@mailwoman/corpus 10.4.0 → 10.6.0

package/out/gb/adapters/os-open/adapter.js

@@ -0,0 +1,298 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `gb-os-open`: Great Britain street addresses joined from three OS OpenData products, each under
+ * the Open Government Licence v3.0.
+ *
+ * - **Code-Point Open** gives every unit postcode in England, Scotland and Wales a grid point.
+ * - **OS Open Roads** gives each named road link a centreline and the roadNameTOID of its name.
+ * - **OS Open Names** gives each roadNameTOID its `NAME1`, `POSTCODE_DISTRICT` and `POPULATED_PLACE`.
+ *
+ * Each postcode is joined to the nearest named road link within 100 m, measured point-to-segment in
+ * EPSG:27700 meters. The street is the Open Names `NAME1` of that link's TOID, the locality is its
+ * `POPULATED_PLACE`. A row carries no region, because the GB layout prints none. Every
+ * component is OS data except `house_number`, which no open GB product publishes and which this
+ * adapter **draws** from local bands seeded by the postcode and TOID, so each row carries a number the
+ * source never stated.
+ *
+ * A postcode is dropped when:
+ *
+ * - its Code-Point coordinate has no position, `Positional_quality_indicator` 90;
+ * - its coordinate is shared with another postcode, so the point does not locate this postcode's street;
+ * - no named link lies within 100 m;
+ * - the joined road's `POSTCODE_DISTRICT` differs from the postcode's own outward code.
+ *
+ * `inputPath` is a directory holding the three products unzipped, each in its own subdirectory:
+ * `code-point-open/` (with `Data/CSV/*.csv`), `open-names/` (with `Data/*.csv`) and `open-roads/`
+ * (with the `.gml` tiles anywhere beneath it). The Code-Point rows are held in memory, the road
+ * tiles stream past them once, and Open Names is read once for the roadNameTOID values the join chose.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { titleCaseGB } from "@mailwoman/codex/gb/place-name";
+import { outwardCode } from "@mailwoman/codex/gb/postcode";
+import { pathExists } from "@mailwoman/core/fs/readers/stat";
+import { stringifyJSON } from "@mailwoman/core/json";
+import { mulberry32 } from "@mailwoman/core/random";
+import { createCodePointParseStats, PQI_NO_COORDINATE, readCodePointCSV } from "@mailwoman/spatial/gb/codepoint";
+import { PathBuilder } from "path-ts";
+import { Globerator } from "spliterator/node/fs";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { stableSourceIDFromParts } from "#adapters/source-id";
+import { NearestLinkJoin } from "#gb/os-open/nearest-link";
+import { createOpenNamesReadStats, populatedPlaceName1, readOpenNamesRoads, } from "#gb/os-open/open/names";
+import { createOpenRoadsReadStats, readOpenRoadsLinks } from "#gb/os-open/open/roads";
+import { SourceRegister } from "#registers";
+import { tieredNumber } from "#synthesizers/utils";
+import { AddressRole, SurfaceOrigin } 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 const GB_OS_OPEN_ADAPTER_ID = "gb-os-open";
+/**
+ * The license on all three products, and on every row.
+ */
+export const GB_OS_OPEN_LICENSE = "OGL-UK-3.0";
+/**
+ * The jurisdictions the adapter emits.
+ *
+ * Code-Point Open covers Great Britain and carries no `BT` postcode.
+ */
+export const GB_OS_OPEN_COUNTRIES = ["GB"];
+/**
+ * The farthest a road link may lie from a postcode's point and still name its street, in EPSG:27700 meters.
+ */
+export const GB_OS_OPEN_JOIN_RADIUS_M = 100;
+/**
+ * The subdirectory of `inputPath` that holds each product.
+ */
+export const GB_OS_OPEN_PRODUCT_DIRECTORIES = {
+    codePointOpen: "code-point-open",
+    openNames: "open-names",
+    openRoads: "open-roads",
+};
+/**
+ * The bands the drawn house number comes from.
+ *
+ * Seven in ten draws fall in 1–60, a quarter in 61–200, and the rest in 201–400.
+ */
+export const GB_OS_OPEN_HOUSE_NUMBER_BANDS = [
+    { cutoff: 0.7, base: 1, span: 60 },
+    { cutoff: 0.95, base: 61, span: 140 },
+    { base: 201, span: 200 },
+];
+/**
+ * The three product directories under `inputPath`.
+ *
+ * @throws Naming the missing path, when a product's subdirectory is absent.
+ */
+export async function gbOSOpenProductPaths(inputPath) {
+    const root = PathBuilder.from(inputPath);
+    const paths = {
+        codePointOpen: root(GB_OS_OPEN_PRODUCT_DIRECTORIES.codePointOpen),
+        openNames: root(GB_OS_OPEN_PRODUCT_DIRECTORIES.openNames),
+        openRoads: root(GB_OS_OPEN_PRODUCT_DIRECTORIES.openRoads),
+    };
+    for (const path of Object.values(paths)) {
+        if (!(await pathExists(path))) {
+            throw new Error(`${GB_OS_OPEN_ADAPTER_ID} adapter: ${path.toString()} is absent; inputPath must hold ${Object.values(GB_OS_OPEN_PRODUCT_DIRECTORIES).join("/, ")}/`);
+        }
+    }
+    return paths;
+}
+/**
+ * The files matching `pattern` under `directory`, sorted so a run reads them in one order.
+ *
+ * @throws When none match, so a wrong layout reports itself rather than reading as an empty product.
+ */
+async function productFiles(directory, pattern) {
+    const names = await Globerator.from(pattern, { cwd: directory.toString(), absolute: false }).toSorted();
+    if (!names.length) {
+        throw new Error(`${GB_OS_OPEN_ADAPTER_ID} adapter: ${directory.toString()} holds no file matching ${pattern}`);
+    }
+    return names.map((name) => directory(name).toString());
+}
+/**
+ * The house number drawn for one postcode and road, from {@linkcode GB_OS_OPEN_HOUSE_NUMBER_BANDS}.
+ *
+ * The seed is the row's source id, so a rerun draws the same number.
+ */
+export function drawGBHouseNumber(sourceID) {
+    const seed = Number.parseInt(sourceID.slice(-8), 16);
+    return tieredNumber(mulberry32(seed), GB_OS_OPEN_HOUSE_NUMBER_BANDS);
+}
+/**
+ * The street name as the row prints it.
+ *
+ * OS writes nearly every road name in mixed case.
+ * A name with no lowercase letter is title-cased.
+ */
+function streetName(name1) {
+    if (/\p{Ll}/u.test(name1))
+        return { street: name1, titleCased: false };
+    return { street: titleCaseGB(name1), titleCased: true };
+}
+/**
+ * Creates the OS OpenData adapter.
+ */
+export function createGBOSOpenAdapter(options = {}) {
+    return {
+        id: GB_OS_OPEN_ADAPTER_ID,
+        defaultLicense: GB_OS_OPEN_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.OrdnanceSurveyOpenData,
+        surface: SurfaceOrigin.Rendered,
+        description: "OS OpenData (Code-Point Open, OS Open Roads, OS Open Names): GB postcodes joined to their nearest named road, with a drawn house number.",
+        async *rows(opts) {
+            if (opts.country && !GB_OS_OPEN_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(GB_OS_OPEN_ADAPTER_ID, GB_OS_OPEN_COUNTRIES, opts.country);
+            }
+            const paths = await gbOSOpenProductPaths(opts.inputPath);
+            const codePointFiles = await productFiles(paths.codePointOpen, "Data/CSV/*.csv");
+            const roadFiles = await productFiles(paths.openRoads, "**/*.gml");
+            const nameFiles = await productFiles(paths.openNames, "Data/*.csv");
+            const summary = {
+                codePointRows: 0,
+                codePointMalformed: 0,
+                pqi: {},
+                postcodesWithCoordinate: 0,
+                joined: 0,
+                dropped: {
+                    noCoordinate: 0,
+                    sharedCoordinate: 0,
+                    noLinkWithinRadius: 0,
+                    noOpenNamesRow: 0,
+                    districtMismatch: 0,
+                },
+                emitted: 0,
+                titleCasedStreets: 0,
+                noLocality: 0,
+                bilingualLocality: 0,
+                roads: createOpenRoadsReadStats(),
+                names: createOpenNamesReadStats(),
+            };
+            // Code-Point: every postcode with a coordinate, and how many postcodes share each coordinate.
+            const codePointStats = createCodePointParseStats();
+            const postcodes = [];
+            const xs = [];
+            const ys = [];
+            const pointCounts = new Map();
+            for (const file of codePointFiles) {
+                for await (const record of readCodePointCSV(file, codePointStats)) {
+                    postcodes.push(record.postcode);
+                    xs.push(record.easting);
+                    ys.push(record.northing);
+                    const key = `${record.easting},${record.northing}`;
+                    pointCounts.set(key, (pointCounts.get(key) ?? 0) + 1);
+                    summary.pqi[record.quality] = (summary.pqi[record.quality] ?? 0) + 1;
+                }
+            }
+            summary.codePointRows = codePointStats.read;
+            summary.codePointMalformed = codePointStats.skippedMalformed;
+            summary.dropped.noCoordinate = codePointStats.skippedNoCoordinate;
+            summary.postcodesWithCoordinate = postcodes.length;
+            if (codePointStats.skippedNoCoordinate > 0) {
+                summary.pqi[PQI_NO_COORDINATE] = codePointStats.skippedNoCoordinate;
+            }
+            // Open Roads: stream every named link past the postcodes once.
+            const join = new NearestLinkJoin({ xs: Float64Array.from(xs), ys: Float64Array.from(ys) }, GB_OS_OPEN_JOIN_RADIUS_M);
+            for (const file of roadFiles) {
+                for await (const link of readOpenRoadsLinks(file, summary.roads, { signal: opts.signal })) {
+                    join.offer({ roadNameTOID: link.roadNameTOID }, link.vertices);
+                }
+            }
+            const wanted = new Set();
+            for (let i = 0; i < postcodes.length; i++) {
+                const match = join.match(i);
+                if (!match)
+                    continue;
+                summary.joined++;
+                wanted.add(match.link.roadNameTOID);
+            }
+            // Open Names: only the roads the join chose.
+            const roads = new Map();
+            for (const file of nameFiles) {
+                for await (const road of readOpenNamesRoads(file, summary.names, wanted)) {
+                    roads.set(road.id, road);
+                }
+            }
+            try {
+                for (let i = 0; i < postcodes.length; i++) {
+                    if (opts.signal?.aborted)
+                        break;
+                    if (opts.limit !== undefined && summary.emitted >= opts.limit)
+                        break;
+                    const postcode = postcodes[i];
+                    if ((pointCounts.get(`${xs[i]},${ys[i]}`) ?? 0) > 1) {
+                        summary.dropped.sharedCoordinate++;
+                        continue;
+                    }
+                    const match = join.match(i);
+                    if (!match) {
+                        summary.dropped.noLinkWithinRadius++;
+                        continue;
+                    }
+                    const road = roads.get(match.link.roadNameTOID);
+                    if (!road) {
+                        summary.dropped.noOpenNamesRow++;
+                        continue;
+                    }
+                    if (road.postcodeDistrict !== outwardCode(postcode)) {
+                        summary.dropped.districtMismatch++;
+                        continue;
+                    }
+                    const { street, titleCased } = streetName(road.name1);
+                    const locality = populatedPlaceName1(road.populatedPlace);
+                    const sourceID = stableSourceIDFromParts(GB_OS_OPEN_ADAPTER_ID, {
+                        postcode,
+                        roadNameTOID: road.id,
+                    });
+                    const components = {
+                        house_number: drawGBHouseNumber(sourceID),
+                        street,
+                    };
+                    if (locality) {
+                        components.locality = locality;
+                    }
+                    components.postcode = postcode;
+                    const rendered = formatAddressRow(components, "GB", { singleLine: true });
+                    if (!rendered)
+                        continue;
+                    if (titleCased) {
+                        summary.titleCasedStreets++;
+                    }
+                    if (!locality) {
+                        summary.noLocality++;
+                    }
+                    if (road.populatedPlace.includes(" / ")) {
+                        summary.bilingualLocality++;
+                    }
+                    yield {
+                        raw: rendered.raw,
+                        components: rendered.components,
+                        country: "GB",
+                        locale: "en-GB",
+                        source: GB_OS_OPEN_ADAPTER_ID,
+                        source_id: sourceID,
+                        corpus_version: "",
+                        license: GB_OS_OPEN_LICENSE,
+                    };
+                    summary.emitted++;
+                }
+            }
+            finally {
+                options.onSummary?.(summary);
+                process.stderr.write(`  ${GB_OS_OPEN_ADAPTER_ID}: ${summary.emitted} emitted of ${summary.codePointRows} Code-Point rows; ` +
+                    `dropped ${stringifyJSON(summary.dropped)}\n`);
+            }
+        },
+    };
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const gbOSOpenAdapter = createGBOSOpenAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/gb/adapters/os-open/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/gb/adapters/os-open/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8BG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAClE,OAAO,EAAE,WAAW,EAAE,MAAM,gCAAgC,CAAA;AAC5D,OAAO,EAAE,WAAW,EAAE,MAAM,8BAA8B,CAAA;AAC1D,OAAO,EAAE,UAAU,EAAE,MAAM,iCAAiC,CAAA;AAC5D,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAA;AACpD,OAAO,EAAE,UAAU,EAAE,MAAM,wBAAwB,CAAA;AACnD,OAAO,EAAE,yBAAyB,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAChH,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,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,eAAe,EAAE,MAAM,0BAA0B,CAAA;AAC1D,OAAO,EACN,wBAAwB,EAExB,mBAAmB,EACnB,kBAAkB,GAClB,MAAM,wBAAwB,CAAA;AAC/B,OAAO,EAAE,wBAAwB,EAAE,kBAAkB,EAAE,MAAM,wBAAwB,CAAA;AACrF,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EAAE,YAAY,EAAyB,MAAM,qBAAqB,CAAA;AACzE,OAAO,EAAE,WAAW,EAA8D,aAAa,EAAE,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,YAAY,CAAA;AAEjD;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,YAAY,CAAA;AAE9C;;;;GAIG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAsB,CAAC,IAAI,CAAC,CAAA;AAE7D;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,GAAG,CAAA;AAE3C;;GAEG;AACH,MAAM,CAAC,MAAM,8BAA8B,GAAG;IAC7C,aAAa,EAAE,iBAAiB;IAChC,SAAS,EAAE,YAAY;IACvB,SAAS,EAAE,YAAY;CACd,CAAA;AAEV;;;;GAIG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAgC;IACzE,EAAE,MAAM,EAAE,GAAG,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE;IAClC,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,EAAE,IAAI,EAAE,GAAG,EAAE;IACrC,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE;CACxB,CAAA;AA8ED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,oBAAoB,CACzC,SAA0B;IAE1B,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAExC,MAAM,KAAK,GAAG;QACb,aAAa,EAAE,IAAI,CAAC,8BAA8B,CAAC,aAAa,CAAC;QACjE,SAAS,EAAE,IAAI,CAAC,8BAA8B,CAAC,SAAS,CAAC;QACzD,SAAS,EAAE,IAAI,CAAC,8BAA8B,CAAC,SAAS,CAAC;KACzD,CAAA;IAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;QACzC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;YAC/B,MAAM,IAAI,KAAK,CACd,GAAG,qBAAqB,aAAa,IAAI,CAAC,QAAQ,EAAE,mCAAmC,MAAM,CAAC,MAAM,CAAC,8BAA8B,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CACnJ,CAAA;QACF,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAA;AACb,CAAC;AAED;;;;GAIG;AACH,KAAK,UAAU,YAAY,CAAC,SAAsB,EAAE,OAAe;IAClE,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,OAAO,EAAE,EAAE,GAAG,EAAE,SAAS,CAAC,QAAQ,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAEvG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QACnB,MAAM,IAAI,KAAK,CAAC,GAAG,qBAAqB,aAAa,SAAS,CAAC,QAAQ,EAAE,2BAA2B,OAAO,EAAE,CAAC,CAAA;IAC/G,CAAC;IAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAA;AACvD,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,QAAgB;IACjD,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAA;IAEpD,OAAO,YAAY,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,6BAA6B,CAAC,CAAA;AACrE,CAAC;AAED;;;;;GAKG;AACH,SAAS,UAAU,CAAC,KAAa;IAChC,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,UAAU,EAAE,KAAK,EAAE,CAAA;IAEtE,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,KAAK,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE,CAAA;AACxD,CAAC;AASD;;GAEG;AACH,MAAM,UAAU,qBAAqB,CAAC,OAAO,GAA2B,EAAE;IACzE,OAAO;QACN,EAAE,EAAE,qBAAqB;QACzB,cAAc,EAAE,kBAAkB;QAClC,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,sBAAsB;QAC/C,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,0IAA0I;QAE3I,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,oBAAoB,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAClE,MAAM,IAAI,uBAAuB,CAAC,qBAAqB,EAAE,oBAAoB,EAAE,IAAI,CAAC,OAAO,CAAC,CAAA;YAC7F,CAAC;YAED,MAAM,KAAK,GAAG,MAAM,oBAAoB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YACxD,MAAM,cAAc,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,aAAa,EAAE,gBAAgB,CAAC,CAAA;YAChF,MAAM,SAAS,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,SAAS,EAAE,UAAU,CAAC,CAAA;YACjE,MAAM,SAAS,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,SAAS,EAAE,YAAY,CAAC,CAAA;YAEnE,MAAM,OAAO,GAAoB;gBAChC,aAAa,EAAE,CAAC;gBAChB,kBAAkB,EAAE,CAAC;gBACrB,GAAG,EAAE,EAAE;gBACP,uBAAuB,EAAE,CAAC;gBAC1B,MAAM,EAAE,CAAC;gBACT,OAAO,EAAE;oBACR,YAAY,EAAE,CAAC;oBACf,gBAAgB,EAAE,CAAC;oBACnB,kBAAkB,EAAE,CAAC;oBACrB,cAAc,EAAE,CAAC;oBACjB,gBAAgB,EAAE,CAAC;iBACnB;gBACD,OAAO,EAAE,CAAC;gBACV,iBAAiB,EAAE,CAAC;gBACpB,UAAU,EAAE,CAAC;gBACb,iBAAiB,EAAE,CAAC;gBACpB,KAAK,EAAE,wBAAwB,EAAE;gBACjC,KAAK,EAAE,wBAAwB,EAAE;aACjC,CAAA;YAED,8FAA8F;YAC9F,MAAM,cAAc,GAAG,yBAAyB,EAAE,CAAA;YAClD,MAAM,SAAS,GAAa,EAAE,CAAA;YAC9B,MAAM,EAAE,GAAa,EAAE,CAAA;YACvB,MAAM,EAAE,GAAa,EAAE,CAAA;YACvB,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB,CAAA;YAE7C,KAAK,MAAM,IAAI,IAAI,cAAc,EAAE,CAAC;gBACnC,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,gBAAgB,CAAC,IAAI,EAAE,cAAc,CAAC,EAAE,CAAC;oBACnE,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAA;oBAC/B,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;oBACvB,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAA;oBAExB,MAAM,GAAG,GAAG,GAAG,MAAM,CAAC,OAAO,IAAI,MAAM,CAAC,QAAQ,EAAE,CAAA;oBAElD,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;oBACrD,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBACrE,CAAC;YACF,CAAC;YAED,OAAO,CAAC,aAAa,GAAG,cAAc,CAAC,IAAI,CAAA;YAC3C,OAAO,CAAC,kBAAkB,GAAG,cAAc,CAAC,gBAAgB,CAAA;YAC5D,OAAO,CAAC,OAAO,CAAC,YAAY,GAAG,cAAc,CAAC,mBAAmB,CAAA;YACjE,OAAO,CAAC,uBAAuB,GAAG,SAAS,CAAC,MAAM,CAAA;YAElD,IAAI,cAAc,CAAC,mBAAmB,GAAG,CAAC,EAAE,CAAC;gBAC5C,OAAO,CAAC,GAAG,CAAC,iBAAiB,CAAC,GAAG,cAAc,CAAC,mBAAmB,CAAA;YACpE,CAAC;YAED,+DAA+D;YAC/D,MAAM,IAAI,GAAG,IAAI,eAAe,CAC/B,EAAE,EAAE,EAAE,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EACxD,wBAAwB,CACxB,CAAA;YAED,KAAK,MAAM,IAAI,IAAI,SAAS,EAAE,CAAC;gBAC9B,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,kBAAkB,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC;oBAC3F,IAAI,CAAC,KAAK,CAAC,EAAE,YAAY,EAAE,IAAI,CAAC,YAAY,EAAE,EAAE,IAAI,CAAC,QAAQ,CAAC,CAAA;gBAC/D,CAAC;YACF,CAAC;YAED,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAA;YAEhC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC3C,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;gBAE3B,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,OAAO,CAAC,MAAM,EAAE,CAAA;gBAChB,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;YACpC,CAAC;YAED,6CAA6C;YAC7C,MAAM,KAAK,GAAG,IAAI,GAAG,EAAyB,CAAA;YAE9C,KAAK,MAAM,IAAI,IAAI,SAAS,EAAE,CAAC;gBAC9B,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,kBAAkB,CAAC,IAAI,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,CAAC,EAAE,CAAC;oBAC1E,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,CAAA;gBACzB,CAAC;YACF,CAAC;YAED,IAAI,CAAC;gBACJ,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;oBAC3C,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,CAAC,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAEpE,MAAM,QAAQ,GAAG,SAAS,CAAC,CAAC,CAAE,CAAA;oBAE9B,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC;wBACrD,OAAO,CAAC,OAAO,CAAC,gBAAgB,EAAE,CAAA;wBAElC,SAAQ;oBACT,CAAC;oBAED,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;oBAE3B,IAAI,CAAC,KAAK,EAAE,CAAC;wBACZ,OAAO,CAAC,OAAO,CAAC,kBAAkB,EAAE,CAAA;wBAEpC,SAAQ;oBACT,CAAC;oBAED,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,YAAY,CAAC,CAAA;oBAE/C,IAAI,CAAC,IAAI,EAAE,CAAC;wBACX,OAAO,CAAC,OAAO,CAAC,cAAc,EAAE,CAAA;wBAEhC,SAAQ;oBACT,CAAC;oBAED,IAAI,IAAI,CAAC,gBAAgB,KAAK,WAAW,CAAC,QAAQ,CAAC,EAAE,CAAC;wBACrD,OAAO,CAAC,OAAO,CAAC,gBAAgB,EAAE,CAAA;wBAElC,SAAQ;oBACT,CAAC;oBAED,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,GAAG,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;oBACrD,MAAM,QAAQ,GAAG,mBAAmB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;oBAEzD,MAAM,QAAQ,GAAG,uBAAuB,CAAC,qBAAqB,EAAE;wBAC/D,QAAQ;wBACR,YAAY,EAAE,IAAI,CAAC,EAAE;qBACrB,CAAC,CAAA;oBAEF,MAAM,UAAU,GAA+B;wBAC9C,YAAY,EAAE,iBAAiB,CAAC,QAAQ,CAAC;wBACzC,MAAM;qBACN,CAAA;oBAED,IAAI,QAAQ,EAAE,CAAC;wBACd,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;oBAC/B,CAAC;oBAED,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;oBAE9B,MAAM,QAAQ,GAAG,gBAAgB,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAA;oBAEzE,IAAI,CAAC,QAAQ;wBAAE,SAAQ;oBAEvB,IAAI,UAAU,EAAE,CAAC;wBAChB,OAAO,CAAC,iBAAiB,EAAE,CAAA;oBAC5B,CAAC;oBAED,IAAI,CAAC,QAAQ,EAAE,CAAC;wBACf,OAAO,CAAC,UAAU,EAAE,CAAA;oBACrB,CAAC;oBAED,IAAI,IAAI,CAAC,cAAc,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;wBACzC,OAAO,CAAC,iBAAiB,EAAE,CAAA;oBAC5B,CAAC;oBAED,MAAM;wBACL,GAAG,EAAE,QAAQ,CAAC,GAAG;wBACjB,UAAU,EAAE,QAAQ,CAAC,UAAU;wBAC/B,OAAO,EAAE,IAAI;wBACb,MAAM,EAAE,OAAO;wBACf,MAAM,EAAE,qBAAqB;wBAC7B,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,kBAAkB;qBAC3B,CAAA;oBAED,OAAO,CAAC,OAAO,EAAE,CAAA;gBAClB,CAAC;YACF,CAAC;oBAAS,CAAC;gBACV,OAAO,CAAC,SAAS,EAAE,CAAC,OAAO,CAAC,CAAA;gBAE5B,OAAO,CAAC,MAAM,CAAC,KAAK,CACnB,KAAK,qBAAqB,KAAK,OAAO,CAAC,OAAO,eAAe,OAAO,CAAC,aAAa,oBAAoB;oBACrG,WAAW,aAAa,CAAC,OAAO,CAAC,OAAO,CAAC,IAAI,CAC9C,CAAA;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,qBAAqB,EAAE,CAAA"}

package/out/gb/os-open/nearest-link.d.ts

@@ -0,0 +1,73 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The nearest line feature to each of a set of points, measured point-to-segment in a planar
+ *   projection.
+ *
+ *   The points sit in a bucket grid and the lines stream past it. Each segment of each line asks the
+ *   grid for the points inside its bounding box widened by the search radius, and each of those
+ *   points keeps the nearest line it has been offered. Memory is bounded by the point count rather
+ *   than by the line count, so a whole country's road network can be read once from disk.
+ *
+ *   Distances are Euclidean in the projection's own units. For EPSG:27700, the British National
+ *   Grid, those are meters, and the grid's scale error stays under 0.04% across Great Britain.
+ */
+/**
+ * A uniform grid of point indices, keyed by cell.
+ */
+export declare class PointBucketIndex {
+    #private;
+    readonly cellSize: number;
+    constructor(cellSize: number);
+    /**
+     * Files the point `index` under the cell holding `(x, y)`.
+     */
+    add(index: number, x: number, y: number): void;
+    /**
+     * Every point index filed in a cell that the box touches.
+     *
+     * The answer holds every point of each touched cell, including points up to one cell outside the box.
+     * A caller measures each one.
+     */
+    within(minX: number, minY: number, maxX: number, maxY: number): Generator<number>;
+}
+/**
+ * The line a point was joined to, and how far away it lies.
+ */
+export interface NearestLinkMatch<T> {
+    link: T;
+    /**
+     * The point-to-segment distance in projection units.
+     */
+    distance: number;
+}
+/**
+ * Joins each point to the nearest line offered to it within `radius`.
+ *
+ * A line offered at exactly the same distance as the current nearest does not replace it,
+ * so the first-offered line wins a tie and the answer follows the order the lines are read in.
+ */
+export declare class NearestLinkJoin<T> {
+    #private;
+    readonly radius: number;
+    /**
+     * @param points The points to join, as parallel coordinate arrays.
+     * A point's position in the arrays is its index in every answer.
+     * @param radius The largest distance at which a line joins, in projection units.
+     */
+    constructor(points: {
+        xs: Float64Array;
+        ys: Float64Array;
+    }, radius: number);
+    /**
+     * Offers one line, as a flat `[x0, y0, x1, y1, …]` vertex array, to every point within reach of it.
+     */
+    offer(link: T, vertices: ArrayLike<number>): void;
+    /**
+     * The nearest line offered to point `index`, or null when none came within the radius.
+     */
+    match(index: number): NearestLinkMatch<T> | null;
+}
+//# sourceMappingURL=nearest-link.d.ts.map

package/out/gb/os-open/nearest-link.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"nearest-link.d.ts","sourceRoot":"","sources":["../../../lib/gb/os-open/nearest-link.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAYH;;GAEG;AACH,qBAAa,gBAAgB;;IAC5B,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAA;IAIzB,YAAY,QAAQ,EAAE,MAAM,EAI3B;IAMD;;OAEG;IACH,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,GAAG,IAAI,CAS7C;IAED;;;;;OAKG;IACF,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,SAAS,CAAC,MAAM,CAAC,CAejF;CACD;AAED;;GAEG;AACH,MAAM,WAAW,gBAAgB,CAAC,CAAC;IAClC,IAAI,EAAE,CAAC,CAAA;IACP;;OAEG;IACH,QAAQ,EAAE,MAAM,CAAA;CAChB;AAED;;;;;GAKG;AACH,qBAAa,eAAe,CAAC,CAAC;;IAC7B,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAA;IAQvB;;;;OAIG;IACH,YAAY,MAAM,EAAE;QAAE,EAAE,EAAE,YAAY,CAAC;QAAC,EAAE,EAAE,YAAY,CAAA;KAAE,EAAE,MAAM,EAAE,MAAM,EAezE;IAED;;OAEG;IACH,KAAK,CAAC,IAAI,EAAE,CAAC,EAAE,QAAQ,EAAE,SAAS,CAAC,MAAM,CAAC,GAAG,IAAI,CAwBhD;IAED;;OAEG;IACH,KAAK,CAAC,KAAK,EAAE,MAAM,GAAG,gBAAgB,CAAC,CAAC,CAAC,GAAG,IAAI,CAI/C;CACD"}

package/out/gb/os-open/nearest-link.js

@@ -0,0 +1,133 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The nearest line feature to each of a set of points, measured point-to-segment in a planar
+ *   projection.
+ *
+ *   The points sit in a bucket grid and the lines stream past it. Each segment of each line asks the
+ *   grid for the points inside its bounding box widened by the search radius, and each of those
+ *   points keeps the nearest line it has been offered. Memory is bounded by the point count rather
+ *   than by the line count, so a whole country's road network can be read once from disk.
+ *
+ *   Distances are Euclidean in the projection's own units. For EPSG:27700, the British National
+ *   Grid, those are meters, and the grid's scale error stays under 0.04% across Great Britain.
+ */
+import { planarSegmentDistanceSquared } from "@mailwoman/spatial/distance";
+/**
+ * The multiplier that packs a cell's column and row into one numeric key.
+ *
+ * A row index must stay below it.
+ * With 100 m cells, 2^21 rows span 209,715 km, so a projected coordinate in meters cannot reach it.
+ */
+const CELL_ROW_SPAN = 2 ** 21;
+/**
+ * A uniform grid of point indices, keyed by cell.
+ */
+export class PointBucketIndex {
+    cellSize;
+    #cells = new Map();
+    constructor(cellSize) {
+        if (!(cellSize > 0))
+            throw new RangeError(`PointBucketIndex: cell size must be positive, got ${cellSize}`);
+        this.cellSize = cellSize;
+    }
+    #key(column, row) {
+        return column * CELL_ROW_SPAN + row;
+    }
+    /**
+     * Files the point `index` under the cell holding `(x, y)`.
+     */
+    add(index, x, y) {
+        const key = this.#key(Math.floor(x / this.cellSize), Math.floor(y / this.cellSize));
+        const bucket = this.#cells.get(key);
+        if (bucket) {
+            bucket.push(index);
+        }
+        else {
+            this.#cells.set(key, [index]);
+        }
+    }
+    /**
+     * Every point index filed in a cell that the box touches.
+     *
+     * The answer holds every point of each touched cell, including points up to one cell outside the box.
+     * A caller measures each one.
+     */
+    *within(minX, minY, maxX, maxY) {
+        const firstColumn = Math.floor(minX / this.cellSize);
+        const lastColumn = Math.floor(maxX / this.cellSize);
+        const firstRow = Math.floor(minY / this.cellSize);
+        const lastRow = Math.floor(maxY / this.cellSize);
+        for (let column = firstColumn; column <= lastColumn; column++) {
+            for (let row = firstRow; row <= lastRow; row++) {
+                const bucket = this.#cells.get(this.#key(column, row));
+                if (bucket) {
+                    yield* bucket;
+                }
+            }
+        }
+    }
+}
+/**
+ * Joins each point to the nearest line offered to it within `radius`.
+ *
+ * A line offered at exactly the same distance as the current nearest does not replace it,
+ * so the first-offered line wins a tie and the answer follows the order the lines are read in.
+ */
+export class NearestLinkJoin {
+    radius;
+    #xs;
+    #ys;
+    #bestSquared;
+    #best;
+    #index;
+    /**
+     * @param points The points to join, as parallel coordinate arrays.
+     * A point's position in the arrays is its index in every answer.
+     * @param radius The largest distance at which a line joins, in projection units.
+     */
+    constructor(points, radius) {
+        if (points.xs.length !== points.ys.length) {
+            throw new RangeError(`NearestLinkJoin: ${points.xs.length} x values against ${points.ys.length} y values`);
+        }
+        this.radius = radius;
+        this.#xs = points.xs;
+        this.#ys = points.ys;
+        this.#bestSquared = new Float64Array(points.xs.length).fill(Number.POSITIVE_INFINITY);
+        this.#best = Array.from({ length: points.xs.length });
+        this.#index = new PointBucketIndex(radius);
+        for (let i = 0; i < points.xs.length; i++) {
+            this.#index.add(i, points.xs[i], points.ys[i]);
+        }
+    }
+    /**
+     * Offers one line, as a flat `[x0, y0, x1, y1, …]` vertex array, to every point within reach of it.
+     */
+    offer(link, vertices) {
+        const radius = this.radius;
+        const radiusSquared = radius * radius;
+        for (let v = 0; v + 3 < vertices.length; v += 2) {
+            const ax = vertices[v];
+            const ay = vertices[v + 1];
+            const bx = vertices[v + 2];
+            const by = vertices[v + 3];
+            for (const i of this.#index.within(Math.min(ax, bx) - radius, Math.min(ay, by) - radius, Math.max(ax, bx) + radius, Math.max(ay, by) + radius)) {
+                const squared = planarSegmentDistanceSquared(this.#xs[i], this.#ys[i], ax, ay, bx, by);
+                if (squared <= radiusSquared && squared < this.#bestSquared[i]) {
+                    this.#bestSquared[i] = squared;
+                    this.#best[i] = link;
+                }
+            }
+        }
+    }
+    /**
+     * The nearest line offered to point `index`, or null when none came within the radius.
+     */
+    match(index) {
+        const link = this.#best[index];
+        return link === undefined ? null : { link, distance: Math.sqrt(this.#bestSquared[index]) };
+    }
+}
+//# sourceMappingURL=nearest-link.js.map

package/out/gb/os-open/nearest-link.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"nearest-link.js","sourceRoot":"","sources":["../../../lib/gb/os-open/nearest-link.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,OAAO,EAAE,4BAA4B,EAAE,MAAM,6BAA6B,CAAA;AAE1E;;;;;GAKG;AACH,MAAM,aAAa,GAAG,CAAC,IAAI,EAAE,CAAA;AAE7B;;GAEG;AACH,MAAM,OAAO,gBAAgB;IACnB,QAAQ,CAAQ;IAEhB,MAAM,GAAG,IAAI,GAAG,EAAoB,CAAA;IAE7C,YAAY,QAAgB;QAC3B,IAAI,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC;YAAE,MAAM,IAAI,UAAU,CAAC,qDAAqD,QAAQ,EAAE,CAAC,CAAA;QAE1G,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAA;IACzB,CAAC;IAED,IAAI,CAAC,MAAc,EAAE,GAAW;QAC/B,OAAO,MAAM,GAAG,aAAa,GAAG,GAAG,CAAA;IACpC,CAAC;IAED;;OAEG;IACH,GAAG,CAAC,KAAa,EAAE,CAAS,EAAE,CAAS;QACtC,MAAM,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAA;QACnF,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAEnC,IAAI,MAAM,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACnB,CAAC;aAAM,CAAC;YACP,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED;;;;;OAKG;IACH,CAAC,MAAM,CAAC,IAAY,EAAE,IAAY,EAAE,IAAY,EAAE,IAAY;QAC7D,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAA;QACpD,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAA;QACnD,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAA;QACjD,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAA;QAEhD,KAAK,IAAI,MAAM,GAAG,WAAW,EAAE,MAAM,IAAI,UAAU,EAAE,MAAM,EAAE,EAAE,CAAC;YAC/D,KAAK,IAAI,GAAG,GAAG,QAAQ,EAAE,GAAG,IAAI,OAAO,EAAE,GAAG,EAAE,EAAE,CAAC;gBAChD,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,CAAA;gBAEtD,IAAI,MAAM,EAAE,CAAC;oBACZ,KAAK,CAAC,CAAC,MAAM,CAAA;gBACd,CAAC;YACF,CAAC;QACF,CAAC;IACF,CAAC;CACD;AAaD;;;;;GAKG;AACH,MAAM,OAAO,eAAe;IAClB,MAAM,CAAQ;IAEd,GAAG,CAAc;IACjB,GAAG,CAAc;IACjB,YAAY,CAAc;IAC1B,KAAK,CAAmB;IACxB,MAAM,CAAkB;IAEjC;;;;OAIG;IACH,YAAY,MAA8C,EAAE,MAAc;QACzE,IAAI,MAAM,CAAC,EAAE,CAAC,MAAM,KAAK,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC;YAC3C,MAAM,IAAI,UAAU,CAAC,oBAAoB,MAAM,CAAC,EAAE,CAAC,MAAM,qBAAqB,MAAM,CAAC,EAAE,CAAC,MAAM,WAAW,CAAC,CAAA;QAC3G,CAAC;QAED,IAAI,CAAC,MAAM,GAAG,MAAM,CAAA;QACpB,IAAI,CAAC,GAAG,GAAG,MAAM,CAAC,EAAE,CAAA;QACpB,IAAI,CAAC,GAAG,GAAG,MAAM,CAAC,EAAE,CAAA;QACpB,IAAI,CAAC,YAAY,GAAG,IAAI,YAAY,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,iBAAiB,CAAC,CAAA;QACrF,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,CAAA;QACrD,IAAI,CAAC,MAAM,GAAG,IAAI,gBAAgB,CAAC,MAAM,CAAC,CAAA;QAE1C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC3C,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC,CAAE,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC,CAAE,CAAC,CAAA;QACjD,CAAC;IACF,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,IAAO,EAAE,QAA2B;QACzC,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAA;QAC1B,MAAM,aAAa,GAAG,MAAM,GAAG,MAAM,CAAA;QAErC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC;YACjD,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,CAAE,CAAA;YACvB,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;YAC3B,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;YAC3B,MAAM,EAAE,GAAG,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;YAE3B,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC,MAAM,CACjC,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,MAAM,EACzB,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,MAAM,EACzB,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,MAAM,EACzB,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,EAAE,CAAC,GAAG,MAAM,CACzB,EAAE,CAAC;gBACH,MAAM,OAAO,GAAG,4BAA4B,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAE,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,CAAC,CAAA;gBAExF,IAAI,OAAO,IAAI,aAAa,IAAI,OAAO,GAAG,IAAI,CAAC,YAAY,CAAC,CAAC,CAAE,EAAE,CAAC;oBACjE,IAAI,CAAC,YAAY,CAAC,CAAC,CAAC,GAAG,OAAO,CAAA;oBAC9B,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,IAAI,CAAA;gBACrB,CAAC;YACF,CAAC;QACF,CAAC;IACF,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,KAAa;QAClB,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;QAE9B,OAAO,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,CAAC,KAAK,CAAE,CAAC,EAAE,CAAA;IAC5F,CAAC;CACD"}

package/out/gb/os-open/open/names.d.ts

@@ -0,0 +1,92 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Reads the road rows of OS Open Names.
+ *
+ *   The CSV product is 819 files, one per 10 km National Grid square, with no header row. OS ships
+ *   the 34 column names in `Doc/OS_Open_Names_Header.csv`, and the first data row of each file opens
+ *   with a byte-order mark, which this reader strips from the `ID` value.
+ *
+ *   A road row's `ID` is the roadNameTOID that OS Open Roads links reference, such as
+ *   `osgb4000000010927776`. Two `LOCAL_TYPE` values are roads: `Named Road`, and
+ *   `Section Of Named Road` for a road that OS splits where it crosses a postcode district or a
+ *   settlement.
+ */
+/**
+ * The 34 columns of an OS Open Names CSV, in file order, as `Doc/OS_Open_Names_Header.csv` names them.
+ */
+export declare const OPEN_NAMES_COLUMNS: readonly ["ID", "NAMES_URI", "NAME1", "NAME1_LANG", "NAME2", "NAME2_LANG", "TYPE", "LOCAL_TYPE", "GEOMETRY_X", "GEOMETRY_Y", "MOST_DETAIL_VIEW_RES", "LEAST_DETAIL_VIEW_RES", "MBR_XMIN", "MBR_YMIN", "MBR_XMAX", "MBR_YMAX", "POSTCODE_DISTRICT", "POSTCODE_DISTRICT_URI", "POPULATED_PLACE", "POPULATED_PLACE_URI", "POPULATED_PLACE_TYPE", "DISTRICT_BOROUGH", "DISTRICT_BOROUGH_URI", "DISTRICT_BOROUGH_TYPE", "COUNTY_UNITARY", "COUNTY_UNITARY_URI", "COUNTY_UNITARY_TYPE", "REGION", "REGION_URI", "COUNTRY", "COUNTRY_URI", "RELATED_SPATIAL_OBJECT", "SAME_AS_DBPEDIA", "SAME_AS_GEONAMES"];
+/**
+ * The `LOCAL_TYPE` values that mark a road row.
+ */
+export declare const OPEN_NAMES_ROAD_TYPES: ReadonlySet<string>;
+/**
+ * One road row of OS Open Names, with the columns the corpus adapter reads.
+ */
+export interface OpenNamesRoad {
+    /**
+     * The roadNameTOID, such as `osgb4000000010927776`.
+     */
+    id: string;
+    /**
+     * The road's name, in the language OS lists first.
+     */
+    name1: string;
+    /**
+     * The road's other-language name, Welsh or Gaelic, or an empty string.
+     */
+    name2: string;
+    /**
+     * The outward code of the postcode district holding the road, such as `LS6`.
+     */
+    postcodeDistrict: string;
+    /**
+     * The settlement holding the road, or an empty string.
+     *
+     * A bilingual settlement is written `<NAME1> / <NAME2>` of that settlement,
+     * such as `Stornoway / Steòrnabhagh`.
+     */
+    populatedPlace: string;
+    districtBorough: string;
+    countyUnitary: string;
+}
+/**
+ * Counts from a read, with a reason for each row that did not yield.
+ */
+export interface OpenNamesReadStats {
+    /**
+     * Rows read across every file.
+     */
+    read: number;
+    /**
+     * Road rows yielded.
+     */
+    roads: number;
+    /**
+     * Road rows skipped because their `ID` was not among the requested ids.
+     */
+    unrequested: number;
+    /**
+     * Rows with a column count other than 34.
+     */
+    malformed: number;
+}
+/**
+ * A stats object with every count at zero.
+ */
+export declare function createOpenNamesReadStats(): OpenNamesReadStats;
+/**
+ * Streams the road rows of one Open Names CSV.
+ *
+ * @param ids When given, only roads whose `ID` is in the set are yielded.
+ */
+export declare function readOpenNamesRoads(csvPath: string, stats: OpenNamesReadStats, ids?: ReadonlySet<string>): AsyncGenerator<OpenNamesRoad>;
+/**
+ * The settlement's first-listed name from a `POPULATED_PLACE` value.
+ *
+ * OS writes a bilingual settlement as `<NAME1> / <NAME2>`, and NAME1 is the form the adapter emits.
+ */
+export declare function populatedPlaceName1(populatedPlace: string): string;
+//# sourceMappingURL=names.d.ts.map

package/out/gb/os-open/open/names.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"names.d.ts","sourceRoot":"","sources":["../../../../lib/gb/os-open/open/names.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAIH;;GAEG;AACH,eAAO,MAAM,kBAAkB,YAC9B,IAAI,EACJ,WAAW,EACX,OAAO,EACP,YAAY,EACZ,OAAO,EACP,YAAY,EACZ,MAAM,EACN,YAAY,EACZ,YAAY,EACZ,YAAY,EACZ,sBAAsB,EACtB,uBAAuB,EACvB,UAAU,EACV,UAAU,EACV,UAAU,EACV,UAAU,EACV,mBAAmB,EACnB,uBAAuB,EACvB,iBAAiB,EACjB,qBAAqB,EACrB,sBAAsB,EACtB,kBAAkB,EAClB,sBAAsB,EACtB,uBAAuB,EACvB,gBAAgB,EAChB,oBAAoB,EACpB,qBAAqB,EACrB,QAAQ,EACR,YAAY,EACZ,SAAS,EACT,aAAa,EACb,wBAAwB,EACxB,iBAAiB,EACjB,kBAAkB,CACT,CAAA;AAOV;;GAEG;AACH,eAAO,MAAM,qBAAqB,EAAE,WAAW,CAAC,MAAM,CAAoD,CAAA;AAE1G;;GAEG;AACH,MAAM,WAAW,aAAa;IAC7B;;OAEG;IACH,EAAE,EAAE,MAAM,CAAA;IACV;;OAEG;IACH,KAAK,EAAE,MAAM,CAAA;IACb;;OAEG;IACH,KAAK,EAAE,MAAM,CAAA;IACb;;OAEG;IACH,gBAAgB,EAAE,MAAM,CAAA;IACxB;;;;;OAKG;IACH,cAAc,EAAE,MAAM,CAAA;IACtB,eAAe,EAAE,MAAM,CAAA;IACvB,aAAa,EAAE,MAAM,CAAA;CACrB;AAED;;GAEG;AACH,MAAM,WAAW,kBAAkB;IAClC;;OAEG;IACH,IAAI,EAAE,MAAM,CAAA;IACZ;;OAEG;IACH,KAAK,EAAE,MAAM,CAAA;IACb;;OAEG;IACH,WAAW,EAAE,MAAM,CAAA;IACnB;;OAEG;IACH,SAAS,EAAE,MAAM,CAAA;CACjB;AAED;;GAEG;AACH,wBAAgB,wBAAwB,IAAI,kBAAkB,CAE7D;AAED;;;;GAIG;AACH,wBAAuB,kBAAkB,CACxC,OAAO,EAAE,MAAM,EACf,KAAK,EAAE,kBAAkB,EACzB,GAAG,CAAC,EAAE,WAAW,CAAC,MAAM,CAAC,GACvB,cAAc,CAAC,aAAa,CAAC,CAgC/B;AAED;;;;GAIG;AACH,wBAAgB,mBAAmB,CAAC,cAAc,EAAE,MAAM,GAAG,MAAM,CAElE"}