@mailwoman/corpus 10.6.0 → 10.7.0

package/out/gu/adapters/bsp/adapter.js

@@ -0,0 +1,346 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `bsp-gu`: the address points of Guam, which the Bureau of Statistics & Plans GIS publishes on the
+ * Government of Guam GISWEB Portal as the ArcGIS layer `Address_Point_(Public_View)`.
+ *
+ * The input is the layer's GeoJSON query pages, one file or a directory of them. The adapter reads
+ * `HOUSE_NUM` as the house number, `PRE_DIR` as the street prefix, `STREETNAME` as the street (or the
+ * assembled `FullRdNm` when `STREETNAME` is blank), `STR_TYPE` as the street suffix, `UNIT` as the unit
+ * and `CITY` (or `PLACENAME` when `CITY` is blank) as the locality. Route roads read `ROUTE 10` with the
+ * type `HWY`, which is how the layer's own `FullRdNm` writes them.
+ *
+ * The layer carries no postcode: its NG911 `Post_Code` column is blank on every feature.
+ *
+ * `HOUSE_NUM` sometimes pairs the number with a building-permit number (`120 / B19000483`); the number is
+ * kept and the permit dropped. `CITY` writes Guam's villages with Chamorro and English spellings
+ * (`HAGÅTÑA`, `HAGATNA`; `INARAJAN`, `INALAHAN`) and with typing slips (`BARRIGDAA`, `MANGIL.AO`), so a
+ * value is kept only when each of its parts is a village, or a community the villages contain,
+ * listed in {@link GUAM_PLACE_NAMES}. A road written `(OFF OF) ROUTE 4A HWY` locates a point near a road
+ * rather than on it, and a street naming two roads (`AGUILAR RD / RT 17`) states no single street, so both
+ * are refused.
+ *
+ * The layer states no license. Under the United States position in `packages/corpus/lib/us/AGENTS.md`
+ * (Feist), a US public body's address file published without terms restricting its use is elected,
+ * and the rows carry the label the US public-body adapters use.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { foldToken } from "@mailwoman/codex/normalize";
+import { readLocalTextFile } from "@mailwoman/core/fs/readers";
+import { tryStat } from "@mailwoman/core/fs/readers/stat";
+import { parseJSONStrict } from "@mailwoman/core/json";
+import { resolvePath } from "path-ts";
+import { Globerator } from "spliterator/node/fs";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { stableSourceIDFromParts } from "#adapters/source-id";
+import { SourceRegister } from "#registers";
+import { AddressRole, countDropped, SurfaceOrigin, } from "#types";
+/**
+ * Registry id for this adapter, stamped into every row it emits.
+ */
+export const BSP_GU_ADAPTER_ID = "bsp-gu";
+/**
+ * The label every row carries, matching the US public-body adapters elected under the Feist reading.
+ */
+export const BSP_GU_LICENSE = "Public Domain";
+/**
+ * Every jurisdiction this adapter emits, checked against a caller's `--country`.
+ */
+export const BSP_GU_COUNTRIES = ["GU"];
+/**
+ * A value with its whitespace collapsed, or null when the layer leaves it blank.
+ * The layer writes an absent string as a single space.
+ */
+function present(value) {
+    const collapsed = value == null ? "" : String(value).replaceAll(/\s+/gu, " ").trim();
+    return collapsed || null;
+}
+/**
+ * Upper-cased, with diacritics removed, so `HAGÅTÑA` and `Hagatna` compare equal.
+ */
+function upperFold(value) {
+    return foldToken(value).toUpperCase();
+}
+/**
+ * Guam's nineteen villages under each name the layer writes for them,
+ * and the communities within a village that `CITY` names on its own
+ * (Tumon and Harmon in Tamuning, Ipan in Talofofo, Paasan in Agana Heights, Anigua in Hagåtña), folded.
+ */
+export const GUAM_PLACE_NAMES = new Set([
+    "AGANA HEIGHTS",
+    "PAASAN",
+    "AGAT",
+    "HAGAT",
+    "ASAN",
+    "MAINA",
+    "ASAN MAINA",
+    "BARRIGADA",
+    "BARRIGADA HEIGHTS",
+    "CHALAN PAGO",
+    "ORDOT",
+    "CHALAN PAGO ORDOT",
+    "DEDEDO",
+    "HAGATNA",
+    "AGANA",
+    "ANIGUA",
+    "INARAJAN",
+    "INALAHAN",
+    "INALAJAN",
+    "MANGILAO",
+    "MERIZO",
+    "MALESSO",
+    "MONGMONG",
+    "TOTO",
+    "MAITE",
+    "MONGMONG TOTO MAITE",
+    "PITI",
+    "SANTA RITA",
+    "SINAJANA",
+    "TALOFOFO",
+    "IPAN",
+    "TAMUNING",
+    "TUMON",
+    "HARMON",
+    "UMATAC",
+    "HUMATAC",
+    "YIGO",
+    "YONA",
+]);
+/**
+ * Whether every part of a place value, split at commas, slashes, hyphens and parentheses,
+ * is a listed place: `CHALAN PAGO - ORDOT` and `AGANA HEIGHTS (PAASAN)` are, `RADIO, BARRIGADA` is not.
+ */
+export function isGuamPlaceName(value) {
+    const parts = upperFold(value)
+        .split(/[,/()-]/u)
+        .map((part) => part.trim())
+        .filter((part) => part.length > 0);
+    return parts.length > 0 && parts.every((part) => GUAM_PLACE_NAMES.has(part));
+}
+/**
+ * A house number: digits, an optional hyphenated half, and an optional letter written
+ * against the number or after a space (`120`, `99-12`, `231A`, `155 A`).
+ */
+const HOUSE_NUMBER = /^\d+(?:-\d+)?(?: ?[A-Z])?$/u;
+/**
+ * A Guam building-permit number, written beside a house number in `HOUSE_NUM` (`B19000483`, `B22-450`).
+ */
+const PERMIT_NUMBER = /^[A-Z]{1,2}\d{2}-?\d{3,}$/iu;
+/**
+ * The house number `HOUSE_NUM` states, with a building-permit number written beside
+ * it removed, or null when the value is not a house number.
+ */
+export function guamHouseNumber(value) {
+    if (HOUSE_NUMBER.test(value))
+        return value;
+    const parts = value
+        .split("/")
+        .map((part) => part.trim())
+        .filter((part) => part.length > 0);
+    const numbers = parts.filter((part) => HOUSE_NUMBER.test(part));
+    if (numbers.length === 1 && parts.every((part) => numbers.includes(part) || PERMIT_NUMBER.test(part))) {
+        return numbers[0];
+    }
+    return null;
+}
+/**
+ * The street types `STR_TYPE` writes, as the layer abbreviates or spells them.
+ */
+const STREET_TYPES = new Set([
+    "ST",
+    "STREET",
+    "HWY",
+    "HIGHWAY",
+    "RD",
+    "ROAD",
+    "CT",
+    "CRT",
+    "COURT",
+    "DR",
+    "DRIVE",
+    "LN",
+    "LANE",
+    "AVE",
+    "AVENUE",
+    "LOOP",
+    "LP",
+    "WAY",
+    "WY",
+    "CIR",
+    "CIRCLE",
+    "BLVD",
+    "BOULEVARD",
+    "PL",
+]);
+const DIRECTIONS = new Set(["N", "S", "E", "W"]);
+/**
+ * Maps one record onto the components the corpus asserts, or returns null when the row is refused.
+ *
+ * Each refusal and each dropped component increments its reason in `dropped`.
+ */
+export function componentsForGuamRecord(record, dropped) {
+    const tally = dropped ?? {};
+    const written = present(record.HOUSE_NUM);
+    if (!written) {
+        countDropped(tally, "row:no-house-number");
+        return null;
+    }
+    const houseNumber = guamHouseNumber(written);
+    if (!houseNumber) {
+        countDropped(tally, "row:house-number-not-a-number");
+        return null;
+    }
+    const name = present(record.STREETNAME);
+    const street = name ?? present(record.FullRdNm);
+    if (!street) {
+        countDropped(tally, "row:no-street");
+        return null;
+    }
+    if (/^\(?OFF OF\b/iu.test(street)) {
+        countDropped(tally, "row:street-off-of");
+        return null;
+    }
+    if (/[/()]/u.test(street)) {
+        countDropped(tally, "row:street-names-two-roads");
+        return null;
+    }
+    const components = { house_number: houseNumber, street };
+    // `FullRdNm` already holds the direction and the type, so they are read beside `STREETNAME` only.
+    if (name) {
+        const direction = present(record.PRE_DIR)?.toUpperCase();
+        if (direction && DIRECTIONS.has(direction)) {
+            if (upperFold(name).startsWith(`${direction} `)) {
+                countDropped(tally, "component:street_prefix:repeats-street");
+            }
+            else {
+                components.street_prefix = direction;
+            }
+        }
+        else if (direction) {
+            countDropped(tally, "component:street_prefix:not-a-direction");
+        }
+        const type = present(record.STR_TYPE)?.toUpperCase().replace(/\.$/u, "");
+        if (type && STREET_TYPES.has(type)) {
+            if (upperFold(name).endsWith(` ${type}`)) {
+                countDropped(tally, "component:street_suffix:repeats-street");
+            }
+            else {
+                components.street_suffix = type;
+            }
+        }
+        else if (type) {
+            countDropped(tally, "component:street_suffix:not-a-street-type");
+        }
+    }
+    const unit = present(record.UNIT);
+    if (unit && /^[A-Z0-9]{1,4}$/u.test(unit)) {
+        // `HOUSE_NUM` sometimes writes the unit letter after the number as well (`155 A` with unit `A`).
+        if (houseNumber.endsWith(` ${unit}`) || (/^[A-Z]$/u.test(unit) && houseNumber.endsWith(unit))) {
+            countDropped(tally, "component:unit:repeats-house-number");
+        }
+        else {
+            components.unit = unit;
+        }
+    }
+    else if (unit) {
+        countDropped(tally, "component:unit:not-a-unit-designator");
+    }
+    const city = present(record.CITY);
+    const place = present(record.PLACENAME);
+    if (city && isGuamPlaceName(city)) {
+        components.locality = city;
+    }
+    else if (city) {
+        countDropped(tally, "component:locality:not-a-guam-place");
+    }
+    else if (place && isGuamPlaceName(place)) {
+        components.locality = place;
+    }
+    if (!components.locality) {
+        countDropped(tally, "row:no-locality");
+        return null;
+    }
+    return components;
+}
+/**
+ * The GeoJSON pages an input names: the file itself, or every `.geojson` in a directory, in name order.
+ */
+async function pageFiles(inputPath) {
+    const stat = await tryStat(inputPath);
+    if (!stat?.isDirectory())
+        return [inputPath];
+    const names = await Globerator.from("*.geojson", { cwd: inputPath, absolute: false }).toSorted();
+    return names.map((name) => `${inputPath}/${name}`);
+}
+export function createGuamAddressPointsAdapter() {
+    return {
+        id: BSP_GU_ADAPTER_ID,
+        defaultLicense: BSP_GU_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.GuamAddressPoints,
+        surface: SurfaceOrigin.Rendered,
+        description: "Address_Point_(Public_View) (Government of Guam, Bureau of Statistics & Plans GIS): Guam's address points.",
+        async *rows(opts) {
+            if (opts.country && !BSP_GU_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(BSP_GU_ADAPTER_ID, BSP_GU_COUNTRIES, opts.country);
+            }
+            let emitted = 0;
+            for (const file of await pageFiles(resolvePath(opts.inputPath))) {
+                // A query page holds at most the layer's 2,000-feature maxRecordCount, so each is parsed whole.
+                const collection = parseJSONStrict(await readLocalTextFile(file));
+                if (!Array.isArray(collection.features)) {
+                    throw new TypeError(`${BSP_GU_ADAPTER_ID} adapter: ${file} holds no GeoJSON features array`);
+                }
+                for (const feature of collection.features) {
+                    if (opts.signal?.aborted)
+                        return;
+                    if (opts.limit !== undefined && emitted >= opts.limit)
+                        return;
+                    const record = feature.properties;
+                    if (!record) {
+                        countDropped(opts, "row:no-properties");
+                        continue;
+                    }
+                    const components = componentsForGuamRecord(record, opts);
+                    if (!components)
+                        continue;
+                    const rendered = formatAddressRow(components, "GU", { singleLine: true });
+                    if (!rendered) {
+                        countDropped(opts, "row:render-failed");
+                        continue;
+                    }
+                    for (const tag of rendered.unplaced) {
+                        countDropped(opts, `component:${tag}:not-rendered`);
+                    }
+                    const { raw, components: aligned } = rendered;
+                    const [lon, lat] = feature.geometry?.coordinates ?? [];
+                    yield {
+                        raw,
+                        components: aligned,
+                        country: "GU",
+                        locale: "en-GU",
+                        source: BSP_GU_ADAPTER_ID,
+                        // `FID` is the layer's object id, unique per feature.
+                        // The point separates the units of one building that the rendered components may not.
+                        source_id: stableSourceIDFromParts(BSP_GU_ADAPTER_ID, {
+                            ...aligned,
+                            fid: present(record.FID) ?? "",
+                            lat: typeof lat === "number" ? String(lat) : "",
+                            lon: typeof lon === "number" ? String(lon) : "",
+                        }),
+                        corpus_version: "",
+                        license: BSP_GU_LICENSE,
+                    };
+                    emitted++;
+                }
+            }
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const guamAddressPointsAdapter = createGuamAddressPointsAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/gu/adapters/bsp/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/gu/adapters/bsp/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAClE,OAAO,EAAE,SAAS,EAAE,MAAM,4BAA4B,CAAA;AACtD,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,OAAO,EAAE,MAAM,iCAAiC,CAAA;AACzD,OAAO,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAA;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,SAAS,CAAA;AACrC,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAEhD,OAAO,EAAE,uBAAuB,EAAE,MAAM,kBAAkB,CAAA;AAC1D,OAAO,EAAE,uBAAuB,EAAE,MAAM,qBAAqB,CAAA;AAC7D,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EACN,WAAW,EAIX,YAAY,EACZ,aAAa,GACb,MAAM,QAAQ,CAAA;AAEf;;GAEG;AACH,MAAM,CAAC,MAAM,iBAAiB,GAAG,QAAQ,CAAA;AAEzC;;GAEG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG,eAAe,CAAA;AAE7C;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAsB,CAAC,IAAI,CAAC,CAAA;AAwBzD;;;GAGG;AACH,SAAS,OAAO,CAAC,KAAyC;IACzD,MAAM,SAAS,GAAG,KAAK,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IAEpF,OAAO,SAAS,IAAI,IAAI,CAAA;AACzB,CAAC;AAED;;GAEG;AACH,SAAS,SAAS,CAAC,KAAa;IAC/B,OAAO,SAAS,CAAC,KAAK,CAAC,CAAC,WAAW,EAAE,CAAA;AACtC,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAwB,IAAI,GAAG,CAAC;IAC5D,eAAe;IACf,QAAQ;IACR,MAAM;IACN,OAAO;IACP,MAAM;IACN,OAAO;IACP,YAAY;IACZ,WAAW;IACX,mBAAmB;IACnB,aAAa;IACb,OAAO;IACP,mBAAmB;IACnB,QAAQ;IACR,SAAS;IACT,OAAO;IACP,QAAQ;IACR,UAAU;IACV,UAAU;IACV,UAAU;IACV,UAAU;IACV,QAAQ;IACR,SAAS;IACT,UAAU;IACV,MAAM;IACN,OAAO;IACP,qBAAqB;IACrB,MAAM;IACN,YAAY;IACZ,UAAU;IACV,UAAU;IACV,MAAM;IACN,UAAU;IACV,OAAO;IACP,QAAQ;IACR,QAAQ;IACR,SAAS;IACT,MAAM;IACN,MAAM;CACN,CAAC,CAAA;AAEF;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa;IAC5C,MAAM,KAAK,GAAG,SAAS,CAAC,KAAK,CAAC;SAC5B,KAAK,CAAC,UAAU,CAAC;SACjB,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEnC,OAAO,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAA;AAC7E,CAAC;AAED;;;GAGG;AACH,MAAM,YAAY,GAAG,6BAA6B,CAAA;AAElD;;GAEG;AACH,MAAM,aAAa,GAAG,6BAA6B,CAAA;AAEnD;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa;IAC5C,IAAI,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,KAAK,CAAA;IAE1C,MAAM,KAAK,GAAG,KAAK;SACjB,KAAK,CAAC,GAAG,CAAC;SACV,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEnC,MAAM,OAAO,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;IAE/D,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACvG,OAAO,OAAO,CAAC,CAAC,CAAE,CAAA;IACnB,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;GAEG;AACH,MAAM,YAAY,GAAwB,IAAI,GAAG,CAAC;IACjD,IAAI;IACJ,QAAQ;IACR,KAAK;IACL,SAAS;IACT,IAAI;IACJ,MAAM;IACN,IAAI;IACJ,KAAK;IACL,OAAO;IACP,IAAI;IACJ,OAAO;IACP,IAAI;IACJ,MAAM;IACN,KAAK;IACL,QAAQ;IACR,MAAM;IACN,IAAI;IACJ,KAAK;IACL,IAAI;IACJ,KAAK;IACL,QAAQ;IACR,MAAM;IACN,WAAW;IACX,IAAI;CACJ,CAAC,CAAA;AAEF,MAAM,UAAU,GAAwB,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAA;AAErE;;;;GAIG;AACH,MAAM,UAAU,uBAAuB,CACtC,MAA8B,EAC9B,OAAyC;IAEzC,MAAM,KAAK,GAAG,OAAO,IAAI,EAAE,CAAA;IAC3B,MAAM,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;IAEzC,IAAI,CAAC,OAAO,EAAE,CAAC;QACd,YAAY,CAAC,KAAK,EAAE,qBAAqB,CAAC,CAAA;QAE1C,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,MAAM,WAAW,GAAG,eAAe,CAAC,OAAO,CAAC,CAAA;IAE5C,IAAI,CAAC,WAAW,EAAE,CAAC;QAClB,YAAY,CAAC,KAAK,EAAE,+BAA+B,CAAC,CAAA;QAEpD,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,UAAU,CAAC,CAAA;IACvC,MAAM,MAAM,GAAG,IAAI,IAAI,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAA;IAE/C,IAAI,CAAC,MAAM,EAAE,CAAC;QACb,YAAY,CAAC,KAAK,EAAE,eAAe,CAAC,CAAA;QAEpC,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,IAAI,gBAAgB,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;QACnC,YAAY,CAAC,KAAK,EAAE,mBAAmB,CAAC,CAAA;QAExC,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,IAAI,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC;QAC3B,YAAY,CAAC,KAAK,EAAE,4BAA4B,CAAC,CAAA;QAEjD,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,MAAM,UAAU,GAA+B,EAAE,YAAY,EAAE,WAAW,EAAE,MAAM,EAAE,CAAA;IAEpF,kGAAkG;IAClG,IAAI,IAAI,EAAE,CAAC;QACV,MAAM,SAAS,GAAG,OAAO,CAAC,MAAM,CAAC,OAAO,CAAC,EAAE,WAAW,EAAE,CAAA;QAExD,IAAI,SAAS,IAAI,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;YAC5C,IAAI,SAAS,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,GAAG,SAAS,GAAG,CAAC,EAAE,CAAC;gBACjD,YAAY,CAAC,KAAK,EAAE,wCAAwC,CAAC,CAAA;YAC9D,CAAC;iBAAM,CAAC;gBACP,UAAU,CAAC,aAAa,GAAG,SAAS,CAAA;YACrC,CAAC;QACF,CAAC;aAAM,IAAI,SAAS,EAAE,CAAC;YACtB,YAAY,CAAC,KAAK,EAAE,yCAAyC,CAAC,CAAA;QAC/D,CAAC;QAED,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,WAAW,EAAE,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA;QAExE,IAAI,IAAI,IAAI,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YACpC,IAAI,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,CAAC,IAAI,IAAI,EAAE,CAAC,EAAE,CAAC;gBAC1C,YAAY,CAAC,KAAK,EAAE,wCAAwC,CAAC,CAAA;YAC9D,CAAC;iBAAM,CAAC;gBACP,UAAU,CAAC,aAAa,GAAG,IAAI,CAAA;YAChC,CAAC;QACF,CAAC;aAAM,IAAI,IAAI,EAAE,CAAC;YACjB,YAAY,CAAC,KAAK,EAAE,2CAA2C,CAAC,CAAA;QACjE,CAAC;IACF,CAAC;IAED,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;IAEjC,IAAI,IAAI,IAAI,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QAC3C,iGAAiG;QACjG,IAAI,WAAW,CAAC,QAAQ,CAAC,IAAI,IAAI,EAAE,CAAC,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,WAAW,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;YAC/F,YAAY,CAAC,KAAK,EAAE,qCAAqC,CAAC,CAAA;QAC3D,CAAC;aAAM,CAAC;YACP,UAAU,CAAC,IAAI,GAAG,IAAI,CAAA;QACvB,CAAC;IACF,CAAC;SAAM,IAAI,IAAI,EAAE,CAAC;QACjB,YAAY,CAAC,KAAK,EAAE,sCAAsC,CAAC,CAAA;IAC5D,CAAC;IAED,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;IACjC,MAAM,KAAK,GAAG,OAAO,CAAC,MAAM,CAAC,SAAS,CAAC,CAAA;IAEvC,IAAI,IAAI,IAAI,eAAe,CAAC,IAAI,CAAC,EAAE,CAAC;QACnC,UAAU,CAAC,QAAQ,GAAG,IAAI,CAAA;IAC3B,CAAC;SAAM,IAAI,IAAI,EAAE,CAAC;QACjB,YAAY,CAAC,KAAK,EAAE,qCAAqC,CAAC,CAAA;IAC3D,CAAC;SAAM,IAAI,KAAK,IAAI,eAAe,CAAC,KAAK,CAAC,EAAE,CAAC;QAC5C,UAAU,CAAC,QAAQ,GAAG,KAAK,CAAA;IAC5B,CAAC;IAED,IAAI,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC;QAC1B,YAAY,CAAC,KAAK,EAAE,iBAAiB,CAAC,CAAA;QAEtC,OAAO,IAAI,CAAA;IACZ,CAAC;IAED,OAAO,UAAU,CAAA;AAClB,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,SAAS,CAAC,SAAiB;IACzC,MAAM,IAAI,GAAG,MAAM,OAAO,CAAC,SAAS,CAAC,CAAA;IAErC,IAAI,CAAC,IAAI,EAAE,WAAW,EAAE;QAAE,OAAO,CAAC,SAAS,CAAC,CAAA;IAE5C,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,WAAW,EAAE,EAAE,GAAG,EAAE,SAAS,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAEhG,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,GAAG,SAAS,IAAI,IAAI,EAAE,CAAC,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,8BAA8B;IAC7C,OAAO;QACN,EAAE,EAAE,iBAAiB;QACrB,cAAc,EAAE,cAAc;QAC9B,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,iBAAiB;QAC1C,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,4GAA4G;QAE7G,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAC9D,MAAM,IAAI,uBAAuB,CAAC,iBAAiB,EAAE,gBAAgB,EAAE,IAAI,CAAC,OAAO,CAAC,CAAA;YACrF,CAAC;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,KAAK,MAAM,IAAI,IAAI,MAAM,SAAS,CAAC,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC;gBACjE,gGAAgG;gBAChG,MAAM,UAAU,GAAG,eAAe,CAAwB,MAAM,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAA;gBAExF,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;oBACzC,MAAM,IAAI,SAAS,CAAC,GAAG,iBAAiB,aAAa,IAAI,kCAAkC,CAAC,CAAA;gBAC7F,CAAC;gBAED,KAAK,MAAM,OAAO,IAAI,UAAU,CAAC,QAAQ,EAAE,CAAC;oBAC3C,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,OAAM;oBAEhC,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,OAAM;oBAE7D,MAAM,MAAM,GAAG,OAAO,CAAC,UAAU,CAAA;oBAEjC,IAAI,CAAC,MAAM,EAAE,CAAC;wBACb,YAAY,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;wBAEvC,SAAQ;oBACT,CAAC;oBAED,MAAM,UAAU,GAAG,uBAAuB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;oBAExD,IAAI,CAAC,UAAU;wBAAE,SAAQ;oBAEzB,MAAM,QAAQ,GAAG,gBAAgB,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAA;oBAEzE,IAAI,CAAC,QAAQ,EAAE,CAAC;wBACf,YAAY,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;wBAEvC,SAAQ;oBACT,CAAC;oBAED,KAAK,MAAM,GAAG,IAAI,QAAQ,CAAC,QAAQ,EAAE,CAAC;wBACrC,YAAY,CAAC,IAAI,EAAE,aAAa,GAAG,eAAe,CAAC,CAAA;oBACpD,CAAC;oBAED,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,QAAQ,CAAA;oBAC7C,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC,GAAG,OAAO,CAAC,QAAQ,EAAE,WAAW,IAAI,EAAE,CAAA;oBAEtD,MAAM;wBACL,GAAG;wBACH,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,IAAI;wBACb,MAAM,EAAE,OAAO;wBACf,MAAM,EAAE,iBAAiB;wBACzB,sDAAsD;wBACtD,sFAAsF;wBACtF,SAAS,EAAE,uBAAuB,CAAC,iBAAiB,EAAE;4BACrD,GAAG,OAAO;4BACV,GAAG,EAAE,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE;4BAC9B,GAAG,EAAE,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE;4BAC/C,GAAG,EAAE,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE;yBAC/C,CAAC;wBACF,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,cAAc;qBACvB,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,8BAA8B,EAAE,CAAA"}

package/out/im/adapters/land-registry/adapter.d.ts

@@ -0,0 +1,68 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `landregistry-im`: the property addresses of the Isle of Man Land Registry's land transactions, which the
+ * Isle of Man Government publishes on gov.im as one CSV of every registered sale since 2000.
+ *
+ * Each row is a sale, so one property appears once per sale. The adapter emits each distinct address
+ * once, comparing the emitted components case-folded and keeping the first. It reads only the address
+ * columns and never the price or date columns.
+ *
+ * `SubUnit_Name` is the flat or apartment (`Flat 2`, `Apartment 5`), `House_Number` the door number,
+ * `House_Name` the house's given name (`Ballacain Farm`) or the building's (`Spectrum Apartments`),
+ * `Street_Name` the street, `Locality` the estate or village inside the town, and `Town` the post town.
+ * The IM layout prints no region, so `Parish_` is not read.
+ *
+ * The register also records sales of land that has no postal address. Such a row describes the land
+ * (`Field North Of Ballawhannel Farm`, `Parcel Of Land`, `Land Adjacent To Mostyn Ville`) in place of
+ * naming a property, and the adapter refuses it, as it refuses a garage, a parking space or a substation
+ * sold on its own.
+ *
+ * The Isle of Man Government publishes the dataset under the Isle of Man Open Government Licence, which
+ * requires the attribution statement below. Every row records that license.
+ */
+import { type AdapterOptions, type CanonicalRow, type CorpusAdapter } from "#types";
+/**
+ * Registry id for this adapter, stamped into every row it emits.
+ */
+export declare const LANDREGISTRY_IM_ADAPTER_ID = "landregistry-im";
+/**
+ * The Isle of Man Open Government Licence, whose version the dataset page does not state.
+ * It has no SPDX identifier.
+ */
+export declare const LANDREGISTRY_IM_LICENSE = "LicenseRef-IM-OGL";
+/**
+ * The attribution statement the license requires.
+ */
+export declare const LANDREGISTRY_IM_ATTRIBUTION = "Contains public sector information licensed under the Isle of Man Open Government Licence";
+/**
+ * Every jurisdiction this adapter emits, checked against a caller's `--country`.
+ */
+export declare const LANDREGISTRY_IM_COUNTRIES: readonly string[];
+/**
+ * The address columns the adapter reads from the land transactions CSV.
+ * The price and date columns are not read.
+ */
+export interface IsleOfManLandTransactionRecord {
+    SubUnit_Name: string;
+    House_Number: string;
+    House_Name: string;
+    Street_Name: string;
+    Locality: string;
+    Town: string;
+    Postcode: string;
+}
+/**
+ * Maps one record onto the components the corpus asserts, or returns null when the row is refused.
+ *
+ * Each refusal and each dropped component increments its reason in `dropped`.
+ */
+export declare function componentsForIsleOfManLandTransactionRecord(record: IsleOfManLandTransactionRecord, dropped?: Pick<AdapterOptions, "dropped">): CanonicalRow["components"] | null;
+export declare function createIsleOfManLandRegistryAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const isleOfManLandRegistryAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/im/adapters/land-registry/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/im/adapters/land-registry/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAWH,OAAO,EAEN,KAAK,cAAc,EACnB,KAAK,YAAY,EACjB,KAAK,aAAa,EAGlB,MAAM,QAAQ,CAAA;AAEf;;GAEG;AACH,eAAO,MAAM,0BAA0B,oBAAoB,CAAA;AAE3D;;;GAGG;AACH,eAAO,MAAM,uBAAuB,sBAAsB,CAAA;AAE1D;;GAEG;AACH,eAAO,MAAM,2BAA2B,8FACoD,CAAA;AAE5F;;GAEG;AACH,eAAO,MAAM,yBAAyB,EAAE,SAAS,MAAM,EAAW,CAAA;AAElE;;;GAGG;AACH,MAAM,WAAW,8BAA8B;IAC9C,YAAY,EAAE,MAAM,CAAA;IACpB,YAAY,EAAE,MAAM,CAAA;IACpB,UAAU,EAAE,MAAM,CAAA;IAClB,WAAW,EAAE,MAAM,CAAA;IACnB,QAAQ,EAAE,MAAM,CAAA;IAChB,IAAI,EAAE,MAAM,CAAA;IACZ,QAAQ,EAAE,MAAM,CAAA;CAChB;AA2DD;;;;GAIG;AACH,wBAAgB,2CAA2C,CAC1D,MAAM,EAAE,8BAA8B,EACtC,OAAO,CAAC,EAAE,IAAI,CAAC,cAAc,EAAE,SAAS,CAAC,GACvC,YAAY,CAAC,YAAY,CAAC,GAAG,IAAI,CAsGnC;AAYD,wBAAgB,kCAAkC,IAAI,aAAa,CA2ElE;AAED;;GAEG;AACH,eAAO,MAAM,4BAA4B,eAAuC,CAAA"}