@mailwoman/corpus 9.2.0 → 9.3.0

package/out/src/adapters/tiger/street-decompose.js

@@ -1,90 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   Decompose a US street name into Stage 3 components: street_prefix, street, street_suffix.
- *
- *   Sources directionals and street types from the curated libpostal/en dictionaries
- *   (`core/data/libpostal/dictionaries/en/{directionals,street_types}.txt`). These are the same
- *   dictionaries the runtime classifiers (StreetPrefixClassifier, StreetSuffixClassifier) use, so
- *   corpus labels and runtime classifications agree on the vocabulary.
- *
- *   Examples: "N Main St" → { prefix: "N", street: "Main", suffix: "St" } "Pennsylvania Avenue NW" →
- *   { prefix: null, street: "Pennsylvania", suffix: "Avenue NW" } "Salmon St" → { prefix: null,
- *   street: "Salmon", suffix: "St" } "SE Hawthorne Blvd" → { prefix: "SE", street: "Hawthorne",
- *   suffix: "Blvd" }
- */
-import { readFileSync } from "node:fs";
-import { resourceDictionaryPath } from "@mailwoman/core/utils";
-import { TextSpliterator } from "spliterator";
-function loadDictionary(filename) {
-    // `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.
-    const text = readFileSync(resourceDictionaryPath("libpostal", "en", filename), "utf8");
-    const set = new Set();
-    // The largest libpostal dictionary is 8.4 KB, and this runs once per process at module load.
-    for (const line of TextSpliterator.from(text)) {
-        const trimmed = line.trim();
-        if (!trimmed || trimmed.startsWith("#"))
-            continue;
-        // libpostal format: canonical|abbr|abbr|... — index all forms
-        for (const form of trimmed.split("|")) {
-            const f = form.trim().toLowerCase();
-            if (f) {
-                set.add(f);
-            }
-        }
-    }
-    return set;
-}
-const DIRECTIONALS = loadDictionary("directionals.txt");
-const STREET_TYPES = loadDictionary("street_types.txt");
-/**
- * Decompose a US street name into prefix/name/suffix components.
- *
- * Conservative — only emits prefix/suffix when there's a clear directional or street-type keyword. Returns the original
- * as `street` if nothing matches.
- */
-export function decomposeStreet(fullname) {
-    const trimmed = fullname.trim();
-    if (!trimmed)
-        return { prefix: null, street: "", suffix: null };
-    const tokens = trimmed.split(/\s+/);
-    if (tokens.length === 1)
-        return { prefix: null, street: trimmed, suffix: null };
-    const norm = (s) => s.toLowerCase().replace(/\.$/, "");
-    let prefix = null;
-    let suffix = null;
-    let startIdx = 0;
-    let endIdx = tokens.length;
-    // Leading directional prefix
-    if (DIRECTIONALS.has(norm(tokens[0])) && tokens.length >= 2) {
-        prefix = tokens[0];
-        startIdx = 1;
-    }
-    // Trailing post-directional combined with street type (e.g. "Pennsylvania Ave NW")
-    const last = norm(tokens[endIdx - 1]);
-    const secondLast = endIdx >= 2 ? norm(tokens[endIdx - 2]) : "";
-    if (DIRECTIONALS.has(last) && STREET_TYPES.has(secondLast)) {
-        suffix = tokens.slice(endIdx - 2, endIdx).join(" ");
-        endIdx -= 2;
-    }
-    else if (STREET_TYPES.has(last) && endIdx - startIdx >= 2) {
-        suffix = tokens[endIdx - 1];
-        endIdx -= 1;
-    }
-    else if (DIRECTIONALS.has(last) && endIdx - startIdx >= 2) {
-        // Post-directional without type
-        suffix = tokens[endIdx - 1];
-        endIdx -= 1;
-    }
-    const street = tokens.slice(startIdx, endIdx).join(" ").trim();
-    if (!street) {
-        return { prefix: null, street: trimmed, suffix: null };
-    }
-    return { prefix, street, suffix };
-}
-//# sourceMappingURL=street-decompose.js.map

package/out/src/adapters/tiger/street-decompose.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"street-decompose.js","sourceRoot":"","sources":["../../../../src/adapters/tiger/street-decompose.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,SAAS,CAAA;AAEtC,OAAO,EAAE,sBAAsB,EAAE,MAAM,uBAAuB,CAAA;AAC9D,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,SAAS,cAAc,CAAC,QAAgB;IACvC,oGAAoG;IACpG,sGAAsG;IACtG,sGAAsG;IACtG,eAAe;IACf,MAAM,IAAI,GAAG,YAAY,CAAC,sBAAsB,CAAC,WAAW,EAAE,IAAI,EAAE,QAAQ,CAAC,EAAE,MAAM,CAAC,CAAA;IACtF,MAAM,GAAG,GAAG,IAAI,GAAG,EAAU,CAAA;IAE7B,6FAA6F;IAC7F,KAAK,MAAM,IAAI,IAAI,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC/C,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAA;QAE3B,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC;YAAE,SAAQ;QAEjD,8DAA8D;QAC9D,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAA;YAEnC,IAAI,CAAC,EAAE,CAAC;gBACP,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAA;YACX,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED,MAAM,YAAY,GAAG,cAAc,CAAC,kBAAkB,CAAC,CAAA;AACvD,MAAM,YAAY,GAAG,cAAc,CAAC,kBAAkB,CAAC,CAAA;AAQvD;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,QAAgB;IAC/C,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,EAAE,CAAA;IAE/B,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAA;IAE/D,MAAM,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;IAEnC,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,CAAA;IAE/E,MAAM,IAAI,GAAG,CAAC,CAAS,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;IAE9D,IAAI,MAAM,GAAkB,IAAI,CAAA;IAChC,IAAI,MAAM,GAAkB,IAAI,CAAA;IAChC,IAAI,QAAQ,GAAG,CAAC,CAAA;IAChB,IAAI,MAAM,GAAG,MAAM,CAAC,MAAM,CAAA;IAE1B,6BAA6B;IAC7B,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC,CAAC,IAAI,MAAM,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;QAC9D,MAAM,GAAG,MAAM,CAAC,CAAC,CAAE,CAAA;QACnB,QAAQ,GAAG,CAAC,CAAA;IACb,CAAC;IAED,mFAAmF;IACnF,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAE,CAAC,CAAA;IACtC,MAAM,UAAU,GAAG,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAE/D,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,YAAY,CAAC,GAAG,CAAC,UAAU,CAAC,EAAE,CAAC;QAC5D,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;QACnD,MAAM,IAAI,CAAC,CAAA;IACZ,CAAC;SAAM,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,MAAM,GAAG,QAAQ,IAAI,CAAC,EAAE,CAAC;QAC7D,MAAM,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,CAAE,CAAA;QAC5B,MAAM,IAAI,CAAC,CAAA;IACZ,CAAC;SAAM,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,MAAM,GAAG,QAAQ,IAAI,CAAC,EAAE,CAAC;QAC7D,gCAAgC;QAChC,MAAM,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,CAAE,CAAA;QAC5B,MAAM,IAAI,CAAC,CAAA;IACZ,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IAE9D,IAAI,CAAC,MAAM,EAAE,CAAC;QACb,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,CAAA;IACvD,CAAC;IAED,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,CAAA;AAClC,CAAC"}

package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts

@@ -1,47 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-hrsa-fqhc`: HRSA "Health Center Service Delivery Site Locations" CSV consumer.
- *
- *   Federally Qualified Health Centers (FQHCs) are HRSA-funded community health programs that
- *   self-report site addresses to the HRSA Data Warehouse. The published CSV (`data.hrsa.gov`)
- *   carries the site name, the postal-formatted street address, and the locality/region/postcode
- *   quad. The source earns its place on adversarial-value-per-row (#22): every facility name is a
- *   human-typed venue string and the addresses pass through enough hands to accumulate the
- *   abbreviation drift + suite designator chaos that pure gazetteer data does not.
- *
- *   The adapter consumes a CSV file the operator pre-downloads. The HRSA data is published as a
- *   single national CSV (~10K rows), small enough that the operator can re-fetch on every corpus
- *   rebuild without an intermediate SQLite step. Column names below match the HRSA Data Warehouse's
- *   "Health Center Service Delivery Site" public dataset. Operators substituting the
- *   closely-related "Site Address" or "Health Center" public extracts may need to remap columns;
- *   the README documents the expected set.
- *
- *   Output: one row per CSV record, with `venue` component carrying the site name and the address
- *   quad on `(house_number, street, locality, region, postcode)`. Component order is critical:
- *   `venue` is inserted FIRST so alignment claims its surface span before `locality` searches for
- *   its own (the kryptonite case "Buffalo Health Clinic, …, Buffalo, NY" relies on `venue`
- *   consuming the first "Buffalo" so locality lands on the second).
- *
- *   License: stamped `"Public Domain"` per the HRSA Data Warehouse'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_HRSA_FQHC_ADAPTER_ID = "usgov-hrsa-fqhc";
-/**
- * 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_HRSA_FQHC_DEFAULT_LICENSE = "Public Domain";
-export declare function createUsgovHrsaFqhcAdapter(): CorpusAdapter;
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export declare const usgovHrsaFqhcAdapter: CorpusAdapter;
-//# sourceMappingURL=adapter.d.ts.map

package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-hrsa-fqhc/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAQH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,0BAA0B,oBAAoB,CAAA;AAC3D;;;GAGG;AACH,eAAO,MAAM,+BAA+B,kBAAkB,CAAA;AAmD9D,wBAAgB,0BAA0B,IAAI,aAAa,CA8E1D;AAED;;GAEG;AACH,eAAO,MAAM,oBAAoB,eAA+B,CAAA"}

package/out/src/adapters/usgov-hrsa-fqhc/adapter.js

@@ -1,139 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-hrsa-fqhc`: HRSA "Health Center Service Delivery Site Locations" CSV consumer.
- *
- *   Federally Qualified Health Centers (FQHCs) are HRSA-funded community health programs that
- *   self-report site addresses to the HRSA Data Warehouse. The published CSV (`data.hrsa.gov`)
- *   carries the site name, the postal-formatted street address, and the locality/region/postcode
- *   quad. The source earns its place on adversarial-value-per-row (#22): every facility name is a
- *   human-typed venue string and the addresses pass through enough hands to accumulate the
- *   abbreviation drift + suite designator chaos that pure gazetteer data does not.
- *
- *   The adapter consumes a CSV file the operator pre-downloads. The HRSA data is published as a
- *   single national CSV (~10K rows), small enough that the operator can re-fetch on every corpus
- *   rebuild without an intermediate SQLite step. Column names below match the HRSA Data Warehouse's
- *   "Health Center Service Delivery Site" public dataset. Operators substituting the
- *   closely-related "Site Address" or "Health Center" public extracts may need to remap columns;
- *   the README documents the expected set.
- *
- *   Output: one row per CSV record, with `venue` component carrying the site name and the address
- *   quad on `(house_number, street, locality, region, postcode)`. Component order is critical:
- *   `venue` is inserted FIRST so alignment claims its surface span before `locality` searches for
- *   its own (the kryptonite case "Buffalo Health Clinic, …, Buffalo, NY" relies on `venue`
- *   consuming the first "Buffalo" so locality lands on the second).
- *
- *   License: stamped `"Public Domain"` per the HRSA Data Warehouse's federal government distribution
- *   terms.
- */
-import { isPresent } from "@mailwoman/core/objects";
-import { reconcileComponents } from "@mailwoman/formatter";
-import { CSVSpliterator } from "spliterator";
-import { 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_HRSA_FQHC_ADAPTER_ID = "usgov-hrsa-fqhc";
-/**
- * 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_HRSA_FQHC_DEFAULT_LICENSE = "Public Domain";
-/**
- * Split a "123 Main St Suite 4" surface form into `(house_number, street)`. The regex tolerates one trailing letter on
- * the number (`"123A Main St"`) and a hyphenated form (`"40-12 Bell Blvd"`); anything else falls back to street-only.
- *
- * Suite / Apt / Unit designators stay on `street` — Mailwoman's `unit` component exists but the address-formatter does
- * not have a clean slot for it, and HRSA addresses do not separate the suite into its own column. Leaving the surface
- * form intact in `street` preserves the adversarial training signal (the model learns that a trailing "Suite 4" is part
- * of the road line in this distribution).
- */
-/**
- * Compose the raw envelope-style address line. Format:
- *
- * "<Site Name>, <house> <street>, <city>, <state> <postcode>"
- *
- * The site name leads (US conventional addressee-then-address ordering) so a downstream model sees the
- * venue-prefix-then-address shape that HRSA users actually type into geocoders.
- */
-function composeRaw(venue, house, street, city, state, postcode) {
-    const streetPart = [house, street].filter(isPresent).join(" ").trim();
-    const cityPart = [city.trim(), [state, postcode].filter(isPresent).join(" ").trim()].filter(isPresent).join(", ");
-    return [venue.trim(), streetPart, cityPart].filter(isPresent).join(", ");
-}
-export function createUsgovHrsaFqhcAdapter() {
-    return {
-        id: USGOV_HRSA_FQHC_ADAPTER_ID,
-        defaultLicense: USGOV_HRSA_FQHC_DEFAULT_LICENSE,
-        description: "HRSA Federally Qualified Health Center site locations (public-domain). Adversarial source: venue + address co-occurrence, hand-entered.",
-        async *rows(opts) {
-            if (opts.country && opts.country !== "US") {
-                throw new Error(`usgov-hrsa-fqhc 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 = (record["Site Name"] ?? "").trim();
-                const split = splitStreetLine(record["Site Address"] ?? "");
-                const city = (record["Site City"] ?? "").trim();
-                const stateAbbr = (record["Site State Abbreviation"] ?? "").trim();
-                const postcode = (record["Site Postal Code"] ?? "").trim();
-                if (!venue || !split || !city || !postcode)
-                    continue;
-                const state = lookupStateAbbreviation(stateAbbr);
-                if (!state)
-                    continue;
-                // Insertion order matters here. `venue` first so alignment claims its span
-                // (which may contain a token like "Buffalo") before `locality` runs its
-                // search — the kryptonite case `Buffalo Health Clinic, Buffalo NY`
-                // otherwise mis-labels the venue's "Buffalo" as locality.
-                const components = {
-                    venue,
-                    ...(split.house_number ? { house_number: split.house_number } : {}),
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode,
-                };
-                const raw = composeRaw(venue, split.house_number, split.street, city, state.abbreviation, postcode);
-                if (!raw)
-                    continue;
-                const aligned = reconcileComponents(components, raw);
-                if (!Object.keys(aligned).length)
-                    continue;
-                const siteID = (record["Site ID"] ?? "").trim();
-                const sourceID = siteID
-                    ? `${USGOV_HRSA_FQHC_ADAPTER_ID}-${siteID}`
-                    : stableSourceID(USGOV_HRSA_FQHC_ADAPTER_ID, aligned);
-                yield {
-                    raw,
-                    components: aligned,
-                    country: "US",
-                    locale: "en-US",
-                    source: USGOV_HRSA_FQHC_ADAPTER_ID,
-                    source_id: sourceID,
-                    corpus_version: "",
-                    license: USGOV_HRSA_FQHC_DEFAULT_LICENSE,
-                };
-                emitted++;
-            }
-        },
-    };
-}
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export const usgovHrsaFqhcAdapter = createUsgovHrsaFqhcAdapter();
-//# sourceMappingURL=adapter.js.map

package/out/src/adapters/usgov-hrsa-fqhc/adapter.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-hrsa-fqhc/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;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,eAAe,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AACjE,OAAO,EAAE,uBAAuB,EAAE,MAAM,sBAAsB,CAAA;AAG9D;;;GAGG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iBAAiB,CAAA;AAC3D;;;GAGG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAG,eAAe,CAAA;AAmB9D;;;;;;;;GAQG;AAEH;;;;;;;GAOG;AACH,SAAS,UAAU,CAClB,KAAa,EACb,KAAyB,EACzB,MAAc,EACd,IAAY,EACZ,KAAa,EACb,QAAgB;IAEhB,MAAM,UAAU,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACrE,MAAM,QAAQ,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEjH,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACzE,CAAC;AAED,MAAM,UAAU,0BAA0B;IACzC,OAAO;QACN,EAAE,EAAE,0BAA0B;QAC9B,cAAc,EAAE,+BAA+B;QAC/C,WAAW,EACV,yIAAyI;QAE1I,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,2DAA2D,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YAC3F,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,IAAkC,EAAE,CAAC;gBAC/D,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,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAChD,MAAM,KAAK,GAAG,eAAe,CAAC,MAAM,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAA;gBAC3D,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC/C,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,yBAAyB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAClE,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,kBAAkB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1D,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,2EAA2E;gBAC3E,wEAAwE;gBACxE,mEAAmE;gBACnE,0DAA0D;gBAC1D,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,KAAK,EAAE,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;gBAEnG,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,MAAM,GAAG,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE/C,MAAM,QAAQ,GAAG,MAAM;oBACtB,CAAC,CAAC,GAAG,0BAA0B,IAAI,MAAM,EAAE;oBAC3C,CAAC,CAAC,cAAc,CAAC,0BAA0B,EAAE,OAAO,CAAC,CAAA;gBAEtD,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,0BAA0B;oBAClC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,+BAA+B;iBACxC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,0BAA0B,EAAE,CAAA"}

package/out/src/adapters/usgov-imls-pls/adapter.d.ts

@@ -1,36 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-imls-pls`: IMLS Public Libraries Survey outlet CSV consumer.
- *
- *   The Institute of Museum and Library Services publishes an annual Public Libraries Survey with one
- *   row per library outlet (~17K rows). Each row carries the library name, street address, city,
- *   ZIP, county, and geocoordinates.
- *
- *   The adapter consumes the outlet CSV the operator pre-downloads via `fetch-imls-pls.ts`. Column
- *   names match the IMLS PLS outlet file header.
- *
- *   Output: one row per outlet with `venue` (library name), `(house_number, street, locality,
- *   subregion, postcode)`, and lat/lon preserved in `source_id` stability.
- *
- *   License: stamped `"Public Domain"` per IMLS 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_IMLS_PLS_ADAPTER_ID = "usgov-imls-pls";
-/**
- * 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_IMLS_PLS_DEFAULT_LICENSE = "Public Domain";
-export declare function createUsgovImlsPlsAdapter(): CorpusAdapter;
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export declare const usgovImlsPlsAdapter: CorpusAdapter;
-//# sourceMappingURL=adapter.d.ts.map

package/out/src/adapters/usgov-imls-pls/adapter.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAQH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,yBAAyB,mBAAmB,CAAA;AACzD;;;GAGG;AACH,eAAO,MAAM,8BAA8B,kBAAkB,CAAA;AAY7D,wBAAgB,yBAAyB,IAAI,aAAa,CAwFzD;AAED;;GAEG;AACH,eAAO,MAAM,mBAAmB,eAA8B,CAAA"}

package/out/src/adapters/usgov-imls-pls/adapter.js

@@ -1,114 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-imls-pls`: IMLS Public Libraries Survey outlet CSV consumer.
- *
- *   The Institute of Museum and Library Services publishes an annual Public Libraries Survey with one
- *   row per library outlet (~17K rows). Each row carries the library name, street address, city,
- *   ZIP, county, and geocoordinates.
- *
- *   The adapter consumes the outlet CSV the operator pre-downloads via `fetch-imls-pls.ts`. Column
- *   names match the IMLS PLS outlet file header.
- *
- *   Output: one row per outlet with `venue` (library name), `(house_number, street, locality,
- *   subregion, postcode)`, and lat/lon preserved in `source_id` stability.
- *
- *   License: stamped `"Public Domain"` per IMLS federal government distribution terms.
- */
-import { isPresent } from "@mailwoman/core/objects";
-import { reconcileComponents } from "@mailwoman/formatter";
-import { CSVSpliterator } from "spliterator";
-import { 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_IMLS_PLS_ADAPTER_ID = "usgov-imls-pls";
-/**
- * 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_IMLS_PLS_DEFAULT_LICENSE = "Public Domain";
-export function createUsgovImlsPlsAdapter() {
-    return {
-        id: USGOV_IMLS_PLS_ADAPTER_ID,
-        defaultLicense: USGOV_IMLS_PLS_DEFAULT_LICENSE,
-        description: "IMLS Public Libraries Survey — ~17K library outlets with venue+address (public-domain).",
-        async *rows(opts) {
-            if (opts.country && opts.country !== "US") {
-                throw new Error(`usgov-imls-pls 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 libName = (record.LIBNAME ?? "").trim();
-                const address = (record.ADDRESS ?? "").trim();
-                const city = (record.CITY ?? "").trim();
-                const zip = (record.ZIP ?? "").trim();
-                const stateAbbr = (record.STABR ?? "").trim();
-                if (!libName || !city || !zip)
-                    continue;
-                const state = lookupStateAbbreviation(stateAbbr);
-                if (!state)
-                    continue;
-                const split = splitStreetLine(address);
-                if (!split)
-                    continue;
-                const components = {
-                    venue: libName,
-                    ...(split.house_number ? { house_number: split.house_number } : {}),
-                    street: split.street,
-                    locality: city,
-                    region: state.abbreviation,
-                    postcode: zip,
-                    // #552: no subregion — US postal addresses don't surface the county, so emitting
-                    // subregion creates a phantom component with no raw-span to align to, quarantining
-                    // ~21% of rows. The county is still available in the source CSV; it just isn't
-                    // a postal-surface component here.
-                };
-                const streetPart = [split.house_number, split.street].filter(isPresent).join(" ").trim();
-                const raw = [
-                    libName,
-                    streetPart,
-                    [city, [stateAbbr, zip].filter(isPresent).join(" ")].filter(isPresent).join(", "),
-                ]
-                    .filter(isPresent)
-                    .join(", ");
-                const aligned = reconcileComponents(components, raw);
-                if (Object.keys(aligned).length <= 2)
-                    continue;
-                const fscsKey = (record.FSCSKEY ?? "").trim();
-                const sourceID = fscsKey
-                    ? `${USGOV_IMLS_PLS_ADAPTER_ID}-${fscsKey}`
-                    : stableSourceID(USGOV_IMLS_PLS_ADAPTER_ID, aligned);
-                yield {
-                    raw,
-                    components: aligned,
-                    country: "US",
-                    locale: "en-US",
-                    source: USGOV_IMLS_PLS_ADAPTER_ID,
-                    source_id: sourceID,
-                    corpus_version: "",
-                    license: USGOV_IMLS_PLS_DEFAULT_LICENSE,
-                };
-                emitted++;
-            }
-        },
-    };
-}
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export const usgovImlsPlsAdapter = createUsgovImlsPlsAdapter();
-//# sourceMappingURL=adapter.js.map

package/out/src/adapters/usgov-imls-pls/adapter.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;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,eAAe,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AACjE,OAAO,EAAE,uBAAuB,EAAE,MAAM,sBAAsB,CAAA;AAG9D;;;GAGG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AACzD;;;GAGG;AACH,MAAM,CAAC,MAAM,8BAA8B,GAAG,eAAe,CAAA;AAY7D,MAAM,UAAU,yBAAyB;IACxC,OAAO;QACN,EAAE,EAAE,yBAAyB;QAC7B,cAAc,EAAE,8BAA8B;QAC9C,WAAW,EAAE,yFAAyF;QAEtG,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,0DAA0D,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YAC1F,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,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACvC,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrC,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAEvC,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;gBAEhD,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,KAAK,GAAG,eAAe,CAAC,OAAO,CAAC,CAAA;gBAEtC,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,UAAU,GAA+B;oBAC9C,KAAK,EAAE,OAAO;oBACd,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,EAAE,GAAG;oBACb,iFAAiF;oBACjF,mFAAmF;oBACnF,+EAA+E;oBAC/E,mCAAmC;iBACnC,CAAA;gBAED,MAAM,UAAU,GAAG,CAAC,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;gBAExF,MAAM,GAAG,GAAG;oBACX,OAAO;oBACP,UAAU;oBACV,CAAC,IAAI,EAAE,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;iBACjF;qBACC,MAAM,CAAC,SAAS,CAAC;qBACjB,IAAI,CAAC,IAAI,CAAC,CAAA;gBAEZ,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,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,MAAM,QAAQ,GAAG,OAAO;oBACvB,CAAC,CAAC,GAAG,yBAAyB,IAAI,OAAO,EAAE;oBAC3C,CAAC,CAAC,cAAc,CAAC,yBAAyB,EAAE,OAAO,CAAC,CAAA;gBAErD,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,yBAAyB;oBACjC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,8BAA8B;iBACvC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,EAAE,CAAA"}

package/out/src/adapters/usgov-irs-bmf/adapter.d.ts

@@ -1,37 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-irs-bmf`: IRS Exempt Organizations Business Master File (EO BMF) CSV consumer.
- *
- *   The EO BMF is the IRS's authoritative registry of US tax-exempt organizations (charities,
- *   churches, foundations, ...), published as per-region CSVs at
- *   `https://www.irs.gov/charities-non-profits/exempt-organizations-business-master-file-extract-eo-bmf`
- *   (`eo1.csv`..`eo4.csv`, `eo_pr.csv`, `eo_xx.csv`). Each row carries an organization NAME plus
- *   its mailing address. It complements `usgov-nppes` with a DIFFERENT venue population
- *   (non-profits vs healthcare providers) and, notably, a high share of PO-box addresses — useful
- *   `po_box`-tag signal (a tag with historically low recall).
- *
- *   Output: one row per record with a usable city + postcode. NAME → `venue`; the street line becomes
- *   `po_box` when it's a PO-box, else `house_number` + `street`; CITY/STATE/ZIP fill the locality
- *   line. STATE is already a USPS abbreviation in the source. License: `"Public Domain"` (US
- *   federal).
- */
-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_IRS_BMF_ADAPTER_ID = "usgov-irs-bmf";
-/**
- * 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_IRS_BMF_DEFAULT_LICENSE = "Public Domain";
-export declare function createUsgovIrsBmfAdapter(): CorpusAdapter;
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export declare const usgovIrsBmfAdapter: CorpusAdapter;
-//# sourceMappingURL=adapter.d.ts.map

package/out/src/adapters/usgov-irs-bmf/adapter.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAOH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,wBAAwB,kBAAkB,CAAA;AACvD;;;GAGG;AACH,eAAO,MAAM,6BAA6B,kBAAkB,CAAA;AA2C5D,wBAAgB,wBAAwB,IAAI,aAAa,CA6ExD;AAED;;GAEG;AACH,eAAO,MAAM,kBAAkB,eAA6B,CAAA"}

package/out/src/adapters/usgov-irs-bmf/adapter.js

@@ -1,124 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `usgov-irs-bmf`: IRS Exempt Organizations Business Master File (EO BMF) CSV consumer.
- *
- *   The EO BMF is the IRS's authoritative registry of US tax-exempt organizations (charities,
- *   churches, foundations, ...), published as per-region CSVs at
- *   `https://www.irs.gov/charities-non-profits/exempt-organizations-business-master-file-extract-eo-bmf`
- *   (`eo1.csv`..`eo4.csv`, `eo_pr.csv`, `eo_xx.csv`). Each row carries an organization NAME plus
- *   its mailing address. It complements `usgov-nppes` with a DIFFERENT venue population
- *   (non-profits vs healthcare providers) and, notably, a high share of PO-box addresses — useful
- *   `po_box`-tag signal (a tag with historically low recall).
- *
- *   Output: one row per record with a usable city + postcode. NAME → `venue`; the street line becomes
- *   `po_box` when it's a PO-box, else `house_number` + `street`; CITY/STATE/ZIP fill the locality
- *   line. STATE is already a USPS abbreviation in the source. License: `"Public Domain"` (US
- *   federal).
- */
-import { isPresent } from "@mailwoman/core/objects";
-import { reconcileComponents } from "@mailwoman/formatter";
-import { CSVSpliterator } from "spliterator";
-import { splitStreetLine, stableSourceID } from "#adapters/utils";
-/**
- * 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_IRS_BMF_ADAPTER_ID = "usgov-irs-bmf";
-/**
- * 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_IRS_BMF_DEFAULT_LICENSE = "Public Domain";
-/**
- * PO box in its many written forms: "PO BOX 12", "P.O. BOX 12", "P O BOX 12", "POB 12", "BOX 12".
- */
-const PO_BOX = /^\s*(?:P\.?\s?O\.?\s*BOX|POB|BOX)\s+\w/i;
-/**
- * Classify the street line into a `po_box` or a `{house_number?, street}` split.
- *
- * BMF mixes street addresses and PO boxes in one `STREET` column, so the PO-box shapes have to be claimed BEFORE the
- * shared house-number split runs — otherwise `splitStreetLine` would hand back `"PO Box 1234"` as a plain street, which
- * is correct for every other US adapter and wrong here.
- */
-function splitStreetLineOrPOBox(street) {
-    const trimmed = street.trim();
-    if (PO_BOX.test(trimmed))
-        return { po_box: trimmed };
-    return splitStreetLine(trimmed);
-}
-function composeRaw(venue, streetPart, city, state, postcode) {
-    const cityPart = [city.trim(), [state, postcode].filter(isPresent).join(" ").trim()].filter(isPresent).join(", ");
-    return [venue, streetPart, cityPart].filter(isPresent).join(", ");
-}
-export function createUsgovIrsBmfAdapter() {
-    return {
-        id: USGOV_IRS_BMF_ADAPTER_ID,
-        defaultLicense: USGOV_IRS_BMF_DEFAULT_LICENSE,
-        description: "IRS Exempt Organizations Business Master File — US non-profit venue+address (public-domain), with strong PO-box coverage.",
-        async *rows(opts) {
-            if (opts.country && opts.country !== "US") {
-                throw new Error(`usgov-irs-bmf 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 ein = (record.EIN ?? "").trim();
-                const venue = (record.NAME ?? "").trim() || undefined;
-                const street = (record.STREET ?? "").trim();
-                const city = (record.CITY ?? "").trim();
-                const state = (record.STATE ?? "").trim();
-                const zipRaw = (record.ZIP ?? "").trim();
-                if (!city || !zipRaw)
-                    continue;
-                const postcode = zipRaw.split("-")[0].trim(); // 5-digit; drop the optional +4
-                const split = splitStreetLineOrPOBox(street);
-                if (!split)
-                    continue;
-                const streetPart = "po_box" in split ? split.po_box : [split.house_number, split.street].filter(isPresent).join(" ");
-                const components = {
-                    ...(venue ? { venue } : {}),
-                    ...("po_box" in split
-                        ? { po_box: split.po_box }
-                        : { ...(split.house_number ? { house_number: split.house_number } : {}), street: split.street }),
-                    locality: city,
-                    ...(state ? { region: state } : {}),
-                    postcode,
-                };
-                const raw = composeRaw(venue, streetPart, city, state, postcode);
-                if (!raw)
-                    continue;
-                const aligned = reconcileComponents(components, raw);
-                if (Object.keys(aligned).length <= 2)
-                    continue;
-                const sourceID = ein ? `${USGOV_IRS_BMF_ADAPTER_ID}-${ein}` : stableSourceID(USGOV_IRS_BMF_ADAPTER_ID, aligned);
-                yield {
-                    raw,
-                    components: aligned,
-                    country: "US",
-                    locale: "en-US",
-                    source: USGOV_IRS_BMF_ADAPTER_ID,
-                    source_id: sourceID,
-                    corpus_version: "",
-                    license: USGOV_IRS_BMF_DEFAULT_LICENSE,
-                };
-                emitted++;
-            }
-        },
-    };
-}
-/**
- * The configured adapter instance registered with the corpus builder.
- */
-export const usgovIrsBmfAdapter = createUsgovIrsBmfAdapter();
-//# sourceMappingURL=adapter.js.map

package/out/src/adapters/usgov-irs-bmf/adapter.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;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,eAAe,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAGjE;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,eAAe,CAAA;AACvD;;;GAGG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAG,eAAe,CAAA;AAE5D;;GAEG;AACH,MAAM,MAAM,GAAG,yCAAyC,CAAA;AAWxD;;;;;;GAMG;AACH,SAAS,sBAAsB,CAAC,MAAc;IAC7C,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,EAAE,CAAA;IAE7B,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;IAEpD,OAAO,eAAe,CAAC,OAAO,CAAC,CAAA;AAChC,CAAC;AAED,SAAS,UAAU,CAClB,KAAyB,EACzB,UAAkB,EAClB,IAAY,EACZ,KAAa,EACb,QAAgB;IAEhB,MAAM,QAAQ,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEjH,OAAO,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AAClE,CAAC;AAED,MAAM,UAAU,wBAAwB;IACvC,OAAO;QACN,EAAE,EAAE,wBAAwB;QAC5B,cAAc,EAAE,6BAA6B;QAC7C,WAAW,EACV,2HAA2H;QAE5H,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,yDAAyD,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YACzF,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,IAAgC,EAAE,CAAC;gBAC7D,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,KAAK,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAA;gBACrD,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC3C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACvC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACzC,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAExC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM;oBAAE,SAAQ;gBAC9B,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAA,CAAC,gCAAgC;gBAE9E,MAAM,KAAK,GAAG,sBAAsB,CAAC,MAAM,CAAC,CAAA;gBAE5C,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,UAAU,GACf,QAAQ,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;gBAElG,MAAM,UAAU,GAA+B;oBAC9C,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC3B,GAAG,CAAC,QAAQ,IAAI,KAAK;wBACpB,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE;wBAC1B,CAAC,CAAC,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC;oBACjG,QAAQ,EAAE,IAAI;oBACd,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnC,QAAQ;iBACR,CAAA;gBAED,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAA;gBAEhE,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,wBAAwB,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC,wBAAwB,EAAE,OAAO,CAAC,CAAA;gBAE/G,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,wBAAwB;oBAChC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,6BAA6B;iBACtC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,wBAAwB,EAAE,CAAA"}