@mailwoman/corpus 9.4.0 → 10.1.0

package/out/adapters/usgov-nppes/adapter.js

@@ -1,118 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-nppes`: CMS National Plan and Provider Enumeration System (NPI registry) CSV consumer.
- *
- *   NPPES is the authoritative US healthcare provider registry, published monthly by CMS. Each row
- *   carries a provider's business practice location address together with their legal business name
- *   or individual name. At ~7M rows it is the single largest venue+address signal source
- *   available.
- *
- *   The adapter consumes the monthly full-replacement CSV (operator pre-downloads via
- *   `fetch-nppes.ts`). Column names match the canonical NPPES "Full Replacement Monthly NPI File"
- *   header published at `https://download.cms.gov/nppes/NPI_Files.html`.
- *
- *   Output: one row per CSV record where the practice location address is populated. Organization
- *   rows carry `venue` from the legal business name; individual rows compose `attention` from
- *   last+first name. Address quad goes on `(house_number, street, locality, region, postcode)`.
- *
- *   License: stamped `"Public Domain"` per CMS's federal government distribution terms.
- */
-import { isPresent } from "@mailwoman/core/objects";
-import { reconcileComponents } from "@mailwoman/formatter";
-import { CSVSpliterator } from "spliterator";
-import { composeRaw, splitStreetLine, stableSourceID } from "#adapters/utils";
-import { lookupStateAbbreviation } from "#codex/us-fips-state";
-/**
- * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
- * it came from.
- */
-export const USGOV_NPPES_ADAPTER_ID = "usgov-nppes";
-/**
- * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
- */
-export const USGOV_NPPES_DEFAULT_LICENSE = "Public Domain";
-export function createUsgovNPPESAdapter() {
-    return {
-        id: USGOV_NPPES_ADAPTER_ID,
-        defaultLicense: USGOV_NPPES_DEFAULT_LICENSE,
-        description: "CMS National Plan and Provider Enumeration System — 7M provider practice locations (public-domain). Venue+address co-occurrence at scale.",
-        async *rows(opts) {
-            if (opts.country && opts.country !== "US") {
-                throw new Error(`usgov-nppes adapter: only US supported, got country=${opts.country}`);
-            }
-            const rows = CSVSpliterator.fromAsync(opts.inputPath, {
-                mode: "object",
-                normalizeKeys: false,
-                enableQuoteHandling: true,
-            });
-            let emitted = 0;
-            for await (const record of rows) {
-                if (opts.signal?.aborted)
-                    break;
-                if (opts.limit !== undefined && emitted >= opts.limit)
-                    break;
-                const npi = (record.NPI ?? "").trim();
-                const orgName = (record["Provider Organization Name (Legal Business Name)"] ?? "").trim();
-                const lastName = (record["Provider Last Name (Legal Name)"] ?? "").trim();
-                const firstName = (record["Provider First Name"] ?? "").trim();
-                const address1 = (record["Provider First Line Business Practice Location Address"] ?? "").trim();
-                const address2 = (record["Provider Second Line Business Practice Location Address"] ?? "").trim();
-                const city = (record["Provider Business Practice Location Address City Name"] ?? "").trim();
-                const stateRaw = (record["Provider Business Practice Location Address State Name"] ?? "").trim();
-                const postcode = (record["Provider Business Practice Location Address Postcode"] ?? "").trim();
-                if (!city || !postcode)
-                    continue;
-                const state = lookupStateAbbreviation(stateRaw);
-                if (!state)
-                    continue;
-                const fullStreet = [address1, address2].filter(isPresent).join(" ");
-                const split = splitStreetLine(fullStreet);
-                if (!split)
-                    continue;
-                const venue = orgName || [firstName, lastName].filter(isPresent).join(" ") || undefined;
-                const components = {
-                    ...(venue ? { venue } : {}),
-                    ...(split.house_number ? { house_number: split.house_number } : {}),
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode,
-                };
-                const raw = composeRaw({
-                    venue,
-                    houseNumber: split.house_number,
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode,
-                });
-                if (!raw)
-                    continue;
-                const aligned = reconcileComponents(components, raw);
-                if (Object.keys(aligned).length <= 2)
-                    continue;
-                const sourceID = npi ? `${USGOV_NPPES_ADAPTER_ID}-${npi}` : stableSourceID(USGOV_NPPES_ADAPTER_ID, aligned);
-                yield {
-                    raw,
-                    components: aligned,
-                    country: "US",
-                    locale: "en-US",
-                    source: USGOV_NPPES_ADAPTER_ID,
-                    source_id: sourceID,
-                    corpus_version: "",
-                    license: USGOV_NPPES_DEFAULT_LICENSE,
-                };
-                emitted++;
-            }
-        },
-    };
-}
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export const usgovNPPESAdapter = createUsgovNPPESAdapter();
-//# sourceMappingURL=adapter.js.map

package/out/adapters/usgov-nppes/adapter.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/usgov-nppes/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAEH,OAAO,EAAE,SAAS,EAAE,MAAM,yBAAyB,CAAA;AACnD,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAC7E,OAAO,EAAE,uBAAuB,EAAE,MAAM,sBAAsB,CAAA;AAG9D;;;GAGG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,aAAa,CAAA;AACnD;;;GAGG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAG,eAAe,CAAA;AAe1D,MAAM,UAAU,uBAAuB;IACtC,OAAO;QACN,EAAE,EAAE,sBAAsB;QAC1B,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EACV,2IAA2I;QAE5I,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,uDAAuD,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YACvF,CAAC;YAED,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE;gBACrD,IAAI,EAAE,QAAQ;gBACd,aAAa,EAAE,KAAK;gBACpB,mBAAmB,EAAE,IAAI;aACzB,CAAC,CAAA;YAEF,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAA+B,EAAE,CAAC;gBAC5D,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrC,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,kDAAkD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACzF,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,iCAAiC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACzE,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,qBAAqB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE9D,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,wDAAwD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAChG,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,yDAAyD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACjG,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,uDAAuD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC3F,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,wDAAwD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAChG,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,sDAAsD,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE9F,IAAI,CAAC,IAAI,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBAEhC,MAAM,KAAK,GAAG,uBAAuB,CAAC,QAAQ,CAAC,CAAA;gBAE/C,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,UAAU,GAAG,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;gBACnE,MAAM,KAAK,GAAG,eAAe,CAAC,UAAU,CAAC,CAAA;gBAEzC,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,KAAK,GAAG,OAAO,IAAI,CAAC,SAAS,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,SAAS,CAAA;gBAEvF,MAAM,UAAU,GAA+B;oBAC9C,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC3B,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ;iBACR,CAAA;gBAED,MAAM,GAAG,GAAG,UAAU,CAAC;oBACtB,KAAK;oBACL,WAAW,EAAE,KAAK,CAAC,YAAY;oBAC/B,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ;iBACR,CAAC,CAAA;gBAEF,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,IAAI,CAAC;oBAAE,SAAQ;gBAE9C,MAAM,QAAQ,GAAG,GAAG,CAAC,CAAC,CAAC,GAAG,sBAAsB,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC,sBAAsB,EAAE,OAAO,CAAC,CAAA;gBAE3G,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,sBAAsB;oBAC9B,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,2BAA2B;iBACpC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,uBAAuB,EAAE,CAAA"}

package/out/adapters/usgov-samhsa-treatment-locator/adapter.d.ts

@@ -1,46 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-samhsa-treatment-locator`: SAMHSA Behavioral Health Treatment Services Locator CSV
- *   consumer.
- *
- *   SAMHSA's Treatment Locator (`findtreatment.gov`) is the federal directory of substance-use and
- *   mental-health treatment facilities. The published CSV carries the facility name, an optional
- *   secondary name (typically the organizational parent), and the postal address quad split into
- *   primary + secondary street lines. Phase 1.6 §1.2 (#22) selects this source for the same reason
- *   it selects HRSA: facility names are hand-typed venue strings and the addresses pass through
- *   enough human + system hands to accumulate the suite-designator + sub-tenant chaos ("Suite C,
- *   behind main building") that pure gazetteer data does not.
- *
- *   SAMHSA's two-line address shape is the key adapter-specific concern. `street1` typically carries
- *   the canonical postal address (`"123 Main St"`); `street2` carries the suite / unit / "second
- *   floor" surface form. The adapter joins them with `", "` into a single `street` component (Phase
- *   1 keeps `unit` as a deferred slot since the OpenCage template doesn't have a clean rendering
- *   for it). Operators wanting a different join policy can subclass the factory.
- *
- *   Column names below match the canonical SAMHSA Behavioral Health Treatment Services Locator CSV
- *   export header. Operators substituting a closely-related extract should rename columns to match;
- *   the README has the mapping cheatsheet.
- *
- *   License: stamped `"Public Domain"` per the SAMHSA Open Data Foundry's federal-government
- *   distribution terms.
- */
-import type { CorpusAdapter } from "#types";
-/**
- * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
- * it came from.
- */
-export declare const USGOV_SAMHSA_ADAPTER_ID = "usgov-samhsa-treatment-locator";
-/**
- * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
- */
-export declare const USGOV_SAMHSA_DEFAULT_LICENSE = "Public Domain";
-export declare function createUsgovSamhsaTreatmentLocatorAdapter(): CorpusAdapter;
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export declare const usgovSamhsaTreatmentLocatorAdapter: CorpusAdapter;
-//# sourceMappingURL=adapter.d.ts.map

package/out/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../lib/adapters/usgov-samhsa-treatment-locator/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAOH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,uBAAuB,mCAAmC,CAAA;AACvE;;;GAGG;AACH,eAAO,MAAM,4BAA4B,kBAAkB,CAAA;AA4D3D,wBAAgB,wCAAwC,IAAI,aAAa,CAiFxE;AAED;;GAEG;AACH,eAAO,MAAM,kCAAkC,eAA6C,CAAA"}

package/out/adapters/usgov-samhsa-treatment-locator/adapter.js

@@ -1,154 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-samhsa-treatment-locator`: SAMHSA Behavioral Health Treatment Services Locator CSV
- *   consumer.
- *
- *   SAMHSA's Treatment Locator (`findtreatment.gov`) is the federal directory of substance-use and
- *   mental-health treatment facilities. The published CSV carries the facility name, an optional
- *   secondary name (typically the organizational parent), and the postal address quad split into
- *   primary + secondary street lines. Phase 1.6 §1.2 (#22) selects this source for the same reason
- *   it selects HRSA: facility names are hand-typed venue strings and the addresses pass through
- *   enough human + system hands to accumulate the suite-designator + sub-tenant chaos ("Suite C,
- *   behind main building") that pure gazetteer data does not.
- *
- *   SAMHSA's two-line address shape is the key adapter-specific concern. `street1` typically carries
- *   the canonical postal address (`"123 Main St"`); `street2` carries the suite / unit / "second
- *   floor" surface form. The adapter joins them with `", "` into a single `street` component (Phase
- *   1 keeps `unit` as a deferred slot since the OpenCage template doesn't have a clean rendering
- *   for it). Operators wanting a different join policy can subclass the factory.
- *
- *   Column names below match the canonical SAMHSA Behavioral Health Treatment Services Locator CSV
- *   export header. Operators substituting a closely-related extract should rename columns to match;
- *   the README has the mapping cheatsheet.
- *
- *   License: stamped `"Public Domain"` per the SAMHSA Open Data Foundry's federal-government
- *   distribution terms.
- */
-import { reconcileComponents } from "@mailwoman/formatter";
-import { CSVSpliterator } from "spliterator";
-import { composeRaw, splitStreetLine, stableSourceID } from "#adapters/utils";
-import { lookupStateAbbreviation } from "#codex/us-fips-state";
-/**
- * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
- * it came from.
- */
-export const USGOV_SAMHSA_ADAPTER_ID = "usgov-samhsa-treatment-locator";
-/**
- * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
- * than having to look them up.
- */
-export const USGOV_SAMHSA_DEFAULT_LICENSE = "Public Domain";
-/**
- * Join the SAMHSA two-line street: primary street + optional secondary line (suite / unit / floor / "behind main
- * building") on `", "`. The combined value is the `street` component surface form. Phase 1 does not break this out into
- * the `unit` component — see the file-level comment.
- */
-function joinTwoLineStreet(street1, street2) {
-    const s1 = street1.trim();
-    const s2 = (street2 ?? "").trim();
-    if (!s1 && !s2)
-        return "";
-    if (!s2)
-        return s1;
-    if (!s1)
-        return s2;
-    return `${s1}, ${s2}`;
-}
-/**
- * Combine `name1` + optional `name2` into a single venue surface form. SAMHSA conventions:
- *
- * - `name1` is the program / clinic name ("Mountain Plains Counseling Services").
- * - `name2` is the parent organization ("Catholic Charities of Wyoming"), if any.
- *
- * Both render together as `"<name1> - <name2>"` when both are present — geocoder users typically type either form, so
- * the model benefits from the joined surface.
- */
-function composeVenue(name1, name2) {
-    const n1 = name1.trim();
-    const n2 = (name2 ?? "").trim();
-    if (!n1 && !n2)
-        return "";
-    if (!n2)
-        return n1;
-    if (!n1)
-        return n2;
-    return `${n1} - ${n2}`;
-}
-export function createUsgovSamhsaTreatmentLocatorAdapter() {
-    return {
-        id: USGOV_SAMHSA_ADAPTER_ID,
-        defaultLicense: USGOV_SAMHSA_DEFAULT_LICENSE,
-        description: "SAMHSA Behavioral Health Treatment Services Locator (public-domain). Adversarial source: venue + two-line address co-occurrence, hand-entered.",
-        async *rows(opts) {
-            if (opts.country && opts.country !== "US") {
-                throw new Error(`usgov-samhsa adapter: only US supported, got country=${opts.country}`);
-            }
-            const rows = CSVSpliterator.fromAsync(opts.inputPath, {
-                mode: "object",
-                normalizeKeys: false,
-                enableQuoteHandling: true,
-            });
-            let emitted = 0;
-            for await (const record of rows) {
-                if (opts.signal?.aborted)
-                    break;
-                if (opts.limit !== undefined && emitted >= opts.limit)
-                    break;
-                const venue = composeVenue(record.name1 ?? "", record.name2);
-                const street = joinTwoLineStreet(record.street1 ?? "", record.street2);
-                const split = splitStreetLine(street);
-                const city = (record.city ?? "").trim();
-                const stateAbbr = (record.state ?? "").trim();
-                const postcode = (record.zip ?? "").trim();
-                if (!venue || !split || !city || !postcode)
-                    continue;
-                const state = lookupStateAbbreviation(stateAbbr);
-                if (!state)
-                    continue;
-                // venue first — same kryptonite-defending insertion order as HRSA.
-                const components = {
-                    venue,
-                    ...(split.house_number ? { house_number: split.house_number } : {}),
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode,
-                };
-                const raw = composeRaw({
-                    venue,
-                    houseNumber: split.house_number,
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode,
-                });
-                if (!raw)
-                    continue;
-                const aligned = reconcileComponents(components, raw);
-                if (!Object.keys(aligned).length)
-                    continue;
-                const frID = (record.frid ?? "").trim();
-                const sourceID = frID ? `${USGOV_SAMHSA_ADAPTER_ID}-${frID}` : stableSourceID(USGOV_SAMHSA_ADAPTER_ID, aligned);
-                yield {
-                    raw,
-                    components: aligned,
-                    country: "US",
-                    locale: "en-US",
-                    source: USGOV_SAMHSA_ADAPTER_ID,
-                    source_id: sourceID,
-                    corpus_version: "",
-                    license: USGOV_SAMHSA_DEFAULT_LICENSE,
-                };
-                emitted++;
-            }
-        },
-    };
-}
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export const usgovSamhsaTreatmentLocatorAdapter = createUsgovSamhsaTreatmentLocatorAdapter();
-//# sourceMappingURL=adapter.js.map

package/out/adapters/usgov-samhsa-treatment-locator/adapter.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/usgov-samhsa-treatment-locator/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAC7E,OAAO,EAAE,uBAAuB,EAAE,MAAM,sBAAsB,CAAA;AAG9D;;;GAGG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAAG,gCAAgC,CAAA;AACvE;;;GAGG;AACH,MAAM,CAAC,MAAM,4BAA4B,GAAG,eAAe,CAAA;AAoB3D;;;;GAIG;AACH,SAAS,iBAAiB,CAAC,OAAe,EAAE,OAA2B;IACtE,MAAM,EAAE,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IACzB,MAAM,EAAE,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAEjC,IAAI,CAAC,EAAE,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAEzB,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAElB,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAElB,OAAO,GAAG,EAAE,KAAK,EAAE,EAAE,CAAA;AACtB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,YAAY,CAAC,KAAa,EAAE,KAAyB;IAC7D,MAAM,EAAE,GAAG,KAAK,CAAC,IAAI,EAAE,CAAA;IACvB,MAAM,EAAE,GAAG,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAE/B,IAAI,CAAC,EAAE,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAEzB,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAElB,IAAI,CAAC,EAAE;QAAE,OAAO,EAAE,CAAA;IAElB,OAAO,GAAG,EAAE,MAAM,EAAE,EAAE,CAAA;AACvB,CAAC;AAED,MAAM,UAAU,wCAAwC;IACvD,OAAO;QACN,EAAE,EAAE,uBAAuB;QAC3B,cAAc,EAAE,4BAA4B;QAC5C,WAAW,EACV,gJAAgJ;QAEjJ,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,wDAAwD,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YACxF,CAAC;YAED,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE;gBACrD,IAAI,EAAE,QAAQ;gBACd,aAAa,EAAE,KAAK;gBACpB,mBAAmB,EAAE,IAAI;aACzB,CAAC,CAAA;YAEF,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAAoC,EAAE,CAAC;gBACjE,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,KAAK,GAAG,YAAY,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,EAAE,MAAM,CAAC,KAAK,CAAC,CAAA;gBAC5D,MAAM,MAAM,GAAG,iBAAiB,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,EAAE,MAAM,CAAC,OAAO,CAAC,CAAA;gBACtE,MAAM,KAAK,GAAG,eAAe,CAAC,MAAM,CAAC,CAAA;gBACrC,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACvC,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1C,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACpD,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;gBAEhD,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,mEAAmE;gBACnE,MAAM,UAAU,GAA+B;oBAC9C,KAAK;oBACL,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ;iBACR,CAAA;gBAED,MAAM,GAAG,GAAG,UAAU,CAAC;oBACtB,KAAK;oBACL,WAAW,EAAE,KAAK,CAAC,YAAY;oBAC/B,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ;iBACR,CAAC,CAAA;gBAEF,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;oBAAE,SAAQ;gBAE1C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAEvC,MAAM,QAAQ,GAAG,IAAI,CAAC,CAAC,CAAC,GAAG,uBAAuB,IAAI,IAAI,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC,uBAAuB,EAAE,OAAO,CAAC,CAAA;gBAE/G,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,uBAAuB;oBAC/B,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,4BAA4B;iBACrC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,kCAAkC,GAAG,wCAAwC,EAAE,CAAA"}

package/out/adapters/utils/index.d.ts

@@ -1,191 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   Adapter framework helpers — the pieces every corpus adapter and the runner share.
- *
- *   This file does **not** define `CorpusAdapter` (that lives in `./types.ts`, which is the single
- *   canonical schema module). It exposes:
- *
- *   - `AdapterRegistry`: a tiny lookup table the CLI + build pipeline use to find adapters by id.
- *   - `InMemoryAdapterRegistry`: the default implementation.
- *   - `stableSourceID(adapterID, components)`: deterministic content-addressed id for adapters whose
- *       source data has no native primary key (CSV, GeoJSON).
- *   - `HOUSE_NUMBER_PREFIX` + `splitStreetLine(line)`: the one house-number/street split every
- *       US CSV adapter uses.
- *   - `composeRaw(parts)`: the envelope-style raw line every US CSV adapter renders.
- *   - `canonicalDedupKey(row)`: normalized signature used to drop near-identical rows during a run.
- *       Adapter-internal dedup; cross-adapter dedup is the runner's job.
- *   - `streamingSha256()`: thin wrapper around `node:crypto` so the runner can hash JSONL output as it
- *       streams (avoids re-reading the slice for the manifest checksum).
- *   - `loadLibpostalDictionary(language, filename)`: the curated libpostal dictionary reader the
- *       street-decompose modules share — the one read this module performs; everything else here is
- *       pure, and other side-effecting code goes in `./runner.ts`.
- */
-import type { ComponentTag } from "@mailwoman/core/types";
-import type { CanonicalRow, CorpusAdapter } from "#types";
-/**
- * Lookup table for corpus adapters.
- *
- * The CLI's `npx mailwoman corpus run <adapter-id>` resolves `<adapter-id>` against this registry; the same registry is
- * iterated by the `corpus build` pipeline. Adapters do not self-register at module load — they're added explicitly so
- * the dependency graph stays traceable.
- */
-export interface AdapterRegistry {
-    /**
-     * Add an adapter. Throws if `adapter.id` is already registered.
-     */
-    register(adapter: CorpusAdapter): void;
-    /**
-     * Return the adapter for `id`, or `undefined`.
-     */
-    get(id: string): CorpusAdapter | undefined;
-    /**
-     * All registered adapters, in insertion order.
-     */
-    list(): readonly CorpusAdapter[];
-    /**
-     * Convenience: ids only, in insertion order.
-     */
-    ids(): readonly string[];
-}
-/**
- * Default in-memory registry. The runner constructs one per invocation; the CLI re-uses a shared singleton
- * (`defaultAdapterRegistry`) populated by `./adapters/index.ts` as adapters come online.
- */
-export declare class InMemoryAdapterRegistry implements AdapterRegistry {
-    #private;
-    register(adapter: CorpusAdapter): void;
-    get(id: string): CorpusAdapter | undefined;
-    list(): readonly CorpusAdapter[];
-    ids(): readonly string[];
-}
-/**
- * Process-wide default registry. Populated by `./adapters/index.ts` as adapters are built; imported by the CLI. Tests
- * should construct their own `InMemoryAdapterRegistry` to avoid cross-test pollution.
- */
-export declare const defaultAdapterRegistry: InMemoryAdapterRegistry;
-/**
- * Deterministic content-addressed source id.
- *
- * For adapters whose upstream source has no native primary key (CSV rows, GeoJSON features), the runner expects a
- * stable id so dedup, holdout manifests, and resumability work across reruns. This helper produces one by hashing the
- * adapter id and a canonical serialization of the components dict (keys sorted, values verbatim).
- *
- * Output format: `<adapterID>-<first-12-hex-chars-of-sha256>`. 48 bits of entropy is enough for ~17M rows per adapter
- * before the expected collision count exceeds 1 (birthday paradox); adapters with more rows should extend the prefix
- * length.
- */
-export declare function stableSourceID(adapterID: string, components: Partial<Record<ComponentTag, string>>): string;
-/**
- * {@link stableSourceID} over arbitrary disambiguator keys — a variant index, a slot number, anything that is not a
- * `ComponentTag`. Every key handed in is sorted and hashed either way; only the key vocabulary differs, and the narrow
- * signature above is what stops an adapter hashing a misspelled component name.
- */
-export declare function stableSourceIDFromParts(adapterID: string, parts: Readonly<Record<string, string | undefined>>): string;
-/**
- * Leading house number on a US-style street line: digits, an optional hyphenated range (Queens `40-12`), an optional
- * single alpha suffix (`101A`), then whitespace, then the rest.
- *
- * This decides where the house number ends and the street begins for every US CSV adapter, so an edge case fixed here
- * is fixed for all of them.
- *
- * The remainder must stay `(\S.*)` and must NOT become `(.+)`: `\s+` and `.` both match a tab, so `\s+(.+)$` lets the
- * engine split a run of tabs between the two groups every possible way before failing — quadratic backtracking on
- * attacker-shaped input. Requiring a non-space start removes the overlap. Group 2 is trimmed by the caller either way,
- * so the two forms are indistinguishable on real input; only the failure cost differs.
- */
-export declare const HOUSE_NUMBER_PREFIX: RegExp;
-/**
- * A street line split into its house number and the remainder.
- */
-export interface SplitStreetLine {
-    house_number?: string;
-    street: string;
-}
-/**
- * Split a US-style street line on {@link HOUSE_NUMBER_PREFIX}.
- *
- * The US CSV sources follow USPS Publication 28 conventions with hand-entry drift. The leading digit run is the house
- * number (`"123 Main St"`, `"6450 W Indian School Rd"`), and the regex tolerates one trailing letter (`"123A Main St"`)
- * plus an optional hyphenated half (`"40-12 Bell Blvd"`, common in NYC and suburban garden-apartment numbering; Hawaii
- * uses it island-wide — `"47-470 Hui Aeko Place"`).
- *
- * Returns `null` for blank input. Anything that does not match the prefix shape (`"PO Box 1234"`, `"RR 2 Box 67"`, `"HC
- * 1"`) becomes a single `street` value rather than being mangled — the model sees the original surface form and
- * downstream classifiers pick it up. Callers that need those forms recognized as something other than a street (see
- * `usgov-irs-bmf`) test for them BEFORE calling this.
- */
-export declare function splitStreetLine(line: string): SplitStreetLine | null;
-/**
- * Parts of the envelope-style raw line {@link composeRaw} renders. Every field is dropped when empty, so a caller only
- * supplies what its source carries.
- */
-export interface ComposeRawParts {
-    venue?: string;
-    houseNumber?: string;
-    street?: string;
-    /**
-     * A street line already joined by the caller (e.g. a PO-box phrase) — used INSTEAD of `houseNumber`/`street`/`unit`
-     * when present.
-     */
-    streetLine?: string;
-    unit?: string;
-    locality: string;
-    region?: string;
-    postcode: string;
-}
-/**
- * Compose the envelope-style raw address line the US CSV adapters share:
- *
- * `"<venue>, <house> <street> <unit>, <city>, <state> <postcode>"`
- *
- * The venue leads (US conventional addressee-then-address ordering) so a downstream model sees the
- * venue-prefix-then-address shape users actually type into geocoders. Empty segments drop out rather than leaving a
- * dangling separator.
- */
-export declare function composeRaw(parts: ComposeRawParts): string;
-/**
- * Load one curated libpostal dictionary (`core/data/libpostal/dictionaries/<language>/<filename>`) as a lower-cased
- * form set. libpostal format: `canonical|abbr|abbr|...` — every form is indexed.
- *
- * `resourceDictionaryPath` already resolves both layouts — `core/data/...` from source and from the packaged `out/`
- * tree. The candidate list this replaced named it TWICE and then guessed a third path off `process.cwd()`, and
- * swallowed every error while probing, so a corrupt dictionary reported as a missing one.
- *
- * The largest libpostal dictionary is 8.4 KB, and each caller runs this once per process at module load.
- */
-export declare function loadLibpostalDictionary(language: string, filename: string): Promise<Set<string>>;
-/**
- * Canonical dedup key for a row.
- *
- * Two rows that share this key are treated as duplicates and only the first wins. The key is built from `country`, the
- * sorted `components` dict, and a normalized `raw` (lower-cased, whitespace collapsed). License and provenance fields
- * are intentionally excluded so the same address from multiple adapters is recognized as a duplicate.
- *
- * Synthetic rows are never deduplicated against natural rows: `synth.method` is folded into the key when present,
- * ensuring each augmentation variant survives.
- */
-export declare function canonicalDedupKey(row: CanonicalRow): string;
-/**
- * Streaming SHA-256 hasher.
- *
- * The runner feeds every JSONL line into one of these so the per-slice checksum can be recorded in `MANIFEST.json`
- * without a second pass over the slice. Implementation is a one-line wrapper, but giving it a name keeps the runner's
- * hash-tracking intent obvious.
- */
-export interface StreamingHasher {
-    update(chunk: string | Uint8Array): void;
-    digest(): string;
-}
-/**
- * Default `StreamingHasher` (SHA-256, hex).
- */
-/**
- * @see {@link StreamingHasher} — this is the incremental counterpart to `sha256Hex`, not a copy of
- *   it. `@mailwoman/core/utils` exposes one-shot digests (`sha256Hex`) and whole-file digests
- *   (`sha256File`); the runner needs neither, because it hashes JSONL lines as they stream past.
- */
-export declare function streamingSha256(): StreamingHasher;
-//# sourceMappingURL=index.d.ts.map

package/out/adapters/utils/index.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../lib/adapters/utils/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAMH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAGzD,OAAO,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzD;;;;;;GAMG;AACH,MAAM,WAAW,eAAe;IAC/B;;OAEG;IACH,QAAQ,CAAC,OAAO,EAAE,aAAa,GAAG,IAAI,CAAA;IAEtC;;OAEG;IACH,GAAG,CAAC,EAAE,EAAE,MAAM,GAAG,aAAa,GAAG,SAAS,CAAA;IAE1C;;OAEG;IACH,IAAI,IAAI,SAAS,aAAa,EAAE,CAAA;IAEhC;;OAEG;IACH,GAAG,IAAI,SAAS,MAAM,EAAE,CAAA;CACxB;AAED;;;GAGG;AACH,qBAAa,uBAAwB,YAAW,eAAe;;IAG9D,QAAQ,CAAC,OAAO,EAAE,aAAa,GAAG,IAAI;IAQtC,GAAG,CAAC,EAAE,EAAE,MAAM,GAAG,aAAa,GAAG,SAAS;IAI1C,IAAI,IAAI,SAAS,aAAa,EAAE;IAIhC,GAAG,IAAI,SAAS,MAAM,EAAE;CAGxB;AAED;;;GAGG;AACH,eAAO,MAAM,sBAAsB,yBAAgC,CAAA;AAEnE;;;;;;;;;;GAUG;AACH,wBAAgB,cAAc,CAAC,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,OAAO,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,GAAG,MAAM,CAE3G;AAED;;;;GAIG;AACH,wBAAgB,uBAAuB,CACtC,SAAS,EAAE,MAAM,EACjB,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS,CAAC,CAAC,GACjD,MAAM,CAQR;AAED;;;;;;;;;;;GAWG;AACH,eAAO,MAAM,mBAAmB,QAAuC,CAAA;AAEvE;;GAEG;AACH,MAAM,WAAW,eAAe;IAC/B,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB,MAAM,EAAE,MAAM,CAAA;CACd;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,eAAe,GAAG,IAAI,CAUpE;AAED;;;GAGG;AACH,MAAM,WAAW,eAAe;IAC/B,KAAK,CAAC,EAAE,MAAM,CAAA;IACd,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB,MAAM,CAAC,EAAE,MAAM,CAAA;IACf;;;OAGG;IACH,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,QAAQ,EAAE,MAAM,CAAA;IAChB,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,QAAQ,EAAE,MAAM,CAAA;CAChB;AAED;;;;;;;;GAQG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,eAAe,GAAG,MAAM,CASzD;AAED;;;;;;;;;GASG;AACH,wBAAsB,uBAAuB,CAAC,QAAQ,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAmBtG;AAED;;;;;;;;;GASG;AACH,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,YAAY,GAAG,MAAM,CAO3D;AAED;;;;;;GAMG;AACH,MAAM,WAAW,eAAe;IAC/B,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,UAAU,GAAG,IAAI,CAAA;IACxC,MAAM,IAAI,MAAM,CAAA;CAChB;AAED;;GAEG;AACH;;;;GAIG;AACH,wBAAgB,eAAe,IAAI,eAAe,CAmBjD"}