@mailwoman/corpus 9.2.0 → 9.4.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +22 -22
- package/data/PROVENANCE.md +7 -7
- package/lib/adapters/ban/adapter.ts +190 -0
- package/lib/adapters/ban/street-decompose.ts +66 -0
- package/lib/adapters/fcc-bdc/adapter.ts +174 -0
- package/lib/adapters/geonames/adapter.ts +199 -0
- package/lib/adapters/geonames-postal/adapter.ts +122 -0
- package/lib/adapters/gnaf/adapter.ts +149 -0
- package/lib/adapters/gnaf/assemble.ts +225 -0
- package/lib/adapters/index.ts +153 -0
- package/lib/adapters/openaddresses/adapter.ts +251 -0
- package/lib/adapters/overture/adapter.ts +162 -0
- package/lib/adapters/state-hi-schools/adapter.ts +170 -0
- package/lib/adapters/state-hi-schools/workbook.ts +35 -0
- package/lib/adapters/synth-po-box/adapter.ts +204 -0
- package/lib/adapters/tiger/adapter.ts +252 -0
- package/lib/adapters/tiger/street-decompose.ts +83 -0
- package/lib/adapters/usgov-hrsa-fqhc/adapter.ts +167 -0
- package/lib/adapters/usgov-imls-pls/adapter.ts +143 -0
- package/lib/adapters/usgov-irs-bmf/adapter.ts +151 -0
- package/lib/adapters/usgov-nad/adapter.ts +334 -0
- package/lib/adapters/usgov-nppes/adapter.ts +151 -0
- package/lib/adapters/usgov-samhsa-treatment-locator/adapter.ts +193 -0
- package/lib/adapters/utils/index.ts +307 -0
- package/lib/adapters/wof-admin-jp/adapter.ts +214 -0
- package/lib/adapters/wof-admin-json/adapter.ts +231 -0
- package/lib/adapters/wof-json-rows.ts +129 -0
- package/lib/adapters/wof-postalcode-json/adapter.ts +176 -0
- package/lib/build.ts +378 -0
- package/lib/env.ts +50 -0
- package/lib/index.ts +13 -0
- package/lib/recipes/anchor-absorption.ts +101 -0
- package/lib/recipes/bare-country.ts +95 -0
- package/lib/recipes/boundary-stress.ts +102 -0
- package/lib/recipes/cn-organizational-units.ts +199 -0
- package/lib/recipes/country-balanced.ts +510 -0
- package/lib/recipes/cz-pcfirst-preposition.ts +112 -0
- package/lib/recipes/fr-admin-split.ts +243 -0
- package/lib/recipes/fr-bare-street.ts +171 -0
- package/lib/recipes/fr-fragment.ts +334 -0
- package/lib/recipes/fr-lieudit.ts +301 -0
- package/lib/recipes/fr-order.ts +246 -0
- package/lib/recipes/german.ts +360 -0
- package/lib/recipes/house-venue.ts +93 -0
- package/lib/recipes/index.ts +97 -0
- package/lib/recipes/intersection.ts +619 -0
- package/lib/recipes/locale.ts +600 -0
- package/lib/recipes/nl-postcode.ts +117 -0
- package/lib/recipes/no-fragment.ts +213 -0
- package/lib/recipes/no-street-led.ts +135 -0
- package/lib/recipes/no-street.ts +91 -0
- package/lib/recipes/po-box-cedex.ts +976 -0
- package/lib/recipes/po-box.ts +124 -0
- package/lib/recipes/reviewed-postcode-tail.ts +190 -0
- package/lib/recipes/scaffold.ts +468 -0
- package/lib/recipes/si-bare-village.ts +99 -0
- package/lib/recipes/street-affix.ts +805 -0
- package/lib/recipes/street-bare.ts +102 -0
- package/lib/recipes/street.ts +83 -0
- package/lib/recipes/sub-venue-sources.ts +878 -0
- package/lib/recipes/sub-venue.ts +998 -0
- package/lib/recipes/trailing-region.ts +178 -0
- package/lib/recipes/unit.ts +304 -0
- package/lib/runner.ts +312 -0
- package/lib/synthesizers/anchor-absorption.ts +288 -0
- package/lib/synthesizers/boundary-stress.ts +532 -0
- package/lib/synthesizers/german.ts +260 -0
- package/lib/synthesizers/house-venue.ts +308 -0
- package/lib/synthesizers/intersection.ts +233 -0
- package/lib/synthesizers/no-street.ts +327 -0
- package/lib/synthesizers/po-box.ts +323 -0
- package/lib/synthesizers/street.ts +269 -0
- package/lib/synthesizers/utils.ts +993 -0
- package/lib/test-kit/corpus-recipe.ts +86 -0
- package/lib/test-kit/index.ts +100 -0
- package/lib/tools/align-slice.ts +68 -0
- package/lib/tools/audit.ts +403 -0
- package/lib/tools/corpus-stats.ts +136 -0
- package/lib/tools/fetch/acra-sg.ts +174 -0
- package/lib/tools/fetch/ban.ts +289 -0
- package/lib/tools/fetch/download.ts +477 -0
- package/lib/tools/fetch/gcis-tw.ts +175 -0
- package/lib/tools/fetch/geonames-dump.ts +257 -0
- package/lib/tools/fetch/geonames-postal.ts +180 -0
- package/lib/tools/fetch/houjin-jp.ts +135 -0
- package/lib/tools/fetch/hrsa.ts +59 -0
- package/lib/tools/fetch/imls-pls.ts +149 -0
- package/lib/tools/fetch/index.ts +174 -0
- package/lib/tools/fetch/juso-kr.ts +165 -0
- package/lib/tools/fetch/localdata-kr.ts +194 -0
- package/lib/tools/fetch/nad.ts +341 -0
- package/lib/tools/fetch/nppes.ts +165 -0
- package/lib/tools/fetch/openaddresses.ts +347 -0
- package/lib/tools/fetch/ourairports.ts +170 -0
- package/lib/tools/fetch/ppd.ts +121 -0
- package/lib/tools/fetch/state-hi-schools.ts +161 -0
- package/lib/tools/fetch/state-sources.ts +146 -0
- package/lib/tools/fetch/tiger-full.ts +308 -0
- package/lib/tools/fetch/wikidata-subvenue.ts +384 -0
- package/lib/tools/golden-expand.ts +574 -0
- package/lib/tools/golden-promote.ts +337 -0
- package/lib/tools/golden-relabel-street.ts +686 -0
- package/lib/tools/index.ts +28 -0
- package/lib/tools/ingest-csv.ts +378 -0
- package/lib/tools/jsonl-to-parquet.ts +207 -0
- package/lib/tools/lint-slice-vocab.ts +388 -0
- package/lib/tools/lint-slice.ts +496 -0
- package/lib/tools/overlay-manifest.ts +148 -0
- package/lib/tools/overture-subvenue.ts +209 -0
- package/lib/tools/postcode-triples.ts +363 -0
- package/lib/tools/slice-kryptonite.ts +137 -0
- package/lib/tools/slice-translit.ts +218 -0
- package/lib/tools/sub-venue/harvest.ts +234 -0
- package/lib/tools/sub-venue/head-nouns.ts +245 -0
- package/lib/tools/sub-venue/surfaces.ts +95 -0
- package/lib/tools/sub-venue/table.ts +281 -0
- package/lib/tools/sub-venue/wikidata.ts +85 -0
- package/lib/tools/sub-venue-lexicon.ts +429 -0
- package/lib/tools/sub-venue-promotions.ts +330 -0
- package/lib/types.ts +247 -0
- package/lib/utils/align.ts +330 -0
- package/lib/utils/golden.ts +189 -0
- package/lib/utils/index.ts +14 -0
- package/lib/utils/parquet.ts +582 -0
- package/lib/utils/slice-stats.ts +95 -0
- package/lib/utils/split.ts +279 -0
- package/lib/utils/tokenize.ts +120 -0
- package/lib/utils/wof-json.ts +247 -0
- package/out/adapters/ban/adapter.d.ts.map +1 -0
- package/out/adapters/ban/adapter.js +149 -0
- package/out/adapters/ban/adapter.js.map +1 -0
- package/out/adapters/ban/street-decompose.d.ts.map +1 -0
- package/out/adapters/ban/street-decompose.js +51 -0
- package/out/adapters/ban/street-decompose.js.map +1 -0
- package/out/adapters/fcc-bdc/adapter.d.ts.map +1 -0
- package/out/adapters/fcc-bdc/adapter.js +144 -0
- package/out/adapters/fcc-bdc/adapter.js.map +1 -0
- package/out/adapters/geonames/adapter.d.ts.map +1 -0
- package/out/adapters/geonames/adapter.js +176 -0
- package/out/adapters/geonames/adapter.js.map +1 -0
- package/out/adapters/geonames-postal/adapter.d.ts +54 -0
- package/out/adapters/geonames-postal/adapter.d.ts.map +1 -0
- package/out/adapters/geonames-postal/adapter.js +107 -0
- package/out/adapters/geonames-postal/adapter.js.map +1 -0
- package/out/adapters/gnaf/adapter.d.ts +49 -0
- package/out/adapters/gnaf/adapter.d.ts.map +1 -0
- package/out/adapters/gnaf/adapter.js +124 -0
- package/out/adapters/gnaf/adapter.js.map +1 -0
- package/out/adapters/gnaf/assemble.d.ts +65 -0
- package/out/adapters/gnaf/assemble.d.ts.map +1 -0
- package/out/adapters/gnaf/assemble.js +154 -0
- package/out/adapters/gnaf/assemble.js.map +1 -0
- package/out/adapters/index.d.ts +51 -0
- package/out/adapters/index.d.ts.map +1 -0
- package/out/adapters/index.js +95 -0
- package/out/adapters/index.js.map +1 -0
- package/out/adapters/openaddresses/adapter.d.ts.map +1 -0
- package/out/adapters/openaddresses/adapter.js +186 -0
- package/out/adapters/openaddresses/adapter.js.map +1 -0
- package/out/adapters/overture/adapter.d.ts +48 -0
- package/out/adapters/overture/adapter.d.ts.map +1 -0
- package/out/adapters/overture/adapter.js +125 -0
- package/out/adapters/overture/adapter.js.map +1 -0
- package/out/adapters/state-hi-schools/adapter.d.ts +46 -0
- package/out/adapters/state-hi-schools/adapter.d.ts.map +1 -0
- package/out/adapters/state-hi-schools/adapter.js +145 -0
- package/out/adapters/state-hi-schools/adapter.js.map +1 -0
- package/out/adapters/state-hi-schools/workbook.d.ts +29 -0
- package/out/adapters/state-hi-schools/workbook.d.ts.map +1 -0
- package/out/adapters/state-hi-schools/workbook.js +20 -0
- package/out/adapters/state-hi-schools/workbook.js.map +1 -0
- package/out/adapters/state-ia-contractors/adapter.d.ts.map +1 -0
- package/out/adapters/state-ia-contractors/adapter.js.map +1 -0
- package/out/adapters/state-ny-notaries/adapter.d.ts.map +1 -0
- package/out/adapters/state-ny-notaries/adapter.js.map +1 -0
- package/out/adapters/state-tx-notaries/adapter.d.ts.map +1 -0
- package/out/adapters/state-tx-notaries/adapter.js.map +1 -0
- package/out/adapters/synth-po-box/adapter.d.ts.map +1 -0
- package/out/adapters/synth-po-box/adapter.js +140 -0
- package/out/adapters/synth-po-box/adapter.js.map +1 -0
- package/out/adapters/tiger/adapter.d.ts +55 -0
- package/out/adapters/tiger/adapter.d.ts.map +1 -0
- package/out/adapters/tiger/adapter.js +201 -0
- package/out/adapters/tiger/adapter.js.map +1 -0
- package/out/adapters/tiger/street-decompose.d.ts.map +1 -0
- package/out/adapters/tiger/street-decompose.js +68 -0
- package/out/adapters/tiger/street-decompose.js.map +1 -0
- package/out/adapters/usgov-hrsa-fqhc/adapter.d.ts +56 -0
- package/out/adapters/usgov-hrsa-fqhc/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-hrsa-fqhc/adapter.js +132 -0
- package/out/adapters/usgov-hrsa-fqhc/adapter.js.map +1 -0
- package/out/adapters/usgov-imls-pls/adapter.d.ts +36 -0
- package/out/adapters/usgov-imls-pls/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-imls-pls/adapter.js +114 -0
- package/out/adapters/usgov-imls-pls/adapter.js.map +1 -0
- package/out/adapters/usgov-irs-bmf/adapter.d.ts +37 -0
- package/out/adapters/usgov-irs-bmf/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-irs-bmf/adapter.js +120 -0
- package/out/adapters/usgov-irs-bmf/adapter.js.map +1 -0
- package/out/adapters/usgov-nad/adapter.d.ts +48 -0
- package/out/adapters/usgov-nad/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-nad/adapter.js +249 -0
- package/out/adapters/usgov-nad/adapter.js.map +1 -0
- package/out/adapters/usgov-nppes/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-nppes/adapter.js +118 -0
- package/out/adapters/usgov-nppes/adapter.js.map +1 -0
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map +1 -0
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.js +154 -0
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.js.map +1 -0
- package/out/adapters/utils/index.d.ts +191 -0
- package/out/adapters/utils/index.d.ts.map +1 -0
- package/out/adapters/utils/index.js +206 -0
- package/out/adapters/utils/index.js.map +1 -0
- package/out/adapters/wof-admin-jp/adapter.d.ts.map +1 -0
- package/out/adapters/wof-admin-jp/adapter.js +161 -0
- package/out/adapters/wof-admin-jp/adapter.js.map +1 -0
- package/out/adapters/wof-admin-json/adapter.d.ts +79 -0
- package/out/adapters/wof-admin-json/adapter.d.ts.map +1 -0
- package/out/adapters/wof-admin-json/adapter.js +197 -0
- package/out/adapters/wof-admin-json/adapter.js.map +1 -0
- package/out/adapters/wof-json-rows.d.ts +62 -0
- package/out/adapters/wof-json-rows.d.ts.map +1 -0
- package/out/adapters/wof-json-rows.js +83 -0
- package/out/adapters/wof-json-rows.js.map +1 -0
- package/out/adapters/wof-postalcode-json/adapter.d.ts +65 -0
- package/out/adapters/wof-postalcode-json/adapter.d.ts.map +1 -0
- package/out/adapters/wof-postalcode-json/adapter.js +151 -0
- package/out/adapters/wof-postalcode-json/adapter.js.map +1 -0
- package/out/build.d.ts +132 -0
- package/out/build.d.ts.map +1 -0
- package/out/build.js +245 -0
- package/out/build.js.map +1 -0
- package/out/codex/us-fips-state.d.ts.map +1 -0
- package/out/codex/us-fips-state.js.map +1 -0
- package/out/env.d.ts +46 -0
- package/out/env.d.ts.map +1 -0
- package/out/env.js +46 -0
- package/out/env.js.map +1 -0
- package/out/index.d.ts +13 -0
- package/out/index.d.ts.map +1 -0
- package/out/index.js +13 -0
- package/out/index.js.map +1 -0
- package/out/recipes/anchor-absorption.d.ts +20 -0
- package/out/recipes/anchor-absorption.d.ts.map +1 -0
- package/out/recipes/anchor-absorption.js +82 -0
- package/out/recipes/anchor-absorption.js.map +1 -0
- package/out/recipes/bare-country.d.ts +25 -0
- package/out/recipes/bare-country.d.ts.map +1 -0
- package/out/recipes/bare-country.js +88 -0
- package/out/recipes/bare-country.js.map +1 -0
- package/out/recipes/boundary-stress.d.ts +20 -0
- package/out/recipes/boundary-stress.d.ts.map +1 -0
- package/out/recipes/boundary-stress.js +84 -0
- package/out/recipes/boundary-stress.js.map +1 -0
- package/out/recipes/cn-organizational-units.d.ts +36 -0
- package/out/recipes/cn-organizational-units.d.ts.map +1 -0
- package/out/recipes/cn-organizational-units.js +159 -0
- package/out/recipes/cn-organizational-units.js.map +1 -0
- package/out/recipes/country-balanced.d.ts +32 -0
- package/out/recipes/country-balanced.d.ts.map +1 -0
- package/out/recipes/country-balanced.js +394 -0
- package/out/recipes/country-balanced.js.map +1 -0
- package/out/recipes/cz-pcfirst-preposition.d.ts +22 -0
- package/out/recipes/cz-pcfirst-preposition.d.ts.map +1 -0
- package/out/recipes/cz-pcfirst-preposition.js +101 -0
- package/out/recipes/cz-pcfirst-preposition.js.map +1 -0
- package/out/recipes/fr-admin-split.d.ts +38 -0
- package/out/recipes/fr-admin-split.d.ts.map +1 -0
- package/out/recipes/fr-admin-split.js +192 -0
- package/out/recipes/fr-admin-split.js.map +1 -0
- package/out/recipes/fr-bare-street.d.ts +28 -0
- package/out/recipes/fr-bare-street.d.ts.map +1 -0
- package/out/recipes/fr-bare-street.js +144 -0
- package/out/recipes/fr-bare-street.js.map +1 -0
- package/out/recipes/fr-fragment.d.ts +77 -0
- package/out/recipes/fr-fragment.d.ts.map +1 -0
- package/out/recipes/fr-fragment.js +277 -0
- package/out/recipes/fr-fragment.js.map +1 -0
- package/out/recipes/fr-lieudit.d.ts +32 -0
- package/out/recipes/fr-lieudit.d.ts.map +1 -0
- package/out/recipes/fr-lieudit.js +227 -0
- package/out/recipes/fr-lieudit.js.map +1 -0
- package/out/recipes/fr-order.d.ts +38 -0
- package/out/recipes/fr-order.d.ts.map +1 -0
- package/out/recipes/fr-order.js +198 -0
- package/out/recipes/fr-order.js.map +1 -0
- package/out/recipes/german.d.ts +44 -0
- package/out/recipes/german.d.ts.map +1 -0
- package/out/recipes/german.js +289 -0
- package/out/recipes/german.js.map +1 -0
- package/out/recipes/house-venue.d.ts +18 -0
- package/out/recipes/house-venue.d.ts.map +1 -0
- package/out/recipes/house-venue.js +74 -0
- package/out/recipes/house-venue.js.map +1 -0
- package/out/recipes/index.d.ts +24 -0
- package/out/recipes/index.d.ts.map +1 -0
- package/out/recipes/index.js +91 -0
- package/out/recipes/index.js.map +1 -0
- package/out/recipes/intersection.d.ts +42 -0
- package/out/recipes/intersection.d.ts.map +1 -0
- package/out/recipes/intersection.js +453 -0
- package/out/recipes/intersection.js.map +1 -0
- package/out/recipes/locale.d.ts +136 -0
- package/out/recipes/locale.d.ts.map +1 -0
- package/out/recipes/locale.js +460 -0
- package/out/recipes/locale.js.map +1 -0
- package/out/recipes/nl-postcode.d.ts +26 -0
- package/out/recipes/nl-postcode.d.ts.map +1 -0
- package/out/recipes/nl-postcode.js +103 -0
- package/out/recipes/nl-postcode.js.map +1 -0
- package/out/recipes/no-fragment.d.ts +43 -0
- package/out/recipes/no-fragment.d.ts.map +1 -0
- package/out/recipes/no-fragment.js +183 -0
- package/out/recipes/no-fragment.js.map +1 -0
- package/out/recipes/no-street-led.d.ts +24 -0
- package/out/recipes/no-street-led.d.ts.map +1 -0
- package/out/recipes/no-street-led.js +114 -0
- package/out/recipes/no-street-led.js.map +1 -0
- package/out/recipes/no-street.d.ts +18 -0
- package/out/recipes/no-street.d.ts.map +1 -0
- package/out/recipes/no-street.js +72 -0
- package/out/recipes/no-street.js.map +1 -0
- package/out/recipes/po-box-cedex.d.ts +38 -0
- package/out/recipes/po-box-cedex.d.ts.map +1 -0
- package/out/recipes/po-box-cedex.js +768 -0
- package/out/recipes/po-box-cedex.js.map +1 -0
- package/out/recipes/po-box.d.ts +17 -0
- package/out/recipes/po-box.d.ts.map +1 -0
- package/out/recipes/po-box.js +101 -0
- package/out/recipes/po-box.js.map +1 -0
- package/out/recipes/reviewed-postcode-tail.d.ts +47 -0
- package/out/recipes/reviewed-postcode-tail.d.ts.map +1 -0
- package/out/recipes/reviewed-postcode-tail.js +125 -0
- package/out/recipes/reviewed-postcode-tail.js.map +1 -0
- package/out/recipes/scaffold.d.ts +323 -0
- package/out/recipes/scaffold.d.ts.map +1 -0
- package/out/recipes/scaffold.js +174 -0
- package/out/recipes/scaffold.js.map +1 -0
- package/out/recipes/si-bare-village.d.ts +32 -0
- package/out/recipes/si-bare-village.d.ts.map +1 -0
- package/out/recipes/si-bare-village.js +89 -0
- package/out/recipes/si-bare-village.js.map +1 -0
- package/out/recipes/street-affix.d.ts +89 -0
- package/out/recipes/street-affix.d.ts.map +1 -0
- package/out/recipes/street-affix.js +563 -0
- package/out/recipes/street-affix.js.map +1 -0
- package/out/recipes/street-bare.d.ts +25 -0
- package/out/recipes/street-bare.d.ts.map +1 -0
- package/out/recipes/street-bare.js +87 -0
- package/out/recipes/street-bare.js.map +1 -0
- package/out/recipes/street.d.ts +15 -0
- package/out/recipes/street.d.ts.map +1 -0
- package/out/recipes/street.js +70 -0
- package/out/recipes/street.js.map +1 -0
- package/out/recipes/sub-venue-sources.d.ts +229 -0
- package/out/recipes/sub-venue-sources.d.ts.map +1 -0
- package/out/recipes/sub-venue-sources.js +654 -0
- package/out/recipes/sub-venue-sources.js.map +1 -0
- package/out/recipes/sub-venue.d.ts +269 -0
- package/out/recipes/sub-venue.d.ts.map +1 -0
- package/out/recipes/sub-venue.js +726 -0
- package/out/recipes/sub-venue.js.map +1 -0
- package/out/recipes/trailing-region.d.ts +74 -0
- package/out/recipes/trailing-region.d.ts.map +1 -0
- package/out/recipes/trailing-region.js +158 -0
- package/out/recipes/trailing-region.js.map +1 -0
- package/out/recipes/unit.d.ts +29 -0
- package/out/recipes/unit.d.ts.map +1 -0
- package/out/recipes/unit.js +231 -0
- package/out/recipes/unit.js.map +1 -0
- package/out/runner.d.ts +130 -0
- package/out/runner.d.ts.map +1 -0
- package/out/runner.js +185 -0
- package/out/runner.js.map +1 -0
- package/out/synthesizers/anchor-absorption.d.ts +63 -0
- package/out/synthesizers/anchor-absorption.d.ts.map +1 -0
- package/out/synthesizers/anchor-absorption.js +221 -0
- package/out/synthesizers/anchor-absorption.js.map +1 -0
- package/out/synthesizers/boundary-stress.d.ts +78 -0
- package/out/synthesizers/boundary-stress.d.ts.map +1 -0
- package/out/synthesizers/boundary-stress.js +455 -0
- package/out/synthesizers/boundary-stress.js.map +1 -0
- package/out/synthesizers/german.d.ts +115 -0
- package/out/synthesizers/german.d.ts.map +1 -0
- package/out/synthesizers/german.js +149 -0
- package/out/synthesizers/german.js.map +1 -0
- package/out/synthesizers/house-venue.d.ts +56 -0
- package/out/synthesizers/house-venue.d.ts.map +1 -0
- package/out/synthesizers/house-venue.js +247 -0
- package/out/synthesizers/house-venue.js.map +1 -0
- package/out/synthesizers/intersection.d.ts +52 -0
- package/out/synthesizers/intersection.d.ts.map +1 -0
- package/out/synthesizers/intersection.js +175 -0
- package/out/synthesizers/intersection.js.map +1 -0
- package/out/synthesizers/no-street.d.ts +72 -0
- package/out/synthesizers/no-street.d.ts.map +1 -0
- package/out/synthesizers/no-street.js +262 -0
- package/out/synthesizers/no-street.js.map +1 -0
- package/out/synthesizers/po-box.d.ts +108 -0
- package/out/synthesizers/po-box.d.ts.map +1 -0
- package/out/synthesizers/po-box.js +250 -0
- package/out/synthesizers/po-box.js.map +1 -0
- package/out/synthesizers/street.d.ts +54 -0
- package/out/synthesizers/street.d.ts.map +1 -0
- package/out/synthesizers/street.js +218 -0
- package/out/synthesizers/street.js.map +1 -0
- package/out/synthesizers/utils.d.ts +241 -0
- package/out/synthesizers/utils.d.ts.map +1 -0
- package/out/synthesizers/utils.js +776 -0
- package/out/synthesizers/utils.js.map +1 -0
- package/out/test-kit/corpus-recipe.d.ts +59 -0
- package/out/test-kit/corpus-recipe.d.ts.map +1 -0
- package/out/test-kit/corpus-recipe.js +39 -0
- package/out/test-kit/corpus-recipe.js.map +1 -0
- package/out/test-kit/index.d.ts +48 -0
- package/out/test-kit/index.d.ts.map +1 -0
- package/out/test-kit/index.js +78 -0
- package/out/test-kit/index.js.map +1 -0
- package/out/tools/align-slice.d.ts +7 -0
- package/out/tools/align-slice.d.ts.map +1 -0
- package/out/tools/align-slice.js +55 -0
- package/out/tools/align-slice.js.map +1 -0
- package/out/tools/audit.d.ts +29 -0
- package/out/tools/audit.d.ts.map +1 -0
- package/out/tools/audit.js +282 -0
- package/out/tools/audit.js.map +1 -0
- package/out/tools/corpus-stats.d.ts +44 -0
- package/out/tools/corpus-stats.d.ts.map +1 -0
- package/out/tools/corpus-stats.js +102 -0
- package/out/tools/corpus-stats.js.map +1 -0
- package/out/tools/fetch/acra-sg.d.ts +23 -0
- package/out/tools/fetch/acra-sg.d.ts.map +1 -0
- package/out/tools/fetch/acra-sg.js +126 -0
- package/out/tools/fetch/acra-sg.js.map +1 -0
- package/out/tools/fetch/ban.d.ts +31 -0
- package/out/tools/fetch/ban.d.ts.map +1 -0
- package/out/tools/fetch/ban.js +239 -0
- package/out/tools/fetch/ban.js.map +1 -0
- package/out/tools/fetch/download.d.ts +182 -0
- package/out/tools/fetch/download.d.ts.map +1 -0
- package/out/tools/fetch/download.js +318 -0
- package/out/tools/fetch/download.js.map +1 -0
- package/out/tools/fetch/gcis-tw.d.ts +23 -0
- package/out/tools/fetch/gcis-tw.d.ts.map +1 -0
- package/out/tools/fetch/gcis-tw.js +135 -0
- package/out/tools/fetch/gcis-tw.js.map +1 -0
- package/out/tools/fetch/geonames-dump.d.ts +91 -0
- package/out/tools/fetch/geonames-dump.d.ts.map +1 -0
- package/out/tools/fetch/geonames-dump.js +171 -0
- package/out/tools/fetch/geonames-dump.js.map +1 -0
- package/out/tools/fetch/geonames-postal.d.ts +76 -0
- package/out/tools/fetch/geonames-postal.d.ts.map +1 -0
- package/out/tools/fetch/geonames-postal.js +126 -0
- package/out/tools/fetch/geonames-postal.js.map +1 -0
- package/out/tools/fetch/houjin-jp.d.ts +24 -0
- package/out/tools/fetch/houjin-jp.d.ts.map +1 -0
- package/out/tools/fetch/houjin-jp.js +102 -0
- package/out/tools/fetch/houjin-jp.js.map +1 -0
- package/out/tools/fetch/hrsa.d.ts +18 -0
- package/out/tools/fetch/hrsa.d.ts.map +1 -0
- package/out/tools/fetch/hrsa.js +46 -0
- package/out/tools/fetch/hrsa.js.map +1 -0
- package/out/tools/fetch/imls-pls.d.ts +25 -0
- package/out/tools/fetch/imls-pls.d.ts.map +1 -0
- package/out/tools/fetch/imls-pls.js +108 -0
- package/out/tools/fetch/imls-pls.js.map +1 -0
- package/out/tools/fetch/index.d.ts +171 -0
- package/out/tools/fetch/index.d.ts.map +1 -0
- package/out/tools/fetch/index.js +170 -0
- package/out/tools/fetch/index.js.map +1 -0
- package/out/tools/fetch/juso-kr.d.ts +30 -0
- package/out/tools/fetch/juso-kr.d.ts.map +1 -0
- package/out/tools/fetch/juso-kr.js +120 -0
- package/out/tools/fetch/juso-kr.js.map +1 -0
- package/out/tools/fetch/localdata-kr.d.ts +30 -0
- package/out/tools/fetch/localdata-kr.d.ts.map +1 -0
- package/out/tools/fetch/localdata-kr.js +139 -0
- package/out/tools/fetch/localdata-kr.js.map +1 -0
- package/out/tools/fetch/nad.d.ts +70 -0
- package/out/tools/fetch/nad.d.ts.map +1 -0
- package/out/tools/fetch/nad.js +234 -0
- package/out/tools/fetch/nad.js.map +1 -0
- package/out/tools/fetch/nppes.d.ts +26 -0
- package/out/tools/fetch/nppes.d.ts.map +1 -0
- package/out/tools/fetch/nppes.js +126 -0
- package/out/tools/fetch/nppes.js.map +1 -0
- package/out/tools/fetch/openaddresses.d.ts +9 -0
- package/out/tools/fetch/openaddresses.d.ts.map +1 -0
- package/out/tools/fetch/openaddresses.js +279 -0
- package/out/tools/fetch/openaddresses.js.map +1 -0
- package/out/tools/fetch/ourairports.d.ts +45 -0
- package/out/tools/fetch/ourairports.d.ts.map +1 -0
- package/out/tools/fetch/ourairports.js +128 -0
- package/out/tools/fetch/ourairports.js.map +1 -0
- package/out/tools/fetch/ppd.d.ts.map +1 -0
- package/out/tools/fetch/ppd.js +85 -0
- package/out/tools/fetch/ppd.js.map +1 -0
- package/out/tools/fetch/state-hi-schools.d.ts +26 -0
- package/out/tools/fetch/state-hi-schools.d.ts.map +1 -0
- package/out/tools/fetch/state-hi-schools.js +112 -0
- package/out/tools/fetch/state-hi-schools.js.map +1 -0
- package/out/tools/fetch/state-sources.d.ts +26 -0
- package/out/tools/fetch/state-sources.d.ts.map +1 -0
- package/out/tools/fetch/state-sources.js +111 -0
- package/out/tools/fetch/state-sources.js.map +1 -0
- package/out/tools/fetch/tiger-full.d.ts +22 -0
- package/out/tools/fetch/tiger-full.d.ts.map +1 -0
- package/out/tools/fetch/tiger-full.js +217 -0
- package/out/tools/fetch/tiger-full.js.map +1 -0
- package/out/tools/fetch/wikidata-subvenue.d.ts +171 -0
- package/out/tools/fetch/wikidata-subvenue.d.ts.map +1 -0
- package/out/tools/fetch/wikidata-subvenue.js +274 -0
- package/out/tools/fetch/wikidata-subvenue.js.map +1 -0
- package/out/tools/gb-title-case.d.ts.map +1 -0
- package/out/tools/gb-title-case.js.map +1 -0
- package/out/tools/golden-expand.d.ts +90 -0
- package/out/tools/golden-expand.d.ts.map +1 -0
- package/out/tools/golden-expand.js +406 -0
- package/out/tools/golden-expand.js.map +1 -0
- package/out/tools/golden-promote.d.ts.map +1 -0
- package/out/tools/golden-promote.js +240 -0
- package/out/tools/golden-promote.js.map +1 -0
- package/out/tools/golden-relabel-street.d.ts +196 -0
- package/out/tools/golden-relabel-street.d.ts.map +1 -0
- package/out/tools/golden-relabel-street.js +439 -0
- package/out/tools/golden-relabel-street.js.map +1 -0
- package/out/tools/index.d.ts +28 -0
- package/out/tools/index.d.ts.map +1 -0
- package/out/tools/index.js +28 -0
- package/out/tools/index.js.map +1 -0
- package/out/tools/ingest-csv.d.ts.map +1 -0
- package/out/tools/ingest-csv.js +265 -0
- package/out/tools/ingest-csv.js.map +1 -0
- package/out/tools/jsonl-to-parquet.d.ts +61 -0
- package/out/tools/jsonl-to-parquet.d.ts.map +1 -0
- package/out/tools/jsonl-to-parquet.js +150 -0
- package/out/tools/jsonl-to-parquet.js.map +1 -0
- package/out/tools/lint-slice-vocab.d.ts +90 -0
- package/out/tools/lint-slice-vocab.d.ts.map +1 -0
- package/out/tools/lint-slice-vocab.js +267 -0
- package/out/tools/lint-slice-vocab.js.map +1 -0
- package/out/tools/lint-slice.d.ts +93 -0
- package/out/tools/lint-slice.d.ts.map +1 -0
- package/out/tools/lint-slice.js +322 -0
- package/out/tools/lint-slice.js.map +1 -0
- package/out/tools/overlay-manifest.d.ts +39 -0
- package/out/tools/overlay-manifest.d.ts.map +1 -0
- package/out/tools/overlay-manifest.js +90 -0
- package/out/tools/overlay-manifest.js.map +1 -0
- package/out/tools/overture-subvenue.d.ts +112 -0
- package/out/tools/overture-subvenue.d.ts.map +1 -0
- package/out/tools/overture-subvenue.js +139 -0
- package/out/tools/overture-subvenue.js.map +1 -0
- package/out/tools/postcode-triples.d.ts +174 -0
- package/out/tools/postcode-triples.d.ts.map +1 -0
- package/out/tools/postcode-triples.js +292 -0
- package/out/tools/postcode-triples.js.map +1 -0
- package/out/tools/slice-kryptonite.d.ts +32 -0
- package/out/tools/slice-kryptonite.d.ts.map +1 -0
- package/out/tools/slice-kryptonite.js +99 -0
- package/out/tools/slice-kryptonite.js.map +1 -0
- package/out/tools/slice-translit.d.ts +51 -0
- package/out/tools/slice-translit.d.ts.map +1 -0
- package/out/tools/slice-translit.js +166 -0
- package/out/tools/slice-translit.js.map +1 -0
- package/out/tools/sub-venue/harvest.d.ts +100 -0
- package/out/tools/sub-venue/harvest.d.ts.map +1 -0
- package/out/tools/sub-venue/harvest.js +168 -0
- package/out/tools/sub-venue/harvest.js.map +1 -0
- package/out/tools/sub-venue/head-nouns.d.ts +50 -0
- package/out/tools/sub-venue/head-nouns.d.ts.map +1 -0
- package/out/tools/sub-venue/head-nouns.js +213 -0
- package/out/tools/sub-venue/head-nouns.js.map +1 -0
- package/out/tools/sub-venue/surfaces.d.ts +45 -0
- package/out/tools/sub-venue/surfaces.d.ts.map +1 -0
- package/out/tools/sub-venue/surfaces.js +77 -0
- package/out/tools/sub-venue/surfaces.js.map +1 -0
- package/out/tools/sub-venue/table.d.ts +229 -0
- package/out/tools/sub-venue/table.d.ts.map +1 -0
- package/out/tools/sub-venue/table.js +108 -0
- package/out/tools/sub-venue/table.js.map +1 -0
- package/out/tools/sub-venue/wikidata.d.ts +26 -0
- package/out/tools/sub-venue/wikidata.d.ts.map +1 -0
- package/out/tools/sub-venue/wikidata.js +62 -0
- package/out/tools/sub-venue/wikidata.js.map +1 -0
- package/out/tools/sub-venue-lexicon.d.ts +156 -0
- package/out/tools/sub-venue-lexicon.d.ts.map +1 -0
- package/out/tools/sub-venue-lexicon.js +283 -0
- package/out/tools/sub-venue-lexicon.js.map +1 -0
- package/out/tools/sub-venue-promotions.d.ts +94 -0
- package/out/tools/sub-venue-promotions.d.ts.map +1 -0
- package/out/tools/sub-venue-promotions.js +266 -0
- package/out/tools/sub-venue-promotions.js.map +1 -0
- package/out/types.d.ts +221 -0
- package/out/types.d.ts.map +1 -0
- package/out/types.js.map +1 -0
- package/out/utils/align.d.ts +93 -0
- package/out/utils/align.d.ts.map +1 -0
- package/out/utils/align.js +234 -0
- package/out/utils/align.js.map +1 -0
- package/out/utils/golden.d.ts +85 -0
- package/out/utils/golden.d.ts.map +1 -0
- package/out/utils/golden.js +127 -0
- package/out/utils/golden.js.map +1 -0
- package/out/utils/index.d.ts +14 -0
- package/out/utils/index.d.ts.map +1 -0
- package/out/utils/index.js +14 -0
- package/out/utils/index.js.map +1 -0
- package/out/utils/license.d.ts.map +1 -0
- package/out/utils/license.js.map +1 -0
- package/out/utils/parquet.d.ts +178 -0
- package/out/utils/parquet.d.ts.map +1 -0
- package/out/utils/parquet.js +371 -0
- package/out/utils/parquet.js.map +1 -0
- package/out/utils/slice-stats.d.ts +52 -0
- package/out/utils/slice-stats.d.ts.map +1 -0
- package/out/utils/slice-stats.js +58 -0
- package/out/utils/slice-stats.js.map +1 -0
- package/out/utils/split.d.ts +128 -0
- package/out/utils/split.d.ts.map +1 -0
- package/out/utils/split.js +204 -0
- package/out/utils/split.js.map +1 -0
- package/out/utils/tokenize.d.ts +64 -0
- package/out/utils/tokenize.d.ts.map +1 -0
- package/out/utils/tokenize.js +82 -0
- package/out/utils/tokenize.js.map +1 -0
- package/out/utils/wof-json.d.ts +113 -0
- package/out/utils/wof-json.d.ts.map +1 -0
- package/out/utils/wof-json.js +173 -0
- package/out/utils/wof-json.js.map +1 -0
- package/package.json +259 -194
- package/out/src/adapters/ban/adapter.d.ts.map +0 -1
- package/out/src/adapters/ban/adapter.js +0 -149
- package/out/src/adapters/ban/adapter.js.map +0 -1
- package/out/src/adapters/ban/street-decompose.d.ts.map +0 -1
- package/out/src/adapters/ban/street-decompose.js +0 -75
- package/out/src/adapters/ban/street-decompose.js.map +0 -1
- package/out/src/adapters/fcc-bdc/adapter.d.ts.map +0 -1
- package/out/src/adapters/fcc-bdc/adapter.js +0 -144
- package/out/src/adapters/fcc-bdc/adapter.js.map +0 -1
- package/out/src/adapters/geonames/adapter.d.ts.map +0 -1
- package/out/src/adapters/geonames/adapter.js +0 -176
- package/out/src/adapters/geonames/adapter.js.map +0 -1
- package/out/src/adapters/geonames-postal/adapter.d.ts +0 -41
- package/out/src/adapters/geonames-postal/adapter.d.ts.map +0 -1
- package/out/src/adapters/geonames-postal/adapter.js +0 -104
- package/out/src/adapters/geonames-postal/adapter.js.map +0 -1
- package/out/src/adapters/gnaf/adapter.d.ts +0 -49
- package/out/src/adapters/gnaf/adapter.d.ts.map +0 -1
- package/out/src/adapters/gnaf/adapter.js +0 -124
- package/out/src/adapters/gnaf/adapter.js.map +0 -1
- package/out/src/adapters/gnaf/assemble.d.ts +0 -65
- package/out/src/adapters/gnaf/assemble.d.ts.map +0 -1
- package/out/src/adapters/gnaf/assemble.js +0 -154
- package/out/src/adapters/gnaf/assemble.js.map +0 -1
- package/out/src/adapters/index.d.ts +0 -51
- package/out/src/adapters/index.d.ts.map +0 -1
- package/out/src/adapters/index.js +0 -95
- package/out/src/adapters/index.js.map +0 -1
- package/out/src/adapters/openaddresses/adapter.d.ts.map +0 -1
- package/out/src/adapters/openaddresses/adapter.js +0 -186
- package/out/src/adapters/openaddresses/adapter.js.map +0 -1
- package/out/src/adapters/overture/adapter.d.ts +0 -48
- package/out/src/adapters/overture/adapter.d.ts.map +0 -1
- package/out/src/adapters/overture/adapter.js +0 -125
- package/out/src/adapters/overture/adapter.js.map +0 -1
- package/out/src/adapters/state-hi-schools/adapter.d.ts +0 -47
- package/out/src/adapters/state-hi-schools/adapter.d.ts.map +0 -1
- package/out/src/adapters/state-hi-schools/adapter.js +0 -133
- package/out/src/adapters/state-hi-schools/adapter.js.map +0 -1
- package/out/src/adapters/state-ia-contractors/adapter.d.ts.map +0 -1
- package/out/src/adapters/state-ia-contractors/adapter.js.map +0 -1
- package/out/src/adapters/state-ny-notaries/adapter.d.ts.map +0 -1
- package/out/src/adapters/state-ny-notaries/adapter.js.map +0 -1
- package/out/src/adapters/state-tx-notaries/adapter.d.ts.map +0 -1
- package/out/src/adapters/state-tx-notaries/adapter.js.map +0 -1
- package/out/src/adapters/synth-po-box/adapter.d.ts.map +0 -1
- package/out/src/adapters/synth-po-box/adapter.js +0 -140
- package/out/src/adapters/synth-po-box/adapter.js.map +0 -1
- package/out/src/adapters/tiger/adapter.d.ts +0 -55
- package/out/src/adapters/tiger/adapter.d.ts.map +0 -1
- package/out/src/adapters/tiger/adapter.js +0 -201
- package/out/src/adapters/tiger/adapter.js.map +0 -1
- package/out/src/adapters/tiger/street-decompose.d.ts.map +0 -1
- package/out/src/adapters/tiger/street-decompose.js +0 -90
- package/out/src/adapters/tiger/street-decompose.js.map +0 -1
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts +0 -47
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.js +0 -139
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.js.map +0 -1
- package/out/src/adapters/usgov-imls-pls/adapter.d.ts +0 -36
- package/out/src/adapters/usgov-imls-pls/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-imls-pls/adapter.js +0 -114
- package/out/src/adapters/usgov-imls-pls/adapter.js.map +0 -1
- package/out/src/adapters/usgov-irs-bmf/adapter.d.ts +0 -37
- package/out/src/adapters/usgov-irs-bmf/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-irs-bmf/adapter.js +0 -124
- package/out/src/adapters/usgov-irs-bmf/adapter.js.map +0 -1
- package/out/src/adapters/usgov-nad/adapter.d.ts +0 -48
- package/out/src/adapters/usgov-nad/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-nad/adapter.js +0 -252
- package/out/src/adapters/usgov-nad/adapter.js.map +0 -1
- package/out/src/adapters/usgov-nppes/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-nppes/adapter.js +0 -116
- package/out/src/adapters/usgov-nppes/adapter.js.map +0 -1
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map +0 -1
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js +0 -156
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js.map +0 -1
- package/out/src/adapters/utils/index.d.ts +0 -150
- package/out/src/adapters/utils/index.d.ts.map +0 -1
- package/out/src/adapters/utils/index.js +0 -159
- package/out/src/adapters/utils/index.js.map +0 -1
- package/out/src/adapters/wof-admin-jp/adapter.d.ts.map +0 -1
- package/out/src/adapters/wof-admin-jp/adapter.js +0 -164
- package/out/src/adapters/wof-admin-jp/adapter.js.map +0 -1
- package/out/src/adapters/wof-admin-json/adapter.d.ts +0 -82
- package/out/src/adapters/wof-admin-json/adapter.d.ts.map +0 -1
- package/out/src/adapters/wof-admin-json/adapter.js +0 -224
- package/out/src/adapters/wof-admin-json/adapter.js.map +0 -1
- package/out/src/adapters/wof-json-rows.d.ts +0 -32
- package/out/src/adapters/wof-json-rows.d.ts.map +0 -1
- package/out/src/adapters/wof-json-rows.js +0 -45
- package/out/src/adapters/wof-json-rows.js.map +0 -1
- package/out/src/adapters/wof-postalcode-json/adapter.d.ts +0 -65
- package/out/src/adapters/wof-postalcode-json/adapter.d.ts.map +0 -1
- package/out/src/adapters/wof-postalcode-json/adapter.js +0 -160
- package/out/src/adapters/wof-postalcode-json/adapter.js.map +0 -1
- package/out/src/build.d.ts +0 -132
- package/out/src/build.d.ts.map +0 -1
- package/out/src/build.js +0 -251
- package/out/src/build.js.map +0 -1
- package/out/src/codex/us-fips-state.d.ts.map +0 -1
- package/out/src/codex/us-fips-state.js.map +0 -1
- package/out/src/index.d.ts +0 -14
- package/out/src/index.d.ts.map +0 -1
- package/out/src/index.js +0 -14
- package/out/src/index.js.map +0 -1
- package/out/src/name-prone-us-suffixes.d.ts +0 -12
- package/out/src/name-prone-us-suffixes.d.ts.map +0 -1
- package/out/src/name-prone-us-suffixes.js +0 -12
- package/out/src/name-prone-us-suffixes.js.map +0 -1
- package/out/src/parquet-wrapper/index.d.ts +0 -9
- package/out/src/parquet-wrapper/index.d.ts.map +0 -1
- package/out/src/parquet-wrapper/index.js +0 -9
- package/out/src/parquet-wrapper/index.js.map +0 -1
- package/out/src/parquet-wrapper/reader.d.ts +0 -34
- package/out/src/parquet-wrapper/reader.d.ts.map +0 -1
- package/out/src/parquet-wrapper/reader.js +0 -69
- package/out/src/parquet-wrapper/reader.js.map +0 -1
- package/out/src/parquet-wrapper/schema.d.ts +0 -49
- package/out/src/parquet-wrapper/schema.d.ts.map +0 -1
- package/out/src/parquet-wrapper/schema.js +0 -55
- package/out/src/parquet-wrapper/schema.js.map +0 -1
- package/out/src/parquet-wrapper/writer.d.ts +0 -46
- package/out/src/parquet-wrapper/writer.d.ts.map +0 -1
- package/out/src/parquet-wrapper/writer.js +0 -76
- package/out/src/parquet-wrapper/writer.js.map +0 -1
- package/out/src/runner.d.ts +0 -126
- package/out/src/runner.d.ts.map +0 -1
- package/out/src/runner.js +0 -189
- package/out/src/runner.js.map +0 -1
- package/out/src/shard-recipes/anchor-absorption.d.ts +0 -20
- package/out/src/shard-recipes/anchor-absorption.d.ts.map +0 -1
- package/out/src/shard-recipes/anchor-absorption.js +0 -82
- package/out/src/shard-recipes/anchor-absorption.js.map +0 -1
- package/out/src/shard-recipes/bare-country.d.ts +0 -25
- package/out/src/shard-recipes/bare-country.d.ts.map +0 -1
- package/out/src/shard-recipes/bare-country.js +0 -87
- package/out/src/shard-recipes/bare-country.js.map +0 -1
- package/out/src/shard-recipes/boundary-stress.d.ts +0 -20
- package/out/src/shard-recipes/boundary-stress.d.ts.map +0 -1
- package/out/src/shard-recipes/boundary-stress.js +0 -83
- package/out/src/shard-recipes/boundary-stress.js.map +0 -1
- package/out/src/shard-recipes/country-balanced.d.ts +0 -32
- package/out/src/shard-recipes/country-balanced.d.ts.map +0 -1
- package/out/src/shard-recipes/country-balanced.js +0 -406
- package/out/src/shard-recipes/country-balanced.js.map +0 -1
- package/out/src/shard-recipes/cz-pcfirst-preposition.d.ts +0 -22
- package/out/src/shard-recipes/cz-pcfirst-preposition.d.ts.map +0 -1
- package/out/src/shard-recipes/cz-pcfirst-preposition.js +0 -100
- package/out/src/shard-recipes/cz-pcfirst-preposition.js.map +0 -1
- package/out/src/shard-recipes/fr-admin-split.d.ts +0 -38
- package/out/src/shard-recipes/fr-admin-split.d.ts.map +0 -1
- package/out/src/shard-recipes/fr-admin-split.js +0 -188
- package/out/src/shard-recipes/fr-admin-split.js.map +0 -1
- package/out/src/shard-recipes/fr-bare-street.d.ts +0 -28
- package/out/src/shard-recipes/fr-bare-street.d.ts.map +0 -1
- package/out/src/shard-recipes/fr-bare-street.js +0 -143
- package/out/src/shard-recipes/fr-bare-street.js.map +0 -1
- package/out/src/shard-recipes/fr-fragment.d.ts +0 -77
- package/out/src/shard-recipes/fr-fragment.d.ts.map +0 -1
- package/out/src/shard-recipes/fr-fragment.js +0 -276
- package/out/src/shard-recipes/fr-fragment.js.map +0 -1
- package/out/src/shard-recipes/fr-lieudit.d.ts +0 -32
- package/out/src/shard-recipes/fr-lieudit.d.ts.map +0 -1
- package/out/src/shard-recipes/fr-lieudit.js +0 -226
- package/out/src/shard-recipes/fr-lieudit.js.map +0 -1
- package/out/src/shard-recipes/fr-order.d.ts +0 -38
- package/out/src/shard-recipes/fr-order.d.ts.map +0 -1
- package/out/src/shard-recipes/fr-order.js +0 -208
- package/out/src/shard-recipes/fr-order.js.map +0 -1
- package/out/src/shard-recipes/german.d.ts +0 -23
- package/out/src/shard-recipes/german.d.ts.map +0 -1
- package/out/src/shard-recipes/german.js +0 -132
- package/out/src/shard-recipes/german.js.map +0 -1
- package/out/src/shard-recipes/house-venue.d.ts +0 -18
- package/out/src/shard-recipes/house-venue.d.ts.map +0 -1
- package/out/src/shard-recipes/house-venue.js +0 -74
- package/out/src/shard-recipes/house-venue.js.map +0 -1
- package/out/src/shard-recipes/index.d.ts +0 -24
- package/out/src/shard-recipes/index.d.ts.map +0 -1
- package/out/src/shard-recipes/index.js +0 -89
- package/out/src/shard-recipes/index.js.map +0 -1
- package/out/src/shard-recipes/intersection.d.ts +0 -42
- package/out/src/shard-recipes/intersection.d.ts.map +0 -1
- package/out/src/shard-recipes/intersection.js +0 -460
- package/out/src/shard-recipes/intersection.js.map +0 -1
- package/out/src/shard-recipes/locale.d.ts +0 -135
- package/out/src/shard-recipes/locale.d.ts.map +0 -1
- package/out/src/shard-recipes/locale.js +0 -459
- package/out/src/shard-recipes/locale.js.map +0 -1
- package/out/src/shard-recipes/nl-postcode.d.ts +0 -26
- package/out/src/shard-recipes/nl-postcode.d.ts.map +0 -1
- package/out/src/shard-recipes/nl-postcode.js +0 -102
- package/out/src/shard-recipes/nl-postcode.js.map +0 -1
- package/out/src/shard-recipes/no-fragment.d.ts +0 -43
- package/out/src/shard-recipes/no-fragment.d.ts.map +0 -1
- package/out/src/shard-recipes/no-fragment.js +0 -188
- package/out/src/shard-recipes/no-fragment.js.map +0 -1
- package/out/src/shard-recipes/no-street-led.d.ts +0 -24
- package/out/src/shard-recipes/no-street-led.d.ts.map +0 -1
- package/out/src/shard-recipes/no-street-led.js +0 -121
- package/out/src/shard-recipes/no-street-led.js.map +0 -1
- package/out/src/shard-recipes/no-street.d.ts +0 -18
- package/out/src/shard-recipes/no-street.d.ts.map +0 -1
- package/out/src/shard-recipes/no-street.js +0 -72
- package/out/src/shard-recipes/no-street.js.map +0 -1
- package/out/src/shard-recipes/po-box-cedex.d.ts +0 -38
- package/out/src/shard-recipes/po-box-cedex.d.ts.map +0 -1
- package/out/src/shard-recipes/po-box-cedex.js +0 -726
- package/out/src/shard-recipes/po-box-cedex.js.map +0 -1
- package/out/src/shard-recipes/po-box.d.ts +0 -17
- package/out/src/shard-recipes/po-box.d.ts.map +0 -1
- package/out/src/shard-recipes/po-box.js +0 -101
- package/out/src/shard-recipes/po-box.js.map +0 -1
- package/out/src/shard-recipes/reviewed-postcode-tail.d.ts +0 -47
- package/out/src/shard-recipes/reviewed-postcode-tail.d.ts.map +0 -1
- package/out/src/shard-recipes/reviewed-postcode-tail.js +0 -125
- package/out/src/shard-recipes/reviewed-postcode-tail.js.map +0 -1
- package/out/src/shard-recipes/scaffold.d.ts +0 -246
- package/out/src/shard-recipes/scaffold.d.ts.map +0 -1
- package/out/src/shard-recipes/scaffold.js +0 -137
- package/out/src/shard-recipes/scaffold.js.map +0 -1
- package/out/src/shard-recipes/si-bare-village.d.ts +0 -32
- package/out/src/shard-recipes/si-bare-village.d.ts.map +0 -1
- package/out/src/shard-recipes/si-bare-village.js +0 -88
- package/out/src/shard-recipes/si-bare-village.js.map +0 -1
- package/out/src/shard-recipes/street-affix.d.ts +0 -89
- package/out/src/shard-recipes/street-affix.d.ts.map +0 -1
- package/out/src/shard-recipes/street-affix.js +0 -578
- package/out/src/shard-recipes/street-affix.js.map +0 -1
- package/out/src/shard-recipes/street-bare.d.ts +0 -25
- package/out/src/shard-recipes/street-bare.d.ts.map +0 -1
- package/out/src/shard-recipes/street-bare.js +0 -86
- package/out/src/shard-recipes/street-bare.js.map +0 -1
- package/out/src/shard-recipes/street.d.ts +0 -15
- package/out/src/shard-recipes/street.d.ts.map +0 -1
- package/out/src/shard-recipes/street.js +0 -69
- package/out/src/shard-recipes/street.js.map +0 -1
- package/out/src/shard-recipes/sub-venue-sources.d.ts +0 -228
- package/out/src/shard-recipes/sub-venue-sources.d.ts.map +0 -1
- package/out/src/shard-recipes/sub-venue-sources.js +0 -664
- package/out/src/shard-recipes/sub-venue-sources.js.map +0 -1
- package/out/src/shard-recipes/sub-venue.d.ts +0 -269
- package/out/src/shard-recipes/sub-venue.d.ts.map +0 -1
- package/out/src/shard-recipes/sub-venue.js +0 -710
- package/out/src/shard-recipes/sub-venue.js.map +0 -1
- package/out/src/shard-recipes/trailing-region.d.ts +0 -74
- package/out/src/shard-recipes/trailing-region.d.ts.map +0 -1
- package/out/src/shard-recipes/trailing-region.js +0 -157
- package/out/src/shard-recipes/trailing-region.js.map +0 -1
- package/out/src/shard-recipes/unit.d.ts +0 -29
- package/out/src/shard-recipes/unit.d.ts.map +0 -1
- package/out/src/shard-recipes/unit.js +0 -243
- package/out/src/shard-recipes/unit.js.map +0 -1
- package/out/src/synthesizers/anchor-absorption.d.ts +0 -63
- package/out/src/synthesizers/anchor-absorption.d.ts.map +0 -1
- package/out/src/synthesizers/anchor-absorption.js +0 -223
- package/out/src/synthesizers/anchor-absorption.js.map +0 -1
- package/out/src/synthesizers/boundary-stress.d.ts +0 -78
- package/out/src/synthesizers/boundary-stress.d.ts.map +0 -1
- package/out/src/synthesizers/boundary-stress.js +0 -455
- package/out/src/synthesizers/boundary-stress.js.map +0 -1
- package/out/src/synthesizers/german.d.ts +0 -107
- package/out/src/synthesizers/german.d.ts.map +0 -1
- package/out/src/synthesizers/german.js +0 -148
- package/out/src/synthesizers/german.js.map +0 -1
- package/out/src/synthesizers/house-venue.d.ts +0 -56
- package/out/src/synthesizers/house-venue.d.ts.map +0 -1
- package/out/src/synthesizers/house-venue.js +0 -267
- package/out/src/synthesizers/house-venue.js.map +0 -1
- package/out/src/synthesizers/intersection.d.ts +0 -52
- package/out/src/synthesizers/intersection.d.ts.map +0 -1
- package/out/src/synthesizers/intersection.js +0 -175
- package/out/src/synthesizers/intersection.js.map +0 -1
- package/out/src/synthesizers/no-street.d.ts +0 -72
- package/out/src/synthesizers/no-street.d.ts.map +0 -1
- package/out/src/synthesizers/no-street.js +0 -282
- package/out/src/synthesizers/no-street.js.map +0 -1
- package/out/src/synthesizers/po-box.d.ts +0 -104
- package/out/src/synthesizers/po-box.d.ts.map +0 -1
- package/out/src/synthesizers/po-box.js +0 -266
- package/out/src/synthesizers/po-box.js.map +0 -1
- package/out/src/synthesizers/street.d.ts +0 -54
- package/out/src/synthesizers/street.d.ts.map +0 -1
- package/out/src/synthesizers/street.js +0 -221
- package/out/src/synthesizers/street.js.map +0 -1
- package/out/src/synthesizers/utils.d.ts +0 -203
- package/out/src/synthesizers/utils.d.ts.map +0 -1
- package/out/src/synthesizers/utils.js +0 -708
- package/out/src/synthesizers/utils.js.map +0 -1
- package/out/src/tools/align-shard.d.ts +0 -8
- package/out/src/tools/align-shard.d.ts.map +0 -1
- package/out/src/tools/align-shard.js +0 -56
- package/out/src/tools/align-shard.js.map +0 -1
- package/out/src/tools/audit.d.ts +0 -28
- package/out/src/tools/audit.d.ts.map +0 -1
- package/out/src/tools/audit.js +0 -285
- package/out/src/tools/audit.js.map +0 -1
- package/out/src/tools/corpus-stats.d.ts +0 -44
- package/out/src/tools/corpus-stats.d.ts.map +0 -1
- package/out/src/tools/corpus-stats.js +0 -138
- package/out/src/tools/corpus-stats.js.map +0 -1
- package/out/src/tools/fetch/ban.d.ts +0 -31
- package/out/src/tools/fetch/ban.d.ts.map +0 -1
- package/out/src/tools/fetch/ban.js +0 -239
- package/out/src/tools/fetch/ban.js.map +0 -1
- package/out/src/tools/fetch/download.d.ts +0 -78
- package/out/src/tools/fetch/download.d.ts.map +0 -1
- package/out/src/tools/fetch/download.js +0 -116
- package/out/src/tools/fetch/download.js.map +0 -1
- package/out/src/tools/fetch/geonames-dump.d.ts +0 -87
- package/out/src/tools/fetch/geonames-dump.d.ts.map +0 -1
- package/out/src/tools/fetch/geonames-dump.js +0 -178
- package/out/src/tools/fetch/geonames-dump.js.map +0 -1
- package/out/src/tools/fetch/geonames-postal.d.ts +0 -76
- package/out/src/tools/fetch/geonames-postal.d.ts.map +0 -1
- package/out/src/tools/fetch/geonames-postal.js +0 -125
- package/out/src/tools/fetch/geonames-postal.js.map +0 -1
- package/out/src/tools/fetch/hrsa.d.ts +0 -18
- package/out/src/tools/fetch/hrsa.d.ts.map +0 -1
- package/out/src/tools/fetch/hrsa.js +0 -46
- package/out/src/tools/fetch/hrsa.js.map +0 -1
- package/out/src/tools/fetch/imls-pls.d.ts +0 -25
- package/out/src/tools/fetch/imls-pls.d.ts.map +0 -1
- package/out/src/tools/fetch/imls-pls.js +0 -108
- package/out/src/tools/fetch/imls-pls.js.map +0 -1
- package/out/src/tools/fetch/index.d.ts +0 -144
- package/out/src/tools/fetch/index.d.ts.map +0 -1
- package/out/src/tools/fetch/index.js +0 -143
- package/out/src/tools/fetch/index.js.map +0 -1
- package/out/src/tools/fetch/nad.d.ts +0 -70
- package/out/src/tools/fetch/nad.d.ts.map +0 -1
- package/out/src/tools/fetch/nad.js +0 -234
- package/out/src/tools/fetch/nad.js.map +0 -1
- package/out/src/tools/fetch/nppes.d.ts +0 -26
- package/out/src/tools/fetch/nppes.d.ts.map +0 -1
- package/out/src/tools/fetch/nppes.js +0 -126
- package/out/src/tools/fetch/nppes.js.map +0 -1
- package/out/src/tools/fetch/openaddresses.d.ts +0 -61
- package/out/src/tools/fetch/openaddresses.d.ts.map +0 -1
- package/out/src/tools/fetch/openaddresses.js +0 -317
- package/out/src/tools/fetch/openaddresses.js.map +0 -1
- package/out/src/tools/fetch/ourairports.d.ts +0 -45
- package/out/src/tools/fetch/ourairports.d.ts.map +0 -1
- package/out/src/tools/fetch/ourairports.js +0 -128
- package/out/src/tools/fetch/ourairports.js.map +0 -1
- package/out/src/tools/fetch/ppd.d.ts.map +0 -1
- package/out/src/tools/fetch/ppd.js +0 -100
- package/out/src/tools/fetch/ppd.js.map +0 -1
- package/out/src/tools/fetch/state-hi-schools.d.ts +0 -25
- package/out/src/tools/fetch/state-hi-schools.d.ts.map +0 -1
- package/out/src/tools/fetch/state-hi-schools.js +0 -162
- package/out/src/tools/fetch/state-hi-schools.js.map +0 -1
- package/out/src/tools/fetch/state-sources.d.ts +0 -26
- package/out/src/tools/fetch/state-sources.d.ts.map +0 -1
- package/out/src/tools/fetch/state-sources.js +0 -111
- package/out/src/tools/fetch/state-sources.js.map +0 -1
- package/out/src/tools/fetch/tiger-full.d.ts +0 -42
- package/out/src/tools/fetch/tiger-full.d.ts.map +0 -1
- package/out/src/tools/fetch/tiger-full.js +0 -248
- package/out/src/tools/fetch/tiger-full.js.map +0 -1
- package/out/src/tools/fetch/wikidata-subvenue.d.ts +0 -171
- package/out/src/tools/fetch/wikidata-subvenue.d.ts.map +0 -1
- package/out/src/tools/fetch/wikidata-subvenue.js +0 -275
- package/out/src/tools/fetch/wikidata-subvenue.js.map +0 -1
- package/out/src/tools/gb-title-case.d.ts.map +0 -1
- package/out/src/tools/gb-title-case.js.map +0 -1
- package/out/src/tools/golden-expand.d.ts +0 -90
- package/out/src/tools/golden-expand.d.ts.map +0 -1
- package/out/src/tools/golden-expand.js +0 -413
- package/out/src/tools/golden-expand.js.map +0 -1
- package/out/src/tools/golden-promote.d.ts.map +0 -1
- package/out/src/tools/golden-promote.js +0 -244
- package/out/src/tools/golden-promote.js.map +0 -1
- package/out/src/tools/golden-relabel-street.d.ts +0 -196
- package/out/src/tools/golden-relabel-street.d.ts.map +0 -1
- package/out/src/tools/golden-relabel-street.js +0 -440
- package/out/src/tools/golden-relabel-street.js.map +0 -1
- package/out/src/tools/index.d.ts +0 -28
- package/out/src/tools/index.d.ts.map +0 -1
- package/out/src/tools/index.js +0 -28
- package/out/src/tools/index.js.map +0 -1
- package/out/src/tools/ingest-csv.d.ts.map +0 -1
- package/out/src/tools/ingest-csv.js +0 -296
- package/out/src/tools/ingest-csv.js.map +0 -1
- package/out/src/tools/jsonl-to-parquet.d.ts +0 -61
- package/out/src/tools/jsonl-to-parquet.d.ts.map +0 -1
- package/out/src/tools/jsonl-to-parquet.js +0 -164
- package/out/src/tools/jsonl-to-parquet.js.map +0 -1
- package/out/src/tools/lint-shard-vocab.d.ts +0 -90
- package/out/src/tools/lint-shard-vocab.d.ts.map +0 -1
- package/out/src/tools/lint-shard-vocab.js +0 -281
- package/out/src/tools/lint-shard-vocab.js.map +0 -1
- package/out/src/tools/lint-shard.d.ts +0 -93
- package/out/src/tools/lint-shard.d.ts.map +0 -1
- package/out/src/tools/lint-shard.js +0 -354
- package/out/src/tools/lint-shard.js.map +0 -1
- package/out/src/tools/overlay-manifest.d.ts +0 -39
- package/out/src/tools/overlay-manifest.d.ts.map +0 -1
- package/out/src/tools/overlay-manifest.js +0 -98
- package/out/src/tools/overlay-manifest.js.map +0 -1
- package/out/src/tools/overture-subvenue.d.ts +0 -112
- package/out/src/tools/overture-subvenue.d.ts.map +0 -1
- package/out/src/tools/overture-subvenue.js +0 -144
- package/out/src/tools/overture-subvenue.js.map +0 -1
- package/out/src/tools/postcode-triples.d.ts +0 -175
- package/out/src/tools/postcode-triples.d.ts.map +0 -1
- package/out/src/tools/postcode-triples.js +0 -304
- package/out/src/tools/postcode-triples.js.map +0 -1
- package/out/src/tools/shard-kryptonite.d.ts +0 -32
- package/out/src/tools/shard-kryptonite.d.ts.map +0 -1
- package/out/src/tools/shard-kryptonite.js +0 -100
- package/out/src/tools/shard-kryptonite.js.map +0 -1
- package/out/src/tools/shard-translit.d.ts +0 -51
- package/out/src/tools/shard-translit.d.ts.map +0 -1
- package/out/src/tools/shard-translit.js +0 -175
- package/out/src/tools/shard-translit.js.map +0 -1
- package/out/src/tools/sub-venue/harvest.d.ts +0 -100
- package/out/src/tools/sub-venue/harvest.d.ts.map +0 -1
- package/out/src/tools/sub-venue/harvest.js +0 -168
- package/out/src/tools/sub-venue/harvest.js.map +0 -1
- package/out/src/tools/sub-venue/head-nouns.d.ts +0 -50
- package/out/src/tools/sub-venue/head-nouns.d.ts.map +0 -1
- package/out/src/tools/sub-venue/head-nouns.js +0 -210
- package/out/src/tools/sub-venue/head-nouns.js.map +0 -1
- package/out/src/tools/sub-venue/surfaces.d.ts +0 -45
- package/out/src/tools/sub-venue/surfaces.d.ts.map +0 -1
- package/out/src/tools/sub-venue/surfaces.js +0 -77
- package/out/src/tools/sub-venue/surfaces.js.map +0 -1
- package/out/src/tools/sub-venue/table.d.ts +0 -229
- package/out/src/tools/sub-venue/table.d.ts.map +0 -1
- package/out/src/tools/sub-venue/table.js +0 -108
- package/out/src/tools/sub-venue/table.js.map +0 -1
- package/out/src/tools/sub-venue/wikidata.d.ts +0 -26
- package/out/src/tools/sub-venue/wikidata.d.ts.map +0 -1
- package/out/src/tools/sub-venue/wikidata.js +0 -62
- package/out/src/tools/sub-venue/wikidata.js.map +0 -1
- package/out/src/tools/sub-venue-lexicon.d.ts +0 -156
- package/out/src/tools/sub-venue-lexicon.d.ts.map +0 -1
- package/out/src/tools/sub-venue-lexicon.js +0 -282
- package/out/src/tools/sub-venue-lexicon.js.map +0 -1
- package/out/src/tools/sub-venue-promotions.d.ts +0 -94
- package/out/src/tools/sub-venue-promotions.d.ts.map +0 -1
- package/out/src/tools/sub-venue-promotions.js +0 -266
- package/out/src/tools/sub-venue-promotions.js.map +0 -1
- package/out/src/types.d.ts +0 -221
- package/out/src/types.d.ts.map +0 -1
- package/out/src/types.js.map +0 -1
- package/out/src/utils/align.d.ts +0 -93
- package/out/src/utils/align.d.ts.map +0 -1
- package/out/src/utils/align.js +0 -234
- package/out/src/utils/align.js.map +0 -1
- package/out/src/utils/golden.d.ts +0 -69
- package/out/src/utils/golden.d.ts.map +0 -1
- package/out/src/utils/golden.js +0 -127
- package/out/src/utils/golden.js.map +0 -1
- package/out/src/utils/index.d.ts +0 -14
- package/out/src/utils/index.d.ts.map +0 -1
- package/out/src/utils/index.js +0 -14
- package/out/src/utils/index.js.map +0 -1
- package/out/src/utils/license.d.ts.map +0 -1
- package/out/src/utils/license.js.map +0 -1
- package/out/src/utils/parquet.d.ts +0 -162
- package/out/src/utils/parquet.d.ts.map +0 -1
- package/out/src/utils/parquet.js +0 -262
- package/out/src/utils/parquet.js.map +0 -1
- package/out/src/utils/split.d.ts +0 -127
- package/out/src/utils/split.d.ts.map +0 -1
- package/out/src/utils/split.js +0 -204
- package/out/src/utils/split.js.map +0 -1
- package/out/src/utils/tokenize.d.ts +0 -56
- package/out/src/utils/tokenize.d.ts.map +0 -1
- package/out/src/utils/tokenize.js +0 -48
- package/out/src/utils/tokenize.js.map +0 -1
- package/out/src/utils/wof-json.d.ts +0 -109
- package/out/src/utils/wof-json.d.ts.map +0 -1
- package/out/src/utils/wof-json.js +0 -173
- package/out/src/utils/wof-json.js.map +0 -1
- package/src/adapters/ban/adapter.ts +0 -191
- package/src/adapters/ban/street-decompose.ts +0 -96
- package/src/adapters/fcc-bdc/adapter.ts +0 -174
- package/src/adapters/geonames/adapter.ts +0 -200
- package/src/adapters/geonames-postal/adapter.ts +0 -119
- package/src/adapters/gnaf/adapter.ts +0 -149
- package/src/adapters/gnaf/assemble.ts +0 -226
- package/src/adapters/index.ts +0 -149
- package/src/adapters/openaddresses/adapter.ts +0 -251
- package/src/adapters/overture/adapter.ts +0 -162
- package/src/adapters/state-hi-schools/adapter.ts +0 -164
- package/src/adapters/synth-po-box/adapter.ts +0 -204
- package/src/adapters/tiger/adapter.ts +0 -253
- package/src/adapters/tiger/street-decompose.ts +0 -111
- package/src/adapters/usgov-hrsa-fqhc/adapter.ts +0 -183
- package/src/adapters/usgov-imls-pls/adapter.ts +0 -143
- package/src/adapters/usgov-irs-bmf/adapter.ts +0 -163
- package/src/adapters/usgov-nad/adapter.ts +0 -348
- package/src/adapters/usgov-nppes/adapter.ts +0 -159
- package/src/adapters/usgov-samhsa-treatment-locator/adapter.ts +0 -204
- package/src/adapters/utils/index.ts +0 -233
- package/src/adapters/wof-admin-jp/adapter.ts +0 -216
- package/src/adapters/wof-admin-json/adapter.ts +0 -256
- package/src/adapters/wof-json-rows.ts +0 -79
- package/src/adapters/wof-postalcode-json/adapter.ts +0 -183
- package/src/build.ts +0 -386
- package/src/index.ts +0 -14
- package/src/name-prone-us-suffixes.ts +0 -12
- package/src/parquet-wrapper/index.ts +0 -9
- package/src/parquet-wrapper/reader.ts +0 -97
- package/src/parquet-wrapper/schema.ts +0 -88
- package/src/parquet-wrapper/writer.ts +0 -110
- package/src/runner.ts +0 -317
- package/src/shard-recipes/anchor-absorption.ts +0 -103
- package/src/shard-recipes/bare-country.ts +0 -95
- package/src/shard-recipes/boundary-stress.ts +0 -101
- package/src/shard-recipes/country-balanced.ts +0 -524
- package/src/shard-recipes/cz-pcfirst-preposition.ts +0 -111
- package/src/shard-recipes/fr-admin-split.ts +0 -239
- package/src/shard-recipes/fr-bare-street.ts +0 -170
- package/src/shard-recipes/fr-fragment.ts +0 -332
- package/src/shard-recipes/fr-lieudit.ts +0 -302
- package/src/shard-recipes/fr-order.ts +0 -258
- package/src/shard-recipes/german.ts +0 -170
- package/src/shard-recipes/house-venue.ts +0 -88
- package/src/shard-recipes/index.ts +0 -95
- package/src/shard-recipes/intersection.ts +0 -630
- package/src/shard-recipes/locale.ts +0 -599
- package/src/shard-recipes/nl-postcode.ts +0 -115
- package/src/shard-recipes/no-fragment.ts +0 -219
- package/src/shard-recipes/no-street-led.ts +0 -143
- package/src/shard-recipes/no-street.ts +0 -86
- package/src/shard-recipes/po-box-cedex.ts +0 -920
- package/src/shard-recipes/po-box.ts +0 -119
- package/src/shard-recipes/reviewed-postcode-tail.ts +0 -189
- package/src/shard-recipes/scaffold.ts +0 -360
- package/src/shard-recipes/si-bare-village.ts +0 -97
- package/src/shard-recipes/street-affix.ts +0 -826
- package/src/shard-recipes/street-bare.ts +0 -101
- package/src/shard-recipes/street.ts +0 -82
- package/src/shard-recipes/sub-venue-sources.ts +0 -885
- package/src/shard-recipes/sub-venue.ts +0 -984
- package/src/shard-recipes/trailing-region.ts +0 -183
- package/src/shard-recipes/unit.ts +0 -321
- package/src/synthesizers/anchor-absorption.ts +0 -290
- package/src/synthesizers/boundary-stress.ts +0 -531
- package/src/synthesizers/german.ts +0 -251
- package/src/synthesizers/house-venue.ts +0 -330
- package/src/synthesizers/intersection.ts +0 -232
- package/src/synthesizers/no-street.ts +0 -349
- package/src/synthesizers/po-box.ts +0 -339
- package/src/synthesizers/street.ts +0 -274
- package/src/synthesizers/utils.ts +0 -903
- package/src/tools/align-shard.ts +0 -70
- package/src/tools/audit.ts +0 -399
- package/src/tools/corpus-stats.ts +0 -184
- package/src/tools/fetch/ban.ts +0 -290
- package/src/tools/fetch/download.ts +0 -176
- package/src/tools/fetch/geonames-dump.ts +0 -267
- package/src/tools/fetch/geonames-postal.ts +0 -180
- package/src/tools/fetch/hrsa.ts +0 -68
- package/src/tools/fetch/imls-pls.ts +0 -158
- package/src/tools/fetch/index.ts +0 -147
- package/src/tools/fetch/nad.ts +0 -342
- package/src/tools/fetch/nppes.ts +0 -174
- package/src/tools/fetch/openaddresses.ts +0 -404
- package/src/tools/fetch/ourairports.ts +0 -171
- package/src/tools/fetch/ppd.ts +0 -137
- package/src/tools/fetch/state-hi-schools.ts +0 -208
- package/src/tools/fetch/state-sources.ts +0 -155
- package/src/tools/fetch/tiger-full.ts +0 -354
- package/src/tools/fetch/wikidata-subvenue.ts +0 -386
- package/src/tools/golden-expand.ts +0 -584
- package/src/tools/golden-promote.ts +0 -353
- package/src/tools/golden-relabel-street.ts +0 -685
- package/src/tools/index.ts +0 -28
- package/src/tools/ingest-csv.ts +0 -422
- package/src/tools/jsonl-to-parquet.ts +0 -221
- package/src/tools/lint-shard-vocab.ts +0 -409
- package/src/tools/lint-shard.ts +0 -543
- package/src/tools/overlay-manifest.ts +0 -157
- package/src/tools/overture-subvenue.ts +0 -215
- package/src/tools/postcode-triples.ts +0 -375
- package/src/tools/shard-kryptonite.ts +0 -139
- package/src/tools/shard-translit.ts +0 -245
- package/src/tools/sub-venue/harvest.ts +0 -235
- package/src/tools/sub-venue/head-nouns.ts +0 -242
- package/src/tools/sub-venue/surfaces.ts +0 -95
- package/src/tools/sub-venue/table.ts +0 -281
- package/src/tools/sub-venue/wikidata.ts +0 -85
- package/src/tools/sub-venue-lexicon.ts +0 -431
- package/src/tools/sub-venue-promotions.ts +0 -330
- package/src/types.ts +0 -247
- package/src/utils/align.ts +0 -331
- package/src/utils/golden.ts +0 -176
- package/src/utils/index.ts +0 -14
- package/src/utils/parquet.ts +0 -391
- package/src/utils/split.ts +0 -279
- package/src/utils/tokenize.ts +0 -78
- package/src/utils/wof-json.ts +0 -244
- /package/{src → lib}/adapters/state-ia-contractors/adapter.ts +0 -0
- /package/{src → lib}/adapters/state-ny-notaries/adapter.ts +0 -0
- /package/{src → lib}/adapters/state-tx-notaries/adapter.ts +0 -0
- /package/{src → lib}/codex/us-fips-state.ts +0 -0
- /package/{src → lib}/tools/gb-title-case.ts +0 -0
- /package/{src → lib}/utils/license.ts +0 -0
- /package/out/{src/adapters → adapters}/ban/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/ban/street-decompose.d.ts +0 -0
- /package/out/{src/adapters → adapters}/fcc-bdc/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/geonames/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/openaddresses/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/state-ia-contractors/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/state-ia-contractors/adapter.js +0 -0
- /package/out/{src/adapters → adapters}/state-ny-notaries/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/state-ny-notaries/adapter.js +0 -0
- /package/out/{src/adapters → adapters}/state-tx-notaries/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/state-tx-notaries/adapter.js +0 -0
- /package/out/{src/adapters → adapters}/synth-po-box/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/tiger/street-decompose.d.ts +0 -0
- /package/out/{src/adapters → adapters}/usgov-nppes/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/usgov-samhsa-treatment-locator/adapter.d.ts +0 -0
- /package/out/{src/adapters → adapters}/wof-admin-jp/adapter.d.ts +0 -0
- /package/out/{src/codex → codex}/us-fips-state.d.ts +0 -0
- /package/out/{src/codex → codex}/us-fips-state.js +0 -0
- /package/out/{src/tools → tools}/fetch/ppd.d.ts +0 -0
- /package/out/{src/tools → tools}/gb-title-case.d.ts +0 -0
- /package/out/{src/tools → tools}/gb-title-case.js +0 -0
- /package/out/{src/tools → tools}/golden-promote.d.ts +0 -0
- /package/out/{src/tools → tools}/ingest-csv.d.ts +0 -0
- /package/out/{src/types.js → types.js} +0 -0
- /package/out/{src/utils → utils}/license.d.ts +0 -0
- /package/out/{src/utils → utils}/license.js +0 -0
|
@@ -0,0 +1,453 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
* @license AGPL-3.0
|
|
4
|
+
* @author Teffen Ellis, et al.
|
|
5
|
+
*
|
|
6
|
+
* `intersection` slice recipe — the REAL-pair intersection training slice (#487). The model scored
|
|
7
|
+
* 0.0 on intersection_a/b because the training mix had ZERO intersection-labeled rows. This is
|
|
8
|
+
* the missing data. Ported from scripts/build-intersection-slice.mjs.
|
|
9
|
+
*
|
|
10
|
+
* STREET PAIRS ARE REAL: the same TIGER 2023 EDGES extraction as the eval builder
|
|
11
|
+
* (scripts/eval/build-intersection-real.ts) — a node where two road edges (MTFCC S1*) with
|
|
12
|
+
* distinct FULLNAMEs meet is a real crossing. Real pairs avoid teaching fake street-street
|
|
13
|
+
* co-occurrences.
|
|
14
|
+
*
|
|
15
|
+
* LEAKAGE POLICY (mirrors the affix slice's VT discipline):
|
|
16
|
+
*
|
|
17
|
+
* - TRAIN counties: Cook IL (grid city) + Morris NJ (suburb).
|
|
18
|
+
* - GOLDEN (`--golden`) county: Washington VT (rural) ONLY — the corpus defaultHoldout state.
|
|
19
|
+
* - Every crossing in data/eval/external/intersection-real.jsonl is excluded from BOTH modes, by node
|
|
20
|
+
* id AND by order-insensitive name pair (the eval shares all three counties).
|
|
21
|
+
*
|
|
22
|
+
* RENDERING: junction-format variety — padded/TIGHT `&` and `/`, `and`, `at`, `@`, leading-phrase
|
|
23
|
+
* `corner of` / `intersection of` — crossed with tails (bare / `, ST` / `, ST ZIP` / `, City, ST
|
|
24
|
+
* [ZIP]`) and case variants. ZIPs are the crossing's own TIGER edge ZIPL (real); the locality
|
|
25
|
+
* tail comes from the OA Cook-county ZIP→city majority map.
|
|
26
|
+
*
|
|
27
|
+
* AUDIT: every emitted row is label-checked on the RAW SURFACE via the #519 char-offset span
|
|
28
|
+
* triple. Any violation FAILS the build (throws). A JSON audit report lands next to the output.
|
|
29
|
+
*
|
|
30
|
+
* External inputs (`--edges-dir`, opts.edgesDir; both already on disk — do not re-download):
|
|
31
|
+
*
|
|
32
|
+
* - <edges-dir>/tl_2023_{17031,34027,50023}_edges.shp (unzipped TIGER 2023 EDGES; default
|
|
33
|
+
* `$MAILWOMAN_DATA_ROOT/census/tiger2023-edges`, where `mailwoman situs interpolation` puts them)
|
|
34
|
+
* - `$MAILWOMAN_DATA_ROOT/oa-cache/us__il__cook.zip` (ZIP→city tails)
|
|
35
|
+
*/
|
|
36
|
+
import { dataRootPath } from "@mailwoman/core/data-root";
|
|
37
|
+
import { pathExists } from "@mailwoman/core/fs/readers";
|
|
38
|
+
import { writeLocalJSONFile } from "@mailwoman/core/fs/writers";
|
|
39
|
+
import { repoRootPath } from "@mailwoman/core/paths";
|
|
40
|
+
import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
|
|
41
|
+
import { JSONSpliterator } from "spliterator";
|
|
42
|
+
import { stableSourceID } from "#adapters/utils";
|
|
43
|
+
import { readZippedCSVRecords } from "#recipes/scaffold";
|
|
44
|
+
import { pick, weightedPick } from "#synthesizers/utils";
|
|
45
|
+
import { alignRow } from "#utils";
|
|
46
|
+
import { connectDuckDB } from "#utils/parquet";
|
|
47
|
+
const TRAIN_COUNTIES = [
|
|
48
|
+
{ fips: "17031", state: "IL", regime: "grid-city" },
|
|
49
|
+
{ fips: "34027", state: "NJ", regime: "suburb" },
|
|
50
|
+
];
|
|
51
|
+
const GOLDEN_COUNTIES = [{ fips: "50023", state: "VT", regime: "rural" }];
|
|
52
|
+
const EVAL_GOLD_PATH = repoRootPath("data", "eval", "external", "intersection-real.jsonl");
|
|
53
|
+
/**
|
|
54
|
+
* Where `mailwoman situs interpolation` unpacks the per-county TIGER EDGES shapefiles — a Census download, so it sits
|
|
55
|
+
* beside the other Census vintages under `census/`. `--edges-dir` overrides it.
|
|
56
|
+
*/
|
|
57
|
+
const DEFAULT_EDGES_DIR = dataRootPath("census", "tiger2023-edges");
|
|
58
|
+
const OA_COOK = { zip: dataRootPath("oa-cache", "us__il__cook.zip"), csv: "us/il/cook.csv" };
|
|
59
|
+
const FORMS = [
|
|
60
|
+
{ id: "amp", w: 0.2, render: (a, b) => `${a} & ${b}` },
|
|
61
|
+
{ id: "and", w: 0.2, render: (a, b) => `${a} and ${b}` },
|
|
62
|
+
{ id: "at", w: 0.12, render: (a, b) => `${a} at ${b}` },
|
|
63
|
+
{ id: "slash", w: 0.08, render: (a, b) => `${a} / ${b}` },
|
|
64
|
+
{ id: "slash-tight", w: 0.06, render: (a, b) => `${a}/${b}` },
|
|
65
|
+
{ id: "amp-tight", w: 0.05, render: (a, b) => `${a}&${b}` },
|
|
66
|
+
{ id: "at-sign", w: 0.06, render: (a, b) => `${a} @ ${b}` },
|
|
67
|
+
{ id: "corner-of", w: 0.115, render: (a, b) => `corner of ${a} and ${b}` },
|
|
68
|
+
{ id: "intersection-of", w: 0.115, render: (a, b) => `intersection of ${a} and ${b}` },
|
|
69
|
+
];
|
|
70
|
+
const TAILS = [
|
|
71
|
+
{ id: "bare", w: 0.55 },
|
|
72
|
+
{ id: "region", w: 0.16 },
|
|
73
|
+
{ id: "region-zip", w: 0.09 },
|
|
74
|
+
{ id: "city-region", w: 0.08 },
|
|
75
|
+
{ id: "city-region-zip", w: 0.12 },
|
|
76
|
+
];
|
|
77
|
+
const CASES = [
|
|
78
|
+
{ id: "as-is", w: 0.82, apply: (s) => s },
|
|
79
|
+
{ id: "upper", w: 0.12, apply: (s) => s.toUpperCase() },
|
|
80
|
+
{ id: "lower", w: 0.06, apply: (s) => s.toLowerCase() },
|
|
81
|
+
];
|
|
82
|
+
/**
|
|
83
|
+
* Words a connector may contribute as O tokens. The audit rejects any O token outside this set — an unlabeled
|
|
84
|
+
* street/locality token would surface here.
|
|
85
|
+
*/
|
|
86
|
+
const CONNECTOR_O_TOKENS = new Set(["and", "at", "of", "corner", "intersection"]);
|
|
87
|
+
/**
|
|
88
|
+
* Street names that would make the connector ambiguous or break verbatim alignment: embedded connector punctuation, or
|
|
89
|
+
* a standalone "and"/"at" word.
|
|
90
|
+
*/
|
|
91
|
+
const BAD_NAME = /[,&@/]|\b(and|at)\b/i;
|
|
92
|
+
/**
|
|
93
|
+
* Punctuation a connector form may leave between spans (besides whitespace): `, & @ /`.
|
|
94
|
+
*/
|
|
95
|
+
const CONNECTOR_PUNCT_RE = /^[\s,&@/]*$/;
|
|
96
|
+
/**
|
|
97
|
+
* Order-insensitive crossing key, for eval-leakage exclusion + pair dedup.
|
|
98
|
+
*/
|
|
99
|
+
const pairKey = (a, b) => [a.toLowerCase(), b.toLowerCase()].toSorted().join("\u001F");
|
|
100
|
+
/**
|
|
101
|
+
* Load the eval's crossings so neither train nor golden ever sees them.
|
|
102
|
+
*/
|
|
103
|
+
async function readEvalExclusions() {
|
|
104
|
+
const nodes = new Set();
|
|
105
|
+
const pairs = new Set();
|
|
106
|
+
if (!(await pathExists(EVAL_GOLD_PATH))) {
|
|
107
|
+
console.error(` WARN: eval gold not found at ${EVAL_GOLD_PATH} — no eval-leakage exclusion applied`);
|
|
108
|
+
return { nodes, pairs };
|
|
109
|
+
}
|
|
110
|
+
for await (const row of JSONSpliterator.fromAsync(EVAL_GOLD_PATH)) {
|
|
111
|
+
nodes.add(Number(row.node));
|
|
112
|
+
pairs.add(pairKey(row.components.intersection_a, row.components.intersection_b));
|
|
113
|
+
}
|
|
114
|
+
return { nodes, pairs };
|
|
115
|
+
}
|
|
116
|
+
/**
|
|
117
|
+
* Extract real crossings from one county's TIGER EDGES shapefile. Same query shape as the eval builder (2 incident
|
|
118
|
+
* distinct S1* FULLNAMEs at a node, both names >=6 chars), plus the edge ZIPL so tails can carry the crossing's own
|
|
119
|
+
* ZIP. Hash-ordered for seed-stable determinism.
|
|
120
|
+
*/
|
|
121
|
+
async function extractCrossings(db, edgesDir, county, seed) {
|
|
122
|
+
const shp = `${edgesDir}/tl_2023_${county.fips}_edges.shp`;
|
|
123
|
+
const result = await db.runAndReadAll(`
|
|
124
|
+
WITH incidence AS (
|
|
125
|
+
SELECT TNIDF AS node, FULLNAME AS name, ZIPL AS zip
|
|
126
|
+
FROM ST_Read('${shp}') WHERE MTFCC LIKE 'S1%' AND FULLNAME IS NOT NULL
|
|
127
|
+
UNION ALL
|
|
128
|
+
SELECT TNIDT AS node, FULLNAME AS name, ZIPL AS zip
|
|
129
|
+
FROM ST_Read('${shp}') WHERE MTFCC LIKE 'S1%' AND FULLNAME IS NOT NULL
|
|
130
|
+
),
|
|
131
|
+
nodes AS (
|
|
132
|
+
SELECT node,
|
|
133
|
+
list_sort(list_distinct(list(name))) AS names,
|
|
134
|
+
max(zip) AS zip
|
|
135
|
+
FROM incidence GROUP BY node
|
|
136
|
+
HAVING len(list_distinct(list(name))) = 2
|
|
137
|
+
)
|
|
138
|
+
SELECT node, names[1] AS a, names[2] AS b, zip,
|
|
139
|
+
hash(node::VARCHAR || '${seed}') AS h
|
|
140
|
+
FROM nodes
|
|
141
|
+
WHERE len(names[1]) >= 6 AND len(names[2]) >= 6
|
|
142
|
+
ORDER BY h
|
|
143
|
+
`);
|
|
144
|
+
const out = [];
|
|
145
|
+
for (const r of result.getRowObjects()) {
|
|
146
|
+
out.push({
|
|
147
|
+
a: String(r.a),
|
|
148
|
+
b: String(r.b),
|
|
149
|
+
zip: r.zip == null ? null : String(r.zip),
|
|
150
|
+
node: Number(r.node),
|
|
151
|
+
fips: county.fips,
|
|
152
|
+
state: county.state,
|
|
153
|
+
});
|
|
154
|
+
}
|
|
155
|
+
return out;
|
|
156
|
+
}
|
|
157
|
+
/**
|
|
158
|
+
* ZIP → majority city from the cached OA Cook-county CSV (real ZIP/city pairings).
|
|
159
|
+
*/
|
|
160
|
+
async function buildZipCityMap() {
|
|
161
|
+
const counts = new Map();
|
|
162
|
+
// zip → Map(city → n)
|
|
163
|
+
for await (const row of readZippedCSVRecords(OA_COOK.zip, OA_COOK.csv)) {
|
|
164
|
+
const city = row.city ?? "";
|
|
165
|
+
const zip = row.postcode ?? "";
|
|
166
|
+
if (!city || !/^\d{5}$/.test(zip) || BAD_NAME.test(city))
|
|
167
|
+
continue;
|
|
168
|
+
let byCity = counts.get(zip);
|
|
169
|
+
if (!byCity) {
|
|
170
|
+
counts.set(zip, (byCity = new Map()));
|
|
171
|
+
}
|
|
172
|
+
byCity.set(city, (byCity.get(city) ?? 0) + 1);
|
|
173
|
+
}
|
|
174
|
+
const map = new Map();
|
|
175
|
+
for (const [zip, byCity] of counts) {
|
|
176
|
+
let best = null;
|
|
177
|
+
for (const [city, n] of byCity)
|
|
178
|
+
if (!best || n > best.n) {
|
|
179
|
+
best = { city, n };
|
|
180
|
+
}
|
|
181
|
+
if (best) {
|
|
182
|
+
map.set(zip, best.city);
|
|
183
|
+
}
|
|
184
|
+
}
|
|
185
|
+
return map;
|
|
186
|
+
}
|
|
187
|
+
/**
|
|
188
|
+
* Render one crossing → { raw, components, formID, tailID, caseID }. Components are inserted in claim order (streets
|
|
189
|
+
* first) so alignment can't grab a region/postcode lookalike inside a street.
|
|
190
|
+
*/
|
|
191
|
+
function renderRow(random, crossing, zipCity) {
|
|
192
|
+
const form = weightedPick(FORMS, random, (f) => f.w);
|
|
193
|
+
const body = form.render(crossing.a, crossing.b);
|
|
194
|
+
let tail = weightedPick(TAILS, random, (t) => t.w);
|
|
195
|
+
const city = crossing.zip ? (zipCity.get(crossing.zip) ?? null) : null;
|
|
196
|
+
// Downgrade unsatisfiable tails (no ZIP on the edge / no city for the ZIP) to the region tail.
|
|
197
|
+
if ((tail.id === "region-zip" || tail.id === "city-region-zip") && !crossing.zip) {
|
|
198
|
+
tail = TAILS[1];
|
|
199
|
+
}
|
|
200
|
+
if ((tail.id === "city-region" || tail.id === "city-region-zip") && !city) {
|
|
201
|
+
tail = TAILS[1];
|
|
202
|
+
}
|
|
203
|
+
const components = { intersection_a: crossing.a, intersection_b: crossing.b };
|
|
204
|
+
let raw = body;
|
|
205
|
+
if (tail.id === "region") {
|
|
206
|
+
raw = `${body}, ${crossing.state}`;
|
|
207
|
+
components.region = crossing.state;
|
|
208
|
+
}
|
|
209
|
+
else if (tail.id === "region-zip") {
|
|
210
|
+
raw = `${body}, ${crossing.state} ${crossing.zip}`;
|
|
211
|
+
components.region = crossing.state;
|
|
212
|
+
components.postcode = crossing.zip;
|
|
213
|
+
}
|
|
214
|
+
else if (tail.id === "city-region") {
|
|
215
|
+
raw = `${body}, ${city}, ${crossing.state}`;
|
|
216
|
+
components.locality = city;
|
|
217
|
+
components.region = crossing.state;
|
|
218
|
+
}
|
|
219
|
+
else if (tail.id === "city-region-zip") {
|
|
220
|
+
raw = `${body}, ${city}, ${crossing.state} ${crossing.zip}`;
|
|
221
|
+
components.locality = city;
|
|
222
|
+
components.region = crossing.state;
|
|
223
|
+
components.postcode = crossing.zip;
|
|
224
|
+
}
|
|
225
|
+
const casing = weightedPick(CASES, random, (c) => c.w);
|
|
226
|
+
raw = casing.apply(raw);
|
|
227
|
+
// Components keep their original case; alignRow matches case-insensitively and labels the
|
|
228
|
+
// tokens of the (cased) raw — the parquet row carries tokens+labels only.
|
|
229
|
+
return { raw, components, formID: form.id, tailID: tail.id, caseID: casing.id };
|
|
230
|
+
}
|
|
231
|
+
/**
|
|
232
|
+
* Label-correctness audit for one aligned row, on the RAW SURFACE via the #519 span triple. Returns a list of
|
|
233
|
+
* violations (empty = clean). Re-derives the span checks independent of `alignRow`'s own assertion, so a builder bug
|
|
234
|
+
* can't vouch for itself.
|
|
235
|
+
*/
|
|
236
|
+
function auditRow(row, components) {
|
|
237
|
+
const errors = [];
|
|
238
|
+
const { raw, tokens, labels, span_starts, span_ends, span_tags } = row;
|
|
239
|
+
if (tokens.length !== labels.length) {
|
|
240
|
+
errors.push("tokens/labels length mismatch");
|
|
241
|
+
}
|
|
242
|
+
if (!span_starts || !span_ends || !span_tags) {
|
|
243
|
+
errors.push("missing the char-offset span triple (#519)");
|
|
244
|
+
return errors;
|
|
245
|
+
}
|
|
246
|
+
if (span_starts.length !== span_ends.length || span_starts.length !== span_tags.length) {
|
|
247
|
+
errors.push(`span triple not parallel: ${span_starts.length}/${span_ends.length}/${span_tags.length}`);
|
|
248
|
+
return errors;
|
|
249
|
+
}
|
|
250
|
+
for (let i = 0; i < span_starts.length; i++) {
|
|
251
|
+
if (!(span_starts[i] >= 0 && span_starts[i] < span_ends[i] && span_ends[i] <= raw.length)) {
|
|
252
|
+
errors.push(`span ${span_tags[i]}@[${span_starts[i]}, ${span_ends[i]}) out of bounds`);
|
|
253
|
+
}
|
|
254
|
+
if (i > 0 && span_starts[i] < span_ends[i - 1]) {
|
|
255
|
+
errors.push(`spans unsorted/overlapping at index ${i}`);
|
|
256
|
+
}
|
|
257
|
+
}
|
|
258
|
+
if (errors.length)
|
|
259
|
+
return errors;
|
|
260
|
+
const compCount = Object.keys(components).length;
|
|
261
|
+
if (span_tags.length !== compCount) {
|
|
262
|
+
errors.push(`span count ${span_tags.length} != components ${compCount}`);
|
|
263
|
+
}
|
|
264
|
+
// Raw-surface reconstruction: each component's single span slices raw to the component verbatim.
|
|
265
|
+
for (const [tag, value] of Object.entries(components)) {
|
|
266
|
+
if (value == null)
|
|
267
|
+
continue;
|
|
268
|
+
const indices = span_tags.map((t, i) => (t === tag ? i : -1)).filter((i) => i >= 0);
|
|
269
|
+
if (indices.length !== 1) {
|
|
270
|
+
errors.push(`${tag}: expected 1 span, got ${indices.length}`);
|
|
271
|
+
continue;
|
|
272
|
+
}
|
|
273
|
+
const idx = indices[0];
|
|
274
|
+
const got = raw.slice(span_starts[idx], span_ends[idx]);
|
|
275
|
+
if (got.toLowerCase() !== value.toLowerCase()) {
|
|
276
|
+
errors.push(`${tag} span "${got}" != component "${value}"`);
|
|
277
|
+
}
|
|
278
|
+
}
|
|
279
|
+
// Negative space: every char outside the spans must be connector material.
|
|
280
|
+
let cursor = 0;
|
|
281
|
+
const uncovered = [];
|
|
282
|
+
for (let i = 0; i < span_starts.length; i++) {
|
|
283
|
+
if (span_starts[i] > cursor) {
|
|
284
|
+
uncovered.push(raw.slice(cursor, span_starts[i]));
|
|
285
|
+
}
|
|
286
|
+
cursor = span_ends[i];
|
|
287
|
+
}
|
|
288
|
+
if (cursor < raw.length) {
|
|
289
|
+
uncovered.push(raw.slice(cursor));
|
|
290
|
+
}
|
|
291
|
+
for (const segment of uncovered) {
|
|
292
|
+
const words = segment.match(/[\p{L}\p{N}]+/gu) ?? [];
|
|
293
|
+
for (const word of words) {
|
|
294
|
+
if (!CONNECTOR_O_TOKENS.has(word.toLowerCase())) {
|
|
295
|
+
errors.push(`illegal uncovered word "${word}"`);
|
|
296
|
+
}
|
|
297
|
+
}
|
|
298
|
+
const punctOnly = segment.replaceAll(/[\p{L}\p{N}]+/gu, "");
|
|
299
|
+
if (!CONNECTOR_PUNCT_RE.test(punctOnly)) {
|
|
300
|
+
errors.push(`illegal uncovered punctuation in "${segment}"`);
|
|
301
|
+
}
|
|
302
|
+
}
|
|
303
|
+
return errors;
|
|
304
|
+
}
|
|
305
|
+
/**
|
|
306
|
+
* Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
|
|
307
|
+
* and `description` below for the surface form it generates.
|
|
308
|
+
*/
|
|
309
|
+
export const intersectionRecipe = {
|
|
310
|
+
name: "intersection",
|
|
311
|
+
description: "Real-pair intersection rows (US): TIGER 2023 EDGES crossings → audited intersection_a/b labels",
|
|
312
|
+
mode: "generate",
|
|
313
|
+
options: [
|
|
314
|
+
{
|
|
315
|
+
flag: "--edges-dir <dir>",
|
|
316
|
+
description: `Unzipped TIGER 2023 EDGES dir. Default ${DEFAULT_EDGES_DIR}`,
|
|
317
|
+
},
|
|
318
|
+
],
|
|
319
|
+
async run(opts, write) {
|
|
320
|
+
// Legacy build-intersection-slice.mjs seeded `mulberry32(opts.seed)`.
|
|
321
|
+
const random = makeMulberry32(opts.seed);
|
|
322
|
+
const count = opts.count ?? 40_000;
|
|
323
|
+
const source = opts.sourceName ?? "synth-intersection";
|
|
324
|
+
const edgesDir = opts.edgesDir ?? DEFAULT_EDGES_DIR;
|
|
325
|
+
const counties = opts.golden ? GOLDEN_COUNTIES : TRAIN_COUNTIES;
|
|
326
|
+
const exclusions = await readEvalExclusions();
|
|
327
|
+
console.error(` eval exclusions: ${exclusions.nodes.size} nodes, ${exclusions.pairs.size} pairs`);
|
|
328
|
+
const db = await connectDuckDB();
|
|
329
|
+
await db.run("INSTALL spatial; LOAD spatial;");
|
|
330
|
+
// Pool real crossings: eval-excluded, connector-safe names, one crossing per distinct pair.
|
|
331
|
+
const pool = [];
|
|
332
|
+
const seenPairs = new Set();
|
|
333
|
+
const stats = { evalExcluded: 0, badName: 0, dupPair: 0 };
|
|
334
|
+
for (const county of counties) {
|
|
335
|
+
const crossings = await extractCrossings(db, edgesDir, county, opts.seed);
|
|
336
|
+
let kept = 0;
|
|
337
|
+
for (const c of crossings) {
|
|
338
|
+
const key = pairKey(c.a, c.b);
|
|
339
|
+
if (exclusions.nodes.has(c.node) || exclusions.pairs.has(key)) {
|
|
340
|
+
stats.evalExcluded++;
|
|
341
|
+
continue;
|
|
342
|
+
}
|
|
343
|
+
if (BAD_NAME.test(c.a) || BAD_NAME.test(c.b) || c.a.includes(c.b) || c.b.includes(c.a)) {
|
|
344
|
+
stats.badName++;
|
|
345
|
+
continue;
|
|
346
|
+
}
|
|
347
|
+
if (seenPairs.has(key)) {
|
|
348
|
+
stats.dupPair++;
|
|
349
|
+
continue;
|
|
350
|
+
}
|
|
351
|
+
seenPairs.add(key);
|
|
352
|
+
pool.push(c);
|
|
353
|
+
kept++;
|
|
354
|
+
}
|
|
355
|
+
console.error(` ${county.fips} (${county.state}, ${county.regime}): ${crossings.length} crossings, ${kept} kept`);
|
|
356
|
+
}
|
|
357
|
+
if (!pool.length) {
|
|
358
|
+
throw new Error(`No crossings found — are the TIGER EDGES shapefiles present in ${edgesDir}?`);
|
|
359
|
+
}
|
|
360
|
+
const zipCity = opts.golden ? new Map() : await buildZipCityMap();
|
|
361
|
+
if (!opts.golden) {
|
|
362
|
+
console.error(` zip→city map: ${zipCity.size} ZIPs (OA Cook)`);
|
|
363
|
+
}
|
|
364
|
+
let emitted = 0;
|
|
365
|
+
let skipped = 0;
|
|
366
|
+
let guard = 0;
|
|
367
|
+
const formCounts = {};
|
|
368
|
+
const tailCounts = {};
|
|
369
|
+
const caseCounts = {};
|
|
370
|
+
const countyCounts = {};
|
|
371
|
+
const usedCrossings = new Set();
|
|
372
|
+
const seenRaw = new Set();
|
|
373
|
+
const auditErrors = [];
|
|
374
|
+
const samples = [];
|
|
375
|
+
while (emitted < count && guard++ < count * 10) {
|
|
376
|
+
const crossing = pick(pool, random);
|
|
377
|
+
const { raw, components, formID, tailID, caseID } = renderRow(random, crossing, zipCity);
|
|
378
|
+
if (seenRaw.has(raw)) {
|
|
379
|
+
skipped++;
|
|
380
|
+
continue;
|
|
381
|
+
}
|
|
382
|
+
if (opts.golden) {
|
|
383
|
+
seenRaw.add(raw);
|
|
384
|
+
write(JSON.stringify({ raw, components, country: "US", form: formID }) + "\n");
|
|
385
|
+
formCounts[formID] = (formCounts[formID] ?? 0) + 1;
|
|
386
|
+
tailCounts[tailID] = (tailCounts[tailID] ?? 0) + 1;
|
|
387
|
+
caseCounts[caseID] = (caseCounts[caseID] ?? 0) + 1;
|
|
388
|
+
countyCounts[crossing.fips] = (countyCounts[crossing.fips] ?? 0) + 1;
|
|
389
|
+
usedCrossings.add(crossing.node);
|
|
390
|
+
emitted++;
|
|
391
|
+
continue;
|
|
392
|
+
}
|
|
393
|
+
const canonical = {
|
|
394
|
+
raw,
|
|
395
|
+
components,
|
|
396
|
+
country: "US",
|
|
397
|
+
locale: "en-US",
|
|
398
|
+
source,
|
|
399
|
+
source_id: stableSourceID(source, components),
|
|
400
|
+
corpus_version: "0.4.0",
|
|
401
|
+
license: "TIGER/Line 2023 EDGES (US Census, public domain) real street pairs; OA Cook IL zip-to-city tails",
|
|
402
|
+
};
|
|
403
|
+
// Verbatim-only alignment: raw is built from the component values, so a fuzzy fallback could
|
|
404
|
+
// only ever mislabel (e.g. claim a lookalike window for a near-duplicate street).
|
|
405
|
+
const aligned = alignRow(canonical, { maxEditDistance: 0 });
|
|
406
|
+
if (aligned.kind !== "labeled" || !aligned.row) {
|
|
407
|
+
skipped++;
|
|
408
|
+
continue;
|
|
409
|
+
}
|
|
410
|
+
const violations = auditRow(aligned.row, components);
|
|
411
|
+
if (violations.length) {
|
|
412
|
+
auditErrors.push({ raw, violations });
|
|
413
|
+
continue;
|
|
414
|
+
}
|
|
415
|
+
seenRaw.add(raw);
|
|
416
|
+
write(JSON.stringify({ ...aligned.row, synth_method: "intersection", synth_base_id: null }) + "\n");
|
|
417
|
+
formCounts[formID] = (formCounts[formID] ?? 0) + 1;
|
|
418
|
+
tailCounts[tailID] = (tailCounts[tailID] ?? 0) + 1;
|
|
419
|
+
caseCounts[caseID] = (caseCounts[caseID] ?? 0) + 1;
|
|
420
|
+
countyCounts[crossing.fips] = (countyCounts[crossing.fips] ?? 0) + 1;
|
|
421
|
+
usedCrossings.add(crossing.node);
|
|
422
|
+
if (samples.length < FORMS.length && !samples.some((s) => s.form === formID)) {
|
|
423
|
+
samples.push({ form: formID, raw, tokens: aligned.row.tokens, labels: aligned.row.labels });
|
|
424
|
+
}
|
|
425
|
+
emitted++;
|
|
426
|
+
}
|
|
427
|
+
const report = {
|
|
428
|
+
mode: opts.golden ? "golden" : "train",
|
|
429
|
+
rows: emitted,
|
|
430
|
+
skipped,
|
|
431
|
+
pool: { crossings: pool.length, used: usedCrossings.size, ...stats },
|
|
432
|
+
per_county: countyCounts,
|
|
433
|
+
forms: formCounts,
|
|
434
|
+
tails: tailCounts,
|
|
435
|
+
cases: caseCounts,
|
|
436
|
+
audit: { errors: auditErrors.length, examples: auditErrors.slice(0, 10) },
|
|
437
|
+
seed: opts.seed,
|
|
438
|
+
source: "TIGER2023 EDGES via DuckDB ST_Read; node = 2 distinct S1* FULLNAMEs; eval crossings excluded",
|
|
439
|
+
samples,
|
|
440
|
+
};
|
|
441
|
+
await writeLocalJSONFile(report, opts.output.replace(/\.jsonl$/, ".report.json"));
|
|
442
|
+
console.error(`Done: emitted ${emitted} rows (skipped ${skipped}) from ${usedCrossings.size}/${pool.length} real crossings. → ${opts.output}\n` +
|
|
443
|
+
` forms: ${JSON.stringify(formCounts)}\n` +
|
|
444
|
+
` tails: ${JSON.stringify(tailCounts)}\n` +
|
|
445
|
+
` cases: ${JSON.stringify(caseCounts)}\n` +
|
|
446
|
+
` audit: ${auditErrors.length} violation(s)`);
|
|
447
|
+
if (auditErrors.length) {
|
|
448
|
+
throw new Error(`AUDIT FAILED — first violation: ${JSON.stringify(auditErrors[0])}`);
|
|
449
|
+
}
|
|
450
|
+
return { emitted, skipped };
|
|
451
|
+
},
|
|
452
|
+
};
|
|
453
|
+
//# sourceMappingURL=intersection.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"intersection.js","sourceRoot":"","sources":["../../lib/recipes/intersection.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkCG;AAGH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAA;AACvD,OAAO,EAAE,kBAAkB,EAAE,MAAM,4BAA4B,CAAA;AAC/D,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AAEpD,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAChD,OAAO,EAAE,oBAAoB,EAAqB,MAAM,mBAAmB,CAAA;AAC3E,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAA;AAExD,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AACjC,OAAO,EAAE,aAAa,EAAE,MAAM,gBAAgB,CAAA;AAQ9C,MAAM,cAAc,GAAsB;IACzC,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,WAAW,EAAE;IACnD,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE;CAChD,CAAA;AAED,MAAM,eAAe,GAAsB,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC,CAAA;AAE5F,MAAM,cAAc,GAAG,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,yBAAyB,CAAC,CAAA;AAC1F;;;GAGG;AACH,MAAM,iBAAiB,GAAG,YAAY,CAAC,QAAQ,EAAE,iBAAiB,CAAC,CAAA;AAEnE,MAAM,OAAO,GAAG,EAAE,GAAG,EAAE,YAAY,CAAC,UAAU,EAAE,kBAAkB,CAAC,EAAE,GAAG,EAAE,gBAAgB,EAAE,CAAA;AAwB5F,MAAM,KAAK,GAAoB;IAC9B,EAAE,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,EAAE;IACtD,EAAE,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,GAAG,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,EAAE;IACxD,EAAE,EAAE,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,EAAE;IACvD,EAAE,EAAE,EAAE,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,EAAE;IACzD,EAAE,EAAE,EAAE,aAAa,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE;IAC7D,EAAE,EAAE,EAAE,WAAW,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE;IAC3D,EAAE,EAAE,EAAE,SAAS,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,EAAE;IAC3D,EAAE,EAAE,EAAE,WAAW,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,aAAa,CAAC,QAAQ,CAAC,EAAE,EAAE;IAC1E,EAAE,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,mBAAmB,CAAC,QAAQ,CAAC,EAAE,EAAE;CACtF,CAAA;AAYD,MAAM,KAAK,GAAoB;IAC9B,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,EAAE,IAAI,EAAE;IACvB,EAAE,EAAE,EAAE,QAAQ,EAAE,CAAC,EAAE,IAAI,EAAE;IACzB,EAAE,EAAE,EAAE,YAAY,EAAE,CAAC,EAAE,IAAI,EAAE;IAC7B,EAAE,EAAE,EAAE,aAAa,EAAE,CAAC,EAAE,IAAI,EAAE;IAC9B,EAAE,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,IAAI,EAAE;CAClC,CAAA;AAQD,MAAM,KAAK,GAAsB;IAChC,EAAE,EAAE,EAAE,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE;IACzC,EAAE,EAAE,EAAE,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,EAAE;IACvD,EAAE,EAAE,EAAE,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,EAAE;CACvD,CAAA;AAED;;;GAGG;AACH,MAAM,kBAAkB,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,QAAQ,EAAE,cAAc,CAAC,CAAC,CAAA;AAEjF;;;GAGG;AACH,MAAM,QAAQ,GAAG,sBAAsB,CAAA;AAEvC;;GAEG;AACH,MAAM,kBAAkB,GAAG,aAAa,CAAA;AAExC;;GAEG;AACH,MAAM,OAAO,GAAG,CAAC,CAAS,EAAE,CAAS,EAAU,EAAE,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;AAU9G;;GAEG;AACH,KAAK,UAAU,kBAAkB;IAChC,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAA;IAC/B,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAA;IAE/B,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,cAAc,CAAC,CAAC,EAAE,CAAC;QACzC,OAAO,CAAC,KAAK,CAAC,kCAAkC,cAAc,sCAAsC,CAAC,CAAA;QAErG,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,CAAA;IACxB,CAAC;IAED,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAAc,cAAc,CAAC,EAAE,CAAC;QAChF,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAA;QAC3B,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,UAAU,CAAC,cAAc,EAAE,GAAG,CAAC,UAAU,CAAC,cAAc,CAAC,CAAC,CAAA;IACjF,CAAC;IAED,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,CAAA;AACxB,CAAC;AAED;;;;GAIG;AACH,KAAK,UAAU,gBAAgB,CAC9B,EAAoB,EACpB,QAAyB,EACzB,MAAc,EACd,IAAY;IAEZ,MAAM,GAAG,GAAG,GAAG,QAAQ,YAAY,MAAM,CAAC,IAAI,YAAY,CAAA;IAE1D,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC,aAAa,CAAC;;;mBAGpB,GAAG;;;mBAGH,GAAG;;;;;;;;;;4BAUM,IAAI;;;;EAI9B,CAAC,CAAA;IAEF,MAAM,GAAG,GAAe,EAAE,CAAA;IAE1B,KAAK,MAAM,CAAC,IAAI,MAAM,CAAC,aAAa,EAAE,EAAE,CAAC;QACxC,GAAG,CAAC,IAAI,CAAC;YACR,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;YACd,CAAC,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;YACd,GAAG,EAAE,CAAC,CAAC,GAAG,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC;YACzC,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC;YACpB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,KAAK,EAAE,MAAM,CAAC,KAAK;SACnB,CAAC,CAAA;IACH,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,eAAe;IAC7B,MAAM,MAAM,GAAG,IAAI,GAAG,EAA+B,CAAA;IAErD,sBAAsB;IACtB,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,oBAAoB,CAAC,OAAO,CAAC,GAAG,EAAE,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACxE,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,IAAI,EAAE,CAAA;QAC3B,MAAM,GAAG,GAAG,GAAG,CAAC,QAAQ,IAAI,EAAE,CAAA;QAE9B,IAAI,CAAC,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,SAAQ;QAClE,IAAI,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAE5B,IAAI,CAAC,MAAM,EAAE,CAAC;YACb,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,MAAM,GAAG,IAAI,GAAG,EAAE,CAAC,CAAC,CAAA;QACtC,CAAC;QAED,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;IAC9C,CAAC;IAED,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IAErC,KAAK,MAAM,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;QACpC,IAAI,IAAI,GAAuC,IAAI,CAAA;QAEnD,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,MAAM;YAC7B,IAAI,CAAC,IAAI,IAAI,CAAC,GAAG,IAAI,CAAC,CAAC,EAAE,CAAC;gBACzB,IAAI,GAAG,EAAE,IAAI,EAAE,CAAC,EAAE,CAAA;YACnB,CAAC;QAEF,IAAI,IAAI,EAAE,CAAC;YACV,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,CAAC,CAAA;QACxB,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;GAGG;AACH,SAAS,SAAS,CACjB,MAAoB,EACpB,QAAkB,EAClB,OAA4B;IAE5B,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IACpD,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC,CAAC,CAAA;IAEhD,IAAI,IAAI,GAAG,YAAY,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IAClD,MAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAEtE,+FAA+F;IAC/F,IAAI,CAAC,IAAI,CAAC,EAAE,KAAK,YAAY,IAAI,IAAI,CAAC,EAAE,KAAK,iBAAiB,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,EAAE,CAAC;QAClF,IAAI,GAAG,KAAK,CAAC,CAAC,CAAE,CAAA;IACjB,CAAC;IAED,IAAI,CAAC,IAAI,CAAC,EAAE,KAAK,aAAa,IAAI,IAAI,CAAC,EAAE,KAAK,iBAAiB,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAC3E,IAAI,GAAG,KAAK,CAAC,CAAC,CAAE,CAAA;IACjB,CAAC;IAED,MAAM,UAAU,GAA0C,EAAE,cAAc,EAAE,QAAQ,CAAC,CAAC,EAAE,cAAc,EAAE,QAAQ,CAAC,CAAC,EAAE,CAAA;IACpH,IAAI,GAAG,GAAG,IAAI,CAAA;IAEd,IAAI,IAAI,CAAC,EAAE,KAAK,QAAQ,EAAE,CAAC;QAC1B,GAAG,GAAG,GAAG,IAAI,KAAK,QAAQ,CAAC,KAAK,EAAE,CAAA;QAClC,UAAU,CAAC,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAA;IACnC,CAAC;SAAM,IAAI,IAAI,CAAC,EAAE,KAAK,YAAY,EAAE,CAAC;QACrC,GAAG,GAAG,GAAG,IAAI,KAAK,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,GAAG,EAAE,CAAA;QAClD,UAAU,CAAC,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAA;QAClC,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAC,GAAI,CAAA;IACpC,CAAC;SAAM,IAAI,IAAI,CAAC,EAAE,KAAK,aAAa,EAAE,CAAC;QACtC,GAAG,GAAG,GAAG,IAAI,KAAK,IAAI,KAAK,QAAQ,CAAC,KAAK,EAAE,CAAA;QAC3C,UAAU,CAAC,QAAQ,GAAG,IAAK,CAAA;QAC3B,UAAU,CAAC,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAA;IACnC,CAAC;SAAM,IAAI,IAAI,CAAC,EAAE,KAAK,iBAAiB,EAAE,CAAC;QAC1C,GAAG,GAAG,GAAG,IAAI,KAAK,IAAI,KAAK,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,GAAG,EAAE,CAAA;QAC3D,UAAU,CAAC,QAAQ,GAAG,IAAK,CAAA;QAC3B,UAAU,CAAC,MAAM,GAAG,QAAQ,CAAC,KAAK,CAAA;QAClC,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAC,GAAI,CAAA;IACpC,CAAC;IAED,MAAM,MAAM,GAAG,YAAY,CAAC,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IACtD,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;IAEvB,0FAA0F;IAC1F,0EAA0E;IAC1E,OAAO,EAAE,GAAG,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,EAAE,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,EAAE,CAAA;AAChF,CAAC;AAED;;;;GAIG;AACH,SAAS,QAAQ,CAAC,GAAe,EAAE,UAAiD;IACnF,MAAM,MAAM,GAAa,EAAE,CAAA;IAC3B,MAAM,EAAE,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,SAAS,EAAE,SAAS,EAAE,GAAG,GAAG,CAAA;IAEtE,IAAI,MAAM,CAAC,MAAM,KAAK,MAAM,CAAC,MAAM,EAAE,CAAC;QACrC,MAAM,CAAC,IAAI,CAAC,+BAA+B,CAAC,CAAA;IAC7C,CAAC;IAED,IAAI,CAAC,WAAW,IAAI,CAAC,SAAS,IAAI,CAAC,SAAS,EAAE,CAAC;QAC9C,MAAM,CAAC,IAAI,CAAC,4CAA4C,CAAC,CAAA;QAEzD,OAAO,MAAM,CAAA;IACd,CAAC;IAED,IAAI,WAAW,CAAC,MAAM,KAAK,SAAS,CAAC,MAAM,IAAI,WAAW,CAAC,MAAM,KAAK,SAAS,CAAC,MAAM,EAAE,CAAC;QACxF,MAAM,CAAC,IAAI,CAAC,6BAA6B,WAAW,CAAC,MAAM,IAAI,SAAS,CAAC,MAAM,IAAI,SAAS,CAAC,MAAM,EAAE,CAAC,CAAA;QAEtG,OAAO,MAAM,CAAA;IACd,CAAC;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC7C,IAAI,CAAC,CAAC,WAAW,CAAC,CAAC,CAAE,IAAI,CAAC,IAAI,WAAW,CAAC,CAAC,CAAE,GAAG,SAAS,CAAC,CAAC,CAAE,IAAI,SAAS,CAAC,CAAC,CAAE,IAAI,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;YAC/F,MAAM,CAAC,IAAI,CAAC,QAAQ,SAAS,CAAC,CAAC,CAAC,KAAK,WAAW,CAAC,CAAC,CAAC,KAAK,SAAS,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAA;QACvF,CAAC;QAED,IAAI,CAAC,GAAG,CAAC,IAAI,WAAW,CAAC,CAAC,CAAE,GAAG,SAAS,CAAC,CAAC,GAAG,CAAC,CAAE,EAAE,CAAC;YAClD,MAAM,CAAC,IAAI,CAAC,uCAAuC,CAAC,EAAE,CAAC,CAAA;QACxD,CAAC;IACF,CAAC;IAED,IAAI,MAAM,CAAC,MAAM;QAAE,OAAO,MAAM,CAAA;IAEhC,MAAM,SAAS,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,MAAM,CAAA;IAEhD,IAAI,SAAS,CAAC,MAAM,KAAK,SAAS,EAAE,CAAC;QACpC,MAAM,CAAC,IAAI,CAAC,cAAc,SAAS,CAAC,MAAM,kBAAkB,SAAS,EAAE,CAAC,CAAA;IACzE,CAAC;IAED,iGAAiG;IACjG,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,CAAC;QACvD,IAAI,KAAK,IAAI,IAAI;YAAE,SAAQ;QAC3B,MAAM,OAAO,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAA;QAEnF,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC1B,MAAM,CAAC,IAAI,CAAC,GAAG,GAAG,0BAA0B,OAAO,CAAC,MAAM,EAAE,CAAC,CAAA;YAE7D,SAAQ;QACT,CAAC;QAED,MAAM,GAAG,GAAG,OAAO,CAAC,CAAC,CAAE,CAAA;QACvB,MAAM,GAAG,GAAG,GAAG,CAAC,KAAK,CAAC,WAAW,CAAC,GAAG,CAAE,EAAE,SAAS,CAAC,GAAG,CAAE,CAAC,CAAA;QAEzD,IAAI,GAAG,CAAC,WAAW,EAAE,KAAK,KAAK,CAAC,WAAW,EAAE,EAAE,CAAC;YAC/C,MAAM,CAAC,IAAI,CAAC,GAAG,GAAG,UAAU,GAAG,mBAAmB,KAAK,GAAG,CAAC,CAAA;QAC5D,CAAC;IACF,CAAC;IAED,2EAA2E;IAC3E,IAAI,MAAM,GAAG,CAAC,CAAA;IACd,MAAM,SAAS,GAAa,EAAE,CAAA;IAE9B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC7C,IAAI,WAAW,CAAC,CAAC,CAAE,GAAG,MAAM,EAAE,CAAC;YAC9B,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,WAAW,CAAC,CAAC,CAAE,CAAC,CAAC,CAAA;QACnD,CAAC;QAED,MAAM,GAAG,SAAS,CAAC,CAAC,CAAE,CAAA;IACvB,CAAC;IAED,IAAI,MAAM,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC;QACzB,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAA;IAClC,CAAC;IAED,KAAK,MAAM,OAAO,IAAI,SAAS,EAAE,CAAC;QACjC,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,iBAAiB,CAAC,IAAI,EAAE,CAAA;QAEpD,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YAC1B,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC;gBACjD,MAAM,CAAC,IAAI,CAAC,2BAA2B,IAAI,GAAG,CAAC,CAAA;YAChD,CAAC;QACF,CAAC;QAED,MAAM,SAAS,GAAG,OAAO,CAAC,UAAU,CAAC,iBAAiB,EAAE,EAAE,CAAC,CAAA;QAE3D,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;YACzC,MAAM,CAAC,IAAI,CAAC,qCAAqC,OAAO,GAAG,CAAC,CAAA;QAC7D,CAAC;IACF,CAAC;IAED,OAAO,MAAM,CAAA;AACd,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAiB;IAC/C,IAAI,EAAE,cAAc;IACpB,WAAW,EAAE,gGAAgG;IAC7G,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE;QACR;YACC,IAAI,EAAE,mBAAmB;YACzB,WAAW,EAAE,0CAA0C,iBAAiB,EAAE;SAC1E;KACD;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,sEAAsE;QACtE,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,MAAM,CAAA;QAClC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,oBAAoB,CAAA;QACtD,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,iBAAiB,CAAA;QACnD,MAAM,QAAQ,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,cAAc,CAAA;QAC/D,MAAM,UAAU,GAAG,MAAM,kBAAkB,EAAE,CAAA;QAE7C,OAAO,CAAC,KAAK,CAAC,sBAAsB,UAAU,CAAC,KAAK,CAAC,IAAI,WAAW,UAAU,CAAC,KAAK,CAAC,IAAI,QAAQ,CAAC,CAAA;QAElG,MAAM,EAAE,GAAG,MAAM,aAAa,EAAE,CAAA;QAChC,MAAM,EAAE,CAAC,GAAG,CAAC,gCAAgC,CAAC,CAAA;QAE9C,4FAA4F;QAC5F,MAAM,IAAI,GAAe,EAAE,CAAA;QAC3B,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAA;QACnC,MAAM,KAAK,GAAG,EAAE,YAAY,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,CAAA;QAEzD,KAAK,MAAM,MAAM,IAAI,QAAQ,EAAE,CAAC;YAC/B,MAAM,SAAS,GAAG,MAAM,gBAAgB,CAAC,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,CAAC,IAAI,CAAC,CAAA;YACzE,IAAI,IAAI,GAAG,CAAC,CAAA;YAEZ,KAAK,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;gBAC3B,MAAM,GAAG,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAA;gBAE7B,IAAI,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;oBAC/D,KAAK,CAAC,YAAY,EAAE,CAAA;oBAEpB,SAAQ;gBACT,CAAC;gBAED,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;oBACxF,KAAK,CAAC,OAAO,EAAE,CAAA;oBAEf,SAAQ;gBACT,CAAC;gBAED,IAAI,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;oBACxB,KAAK,CAAC,OAAO,EAAE,CAAA;oBAEf,SAAQ;gBACT,CAAC;gBAED,SAAS,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;gBAClB,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;gBAEZ,IAAI,EAAE,CAAA;YACP,CAAC;YAED,OAAO,CAAC,KAAK,CAAC,KAAK,MAAM,CAAC,IAAI,KAAK,MAAM,CAAC,KAAK,KAAK,MAAM,CAAC,MAAM,MAAM,SAAS,CAAC,MAAM,eAAe,IAAI,OAAO,CAAC,CAAA;QACnH,CAAC;QAED,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;YAClB,MAAM,IAAI,KAAK,CAAC,kEAAkE,QAAQ,GAAG,CAAC,CAAA;QAC/F,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,GAAG,EAAkB,CAAC,CAAC,CAAC,MAAM,eAAe,EAAE,CAAA;QAEjF,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;YAClB,OAAO,CAAC,KAAK,CAAC,mBAAmB,OAAO,CAAC,IAAI,iBAAiB,CAAC,CAAA;QAChE,CAAC;QAED,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,KAAK,GAAG,CAAC,CAAA;QACb,MAAM,UAAU,GAA2B,EAAE,CAAA;QAC7C,MAAM,UAAU,GAA2B,EAAE,CAAA;QAC7C,MAAM,UAAU,GAA2B,EAAE,CAAA;QAC7C,MAAM,YAAY,GAA2B,EAAE,CAAA;QAC/C,MAAM,aAAa,GAAG,IAAI,GAAG,EAAU,CAAA;QACvC,MAAM,OAAO,GAAG,IAAI,GAAG,EAAU,CAAA;QACjC,MAAM,WAAW,GAAiD,EAAE,CAAA;QACpE,MAAM,OAAO,GAA+F,EAAE,CAAA;QAE9G,OAAO,OAAO,GAAG,KAAK,IAAI,KAAK,EAAE,GAAG,KAAK,GAAG,EAAE,EAAE,CAAC;YAChD,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,EAAE,MAAM,CAAC,CAAA;YACnC,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,SAAS,CAAC,MAAM,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAA;YAExF,IAAI,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC;gBACtB,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;gBAChB,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;gBAC9E,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBAClD,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBAClD,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBAClD,YAAY,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;gBACpE,aAAa,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAA;gBAEhC,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,SAAS,GAAiB;gBAC/B,GAAG;gBACH,UAAU;gBACV,OAAO,EAAE,IAAI;gBACb,MAAM,EAAE,OAAO;gBACf,MAAM;gBACN,SAAS,EAAE,cAAc,CAAC,MAAM,EAAE,UAAU,CAAC;gBAC7C,cAAc,EAAE,OAAO;gBACvB,OAAO,EAAE,kGAAkG;aAC3G,CAAA;YAED,6FAA6F;YAC7F,kFAAkF;YAClF,MAAM,OAAO,GAAG,QAAQ,CAAC,SAAS,EAAE,EAAE,eAAe,EAAE,CAAC,EAAE,CAAC,CAAA;YAE3D,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,UAAU,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,UAAU,CAAC,CAAA;YAEpD,IAAI,UAAU,CAAC,MAAM,EAAE,CAAC;gBACvB,WAAW,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,UAAU,EAAE,CAAC,CAAA;gBAErC,SAAQ;YACT,CAAC;YAED,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;YAChB,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,cAAc,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;YACnG,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;YAClD,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;YAClD,UAAU,CAAC,MAAM,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;YAClD,YAAY,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAA;YACpE,aAAa,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAA;YAEhC,IAAI,OAAO,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,MAAM,CAAC,EAAE,CAAC;gBAC9E,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAA;YAC5F,CAAC;YAED,OAAO,EAAE,CAAA;QACV,CAAC;QAED,MAAM,MAAM,GAAG;YACd,IAAI,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO;YACtC,IAAI,EAAE,OAAO;YACb,OAAO;YACP,IAAI,EAAE,EAAE,SAAS,EAAE,IAAI,CAAC,MAAM,EAAE,IAAI,EAAE,aAAa,CAAC,IAAI,EAAE,GAAG,KAAK,EAAE;YACpE,UAAU,EAAE,YAAY;YACxB,KAAK,EAAE,UAAU;YACjB,KAAK,EAAE,UAAU;YACjB,KAAK,EAAE,UAAU;YACjB,KAAK,EAAE,EAAE,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,QAAQ,EAAE,WAAW,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE;YACzE,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,MAAM,EAAE,8FAA8F;YACtG,OAAO;SACP,CAAA;QAED,MAAM,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,UAAU,EAAE,cAAc,CAAC,CAAC,CAAA;QAEjF,OAAO,CAAC,KAAK,CACZ,iBAAiB,OAAO,kBAAkB,OAAO,UAAU,aAAa,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,sBAAsB,IAAI,CAAC,MAAM,IAAI;YAChI,YAAY,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,IAAI;YAC1C,YAAY,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,IAAI;YAC1C,YAAY,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,IAAI;YAC1C,YAAY,WAAW,CAAC,MAAM,eAAe,CAC9C,CAAA;QAED,IAAI,WAAW,CAAC,MAAM,EAAE,CAAC;YACxB,MAAM,IAAI,KAAK,CAAC,mCAAmC,IAAI,CAAC,SAAS,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;QACrF,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}
|
|
@@ -0,0 +1,136 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
* @license AGPL-3.0
|
|
4
|
+
* @author Teffen Ellis, et al.
|
|
5
|
+
*
|
|
6
|
+
* `locale` slice recipe — the multi-locale generalization of the `german` recipe. Reads REAL
|
|
7
|
+
* OpenAddresses tuples for a `--country` (DE/FR/NL/IT/ES), renders each via
|
|
8
|
+
* {@link synthesizeLocaleRow} in BOTH orders (`--intl-fraction`, default 0.4 international / the
|
|
9
|
+
* rest country-native), aligns to BIO, and emits a labeled JSONL. Generate-mode: it STREAMS each
|
|
10
|
+
* source CSV (a streamed zip member for cached zips, plain `createReadStream` for extracted CSVs) and
|
|
11
|
+
* reservoir-samples to {@link RESERVOIR_CAP} (so FR/ES countrywide work in bounded memory), then
|
|
12
|
+
* draws `--count` rows from the pool with the passed `random`. Ported from
|
|
13
|
+
* scripts/build-locale-slice.mjs.
|
|
14
|
+
*
|
|
15
|
+
* The reservoir uses its OWN seeded PRNG ({@link makeMulberry32}, per part), independent of the
|
|
16
|
+
* emit `random`, so the input sample is reproducible WITHOUT perturbing the synth/order draws.
|
|
17
|
+
*
|
|
18
|
+
* SURFACE DIVERSITY (#241): two per-country shape draws ride the emit loop, sized by the
|
|
19
|
+
* 2026-07-02 format-diversity audit against the `openaddresses-{es,nl,it}-sample.jsonl` observed
|
|
20
|
+
* forms. ES: the OpenCage template comma-joins the house number (`CALLE MAYOR, 12`) but all 3,000
|
|
21
|
+
* eval rows space-join (`CALLE MAYOR 12`) — {@link ES_SPACE_JOIN_FRACTION} of native rows collapse
|
|
22
|
+
* the comma. NL: OA (and the eval, 3,000/3,000) glue the postcode (`1187LM`) while the national
|
|
23
|
+
* convention spaces it (`1187 LM`) — {@link NL_GLUED_POSTCODE_FRACTION} of rows keep the glued
|
|
24
|
+
* source shape, the rest the spaced conventional one. These draws are consumed ONLY for their
|
|
25
|
+
* country, so DE/FR emit streams are unchanged for a given seed.
|
|
26
|
+
*/
|
|
27
|
+
import type { PathBuilderLike } from "path-ts";
|
|
28
|
+
import type { CorpusRecipe } from "#recipes/scaffold";
|
|
29
|
+
import { type LocaleBaseTuple, type SynthesizedLocaleRow } from "#synthesizers/german";
|
|
30
|
+
/**
|
|
31
|
+
* One per-country OA source part: either a cached `zip` + `csv` member (streamed out of the archive) or an extracted
|
|
32
|
+
* plain `path` (streamed via `createReadStream`). Both carry the standard OA header
|
|
33
|
+
* (LON,LAT,NUMBER,STREET,UNIT,CITY,DISTRICT,REGION,POSTCODE,ID,HASH). An optional `region` fallback covers countries
|
|
34
|
+
* whose REGION column is empty (DE — the Bundesland is implied by the per-state file).
|
|
35
|
+
*/
|
|
36
|
+
export interface LocalePart {
|
|
37
|
+
zip?: PathBuilderLike;
|
|
38
|
+
csv?: string;
|
|
39
|
+
path?: PathBuilderLike;
|
|
40
|
+
region?: string;
|
|
41
|
+
/**
|
|
42
|
+
* NZ — the OA DISTRICT column holds the city (`Auckland`) and CITY holds the suburb (`Birkenhead`). When set, map
|
|
43
|
+
* DISTRICT→locality and CITY→dependent_locality (falling back to CITY→locality when DISTRICT is empty, ~18% of NZ
|
|
44
|
+
* rows). Without this, the default CITY→locality mapping wrongly trains the suburb as the locality.
|
|
45
|
+
*/
|
|
46
|
+
districtAsLocality?: boolean;
|
|
47
|
+
/**
|
|
48
|
+
* ES pedanía part only — the header is the RAW (un-conformed) CNIG export schema (`numero`, `tipo_vial`,
|
|
49
|
+
* `nombre_via`, `poblacion`, `municipio`, `comunidad_autonoma`, `cod_postal`), NOT the standard OA
|
|
50
|
+
* NUMBER/STREET/CITY/DISTRICT/REGION/POSTCODE header every other part uses. Verified 2026-07-22 by exact- coordinate
|
|
51
|
+
* cross-check: the OA-conformed `extracted/es/countrywide.csv` collapses CITY to `municipio` and drops `poblacion`
|
|
52
|
+
* (Spain's below-municipio núcleo/pedanía name) entirely, so the pedanía signal survives ONLY in this raw export.
|
|
53
|
+
* `street` is reconstructed as `tipo_vial + " " + nombre_via` (verified byte-identical to the conformed STREET column
|
|
54
|
+
* for the same row). CITY-analog = `poblacion`, DISTRICT-analog = `municipio`.
|
|
55
|
+
*/
|
|
56
|
+
cnigRaw?: boolean;
|
|
57
|
+
}
|
|
58
|
+
export interface LocaleCountrySource {
|
|
59
|
+
source: string;
|
|
60
|
+
parts: LocalePart[];
|
|
61
|
+
/**
|
|
62
|
+
* The `corpus_version` stamped on emitted rows. DE/FR keep the historical `0.4.0` (regenerating those slices must
|
|
63
|
+
* stay lineage-identical); ES/IT/NL are the #241 staging lineage (`v0.9.9-es-it-nl`).
|
|
64
|
+
*/
|
|
65
|
+
corpusVersion: string;
|
|
66
|
+
/**
|
|
67
|
+
* An ALTERNATE part list, read instead of {@link parts} when the `--district-as-locality` override is explicitly
|
|
68
|
+
* `true` for this invocation (see `run()`). ES-only for now — the standard `parts` entry can't supply real
|
|
69
|
+
* dependent-locality signal (its OA-conformed CSV drops `poblacion`; see {@link LocalePart.cnigRaw}), so the pedanía
|
|
70
|
+
* build reads a wholly different raw source instead of flipping the standard CITY/DISTRICT columns. `undefined` for
|
|
71
|
+
* every other country — the override then just forces `districtAsLocality` on the normal `parts`, as GB/NZ already do
|
|
72
|
+
* per-part.
|
|
73
|
+
*/
|
|
74
|
+
pedaniaParts?: LocalePart[];
|
|
75
|
+
}
|
|
76
|
+
/**
|
|
77
|
+
* OA CITY-noise normalization (#241) — the documented cleaning step, derived from the 2026-07-02 FULL-STREAM audit of
|
|
78
|
+
* the ES (15.6M rows), IT (13.9M), and NL (9.1M) sources (not a hand-list). Returns the cleaned city, or `null` to drop
|
|
79
|
+
* the tuple.
|
|
80
|
+
*
|
|
81
|
+
* Cleaned classes:
|
|
82
|
+
*
|
|
83
|
+
* 1. DROP pseudo-localities — the ES cadastral aggregates (`Comunidad de 09076, 09150 y 09578`, `Ledanía de …`; 0.06% of
|
|
84
|
+
* ES rows): any CITY containing a comma or a ≥4-digit run is a land-register aggregate, not a renderable city.
|
|
85
|
+
* Structural, locale-safe — NL's genuine `2e Valthermond` (one digit) survives; IT/NL have zero hits.
|
|
86
|
+
* 2. STRIP a trailing parenthesized 1–3-letter admin code — the NL BAG province disambiguator (`Bergen (NH)`, `Rijswijk
|
|
87
|
+
* (GLD)` → `Bergen`, `Rijswijk`; 0.13% of NL rows). The analogue of the German Kreis/region-suffix class (#241 names
|
|
88
|
+
* `Rabenau Sachs` / `Weißwasser /O.L.`): an admin-region gloss glued onto the locality value that dirties locality
|
|
89
|
+
* labels.
|
|
90
|
+
*
|
|
91
|
+
* Audit-verified NON-noise, deliberately NOT cleaned (a naive suffix rule would mangle real names):
|
|
92
|
+
*
|
|
93
|
+
* - ES/IT city-ends-with-province (`Alhama de Almería`, `GENZANO DI ROMA`; ~0.8% each): genuine toponyms whose linking
|
|
94
|
+
* `de`/`di` makes them full names, unlike the German glued-abbreviation class.
|
|
95
|
+
* - ES bilingual slash names (`Laudio/Llodio`; 2.16%): official co-names — the eval expects them verbatim.
|
|
96
|
+
* - IT ALL-CAPS city casing (98.79% of the source, and the eval's observed form): casing is the #829 case-augmentation
|
|
97
|
+
* change, not this slice's.
|
|
98
|
+
*/
|
|
99
|
+
export declare function cleanCityNoise(city: string): string | null;
|
|
100
|
+
/**
|
|
101
|
+
* Stream real tuples out of an OA source part and reservoir-sample to {@link RESERVOIR_CAP}. Reads the CSV row-by-row —
|
|
102
|
+
* `readZipEntry | CSVSpliterator` for zip parts, `createReadStream | CSVSpliterator` for extracted parts (both bounded
|
|
103
|
+
* memory) — and keeps a uniform random sample (Algorithm R) seeded by `rng`, separate from the emit loop's PRNG. NO
|
|
104
|
+
* global dedup (a 25M-key Set would OOM; OA rows are near-unique). The city passes through {@link cleanCityNoise}; the
|
|
105
|
+
* region falls back to `part.region` when the row's REGION cell is empty (DE).
|
|
106
|
+
*
|
|
107
|
+
* Exported for {@link locale.test.ts} — the CSV read path (quote handling, CRLF, region fallback) has no other test.
|
|
108
|
+
*/
|
|
109
|
+
export declare function readTuples(part: LocalePart, rng: () => number): Promise<LocaleBaseTuple[]>;
|
|
110
|
+
/**
|
|
111
|
+
* Country-append fraction (the fr-admin-split #728 pattern, generalized to the locale recipe): mutates `synth` in
|
|
112
|
+
* place, `countryFraction` of the time appending an explicit country surface form ("United Kingdom") to `raw` + a
|
|
113
|
+
* `country` component — the model relearns to emit country WHEN the token is present without over-firing it on the
|
|
114
|
+
* (still-majority) country-less rows. `countryFraction <= 0` (the default) short-circuits the `random()` draw away
|
|
115
|
+
* entirely — no `synth` mutation and no RNG consumption — so every existing locale's emit stream stays byte-identical
|
|
116
|
+
* to before this option existed. Exported for {@link locale.test.ts}.
|
|
117
|
+
*/
|
|
118
|
+
export declare function applyCountryAppend(synth: SynthesizedLocaleRow, country: string, countryFraction: number, random: () => number): void;
|
|
119
|
+
/**
|
|
120
|
+
* Merge the `--district-as-locality` CLI override onto one part. `undefined` (flag absent) returns `part` unchanged
|
|
121
|
+
* (same object — no allocation, no behavior change); `true`/`false` returns a shallow copy with `districtAsLocality`
|
|
122
|
+
* forced to that value for this invocation only. Exported for {@link locale.test.ts}.
|
|
123
|
+
*/
|
|
124
|
+
export declare function applyDistrictAsLocalityOverride(part: LocalePart, override: boolean | undefined): LocalePart;
|
|
125
|
+
/**
|
|
126
|
+
* Pick which part list a `--country` run reads: {@link LocaleCountrySource.pedaniaParts} when the override is explicitly
|
|
127
|
+
* `true` AND the country registers one (ES only, so far), else the default `parts` — unchanged for every other
|
|
128
|
+
* country/override combination. Exported for {@link locale.test.ts}.
|
|
129
|
+
*/
|
|
130
|
+
export declare function resolveLocaleParts(countrySource: LocaleCountrySource, override: boolean | undefined): LocalePart[];
|
|
131
|
+
/**
|
|
132
|
+
* Slice recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
|
|
133
|
+
* and `description` below for the surface form it generates.
|
|
134
|
+
*/
|
|
135
|
+
export declare const localeRecipe: CorpusRecipe;
|
|
136
|
+
//# sourceMappingURL=locale.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"locale.d.ts","sourceRoot":"","sources":["../../lib/recipes/locale.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AAQH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAA;AAI9C,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAA;AACrD,OAAO,EAAE,KAAK,eAAe,EAAE,KAAK,oBAAoB,EAAuB,MAAM,sBAAsB,CAAA;AAI3G;;;;;GAKG;AACH,MAAM,WAAW,UAAU;IAC1B,GAAG,CAAC,EAAE,eAAe,CAAA;IACrB,GAAG,CAAC,EAAE,MAAM,CAAA;IACZ,IAAI,CAAC,EAAE,eAAe,CAAA;IACtB,MAAM,CAAC,EAAE,MAAM,CAAA;IACf;;;;OAIG;IACH,kBAAkB,CAAC,EAAE,OAAO,CAAA;IAC5B;;;;;;;;OAQG;IACH,OAAO,CAAC,EAAE,OAAO,CAAA;CACjB;AAED,MAAM,WAAW,mBAAmB;IACnC,MAAM,EAAE,MAAM,CAAA;IACd,KAAK,EAAE,UAAU,EAAE,CAAA;IACnB;;;OAGG;IACH,aAAa,EAAE,MAAM,CAAA;IACrB;;;;;;;OAOG;IACH,YAAY,CAAC,EAAE,UAAU,EAAE,CAAA;CAC3B;AA6FD;;;;;;;;;;;;;;;;;;;;;;GAsBG;AACH,wBAAgB,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAM1D;AAgBD;;;;;;;;GAQG;AACH,wBAAsB,UAAU,CAAC,IAAI,EAAE,UAAU,EAAE,GAAG,EAAE,MAAM,MAAM,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAkJhG;AAED;;;;;;;GAOG;AACH,wBAAgB,kBAAkB,CACjC,KAAK,EAAE,oBAAoB,EAC3B,OAAO,EAAE,MAAM,EACf,eAAe,EAAE,MAAM,EACvB,MAAM,EAAE,MAAM,MAAM,GAClB,IAAI,CAgBN;AAED;;;;GAIG;AACH,wBAAgB,+BAA+B,CAAC,IAAI,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,GAAG,SAAS,GAAG,UAAU,CAE3G;AAED;;;;GAIG;AACH,wBAAgB,kBAAkB,CAAC,aAAa,EAAE,mBAAmB,EAAE,QAAQ,EAAE,OAAO,GAAG,SAAS,GAAG,UAAU,EAAE,CAElH;AAED;;;GAGG;AACH,eAAO,MAAM,YAAY,EAAE,YAmK1B,CAAA"}
|