@mailwoman/corpus 5.9.0 → 5.10.0

package/out/src/tools/fetch/nad.js

@@ -0,0 +1,225 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch the US DOT National Address Database (NAD) — ~97 million structured US address-point
+ *   records aggregated from state and local authorities. Source for the `usgov-nad` adapter (#30).
+ *   US Public Domain (17 U.S.C. § 105).
+ *
+ *   - Bounded per-chunk page concurrency (4× speedup at safe pressure)
+ *   - 5× larger page size (5 000 vs. the old bash fetcher's 2 000) — fewer round-trips per chunk
+ *   - Honest `complete: true` flag: only set when every page in the chunk fetched cleanly
+ *   - Built-in fetch with gzip/brotli decompression (no curl + Python subprocess tax)
+ *   - Per-chunk manifest with sha256 + record count + error count
+ *
+ *   ## Source layout
+ *
+ *   The ArcGIS FeatureServer is the only fully-automated path. As of 2026-05:
+ *
+ *   - **`bulk`** mode requires a pre-signed S3 URL (Akamai blocks scripted curl on the DOT page). Pass
+ *       `--nad-url <presigned>` from a browser visit to
+ *       [https://www.transportation.gov/gis/national-address-database](https://www.transportation.gov/gis/national-address-database).
+ *   - **`featureserver`** mode (default) pages the live FeatureService via OBJECTID ranges, writing
+ *       NDJSON chunks into `<outRoot>/usgov-nad/featureserver/`.
+ *
+ *   ## Usage
+ *
+ *   ```sh
+ *   mailwoman corpus fetch nad --out-root /mnt/playpen/mailwoman-data/corpus/sources
+ *
+ *   # Resume from an OID
+ *   mailwoman corpus fetch nad --start-oid 34400001
+ *
+ *   # Increase concurrency on a fast link
+ *   mailwoman corpus fetch nad --concurrency 8 --page-size 10000
+ *   ```
+ */
+import { existsSync, mkdirSync, statSync } from "node:fs";
+import { writeFile } from "node:fs/promises";
+import { join } from "node:path";
+import { sha256File } from "@mailwoman/core/utils";
+import { downloadToFile, readManifest, writeManifest } from "./download.js";
+const SLUG = "usgov-nad";
+const FEATURE_SERVICE_URL = "https://services.arcgis.com/xOi1kZaI0eWDREZv/ArcGIS/rest/services/Address_Points_from_National_Address_Database_view/FeatureServer/0";
+async function fetchPage(startOID, endOID, pageSize) {
+    const url = new URL(`${FEATURE_SERVICE_URL}/query`);
+    url.searchParams.set("where", `OBJECTID BETWEEN ${startOID} AND ${endOID}`);
+    url.searchParams.set("outFields", "*");
+    url.searchParams.set("f", "json");
+    url.searchParams.set("resultRecordCount", String(pageSize));
+    const res = await fetch(url, {
+        headers: { "Accept-Encoding": "gzip, br" },
+        signal: AbortSignal.timeout(120_000),
+    });
+    if (!res.ok)
+        throw new Error(`HTTP ${res.status} ${res.statusText} on OID ${startOID}-${endOID}`);
+    const data = (await res.json());
+    if (data.error)
+        throw new Error(`ArcGIS error on OID ${startOID}-${endOID}: ${data.error.message}`);
+    return (data.features ?? []).map((f) => f.attributes);
+}
+async function discoverTotalCount() {
+    const url = new URL(`${FEATURE_SERVICE_URL}/query`);
+    url.searchParams.set("where", "1=1");
+    url.searchParams.set("returnCountOnly", "true");
+    url.searchParams.set("f", "json");
+    const res = await fetch(url, { signal: AbortSignal.timeout(30_000) });
+    if (!res.ok)
+        throw new Error(`Failed to discover NAD record count: HTTP ${res.status}`);
+    const data = (await res.json());
+    if (typeof data.count !== "number")
+        throw new Error("NAD count query returned no count field");
+    return data.count;
+}
+/**
+ * Fetch a single chunk by paging through its OID range with bounded concurrency. Returns the count of records written
+ * and the count of pages that errored. The caller decides whether to mark the chunk complete based on errors === 0.
+ *
+ * NOTE(phase1): this is a JSON API pager, not a file download — the shared `downloadToFile` doesn't apply here.
+ */
+async function fetchChunk(chunkPath, chunkStart, chunkEnd, pageSize, concurrency, report) {
+    const pageRanges = [];
+    for (let cursor = chunkStart; cursor <= chunkEnd; cursor += pageSize) {
+        pageRanges.push([cursor, Math.min(cursor + pageSize - 1, chunkEnd)]);
+    }
+    // Run bounded-concurrency page fetches. Results indexed by page slot for in-order write.
+    const pageResults = pageRanges.map(() => ({
+        rows: [],
+        error: null,
+    }));
+    let nextSlot = 0;
+    const workers = Array.from({ length: Math.min(concurrency, pageRanges.length) }, async () => {
+        while (true) {
+            const slot = nextSlot++;
+            if (slot >= pageRanges.length)
+                return;
+            const [s, e] = pageRanges[slot];
+            try {
+                pageResults[slot].rows = await fetchPage(s, e, pageSize);
+            }
+            catch (err) {
+                pageResults[slot].error = err;
+                report?.(`    ✗ page ${s}-${e}: ${err.message}`);
+            }
+        }
+    });
+    await Promise.all(workers);
+    // Single-writer phase — write all pages in OID order to keep NDJSON deterministic.
+    const lines = [];
+    let errors = 0;
+    for (const { rows, error } of pageResults) {
+        if (error) {
+            errors++;
+            continue;
+        }
+        for (const row of rows) {
+            lines.push(JSON.stringify(row));
+        }
+    }
+    await writeFile(chunkPath, lines.length === 0 ? "" : lines.join("\n") + "\n");
+    return { recordCount: lines.length, errors };
+}
+async function featureserverMode(options, report) {
+    const chunkSize = options.chunkSize ?? 100_000;
+    const pageSize = options.pageSize ?? 5_000;
+    const concurrency = options.concurrency ?? 4;
+    const startOID = options.startOID ?? 1;
+    const chunkDir = join(options.outRoot, SLUG, "featureserver");
+    mkdirSync(chunkDir, { recursive: true });
+    report?.(`=== ${SLUG} / featureserver`);
+    report?.(`  Discovering record count ...`);
+    const totalCount = await discoverTotalCount();
+    const endOID = options.endOID ?? totalCount;
+    report?.(`  Total records: ${totalCount.toLocaleString()}`);
+    report?.(`  OID range: ${startOID.toLocaleString()} .. ${endOID.toLocaleString()}`);
+    report?.(`  Chunk size: ${chunkSize}, page size: ${pageSize}, concurrency: ${concurrency}`);
+    let fetched = 0;
+    let skipped = 0;
+    let totalRecords = 0;
+    let totalErrors = 0;
+    const failedCodes = [];
+    for (let cursor = startOID; cursor <= endOID; cursor += chunkSize) {
+        const chunkEnd = Math.min(cursor + chunkSize - 1, endOID);
+        const chunkName = `oids_${cursor}-${chunkEnd}`;
+        const chunkPath = join(chunkDir, `${chunkName}.ndjson`);
+        const manifestPath = join(chunkDir, `${chunkName}.manifest.json`);
+        // Idempotency: skip a chunk only if it's marked complete (the bash version's bug was
+        // marking complete on partial-failure runs; we now only set complete after a clean fetch).
+        if (existsSync(chunkPath)) {
+            const recorded = await readManifest(manifestPath);
+            if (recorded?.complete) {
+                skipped++;
+                continue;
+            }
+        }
+        report?.(`  Fetching ${chunkName} ...`);
+        const t0 = Date.now();
+        const { recordCount, errors } = await fetchChunk(chunkPath, cursor, chunkEnd, pageSize, concurrency, report);
+        const elapsed = ((Date.now() - t0) / 1000).toFixed(1);
+        const bytes = statSync(chunkPath).size;
+        const sha = await sha256File(chunkPath);
+        const manifest = {
+            source_url: FEATURE_SERVICE_URL,
+            oid_range: [cursor, chunkEnd],
+            downloaded_at: new Date().toISOString(),
+            filename: `${chunkName}.ndjson`,
+            sha256: sha,
+            bytes,
+            record_count: recordCount,
+            page_errors: errors,
+            complete: errors === 0,
+        };
+        await writeManifest(manifestPath, manifest);
+        const status = errors === 0 ? "✓" : `⚠ ${errors} page errors`;
+        report?.(`    ${status}  ${recordCount.toLocaleString()} records in ${elapsed}s  (${(bytes / 1024 / 1024).toFixed(1)} MB)`);
+        fetched++;
+        totalRecords += recordCount;
+        totalErrors += errors;
+        if (errors > 0) {
+            failedCodes.push(chunkName);
+        }
+    }
+    report?.(`=== featureserver summary ===`);
+    report?.(`chunks fetched: ${fetched}  skipped: ${skipped}`);
+    report?.(`total records: ${totalRecords.toLocaleString()}`);
+    report?.(`page errors:   ${totalErrors}`);
+    report?.(`output:        ${chunkDir}`);
+    // `failed` counts page errors (the old exit-1 condition), `failedCodes` names the dirty chunks.
+    return { fetched, skipped, failed: totalErrors, failedCodes };
+}
+async function bulkMode(options, report) {
+    if (!options.nadURL) {
+        throw new Error(`bulk mode requires --nad-url. The DOT page is Akamai-gated; ` +
+            `visit https://www.transportation.gov/gis/national-address-database in a browser, ` +
+            `accept the disclaimer, and re-run with the pre-signed S3 URL.`);
+    }
+    const destDir = join(options.outRoot, SLUG);
+    mkdirSync(destDir, { recursive: true });
+    const filename = new URL(options.nadURL).pathname.split("/").pop() ?? "NAD.zip";
+    const destPath = join(destDir, filename);
+    report?.(`=== ${SLUG} / ${filename}`);
+    report?.(`  URL: ${options.nadURL.slice(0, 100)}${options.nadURL.length > 100 ? "…" : ""}`);
+    const { bytes } = await downloadToFile({ url: options.nadURL, dest: destPath, timeoutMs: 3 * 3600 * 1000, report });
+    const sha = await sha256File(destPath);
+    await writeManifest(join(destDir, "MANIFEST.json"), {
+        source_url: options.nadURL,
+        downloaded_at: new Date().toISOString(),
+        filename,
+        sha256: sha,
+        bytes,
+    });
+    report?.(`  ✓ ${(bytes / 1024 / 1024 / 1024).toFixed(2)} GB  sha256=${sha}`);
+    return { fetched: 1, skipped: 0, failed: 0, failedCodes: [] };
+}
+export async function fetchNAD(options, report) {
+    const mode = options.mode ?? "featureserver";
+    if (mode === "featureserver") {
+        return featureserverMode(options, report);
+    }
+    if (mode === "bulk") {
+        return bulkMode(options, report);
+    }
+    throw new Error(`unknown mode "${String(mode)}" (expected featureserver|bulk)`);
+}
+//# sourceMappingURL=nad.js.map

package/out/src/tools/fetch/nad.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"nad.js","sourceRoot":"","sources":["../../../../src/tools/fetch/nad.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAoCG;AAEH,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,SAAS,CAAA;AACzD,OAAO,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAA;AAC5C,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAEhC,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAGlD,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,eAAe,CAAA;AAE3E,MAAM,IAAI,GAAG,WAAW,CAAA;AACxB,MAAM,mBAAmB,GACxB,sIAAsI,CAAA;AA+BvI,KAAK,UAAU,SAAS,CAAC,QAAgB,EAAE,MAAc,EAAE,QAAgB;IAC1E,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC,GAAG,mBAAmB,QAAQ,CAAC,CAAA;IACnD,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,OAAO,EAAE,oBAAoB,QAAQ,QAAQ,MAAM,EAAE,CAAC,CAAA;IAC3E,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IACtC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAA;IACjC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,mBAAmB,EAAE,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAA;IAE3D,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE;QAC5B,OAAO,EAAE,EAAE,iBAAiB,EAAE,UAAU,EAAE;QAC1C,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC;KACpC,CAAC,CAAA;IAEF,IAAI,CAAC,GAAG,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,QAAQ,GAAG,CAAC,MAAM,IAAI,GAAG,CAAC,UAAU,WAAW,QAAQ,IAAI,MAAM,EAAE,CAAC,CAAA;IACjG,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAA+E,CAAA;IAE7G,IAAI,IAAI,CAAC,KAAK;QAAE,MAAM,IAAI,KAAK,CAAC,uBAAuB,QAAQ,IAAI,MAAM,KAAK,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC,CAAA;IAEnG,OAAO,CAAC,IAAI,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,CAAA;AACtD,CAAC;AAED,KAAK,UAAU,kBAAkB;IAChC,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC,GAAG,mBAAmB,QAAQ,CAAC,CAAA;IACnD,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,CAAA;IACpC,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,iBAAiB,EAAE,MAAM,CAAC,CAAA;IAC/C,GAAG,CAAC,YAAY,CAAC,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,CAAA;IACjC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE,EAAE,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,CAAA;IAErE,IAAI,CAAC,GAAG,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,6CAA6C,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;IACvF,MAAM,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,IAAI,EAAE,CAAuB,CAAA;IAErD,IAAI,OAAO,IAAI,CAAC,KAAK,KAAK,QAAQ;QAAE,MAAM,IAAI,KAAK,CAAC,yCAAyC,CAAC,CAAA;IAE9F,OAAO,IAAI,CAAC,KAAK,CAAA;AAClB,CAAC;AAED;;;;;GAKG;AACH,KAAK,UAAU,UAAU,CACxB,SAAiB,EACjB,UAAkB,EAClB,QAAgB,EAChB,QAAgB,EAChB,WAAmB,EACnB,MAA+B;IAE/B,MAAM,UAAU,GAA4B,EAAE,CAAA;IAE9C,KAAK,IAAI,MAAM,GAAG,UAAU,EAAE,MAAM,IAAI,QAAQ,EAAE,MAAM,IAAI,QAAQ,EAAE,CAAC;QACtE,UAAU,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,QAAQ,GAAG,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAA;IACrE,CAAC;IAED,yFAAyF;IACzF,MAAM,WAAW,GAAoD,UAAU,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC;QAC1F,IAAI,EAAE,EAAE;QACR,KAAK,EAAE,IAAI;KACX,CAAC,CAAC,CAAA;IACH,IAAI,QAAQ,GAAG,CAAC,CAAA;IAChB,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,IAAI,CAAC,GAAG,CAAC,WAAW,EAAE,UAAU,CAAC,MAAM,CAAC,EAAE,EAAE,KAAK,IAAI,EAAE;QAC3F,OAAO,IAAI,EAAE,CAAC;YACb,MAAM,IAAI,GAAG,QAAQ,EAAE,CAAA;YAEvB,IAAI,IAAI,IAAI,UAAU,CAAC,MAAM;gBAAE,OAAM;YACrC,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,UAAU,CAAC,IAAI,CAAE,CAAA;YAEhC,IAAI,CAAC;gBACJ,WAAW,CAAC,IAAI,CAAE,CAAC,IAAI,GAAG,MAAM,SAAS,CAAC,CAAC,EAAE,CAAC,EAAE,QAAQ,CAAC,CAAA;YAC1D,CAAC;YAAC,OAAO,GAAG,EAAE,CAAC;gBACd,WAAW,CAAC,IAAI,CAAE,CAAC,KAAK,GAAG,GAAY,CAAA;gBACvC,MAAM,EAAE,CAAC,cAAc,CAAC,IAAI,CAAC,KAAM,GAAa,CAAC,OAAO,EAAE,CAAC,CAAA;YAC5D,CAAC;QACF,CAAC;IACF,CAAC,CAAC,CAAA;IACF,MAAM,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IAE1B,mFAAmF;IACnF,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,KAAK,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,WAAW,EAAE,CAAC;QAC3C,IAAI,KAAK,EAAE,CAAC;YACX,MAAM,EAAE,CAAA;YACR,SAAQ;QACT,CAAC;QAED,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACxB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAA;QAChC,CAAC;IACF,CAAC;IACD,MAAM,SAAS,CAAC,SAAS,EAAE,KAAK,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,CAAA;IAE7E,OAAO,EAAE,WAAW,EAAE,KAAK,CAAC,MAAM,EAAE,MAAM,EAAE,CAAA;AAC7C,CAAC;AAED,KAAK,UAAU,iBAAiB,CAAC,OAAwB,EAAE,MAA+B;IACzF,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,IAAI,OAAO,CAAA;IAC9C,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,IAAI,KAAK,CAAA;IAC1C,MAAM,WAAW,GAAG,OAAO,CAAC,WAAW,IAAI,CAAC,CAAA;IAC5C,MAAM,QAAQ,GAAG,OAAO,CAAC,QAAQ,IAAI,CAAC,CAAA;IAEtC,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,EAAE,eAAe,CAAC,CAAA;IAC7D,SAAS,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IAExC,MAAM,EAAE,CAAC,OAAO,IAAI,kBAAkB,CAAC,CAAA;IACvC,MAAM,EAAE,CAAC,gCAAgC,CAAC,CAAA;IAC1C,MAAM,UAAU,GAAG,MAAM,kBAAkB,EAAE,CAAA;IAC7C,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,UAAU,CAAA;IAC3C,MAAM,EAAE,CAAC,oBAAoB,UAAU,CAAC,cAAc,EAAE,EAAE,CAAC,CAAA;IAC3D,MAAM,EAAE,CAAC,gBAAgB,QAAQ,CAAC,cAAc,EAAE,OAAO,MAAM,CAAC,cAAc,EAAE,EAAE,CAAC,CAAA;IACnF,MAAM,EAAE,CAAC,iBAAiB,SAAS,gBAAgB,QAAQ,kBAAkB,WAAW,EAAE,CAAC,CAAA;IAE3F,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,IAAI,YAAY,GAAG,CAAC,CAAA;IACpB,IAAI,WAAW,GAAG,CAAC,CAAA;IACnB,MAAM,WAAW,GAAa,EAAE,CAAA;IAEhC,KAAK,IAAI,MAAM,GAAG,QAAQ,EAAE,MAAM,IAAI,MAAM,EAAE,MAAM,IAAI,SAAS,EAAE,CAAC;QACnE,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,SAAS,GAAG,CAAC,EAAE,MAAM,CAAC,CAAA;QACzD,MAAM,SAAS,GAAG,QAAQ,MAAM,IAAI,QAAQ,EAAE,CAAA;QAC9C,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,EAAE,GAAG,SAAS,SAAS,CAAC,CAAA;QACvD,MAAM,YAAY,GAAG,IAAI,CAAC,QAAQ,EAAE,GAAG,SAAS,gBAAgB,CAAC,CAAA;QAEjE,qFAAqF;QACrF,2FAA2F;QAC3F,IAAI,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;YAC3B,MAAM,QAAQ,GAAG,MAAM,YAAY,CAAgB,YAAY,CAAC,CAAA;YAEhE,IAAI,QAAQ,EAAE,QAAQ,EAAE,CAAC;gBACxB,OAAO,EAAE,CAAA;gBACT,SAAQ;YACT,CAAC;QACF,CAAC;QAED,MAAM,EAAE,CAAC,cAAc,SAAS,MAAM,CAAC,CAAA;QACvC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QACrB,MAAM,EAAE,WAAW,EAAE,MAAM,EAAE,GAAG,MAAM,UAAU,CAAC,SAAS,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,CAAC,CAAA;QAC5G,MAAM,OAAO,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAA;QACrD,MAAM,KAAK,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,IAAI,CAAA;QACtC,MAAM,GAAG,GAAG,MAAM,UAAU,CAAC,SAAS,CAAC,CAAA;QAEvC,MAAM,QAAQ,GAAkB;YAC/B,UAAU,EAAE,mBAAmB;YAC/B,SAAS,EAAE,CAAC,MAAM,EAAE,QAAQ,CAAC;YAC7B,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YACvC,QAAQ,EAAE,GAAG,SAAS,SAAS;YAC/B,MAAM,EAAE,GAAG;YACX,KAAK;YACL,YAAY,EAAE,WAAW;YACzB,WAAW,EAAE,MAAM;YACnB,QAAQ,EAAE,MAAM,KAAK,CAAC;SACtB,CAAA;QACD,MAAM,aAAa,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;QAE3C,MAAM,MAAM,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,MAAM,cAAc,CAAA;QAC7D,MAAM,EAAE,CACP,OAAO,MAAM,KAAK,WAAW,CAAC,cAAc,EAAE,eAAe,OAAO,OAAO,CAAC,KAAK,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CACjH,CAAA;QACD,OAAO,EAAE,CAAA;QACT,YAAY,IAAI,WAAW,CAAA;QAC3B,WAAW,IAAI,MAAM,CAAA;QAErB,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC;YAChB,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;QAC5B,CAAC;IACF,CAAC;IAED,MAAM,EAAE,CAAC,+BAA+B,CAAC,CAAA;IACzC,MAAM,EAAE,CAAC,mBAAmB,OAAO,cAAc,OAAO,EAAE,CAAC,CAAA;IAC3D,MAAM,EAAE,CAAC,kBAAkB,YAAY,CAAC,cAAc,EAAE,EAAE,CAAC,CAAA;IAC3D,MAAM,EAAE,CAAC,kBAAkB,WAAW,EAAE,CAAC,CAAA;IACzC,MAAM,EAAE,CAAC,kBAAkB,QAAQ,EAAE,CAAC,CAAA;IAEtC,gGAAgG;IAChG,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,WAAW,EAAE,CAAA;AAC9D,CAAC;AAED,KAAK,UAAU,QAAQ,CAAC,OAAwB,EAAE,MAA+B;IAChF,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;QACrB,MAAM,IAAI,KAAK,CACd,8DAA8D;YAC7D,mFAAmF;YACnF,+DAA+D,CAChE,CAAA;IACF,CAAC;IACD,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;IAC3C,SAAS,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IACvC,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,SAAS,CAAA;IAC/E,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAA;IAExC,MAAM,EAAE,CAAC,OAAO,IAAI,MAAM,QAAQ,EAAE,CAAC,CAAA;IACrC,MAAM,EAAE,CAAC,UAAU,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAA;IAE3F,MAAM,EAAE,KAAK,EAAE,GAAG,MAAM,cAAc,CAAC,EAAE,GAAG,EAAE,OAAO,CAAC,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC,GAAG,IAAI,GAAG,IAAI,EAAE,MAAM,EAAE,CAAC,CAAA;IAEnH,MAAM,GAAG,GAAG,MAAM,UAAU,CAAC,QAAQ,CAAC,CAAA;IACtC,MAAM,aAAa,CAAC,IAAI,CAAC,OAAO,EAAE,eAAe,CAAC,EAAE;QACnD,UAAU,EAAE,OAAO,CAAC,MAAM;QAC1B,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;QACvC,QAAQ;QACR,MAAM,EAAE,GAAG;QACX,KAAK;KACL,CAAC,CAAA;IACF,MAAM,EAAE,CAAC,OAAO,CAAC,KAAK,GAAG,IAAI,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,GAAG,EAAE,CAAC,CAAA;IAE5E,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAA;AAC9D,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,QAAQ,CAAC,OAAwB,EAAE,MAA+B;IACvF,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,IAAI,eAAe,CAAA;IAE5C,IAAI,IAAI,KAAK,eAAe,EAAE,CAAC;QAC9B,OAAO,iBAAiB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IAC1C,CAAC;IAED,IAAI,IAAI,KAAK,MAAM,EAAE,CAAC;QACrB,OAAO,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IACjC,CAAC;IAED,MAAM,IAAI,KAAK,CAAC,iBAAiB,MAAM,CAAC,IAAI,CAAC,iCAAiC,CAAC,CAAA;AAChF,CAAC"}

package/out/src/tools/fetch/nppes.d.ts

@@ -0,0 +1,26 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Re-fetch the NPPES (National Plan and Provider Enumeration System) full monthly data
+ *   dissemination file. ~7M provider rows with venue+address data. Source for the `usgov-nppes`
+ *   adapter. US Public Domain.
+ *
+ *   The file is published monthly by CMS. This module discovers the current filename by scraping the
+ *   NPI_Files.html index, then downloads the ZIP and extracts only the main registry CSV
+ *   (npidata_pfile_*.csv). The smaller endpoint/othername/pl files stay zipped — we don't need them.
+ *
+ *   Uses Node's built-in fetch (gzip/brotli) to parse the HTML index and download the ZIP, and
+ *   streaming sha256 instead of sha256sum. The ZIP is unpacked with the `unzip` binary via
+ *   `node:child_process` (no clean Node equivalent for member listing + selective extraction). NOTE:
+ *   the old bash fetcher used `curl --continue-at -` to resume a partial download; native fetch has
+ *   no resume, so a partial run re-downloads from the start.
+ *
+ *   Invoke via `mailwoman corpus fetch nppes --out-root <path>`. Idempotent: if dest CSV exists and
+ *   sha256 matches MANIFEST, skips download.
+ */
+import type { BaseFetchOptions, FetchSummary } from "./download.ts";
+export type FetchNPPESOptions = BaseFetchOptions;
+export declare function fetchNPPES(options: FetchNPPESOptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=nppes.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"nppes.d.ts","sourceRoot":"","sources":["../../../../src/tools/fetch/nppes.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAUH,OAAO,KAAK,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,eAAe,CAAA;AASnE,MAAM,MAAM,iBAAiB,GAAG,gBAAgB,CAAA;AA6ChD,wBAAsB,UAAU,CAAC,OAAO,EAAE,iBAAiB,EAAE,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAAG,OAAO,CAAC,YAAY,CAAC,CA2FnH"}

package/out/src/tools/fetch/nppes.js

@@ -0,0 +1,136 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Re-fetch the NPPES (National Plan and Provider Enumeration System) full monthly data
+ *   dissemination file. ~7M provider rows with venue+address data. Source for the `usgov-nppes`
+ *   adapter. US Public Domain.
+ *
+ *   The file is published monthly by CMS. This module discovers the current filename by scraping the
+ *   NPI_Files.html index, then downloads the ZIP and extracts only the main registry CSV
+ *   (npidata_pfile_*.csv). The smaller endpoint/othername/pl files stay zipped — we don't need them.
+ *
+ *   Uses Node's built-in fetch (gzip/brotli) to parse the HTML index and download the ZIP, and
+ *   streaming sha256 instead of sha256sum. The ZIP is unpacked with the `unzip` binary via
+ *   `node:child_process` (no clean Node equivalent for member listing + selective extraction). NOTE:
+ *   the old bash fetcher used `curl --continue-at -` to resume a partial download; native fetch has
+ *   no resume, so a partial run re-downloads from the start.
+ *
+ *   Invoke via `mailwoman corpus fetch nppes --out-root <path>`. Idempotent: if dest CSV exists and
+ *   sha256 matches MANIFEST, skips download.
+ */
+import { execFile } from "node:child_process";
+import { existsSync, mkdirSync, statSync } from "node:fs";
+import { rm } from "node:fs/promises";
+import { join } from "node:path";
+import { promisify } from "node:util";
+import { sha256File } from "@mailwoman/core/utils";
+import { downloadToFile, readManifest, writeManifest } from "./download.js";
+const execFileAsync = promisify(execFile);
+const INDEX_URL = "https://download.cms.gov/nppes/NPI_Files.html";
+const BASE_URL = "https://download.cms.gov/nppes";
+const SLUG = "usgov-nppes";
+/**
+ * Scrape the NPI_Files.html index for the latest full monthly ZIP. Full-replacement files match
+ * `NPPES_Data_Dissemination_<Month>_<Year>*.zip`; weekly files carry a `MMDDYY_MMDDYY` date range, which we exclude.
+ */
+async function discoverLatestZip() {
+    const res = await fetch(INDEX_URL, {
+        headers: { "Accept-Encoding": "gzip, br" },
+        signal: AbortSignal.timeout(60_000),
+    });
+    if (!res.ok)
+        throw new Error(`HTTP ${res.status} ${res.statusText} on ${INDEX_URL}`);
+    const html = await res.text();
+    for (const match of html.matchAll(/NPPES_Data_Dissemination_[A-Za-z]+_\d{4}[^"]*\.zip/g)) {
+        const name = match[0];
+        if (name && !/\d{6}_\d{6}/.test(name))
+            return name;
+    }
+    return undefined;
+}
+/** Extract the main registry CSV name (npidata_pfile_*.csv) from a ZIP's `unzip -l` listing. */
+async function findNpidataCSV(zipPath) {
+    const listing = await execFileAsync("unzip", ["-l", zipPath]);
+    for (const line of listing.stdout.split("\n")) {
+        const match = /npidata_pfile\S+\.csv/i.exec(line);
+        if (match?.[0])
+            return match[0];
+    }
+    return undefined;
+}
+export async function fetchNPPES(options, report) {
+    const destDir = join(options.outRoot, SLUG);
+    mkdirSync(destDir, { recursive: true });
+    const manifestPath = join(destDir, "MANIFEST.json");
+    report?.(`=== ${SLUG}`);
+    report?.(`  Discovering latest full-replacement ZIP from ${INDEX_URL} ...`);
+    const zipFilename = await discoverLatestZip();
+    if (!zipFilename) {
+        report?.(`  ✗ Could not discover ZIP filename from ${INDEX_URL}`);
+        return { fetched: 0, skipped: 0, failed: 1, failedCodes: [SLUG] };
+    }
+    const zipURL = `${BASE_URL}/${zipFilename}`;
+    const zipDest = join(destDir, zipFilename);
+    report?.(`  Latest full file: ${zipFilename}`);
+    // ------------------------------------------------------------------
+    // Idempotency check: if the main CSV already exists and sha matches,
+    // skip re-download.
+    // ------------------------------------------------------------------
+    const recorded = await readManifest(manifestPath);
+    if (recorded?.sha256 && recorded.filename) {
+        const recordedPath = join(destDir, recorded.filename);
+        if (existsSync(recordedPath) && (await sha256File(recordedPath)) === recorded.sha256) {
+            report?.("  ✓ Already current (sha256 matches MANIFEST) — skipping download.");
+            return { fetched: 0, skipped: 1, failed: 0, failedCodes: [] };
+        }
+    }
+    // ------------------------------------------------------------------
+    // Download ZIP (large; 60-minute timeout)
+    // ------------------------------------------------------------------
+    report?.(`  Downloading ${zipURL} ...`);
+    const { bytes: zipSize } = await downloadToFile({
+        url: zipURL,
+        dest: zipDest,
+        timeoutMs: 3_600_000,
+        headers: { "Accept-Encoding": "gzip, br" },
+        report,
+    });
+    report?.(`  Downloaded: ${(zipSize / 1024 / 1024).toFixed(1)} MB`);
+    // ------------------------------------------------------------------
+    // Extract only the main registry CSV (npidata_pfile_*.csv)
+    // ------------------------------------------------------------------
+    report?.("  Extracting npidata_pfile CSV from ZIP ...");
+    const csvName = await findNpidataCSV(zipDest);
+    if (!csvName) {
+        report?.("  ✗ Could not find npidata_pfile CSV inside ZIP");
+        return { fetched: 0, skipped: 0, failed: 1, failedCodes: [SLUG] };
+    }
+    report?.(`  Extracting: ${csvName}`);
+    await execFileAsync("unzip", ["-o", "-j", zipDest, csvName, "-d", destDir]);
+    const csvDest = join(destDir, csvName);
+    const csvSize = statSync(csvDest).size;
+    const csvSha = await sha256File(csvDest);
+    report?.(`  CSV size: ${(csvSize / 1024 / 1024).toFixed(1)} MB`);
+    // ------------------------------------------------------------------
+    // Remove the ZIP to reclaim ~1 GB (the CSV is what adapters consume)
+    // ------------------------------------------------------------------
+    await rm(zipDest, { force: true });
+    report?.("  Removed ZIP (CSV kept)");
+    // ------------------------------------------------------------------
+    // Write MANIFEST (records the extracted CSV, not the ZIP)
+    // ------------------------------------------------------------------
+    const manifest = {
+        source_url: zipURL,
+        downloaded_at: new Date().toISOString(),
+        filename: csvName,
+        sha256: csvSha,
+        bytes: csvSize,
+    };
+    await writeManifest(manifestPath, manifest);
+    report?.(`  ✓ ${(csvSize / 1024 / 1024).toFixed(1)} MB  sha256=${csvSha}`);
+    report?.(`  MANIFEST written to ${manifestPath}`);
+    return { fetched: 1, skipped: 0, failed: 0, failedCodes: [] };
+}
+//# sourceMappingURL=nppes.js.map

package/out/src/tools/fetch/nppes.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"nppes.js","sourceRoot":"","sources":["../../../../src/tools/fetch/nppes.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;GAqBG;AAEH,OAAO,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAA;AAC7C,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,QAAQ,EAAE,MAAM,SAAS,CAAA;AACzD,OAAO,EAAE,EAAE,EAAE,MAAM,kBAAkB,CAAA;AACrC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAChC,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAA;AAErC,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAA;AAGlD,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,eAAe,CAAA;AAE3E,MAAM,aAAa,GAAG,SAAS,CAAC,QAAQ,CAAC,CAAA;AAEzC,MAAM,SAAS,GAAG,+CAA+C,CAAA;AACjE,MAAM,QAAQ,GAAG,gCAAgC,CAAA;AACjD,MAAM,IAAI,GAAG,aAAa,CAAA;AAY1B;;;GAGG;AACH,KAAK,UAAU,iBAAiB;IAC/B,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,EAAE;QAClC,OAAO,EAAE,EAAE,iBAAiB,EAAE,UAAU,EAAE;QAC1C,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,MAAM,CAAC;KACnC,CAAC,CAAA;IAEF,IAAI,CAAC,GAAG,CAAC,EAAE;QAAE,MAAM,IAAI,KAAK,CAAC,QAAQ,GAAG,CAAC,MAAM,IAAI,GAAG,CAAC,UAAU,OAAO,SAAS,EAAE,CAAC,CAAA;IACpF,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,EAAE,CAAA;IAE7B,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,QAAQ,CAAC,qDAAqD,CAAC,EAAE,CAAC;QAC1F,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAA;QAErB,IAAI,IAAI,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,IAAI,CAAA;IACnD,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,gGAAgG;AAChG,KAAK,UAAU,cAAc,CAAC,OAAe;IAC5C,MAAM,OAAO,GAAG,MAAM,aAAa,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAA;IAE7D,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QAC/C,MAAM,KAAK,GAAG,wBAAwB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAEjD,IAAI,KAAK,EAAE,CAAC,CAAC,CAAC;YAAE,OAAO,KAAK,CAAC,CAAC,CAAC,CAAA;IAChC,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,OAA0B,EAAE,MAA+B;IAC3F,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;IAC3C,SAAS,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;IACvC,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,EAAE,eAAe,CAAC,CAAA;IAEnD,MAAM,EAAE,CAAC,OAAO,IAAI,EAAE,CAAC,CAAA;IACvB,MAAM,EAAE,CAAC,kDAAkD,SAAS,MAAM,CAAC,CAAA;IAE3E,MAAM,WAAW,GAAG,MAAM,iBAAiB,EAAE,CAAA;IAE7C,IAAI,CAAC,WAAW,EAAE,CAAC;QAClB,MAAM,EAAE,CAAC,4CAA4C,SAAS,EAAE,CAAC,CAAA;QAEjE,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,WAAW,EAAE,CAAC,IAAI,CAAC,EAAE,CAAA;IAClE,CAAC;IAED,MAAM,MAAM,GAAG,GAAG,QAAQ,IAAI,WAAW,EAAE,CAAA;IAC3C,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,EAAE,WAAW,CAAC,CAAA;IAC1C,MAAM,EAAE,CAAC,uBAAuB,WAAW,EAAE,CAAC,CAAA;IAE9C,qEAAqE;IACrE,qEAAqE;IACrE,oBAAoB;IACpB,qEAAqE;IACrE,MAAM,QAAQ,GAAG,MAAM,YAAY,CAA0B,YAAY,CAAC,CAAA;IAE1E,IAAI,QAAQ,EAAE,MAAM,IAAI,QAAQ,CAAC,QAAQ,EAAE,CAAC;QAC3C,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,EAAE,QAAQ,CAAC,QAAQ,CAAC,CAAA;QAErD,IAAI,UAAU,CAAC,YAAY,CAAC,IAAI,CAAC,MAAM,UAAU,CAAC,YAAY,CAAC,CAAC,KAAK,QAAQ,CAAC,MAAM,EAAE,CAAC;YACtF,MAAM,EAAE,CAAC,oEAAoE,CAAC,CAAA;YAE9E,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAA;QAC9D,CAAC;IACF,CAAC;IAED,qEAAqE;IACrE,0CAA0C;IAC1C,qEAAqE;IACrE,MAAM,EAAE,CAAC,iBAAiB,MAAM,MAAM,CAAC,CAAA;IACvC,MAAM,EAAE,KAAK,EAAE,OAAO,EAAE,GAAG,MAAM,cAAc,CAAC;QAC/C,GAAG,EAAE,MAAM;QACX,IAAI,EAAE,OAAO;QACb,SAAS,EAAE,SAAS;QACpB,OAAO,EAAE,EAAE,iBAAiB,EAAE,UAAU,EAAE;QAC1C,MAAM;KACN,CAAC,CAAA;IACF,MAAM,EAAE,CAAC,iBAAiB,CAAC,OAAO,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,CAAA;IAElE,qEAAqE;IACrE,2DAA2D;IAC3D,qEAAqE;IACrE,MAAM,EAAE,CAAC,6CAA6C,CAAC,CAAA;IACvD,MAAM,OAAO,GAAG,MAAM,cAAc,CAAC,OAAO,CAAC,CAAA;IAE7C,IAAI,CAAC,OAAO,EAAE,CAAC;QACd,MAAM,EAAE,CAAC,iDAAiD,CAAC,CAAA;QAE3D,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,WAAW,EAAE,CAAC,IAAI,CAAC,EAAE,CAAA;IAClE,CAAC;IAED,MAAM,EAAE,CAAC,iBAAiB,OAAO,EAAE,CAAC,CAAA;IACpC,MAAM,aAAa,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,OAAO,CAAC,CAAC,CAAA;IAE3E,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,EAAE,OAAO,CAAC,CAAA;IACtC,MAAM,OAAO,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAC,IAAI,CAAA;IACtC,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,OAAO,CAAC,CAAA;IACxC,MAAM,EAAE,CAAC,eAAe,CAAC,OAAO,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,KAAK,CAAC,CAAA;IAEhE,qEAAqE;IACrE,qEAAqE;IACrE,qEAAqE;IACrE,MAAM,EAAE,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAA;IAClC,MAAM,EAAE,CAAC,0BAA0B,CAAC,CAAA;IAEpC,qEAAqE;IACrE,0DAA0D;IAC1D,qEAAqE;IACrE,MAAM,QAAQ,GAAmB;QAChC,UAAU,EAAE,MAAM;QAClB,aAAa,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;QACvC,QAAQ,EAAE,OAAO;QACjB,MAAM,EAAE,MAAM;QACd,KAAK,EAAE,OAAO;KACd,CAAA;IACD,MAAM,aAAa,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;IAE3C,MAAM,EAAE,CAAC,OAAO,CAAC,OAAO,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,MAAM,EAAE,CAAC,CAAA;IAC1E,MAAM,EAAE,CAAC,yBAAyB,YAAY,EAAE,CAAC,CAAA;IAEjD,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAA;AAC9D,CAAC"}

package/out/src/tools/fetch/openaddresses.d.ts

@@ -0,0 +1,59 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Fetch an OpenAddresses country collection from batch.openaddresses.io.
+ *
+ *   Source: https://batch.openaddresses.io
+ *   License: MIXED — OpenAddresses aggregates hundreds of upstream sources with per-source licenses
+ *   (CC-BY, CC0, PDDL, ODbL, CC-BY-SA, and proprietary attribution-only). The per-row LICENSE filter
+ *   in the openaddresses adapter is essential for proprietary-weights training: Tier-C rows (ODbL,
+ *   CC-BY-SA, CC-SA) are dropped at ingest by default. This module downloads the raw collection; the
+ *   adapter does the license gating.
+ *
+ *   Native `fetch` streams the download to disk (no curl + Python subprocess tax);
+ *   `node:child_process` keeps the genuine shell ops it still needs (`file` magic detection +
+ *   `gunzip` decompression, both nice/ionice-deprioritized).
+ *
+ *   ## Authentication note (2026-05-18)
+ *
+ *   The batch.openaddresses.io download endpoint now requires a registered account. Downloads are
+ *   still free at the "basic" tier (GeoJSON+LD output).
+ *
+ *   1. Register at https://batch.openaddresses.io/register
+ *   2. Log in and go to Profile → "Create Token"
+ *   3. Export the token: `export OA_BATCH_TOKEN=<your-token>`
+ *   4. Re-run the command.
+ *
+ *   The collection URL pattern (verified 2026-05-18):
+ *
+ *   - `POST /api/login {username, password}` → `{token}`
+ *   - `GET  /api/job/{job_id}/output/source.geojson.gz?token={token}`
+ *
+ *   Collections are downloaded as a combined GeoJSON.gz via:
+ *
+ *   - `GET  /api/collections/{collection_id}/download` (returns a redirect to S3)
+ *
+ *   Collection IDs discovered from `/api/collections`:
+ *
+ *   - `id=6  name="ca"  size=2044467556` (~1.9 GiB uncompressed, verified 2026-05-18)
+ *
+ *   ## Usage
+ *
+ *   ```sh
+ *   # With token (preferred). Default country: ca. Supports any OA country code (us-west, fr, …)
+ *   OA_BATCH_TOKEN=<token> mailwoman corpus fetch openaddresses --country ca \
+ *     --out-root /mnt/playpen/mailwoman-data/corpus/sources
+ *
+ *   # Without token (will detect + print instructions, then report the failure):
+ *   mailwoman corpus fetch openaddresses --country ca
+ *   ```
+ */
+import type { BaseFetchOptions, FetchSummary } from "./download.ts";
+export interface FetchOpenAddressesOptions extends BaseFetchOptions {
+    /** OA country collection code. Default `ca`. */
+    country?: string;
+}
+export declare function fetchOpenAddresses(options: FetchOpenAddressesOptions, report?: (line: string) => void): Promise<FetchSummary>;
+//# sourceMappingURL=openaddresses.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"openaddresses.d.ts","sourceRoot":"","sources":["../../../../src/tools/fetch/openaddresses.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmDG;AAaH,OAAO,KAAK,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,eAAe,CAAA;AAoBnE,MAAM,WAAW,yBAA0B,SAAQ,gBAAgB;IAClE,gDAAgD;IAChD,OAAO,CAAC,EAAE,MAAM,CAAA;CAChB;AAkGD,wBAAsB,kBAAkB,CACvC,OAAO,EAAE,yBAAyB,EAClC,MAAM,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,GAC7B,OAAO,CAAC,YAAY,CAAC,CA+LvB"}