "#us/tools/fetch/state/hi-schools";
+import { fetchStateSources } from "#us/tools/fetch/state/sources";
+import { fetchTigerFull } from "#us/tools/fetch/tiger-full";
+export * from "#sg/tools/fetch/acra";
+export * from "#ee/tools/fetch/ads";
+export * from "#be/tools/fetch/brussels";
+export * from "#tools/fetch/vlaanderen-ad";
+export * from "#be/tools/fetch/wallonie";
+export * from "#it/tools/fetch/anac";
+export * from "#fr/tools/fetch/finess";
+export * from "#fr/tools/fetch/annuaire-education";
+export * from "#ro/tools/fetch/retea-scolara";
+export * from "#tools/fetch/cordis";
+export * from "#tools/fetch/gleif";
+export * from "#fr/tools/fetch/ban";
+export * from "#lu/tools/fetch/bd-adresses";
+export * from "#lv/tools/fetch/varis";
+export * from "#es/tools/fetch/bizkaia";
+export * from "#es/tools/fetch/catastro";
+export * from "#cz/tools/fetch/cuzk";
+export * from "#es/tools/fetch/gipuzkoa";
+export * from "#es/tools/fetch/navarra";
+export * from "#pl/tools/fetch/emuia";
+export * from "#dk/tools/fetch/inspire";
+export * from "#sk/tools/fetch/inspire";
+export * from "#tw/tools/fetch/gcis";
+export * from "#tools/fetch/geonames/dump";
+export * from "#tools/fetch/geonames/postal";
+export * from "#jp/tools/fetch/houjin";
+export * from "#us/tools/fetch/hrsa";
+export * from "#us/tools/fetch/imls-pls";
+export * from "#kr/tools/fetch/juso";
+export * from "#nl/tools/fetch/kadaster";
+export * from "#kr/tools/fetch/localdata";
+export * from "#no/tools/fetch/matrikkelen";
+export * from "#us/tools/fetch/nad";
+export * from "#us/tools/fetch/nppes";
+export * from "#tools/fetch/openaddresses";
+export * from "#tools/fetch/ourairports";
+export * from "#fi/tools/fetch/ryhti";
+export * from "#us/tools/fetch/state/hi-schools";
+export * from "#us/tools/fetch/state/sources";
+export * from "#us/tools/fetch/tiger-full";
+export * from "#tools/fetch/wikidata-subvenue";
+/**
+ * The fetcher for each source id that `mailwoman corpus fetch` accepts.
+ */
+export const FETCH_SOURCES = {
+    "acra-sg": fetchACRASG,
+    "ads-ee": fetchADSEE,
+    ban: fetchBan,
+    brussels: fetchBrussels,
+    cordis: fetchCORDIS,
+    "bd-adresses": fetchBDAdresses,
+    "cz-cuzk": fetchCzCuzk,
+    "dk-inspire": fetchDKAddresses,
+    "emuia-pl": fetchEMUiAPL,
+    "es-bizkaia": fetchESBizkaia,
+    "es-catastro": fetchESCatastro,
+    "es-gipuzkoa": fetchESGipuzkoa,
+    "es-navarra": fetchESNavarra,
+    "fr-annuaire-education-overseas": fetchAnnuaireEducationOverseas,
+    "fr-finess-overseas": fetchFinessOverseas,
+    "gcis-tw": fetchGCISTW,
+    gleif: fetchGLEIF,
+    "houjin-jp": fetchHoujinJP,
+    "it-anac": fetchITANAC,
+    "juso-kr": fetchJusoKR,
+    "localdata-kr": fetchLocaldataKR,
+    "lv-varis": fetchVaris,
+    matrikkelen: fetchMatrikkelen,
+    nad: fetchNAD,
+    "nl-kadaster": fetchNLKadaster,
+    "ro-retea-scolara": fetchReteaScolara,
+    ryhti: fetchRyhti,
+    "sk-inspire": fetchSKInspire,
+    "geonames-dump": fetchGeonamesDumps,
+    "geonames-postal": fetchGeonamesPostal,
+    hrsa: fetchHRSA,
+    "imls-pls": fetchIMLSPLS,
+    nppes: fetchNPPES,
+    openaddresses: fetchOpenAddresses,
+    ourairports: fetchOurAirports,
+    vlaanderen: fetchVlaanderenAD,
+    wallonie: fetchWallonie,
+    "state-sources": fetchStateSources,
+    "state-hi-schools": fetchStateHISchools,
+    "tiger-full": fetchTigerFull,
+    "wikidata-subvenue": fetchWikidataSubVenue,
+};
+//# sourceMappingURL=fetch.js.map

package/out/tools/fetch.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"fetch.js","sourceRoot":"","sources":["../../lib/tools/fetch.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACxD,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAA;AAC1D,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAChD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,8BAA8B,EAAE,MAAM,oCAAoC,CAAA;AACnF,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAA;AAC9C,OAAO,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAA;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,aAAa,EAAE,MAAM,wBAAwB,CAAA;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAA;AAC5D,OAAO,EAAE,eAAe,EAAE,MAAM,6BAA6B,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EAAE,gBAAgB,EAAE,MAAM,6BAA6B,CAAA;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AACpD,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAA;AACxD,OAAO,EAAE,WAAW,EAAE,MAAM,qBAAqB,CAAA;AACjD,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,mBAAmB,EAAE,MAAM,8BAA8B,CAAA;AAClE,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAA;AAC/C,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,gBAAgB,EAAE,MAAM,0BAA0B,CAAA;AAC3D,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,qBAAqB,EAAE,MAAM,gCAAgC,CAAA;AACtE,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAA;AAClD,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAA;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,0BAA0B,CAAA;AACvD,OAAO,EAAE,QAAQ,EAAE,MAAM,qBAAqB,CAAA;AAC9C,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAClD,OAAO,EAAE,mBAAmB,EAAE,MAAM,kCAAkC,CAAA;AACtE,OAAO,EAAE,iBAAiB,EAAE,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,cAAc,EAAE,MAAM,4BAA4B,CAAA;AAE3D,cAAc,sBAAsB,CAAA;AACpC,cAAc,qBAAqB,CAAA;AACnC,cAAc,0BAA0B,CAAA;AACxC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,wBAAwB,CAAA;AACtC,cAAc,oCAAoC,CAAA;AAClD,cAAc,+BAA+B,CAAA;AAC7C,cAAc,qBAAqB,CAAA;AACnC,cAAc,oBAAoB,CAAA;AAClC,cAAc,qBAAqB,CAAA;AACnC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,uBAAuB,CAAA;AACrC,cAAc,yBAAyB,CAAA;AACvC,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,yBAAyB,CAAA;AACvC,cAAc,uBAAuB,CAAA;AACrC,cAAc,yBAAyB,CAAA;AACvC,cAAc,yBAAyB,CAAA;AACvC,cAAc,sBAAsB,CAAA;AACpC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,8BAA8B,CAAA;AAC5C,cAAc,wBAAwB,CAAA;AACtC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,sBAAsB,CAAA;AACpC,cAAc,0BAA0B,CAAA;AACxC,cAAc,2BAA2B,CAAA;AACzC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,qBAAqB,CAAA;AACnC,cAAc,uBAAuB,CAAA;AACrC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,0BAA0B,CAAA;AACxC,cAAc,uBAAuB,CAAA;AACrC,cAAc,kCAAkC,CAAA;AAChD,cAAc,+BAA+B,CAAA;AAC7C,cAAc,4BAA4B,CAAA;AAC1C,cAAc,gCAAgC,CAAA;AAE9C;;GAEG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG;IAC5B,SAAS,EAAE,WAAW;IACtB,QAAQ,EAAE,UAAU;IACpB,GAAG,EAAE,QAAQ;IACb,QAAQ,EAAE,aAAa;IACvB,MAAM,EAAE,WAAW;IACnB,aAAa,EAAE,eAAe;IAC9B,SAAS,EAAE,WAAW;IACtB,YAAY,EAAE,gBAAgB;IAC9B,UAAU,EAAE,YAAY;IACxB,YAAY,EAAE,cAAc;IAC5B,aAAa,EAAE,eAAe;IAC9B,aAAa,EAAE,eAAe;IAC9B,YAAY,EAAE,cAAc;IAC5B,gCAAgC,EAAE,8BAA8B;IAChE,oBAAoB,EAAE,mBAAmB;IACzC,SAAS,EAAE,WAAW;IACtB,KAAK,EAAE,UAAU;IACjB,WAAW,EAAE,aAAa;IAC1B,SAAS,EAAE,WAAW;IACtB,SAAS,EAAE,WAAW;IACtB,cAAc,EAAE,gBAAgB;IAChC,UAAU,EAAE,UAAU;IACtB,WAAW,EAAE,gBAAgB;IAC7B,GAAG,EAAE,QAAQ;IACb,aAAa,EAAE,eAAe;IAC9B,kBAAkB,EAAE,iBAAiB;IACrC,KAAK,EAAE,UAAU;IACjB,YAAY,EAAE,cAAc;IAC5B,eAAe,EAAE,kBAAkB;IACnC,iBAAiB,EAAE,mBAAmB;IACtC,IAAI,EAAE,SAAS;IACf,UAAU,EAAE,YAAY;IACxB,KAAK,EAAE,UAAU;IACjB,aAAa,EAAE,kBAAkB;IACjC,WAAW,EAAE,gBAAgB;IAC7B,UAAU,EAAE,iBAAiB;IAC7B,QAAQ,EAAE,aAAa;IACvB,eAAe,EAAE,iBAAiB;IAClC,kBAAkB,EAAE,mBAAmB;IACvC,YAAY,EAAE,cAAc;IAC5B,mBAAmB,EAAE,qBAAqB;CACjC,CAAA"}

package/out/tools/golden/expand.d.ts

@@ -0,0 +1,105 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Generate golden-set candidate entries by LLM-driven surface-form synthesis from a verified-label
+ *   seed pulled out of the corpus's labeled test split.
+ *
+ *   ## Why this approach
+ *
+ *   The Phase 2 golden set has 74 entries. session-notes.md called for ≥500/locale. Manual curation
+ *   doesn't scale. Pure-LLM generation (invent raw + labels from scratch) is too noisy — labels
+ *   would be unverified.
+ *
+ *   This module takes the middle path:
+ *
+ *   1. **Seeds come from the corpus-v0.2.0 test split's parquet files** — already through the alignment pipeline, so
+ *        labels
+ *        are pipeline-verified.
+ *   2. **LLM only varies the surface form** — case, abbreviations, reordering, dropped components. The
+ *        component values (locality string, postcode digits, etc.) are preserved verbatim.
+ *   3. **Programmatic validator drops hallucinations**. Every component value must appear as a
+ *        substring (case-insensitive, whitespace-normalized) of the variant's raw. Failures dropped
+ *        silently. cost is wasted tokens, never bad-labeled golden entries.
+ *
+ *   ## Usage
+ *
+ *   ```sh
+ *   DEEPSEEK_API_KEY=sk-... \
+ *   mailwoman corpus golden expand \
+ *   --count 1000 \
+ *   --variants 5 \
+ *   --output data/eval/golden/candidates/expand-$(date +%Y%m%d-%H%M%S).jsonl
+ *   ```
+ *
+ *   ## Env
+ *
+ *   - `DEEPSEEK_API_KEY` — required for provider `deepseek`
+ *   - `ANTHROPIC_API_KEY` — required for provider `anthropic`
+ *
+ *   ## What this module does not do
+ *
+ *   - Does not commit anything or modify the versioned golden dir. Candidates land in
+ *       `data/eval/golden/candidates/` for operator review (skim, prune, then run
+ *       `mailwoman corpus golden promote`).
+ *   - Does not score the LLM's quality — that's an eyeball job after pilot lands.
+ *   - Does not retry hallucinated candidates. Cost of wasted tokens is trivial (~$0.0006/each).
+ */
+export interface ExpandGoldenOptions {
+    /**
+     * Corpus test-split parquet path(s), comma-separated.
+     *
+     * Default: the v0.2.0 test parquet file under the data root.
+     */
+    corpus?: string;
+    /**
+     * Total seeds to process.
+     *
+     * Default `100` (pilot).
+     */
+    count?: number;
+    /**
+     * Variants requested per seed.
+     *
+     * Default `5`.
+     */
+    variants?: number;
+    /**
+     * Jsonl output path.
+     *
+     * Default `data/eval/golden/candidates/expand-<ts>.jsonl`.
+     */
+    output?: string;
+    /**
+     * LLM provider.
+     *
+     * Default `deepseek`.
+     */
+    provider?: "deepseek" | "anthropic";
+    /**
+     * Model id.
+     *
+     * Default depends on provider.
+     */
+    model?: string;
+    /**
+     * Parallel LLM calls.
+     *
+     * Default `4`.
+     */
+    concurrency?: number;
+    /**
+     * Comma-separated source allow-list.
+     */
+    includeSources?: string;
+}
+export interface ExpandGoldenSummary {
+    seedsProcessed: number;
+    kept: number;
+    dropped: number;
+    errored: number;
+    outputPath: string;
+}
+export declare function expandGolden(options?: ExpandGoldenOptions, report?: (line: string) => void): Promise<ExpandGoldenSummary>;
+//# sourceMappingURL=expand.d.ts.map

package/out/tools/golden/expand.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"expand.d.ts","sourceRoot":"","sources":["../../../lib/tools/golden/expand.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+CG;AAgEH,MAAM,WAAW,mBAAmB;IACnC;;;;OAIG;IACH,MAAM,CAAC,EAAE,MAAM,CAAA;IACf;;;;OAIG;IACH,KAAK,CAAC,EAAE,MAAM,CAAA;IACd;;;;OAIG;IACH,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB;;;;OAIG;IACH,MAAM,CAAC,EAAE,MAAM,CAAA;IACf;;;;OAIG;IACH,QAAQ,CAAC,EAAE,UAAU,GAAG,WAAW,CAAA;IACnC;;;;OAIG;IACH,KAAK,CAAC,EAAE,MAAM,CAAA;IACd;;;;OAIG;IACH,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,cAAc,CAAC,EAAE,MAAM,CAAA;CACvB;AAED,MAAM,WAAW,mBAAmB;IACnC,cAAc,EAAE,MAAM,CAAA;IACtB,IAAI,EAAE,MAAM,CAAA;IACZ,OAAO,EAAE,MAAM,CAAA;IACf,OAAO,EAAE,MAAM,CAAA;IACf,UAAU,EAAE,MAAM,CAAA;CAClB;AA6UD,wBAAsB,YAAY,CACjC,OAAO,GAAE,mBAAwB,EACjC,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,mBAAmB,CAAC,CAsG9B"}