@mailwoman/corpus 9.0.0 → 9.1.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mailwoman/corpus",
-  "version": "9.0.0",
+  "version": "9.1.0",
   "description": "Mailwoman corpus pipeline: BIO-labeled dataset builder for the neural classifier.",
   "license": "AGPL-3.0-only OR LicenseRef-Commercial",
   "repository": {
@@ -9,6 +9,7 @@
     "directory": "corpus"
   },
   "files": [
+    "data/**",
     "out/**/*.js",
     "out/**/*.js.map",
     "out/**/*.d.ts",
@@ -24,6 +25,9 @@
     "!test-kit/**/*"
   ],
   "type": "module",
+  "imports": {
+    "#test-kit": "./test-kit/index.ts"
+  },
   "exports": {
     "./package.json": "./package.json",
     ".": {
@@ -34,25 +38,46 @@
       "types": "./out/src/types.d.ts",
       "default": "./out/src/types.js"
     },
-    "./format": {
-      "types": "./out/src/format.d.ts",
-      "default": "./out/src/format.js"
+    "./data/*": "./data/*",
+    "./utils": {
+      "types": "./out/src/utils/index.d.ts",
+      "default": "./out/src/utils/index.js"
+    },
+    "./parquet-wrapper": {
+      "types": "./out/src/parquet-wrapper/index.d.ts",
+      "default": "./out/src/parquet-wrapper/index.js"
     },
-    "./license": {
-      "types": "./out/src/license.d.ts",
-      "default": "./out/src/license.js"
+    "./runner": {
+      "types": "./out/src/runner.d.ts",
+      "default": "./out/src/runner.js"
     },
     "./build": {
       "types": "./out/src/build.d.ts",
       "default": "./out/src/build.js"
     },
+    "./synthesizers/*": {
+      "types": "./out/src/synthesizers/*.d.ts",
+      "default": "./out/src/synthesizers/*.js"
+    },
+    "./adapters/utils": {
+      "types": "./out/src/adapters/utils/index.d.ts",
+      "default": "./out/src/adapters/utils/index.js"
+    },
     "./adapters": {
       "types": "./out/src/adapters/index.d.ts",
       "default": "./out/src/adapters/index.js"
     },
+    "./adapters/*": {
+      "types": "./out/src/adapters/*.d.ts",
+      "default": "./out/src/adapters/*.js"
+    },
     "./tools": {
       "types": "./out/src/tools/index.d.ts",
       "default": "./out/src/tools/index.js"
+    },
+    "./codex/*": {
+      "types": "./out/src/codex/*.d.ts",
+      "default": "./out/src/codex/*.js"
     }
   },
   "publishConfig": {
@@ -67,39 +92,60 @@
         "types": "./out/src/types.d.ts",
         "default": "./out/src/types.js"
       },
-      "./format": {
-        "types": "./out/src/format.d.ts",
-        "default": "./out/src/format.js"
+      "./data/*": "./data/*",
+      "./utils": {
+        "types": "./out/src/utils/index.d.ts",
+        "default": "./out/src/utils/index.js"
       },
-      "./license": {
-        "types": "./out/src/license.d.ts",
-        "default": "./out/src/license.js"
+      "./parquet-wrapper": {
+        "types": "./out/src/parquet-wrapper/index.d.ts",
+        "default": "./out/src/parquet-wrapper/index.js"
+      },
+      "./runner": {
+        "types": "./out/src/runner.d.ts",
+        "default": "./out/src/runner.js"
       },
       "./build": {
         "types": "./out/src/build.d.ts",
         "default": "./out/src/build.js"
       },
+      "./synthesizers/*": {
+        "types": "./out/src/synthesizers/*.d.ts",
+        "default": "./out/src/synthesizers/*.js"
+      },
+      "./adapters/utils": {
+        "types": "./out/src/adapters/utils/index.d.ts",
+        "default": "./out/src/adapters/utils/index.js"
+      },
       "./adapters": {
         "types": "./out/src/adapters/index.d.ts",
         "default": "./out/src/adapters/index.js"
       },
+      "./adapters/*": {
+        "types": "./out/src/adapters/*.d.ts",
+        "default": "./out/src/adapters/*.js"
+      },
       "./tools": {
         "types": "./out/src/tools/index.d.ts",
         "default": "./out/src/tools/index.js"
+      },
+      "./codex/*": {
+        "types": "./out/src/codex/*.d.ts",
+        "default": "./out/src/codex/*.js"
       }
     }
   },
   "dependencies": {
     "@dsnp/parquetjs": "1.8.9",
-    "@mailwoman/ban": "9.0.0",
-    "@mailwoman/codex": "9.0.0",
-    "@mailwoman/core": "9.0.0",
-    "@mailwoman/formatter": "9.0.0",
+    "@mailwoman/ban": "9.1.0",
+    "@mailwoman/codex": "9.1.0",
+    "@mailwoman/core": "9.1.0",
+    "@mailwoman/formatter": "9.1.0",
     "fast-glob": "^3.3.3",
     "fastest-levenshtein": "^1.0.16",
     "lru-cache": "^11.5.2",
     "path-ts": "^2.1.0",
-    "spliterator": "^6.0.1"
+    "spliterator": "^6.1.0"
   },
   "devDependencies": {
     "@types/node": ">=26.1.2"

package/src/adapters/ban/adapter.ts

@@ -31,11 +31,11 @@
  *   country !== FR).
  */
 
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 import { decomposeFrStreet } from "./street-decompose.ts"
 
 /**

package/src/adapters/fcc-bdc/adapter.ts

@@ -36,10 +36,10 @@
 
 import { DatabaseSync } from "node:sqlite"
 
-import { splitStreetLine } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
+import { splitStreetLine } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
 
 /**
  * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset

package/src/adapters/geonames/adapter.ts

@@ -30,13 +30,12 @@
 
 import { existsSync } from "node:fs"
 
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { dirname, join } from "path-ts"
 import { TSVSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/geonames-postal/adapter.ts

@@ -23,12 +23,11 @@
  *        `"CC-BY-4.0"` per row (attribute "GeoNames").
  */
 
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { TSVSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/gnaf/adapter.ts

@@ -29,12 +29,11 @@
  */
 
 import { tryParsingJSON } from "@mailwoman/core/objects"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { TextSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/gnaf/assemble.ts

@@ -27,7 +27,7 @@
  */
 
 import { createWriteStream } from "node:fs"
-import { readdir, readFile } from "node:fs/promises"
+import { readdir } from "node:fs/promises"
 import { join } from "node:path"
 
 import { tryParsingJSON } from "@mailwoman/core/objects"

package/src/adapters/index.ts

@@ -19,8 +19,9 @@
  *   implementations; the registered ids are unchanged).
  */
 
-import { defaultAdapterRegistry } from "../adapter.ts"
-import type { CorpusAdapter } from "../types.ts"
+import { defaultAdapterRegistry } from "@mailwoman/corpus/adapters/utils"
+import type { CorpusAdapter } from "@mailwoman/corpus/types"
+
 import { banAdapter } from "./ban/adapter.ts"
 import { fccBdcAdapter } from "./fcc-bdc/adapter.ts"
 import { geonamesPostalAdapter } from "./geonames-postal/adapter.ts"

package/src/adapters/openaddresses/adapter.ts

@@ -37,13 +37,12 @@
  */
 
 import { tryParsingJSON } from "@mailwoman/core/objects"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { SHARE_ALIKE_PATTERN } from "@mailwoman/corpus/utils"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
 import { TextSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import { SHARE_ALIKE_PATTERN } from "../../license.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/overture/adapter.ts

@@ -31,12 +31,11 @@
  */
 
 import { tryParsingJSON } from "@mailwoman/core/objects"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
 import { TextSpliterator } from "spliterator"
 
-import { stableSourceID } from "../../adapter.ts"
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/state-hi-schools/adapter.ts

@@ -29,13 +29,12 @@
  *   License: stamped `"Public Domain"` per Hawaii state government open-data terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/state-ia-contractors/adapter.ts

@@ -16,13 +16,12 @@
  *   License: stamped `"Public Domain"` per Iowa state government open-data terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/state-ny-notaries/adapter.ts

@@ -14,13 +14,12 @@
  *   License: stamped `"Public Domain"` per New York state government open-data terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/state-tx-notaries/adapter.ts

@@ -15,13 +15,12 @@
  *   License: stamped `"Public Domain"` per Texas state government open-data terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/synth-po-box/adapter.ts

@@ -7,7 +7,7 @@
  *
  *   Consumes a JSONL stream of (locality, region, postcode, country) tuples — typically extracted
  *   from existing corpus output (TIGER/NAD/BAN/WOF) — and emits synthetic PO box training rows. See
- *   `../../synthesize-po-box.ts` for the per-locale templates and number-noise logic.
+ *   `@mailwoman/corpus/synthesizers/po-box` for the per-locale templates and number-noise logic.
  *
  *   Why an adapter and not an augmenter:
  *
@@ -22,17 +22,16 @@
 
 import { tryParsingJSON } from "@mailwoman/core/objects"
 import { makeLcg } from "@mailwoman/core/utils"
-import { TextSpliterator } from "spliterator"
-
-import { stableSourceID } from "../../adapter.ts"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
 import {
 	countryToLocale,
 	REGION_OPTIONAL_LOCALES,
 	synthesizeMilitaryPoBoxRow,
 	synthesizePoBoxRow,
 	type PoBoxBaseTuple,
-} from "../../synthesize-po-box.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
+} from "@mailwoman/corpus/synthesizers/po-box"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { TextSpliterator } from "spliterator"
 
 /**
  * Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset

package/src/adapters/tiger/adapter.ts

@@ -36,9 +36,10 @@
 
 import { DatabaseSync } from "node:sqlite"
 
-import { lookupFipsState } from "../../codex/us-fips-state.ts"
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
+import { lookupFipsState } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
+
 import { decomposeStreet } from "./street-decompose.ts"
 
 /**

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

@@ -29,13 +29,12 @@
  *   terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

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

@@ -18,13 +18,12 @@
  *   License: stamped `"Public Domain"` per IMLS federal government distribution terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

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

@@ -19,12 +19,11 @@
  *   federal).
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/usgov-nad/adapter.ts

@@ -34,11 +34,10 @@ import { readdir } from "node:fs/promises"
 import { join } from "node:path"
 
 import { tryParsingJSON } from "@mailwoman/core/objects"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { TextSpliterator } from "spliterator"
 
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/usgov-nppes/adapter.ts

@@ -21,13 +21,12 @@
  *   License: stamped `"Public Domain"` per CMS's federal government distribution terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/adapters/usgov-samhsa-treatment-locator/adapter.ts

@@ -28,13 +28,12 @@
  *   distribution terms.
  */
 
+import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { reconcileComponents } from "@mailwoman/formatter"
 import { CSVSpliterator } from "spliterator"
 
-import { splitStreetLine, stableSourceID } from "../../adapter.ts"
-import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
-import { reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-
 /**
  * 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.

package/src/{adapter.ts → adapters/utils/index.ts}

@@ -26,8 +26,7 @@ import { createHash, type Hash } from "node:crypto"
 
 import type { ComponentTag } from "@mailwoman/core/types"
 import { sha256Hex } from "@mailwoman/core/utils"
-
-import type { CanonicalRow, CorpusAdapter } from "./types.ts"
+import type { CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
 
 /**
  * Lookup table for corpus adapters.

package/src/adapters/wof-admin-jp/adapter.ts

@@ -31,7 +31,7 @@
 
 import { DatabaseSync } from "node:sqlite"
 
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
 
 /**
  * Ancestry hops walked upward from a WOF record before giving up — deeper than any real JP admin chain.

package/src/adapters/wof-admin-json/adapter.ts

@@ -41,10 +41,9 @@
 
 import type { WhosOnFirstPlacetype } from "@mailwoman/core/resources/whosonfirst"
 import type { ComponentTag } from "@mailwoman/core/types"
-
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-import { buildAncestryIndex, normalizeNameKey, walkFeatures, type WOFRecord } from "../../wof-json.ts"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { buildAncestryIndex, normalizeNameKey, walkFeatures, type WOFRecord } from "@mailwoman/corpus/utils"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
 
 /**
  * Display name for the country, keyed by ISO 3166-1 alpha-2.

package/src/adapters/wof-postalcode-json/adapter.ts

@@ -39,10 +39,9 @@
 
 import type { WhosOnFirstPlacetype } from "@mailwoman/core/resources/whosonfirst"
 import type { ComponentTag } from "@mailwoman/core/types"
-
-import { formatAddress, reconcileComponents } from "../../format.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
-import { buildAncestryIndex, normalizeNameKey, walkFeatures, type WOFRecord } from "../../wof-json.ts"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "@mailwoman/corpus/types"
+import { buildAncestryIndex, normalizeNameKey, walkFeatures, type WOFRecord } from "@mailwoman/corpus/utils"
+import { formatAddress, reconcileComponents } from "@mailwoman/formatter"
 
 const COUNTRY_DISPLAY_NAME: Record<string, string> = {
 	US: "United States of America",

package/src/build.ts

@@ -53,22 +53,22 @@ import { join } from "node:path"
 
 import { $public } from "@mailwoman/core/env"
 import { parseJSONStrict } from "@mailwoman/core/objects"
-import { JSONSpliterator } from "spliterator"
-
-import { defaultAdapterRegistry } from "./adapter.ts"
-import { alignRow } from "./align.ts"
-import { licenseExcluded } from "./license.ts"
-import { writeShards, type ShardManifest } from "./parquet.ts"
-import { runAdapter, type AdapterRunManifest } from "./runner.ts"
+import { defaultAdapterRegistry } from "@mailwoman/corpus/adapters/utils"
+import { runAdapter, type AdapterRunManifest } from "@mailwoman/corpus/runner"
+import { defaultAugmentationsForCountry, synthesizeRow } from "@mailwoman/corpus/synthesizers/utils"
+import type { AdapterOptions, CanonicalRow, CorpusAdapter, LabeledRow } from "@mailwoman/corpus/types"
 import {
+	alignRow,
+	licenseExcluded,
+	writeShards,
 	defaultHoldouts,
 	splitForRow,
 	writeSplitManifestsFromLabeledFiles,
+	type ShardManifest,
 	type SplitManifest,
 	type SplitName,
-} from "./split.ts"
-import { defaultAugmentationsForCountry, synthesizeRow } from "./synthesize.ts"
-import type { AdapterOptions, CanonicalRow, CorpusAdapter, LabeledRow } from "./types.ts"
+} from "@mailwoman/corpus/utils"
+import { JSONSpliterator } from "spliterator"
 
 /**
  * Stage tags surfaced to `onProgress`.

package/src/index.ts

@@ -4,25 +4,11 @@
  * @author Teffen Ellis, et al.
  */
 
-export * from "./adapter.ts"
+export * from "./adapters/utils/index.ts"
 export * from "./adapters/gnaf/assemble.ts"
 export * from "./adapters/index.ts"
-export * from "./align.ts"
 export * from "./build.ts"
-export * from "./format.ts"
-export * from "./golden.ts"
-export * from "./parquet.ts"
-export * from "./parquet-wrapper/index.ts"
 export * from "./runner.ts"
+export * from "./parquet-wrapper/index.ts"
 export * from "./shard-recipes/index.ts"
-export * from "./split.ts"
-export * from "./synthesize-anchor-absorption.ts"
-export * from "./synthesize-german.ts"
-export * from "./synthesize-house-venue.ts"
-export * from "./synthesize-intersection.ts"
-export * from "./synthesize-no-street.ts"
-export * from "./synthesize-po-box.ts"
-export * from "./synthesize-street.ts"
-export * from "./synthesize.ts"
-export * from "./tokenize.ts"
-export * from "./types.ts"
+export * from "@mailwoman/corpus/types"