@mailwoman/corpus 9.4.0 → 10.0.0
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 +16 -0
- package/lib/adapters/geonames/adapter.ts +3 -3
- package/lib/adapters/geonames/postal/adapter.ts +122 -0
- package/lib/adapters/index.ts +39 -32
- package/lib/adapters/openaddresses/adapter.ts +4 -7
- package/lib/adapters/osm/adapter.ts +278 -0
- package/lib/adapters/overture/adapter.ts +22 -7
- package/lib/adapters/utils/index.ts +1 -1
- package/lib/adapters/wof/admin/json/adapter.ts +239 -0
- package/lib/adapters/wof/json-rows.ts +181 -0
- package/lib/adapters/wof/postalcode-json/adapter.ts +196 -0
- package/lib/au/adapters/gnaf/adapter.ts +152 -0
- package/lib/au/adapters/gnaf/assemble.ts +225 -0
- package/lib/de/recipes/locale.ts +360 -0
- package/lib/fr/adapters/ban/adapter.ts +159 -0
- package/lib/fr/recipes/admin-split.ts +243 -0
- package/lib/fr/recipes/bare-street.ts +171 -0
- package/lib/fr/recipes/fragment.ts +334 -0
- package/lib/fr/recipes/lieudit.ts +274 -0
- package/lib/fr/recipes/order.ts +246 -0
- package/lib/fr/tools/fetch/ban.ts +289 -0
- package/lib/index.ts +1 -1
- package/lib/international/recipes/locale.ts +598 -0
- package/lib/jp/tools/fetch/houjin.ts +141 -0
- package/lib/kr/adapters/juso/label-rows.ts +259 -0
- package/lib/kr/adapters/localdata/align.ts +299 -0
- package/lib/kr/adapters/localdata/key-index.ts +185 -0
- package/lib/kr/adapters/localdata/permit-rows.ts +186 -0
- package/lib/kr/tools/fetch/juso.ts +172 -0
- package/lib/kr/tools/fetch/localdata.ts +206 -0
- package/lib/recipes/bare/postcode/eval.ts +99 -0
- package/lib/recipes/bare/postcode/index.ts +327 -0
- package/lib/recipes/country-balanced.ts +1 -1
- package/lib/recipes/index.ts +28 -21
- package/lib/recipes/intersection.ts +1 -1
- package/lib/recipes/po/box/cedex/au-nz.ts +82 -0
- package/lib/recipes/po/box/cedex/ca.ts +58 -0
- package/lib/recipes/po/box/cedex/fr.ts +76 -0
- package/lib/recipes/po/box/cedex/phrases.ts +108 -0
- package/lib/recipes/po/box/cedex/recipe.ts +203 -0
- package/lib/recipes/po/box/cedex/sources.ts +189 -0
- package/lib/recipes/po/box/cedex/types.ts +35 -0
- package/lib/recipes/po/box/cedex/us.ts +69 -0
- package/lib/recipes/po/box/cedex/vocabulary.ts +106 -0
- package/lib/recipes/po/box/cedex.ts +9 -0
- package/lib/recipes/po/box/index.ts +128 -0
- package/lib/recipes/scaffold.ts +18 -41
- package/lib/recipes/street/affix.ts +807 -0
- package/lib/recipes/sub/venue/context.ts +149 -0
- package/lib/recipes/sub/venue/index.ts +997 -0
- package/lib/recipes/sub/venue/sources.ts +739 -0
- package/lib/recipes/trailing-region.ts +6 -1
- package/lib/recipes/unit.ts +1 -1
- package/lib/sg/recipes/register.ts +270 -0
- package/lib/sg/tools/fetch/acra.ts +179 -0
- package/lib/south-asia/recipes/register.ts +214 -0
- package/lib/synthesizers/anchor-absorption.ts +1 -1
- package/lib/synthesizers/locale.ts +261 -0
- package/lib/synthesizers/street.ts +1 -1
- package/lib/synthesizers/utils.ts +2 -2
- package/lib/tools/audit.ts +3 -2
- package/lib/tools/corpus-stats.ts +3 -3
- package/lib/tools/fetch/download/index.ts +10 -0
- package/lib/tools/fetch/download/manifest.ts +102 -0
- package/lib/tools/fetch/download/network.ts +226 -0
- package/lib/tools/fetch/download/retry.ts +173 -0
- package/lib/tools/fetch/geonames/dump.ts +257 -0
- package/lib/tools/fetch/geonames/postal.ts +180 -0
- package/lib/tools/fetch/index.ts +30 -30
- package/lib/tools/fetch/openaddresses.ts +2 -2
- package/lib/tools/fetch/ourairports.ts +2 -2
- package/lib/tools/fetch/ppd.ts +1 -3
- package/lib/tools/fetch/wikidata-subvenue.ts +2 -2
- package/lib/tools/golden/promote.ts +340 -0
- package/lib/tools/golden/relabel-street/directory.ts +340 -0
- package/lib/tools/golden/relabel-street/index.ts +9 -0
- package/lib/tools/golden/relabel-street/row.ts +361 -0
- package/lib/tools/index.ts +10 -10
- package/lib/tools/ingest-csv.ts +5 -7
- package/lib/tools/lint/slice/vocab.ts +376 -0
- package/lib/tools/overlay-manifest.ts +92 -25
- package/lib/tools/overture-subvenue.ts +1 -1
- package/lib/tools/postcode-triples.ts +138 -14
- package/lib/tools/sub/venue/harvest.ts +234 -0
- package/lib/tools/sub/venue/head-nouns.ts +245 -0
- package/lib/tools/sub/venue/lexicon.ts +429 -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/tw/tools/fetch/gcis.ts +180 -0
- package/lib/types.ts +1 -1
- package/lib/us/adapters/fcc-bdc/adapter.ts +173 -0
- package/lib/us/adapters/state/hi-schools/adapter.ts +160 -0
- package/lib/us/adapters/state/ia-contractors/adapter.ts +136 -0
- package/lib/us/adapters/state/ny-notaries/adapter.ts +133 -0
- package/lib/us/adapters/state/tx-notaries/adapter.ts +149 -0
- package/lib/us/adapters/tiger/adapter.ts +250 -0
- package/lib/us/adapters/usgov/hrsa-fqhc/adapter.ts +156 -0
- package/lib/us/adapters/usgov/imls-pls/adapter.ts +134 -0
- package/lib/us/adapters/usgov/irs-bmf/adapter.ts +145 -0
- package/lib/us/adapters/usgov/nad/adapter.ts +324 -0
- package/lib/us/adapters/usgov/nppes/adapter.ts +143 -0
- package/lib/us/adapters/usgov/samhsa-treatment-locator/adapter.ts +182 -0
- package/lib/us/tools/fetch/hrsa.ts +59 -0
- package/lib/us/tools/fetch/imls-pls.ts +149 -0
- package/lib/us/tools/fetch/nad.ts +341 -0
- package/lib/us/tools/fetch/nppes.ts +165 -0
- package/lib/us/tools/fetch/state/hi-schools.ts +161 -0
- package/lib/us/tools/fetch/state/sources.ts +146 -0
- package/lib/us/tools/fetch/tiger-full.ts +308 -0
- package/lib/utils/align.ts +1 -1
- package/lib/utils/golden.ts +12 -8
- package/lib/utils/wof-json.ts +7 -13
- package/out/adapters/geonames/adapter.d.ts +1 -1
- package/out/adapters/geonames/adapter.js +3 -3
- package/out/adapters/geonames/adapter.js.map +1 -1
- 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/index.d.ts +16 -16
- package/out/adapters/index.d.ts.map +1 -1
- package/out/adapters/index.js +34 -32
- package/out/adapters/index.js.map +1 -1
- package/out/adapters/openaddresses/adapter.d.ts.map +1 -1
- package/out/adapters/openaddresses/adapter.js +4 -6
- package/out/adapters/openaddresses/adapter.js.map +1 -1
- package/out/adapters/osm/adapter.d.ts +95 -0
- package/out/adapters/osm/adapter.d.ts.map +1 -0
- package/out/adapters/osm/adapter.js +211 -0
- package/out/adapters/osm/adapter.js.map +1 -0
- package/out/adapters/overture/adapter.d.ts +10 -0
- package/out/adapters/overture/adapter.d.ts.map +1 -1
- package/out/adapters/overture/adapter.js +20 -7
- package/out/adapters/overture/adapter.js.map +1 -1
- package/out/adapters/synth-po-box/adapter.js.map +1 -1
- package/out/adapters/utils/index.d.ts +1 -1
- package/out/adapters/utils/index.d.ts.map +1 -1
- package/out/adapters/utils/index.js.map +1 -1
- 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 +204 -0
- package/out/adapters/wof/admin/json/adapter.js.map +1 -0
- package/out/adapters/wof/json-rows.d.ts +73 -0
- package/out/adapters/wof/json-rows.d.ts.map +1 -0
- package/out/adapters/wof/json-rows.js +115 -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 +167 -0
- package/out/adapters/wof/postalcode-json/adapter.js.map +1 -0
- package/out/au/adapters/gnaf/adapter.d.ts.map +1 -0
- package/out/au/adapters/gnaf/adapter.js +127 -0
- package/out/au/adapters/gnaf/adapter.js.map +1 -0
- package/out/au/adapters/gnaf/assemble.d.ts +65 -0
- package/out/au/adapters/gnaf/assemble.d.ts.map +1 -0
- package/out/au/adapters/gnaf/assemble.js +154 -0
- package/out/au/adapters/gnaf/assemble.js.map +1 -0
- package/out/cn/recipes/organizational-units.d.ts +36 -0
- package/out/cn/recipes/organizational-units.d.ts.map +1 -0
- package/out/cn/recipes/organizational-units.js +159 -0
- package/out/cn/recipes/organizational-units.js.map +1 -0
- package/out/cz/recipes/pcfirst-preposition.d.ts +22 -0
- package/out/cz/recipes/pcfirst-preposition.d.ts.map +1 -0
- package/out/cz/recipes/pcfirst-preposition.js +101 -0
- package/out/cz/recipes/pcfirst-preposition.js.map +1 -0
- package/out/de/recipes/locale.d.ts +44 -0
- package/out/de/recipes/locale.d.ts.map +1 -0
- package/out/de/recipes/locale.js +289 -0
- package/out/de/recipes/locale.js.map +1 -0
- package/out/env.d.ts +1 -1
- package/out/fr/adapters/ban/adapter.d.ts.map +1 -0
- package/out/fr/adapters/ban/adapter.js +123 -0
- package/out/fr/adapters/ban/adapter.js.map +1 -0
- package/out/fr/adapters/ban/street-decompose.d.ts.map +1 -0
- package/out/fr/adapters/ban/street-decompose.js.map +1 -0
- package/out/fr/recipes/admin-split.d.ts +38 -0
- package/out/fr/recipes/admin-split.d.ts.map +1 -0
- package/out/fr/recipes/admin-split.js +192 -0
- package/out/fr/recipes/admin-split.js.map +1 -0
- package/out/fr/recipes/bare-street.d.ts +28 -0
- package/out/fr/recipes/bare-street.d.ts.map +1 -0
- package/out/fr/recipes/bare-street.js +144 -0
- package/out/fr/recipes/bare-street.js.map +1 -0
- package/out/fr/recipes/fragment.d.ts +77 -0
- package/out/fr/recipes/fragment.d.ts.map +1 -0
- package/out/fr/recipes/fragment.js +277 -0
- package/out/fr/recipes/fragment.js.map +1 -0
- package/out/fr/recipes/lieudit.d.ts +32 -0
- package/out/fr/recipes/lieudit.d.ts.map +1 -0
- package/out/fr/recipes/lieudit.js +211 -0
- package/out/fr/recipes/lieudit.js.map +1 -0
- package/out/fr/recipes/order.d.ts +38 -0
- package/out/fr/recipes/order.d.ts.map +1 -0
- package/out/fr/recipes/order.js +198 -0
- package/out/fr/recipes/order.js.map +1 -0
- package/out/fr/tools/fetch/ban.d.ts +31 -0
- package/out/fr/tools/fetch/ban.d.ts.map +1 -0
- package/out/fr/tools/fetch/ban.js +239 -0
- package/out/fr/tools/fetch/ban.js.map +1 -0
- package/out/index.d.ts +1 -1
- package/out/index.d.ts.map +1 -1
- package/out/index.js +1 -1
- package/out/index.js.map +1 -1
- package/out/international/recipes/locale.d.ts +136 -0
- package/out/international/recipes/locale.d.ts.map +1 -0
- package/out/international/recipes/locale.js +458 -0
- package/out/international/recipes/locale.js.map +1 -0
- package/out/jp/adapters/wof-admin/adapter.d.ts.map +1 -0
- package/out/jp/adapters/wof-admin/adapter.js.map +1 -0
- package/out/jp/tools/fetch/houjin.d.ts +24 -0
- package/out/jp/tools/fetch/houjin.d.ts.map +1 -0
- package/out/jp/tools/fetch/houjin.js +102 -0
- package/out/jp/tools/fetch/houjin.js.map +1 -0
- package/out/kr/adapters/juso/label-rows.d.ts +76 -0
- package/out/kr/adapters/juso/label-rows.d.ts.map +1 -0
- package/out/kr/adapters/juso/label-rows.js +194 -0
- package/out/kr/adapters/juso/label-rows.js.map +1 -0
- package/out/kr/adapters/localdata/align.d.ts +43 -0
- package/out/kr/adapters/localdata/align.d.ts.map +1 -0
- package/out/kr/adapters/localdata/align.js +223 -0
- package/out/kr/adapters/localdata/align.js.map +1 -0
- package/out/kr/adapters/localdata/key-index.d.ts +71 -0
- package/out/kr/adapters/localdata/key-index.d.ts.map +1 -0
- package/out/kr/adapters/localdata/key-index.js +133 -0
- package/out/kr/adapters/localdata/key-index.js.map +1 -0
- package/out/kr/adapters/localdata/permit-rows.d.ts +66 -0
- package/out/kr/adapters/localdata/permit-rows.d.ts.map +1 -0
- package/out/kr/adapters/localdata/permit-rows.js +131 -0
- package/out/kr/adapters/localdata/permit-rows.js.map +1 -0
- package/out/kr/tools/fetch/juso.d.ts +30 -0
- package/out/kr/tools/fetch/juso.d.ts.map +1 -0
- package/out/kr/tools/fetch/juso.js +120 -0
- package/out/kr/tools/fetch/juso.js.map +1 -0
- package/out/kr/tools/fetch/localdata.d.ts +30 -0
- package/out/kr/tools/fetch/localdata.d.ts.map +1 -0
- package/out/kr/tools/fetch/localdata.js +139 -0
- package/out/kr/tools/fetch/localdata.js.map +1 -0
- package/out/nl/recipes/postcode.d.ts +26 -0
- package/out/nl/recipes/postcode.d.ts.map +1 -0
- package/out/nl/recipes/postcode.js +103 -0
- package/out/nl/recipes/postcode.js.map +1 -0
- package/out/no/recipes/fragment.d.ts +43 -0
- package/out/no/recipes/fragment.d.ts.map +1 -0
- package/out/no/recipes/fragment.js +183 -0
- package/out/no/recipes/fragment.js.map +1 -0
- package/out/no/recipes/street/index.d.ts +18 -0
- package/out/no/recipes/street/index.d.ts.map +1 -0
- package/out/no/recipes/street/index.js +72 -0
- package/out/no/recipes/street/index.js.map +1 -0
- package/out/no/recipes/street/led.d.ts +24 -0
- package/out/no/recipes/street/led.d.ts.map +1 -0
- package/out/no/recipes/street/led.js +114 -0
- package/out/no/recipes/street/led.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/bare/postcode/eval.d.ts +29 -0
- package/out/recipes/bare/postcode/eval.d.ts.map +1 -0
- package/out/recipes/bare/postcode/eval.js +85 -0
- package/out/recipes/bare/postcode/eval.js.map +1 -0
- package/out/recipes/bare/postcode/index.d.ts +62 -0
- package/out/recipes/bare/postcode/index.d.ts.map +1 -0
- package/out/recipes/bare/postcode/index.js +267 -0
- package/out/recipes/bare/postcode/index.js.map +1 -0
- package/out/recipes/country-balanced.js.map +1 -1
- package/out/recipes/index.d.ts.map +1 -1
- package/out/recipes/index.js +28 -21
- package/out/recipes/index.js.map +1 -1
- package/out/recipes/intersection.js.map +1 -1
- package/out/recipes/po/box/cedex/au-nz.d.ts +7 -0
- package/out/recipes/po/box/cedex/au-nz.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/au-nz.js +56 -0
- package/out/recipes/po/box/cedex/au-nz.js.map +1 -0
- package/out/recipes/po/box/cedex/ca.d.ts +7 -0
- package/out/recipes/po/box/cedex/ca.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/ca.js +38 -0
- package/out/recipes/po/box/cedex/ca.js.map +1 -0
- package/out/recipes/po/box/cedex/fr.d.ts +7 -0
- package/out/recipes/po/box/cedex/fr.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/fr.js +57 -0
- package/out/recipes/po/box/cedex/fr.js.map +1 -0
- package/out/recipes/po/box/cedex/phrases.d.ts +8 -0
- package/out/recipes/po/box/cedex/phrases.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/phrases.js +76 -0
- package/out/recipes/po/box/cedex/phrases.js.map +1 -0
- package/out/recipes/po/box/cedex/recipe.d.ts +9 -0
- package/out/recipes/po/box/cedex/recipe.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/recipe.js +153 -0
- package/out/recipes/po/box/cedex/recipe.js.map +1 -0
- package/out/recipes/po/box/cedex/sources.d.ts +86 -0
- package/out/recipes/po/box/cedex/sources.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/sources.js +143 -0
- package/out/recipes/po/box/cedex/sources.js.map +1 -0
- package/out/recipes/po/box/cedex/types.d.ts +31 -0
- package/out/recipes/po/box/cedex/types.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/types.js +5 -0
- package/out/recipes/po/box/cedex/types.js.map +1 -0
- package/out/recipes/po/box/cedex/us.d.ts +7 -0
- package/out/recipes/po/box/cedex/us.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/us.js +51 -0
- package/out/recipes/po/box/cedex/us.js.map +1 -0
- package/out/recipes/po/box/cedex/vocabulary.d.ts +68 -0
- package/out/recipes/po/box/cedex/vocabulary.d.ts.map +1 -0
- package/out/recipes/po/box/cedex/vocabulary.js +90 -0
- package/out/recipes/po/box/cedex/vocabulary.js.map +1 -0
- package/out/recipes/po/box/cedex.d.ts +9 -0
- package/out/recipes/po/box/cedex.d.ts.map +1 -0
- package/out/recipes/po/box/cedex.js +9 -0
- package/out/recipes/po/box/cedex.js.map +1 -0
- package/out/recipes/po/box/index.d.ts +17 -0
- package/out/recipes/po/box/index.d.ts.map +1 -0
- package/out/recipes/po/box/index.js +105 -0
- package/out/recipes/po/box/index.js.map +1 -0
- package/out/recipes/scaffold.d.ts +20 -0
- package/out/recipes/scaffold.d.ts.map +1 -1
- package/out/recipes/scaffold.js +18 -37
- package/out/recipes/scaffold.js.map +1 -1
- 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 +565 -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/sub/venue/context.d.ts +29 -0
- package/out/recipes/sub/venue/context.d.ts.map +1 -0
- package/out/recipes/sub/venue/context.js +106 -0
- package/out/recipes/sub/venue/context.js.map +1 -0
- package/out/recipes/sub/venue/index.d.ts +270 -0
- package/out/recipes/sub/venue/index.d.ts.map +1 -0
- package/out/recipes/sub/venue/index.js +728 -0
- package/out/recipes/sub/venue/index.js.map +1 -0
- package/out/recipes/sub/venue/sources.d.ts +209 -0
- package/out/recipes/sub/venue/sources.d.ts.map +1 -0
- package/out/recipes/sub/venue/sources.js +555 -0
- package/out/recipes/sub/venue/sources.js.map +1 -0
- package/out/recipes/trailing-region.d.ts.map +1 -1
- package/out/recipes/trailing-region.js +6 -1
- package/out/recipes/trailing-region.js.map +1 -1
- package/out/recipes/unit.js.map +1 -1
- package/out/sg/recipes/register.d.ts +69 -0
- package/out/sg/recipes/register.d.ts.map +1 -0
- package/out/sg/recipes/register.js +211 -0
- package/out/sg/recipes/register.js.map +1 -0
- package/out/sg/tools/fetch/acra.d.ts +23 -0
- package/out/sg/tools/fetch/acra.d.ts.map +1 -0
- package/out/sg/tools/fetch/acra.js +126 -0
- package/out/sg/tools/fetch/acra.js.map +1 -0
- package/out/si/recipes/bare-village.d.ts +32 -0
- package/out/si/recipes/bare-village.d.ts.map +1 -0
- package/out/si/recipes/bare-village.js +89 -0
- package/out/si/recipes/bare-village.js.map +1 -0
- package/out/south-asia/recipes/register.d.ts +51 -0
- package/out/south-asia/recipes/register.d.ts.map +1 -0
- package/out/south-asia/recipes/register.js +154 -0
- package/out/south-asia/recipes/register.js.map +1 -0
- package/out/synthesizers/anchor-absorption.d.ts +1 -1
- package/out/synthesizers/anchor-absorption.d.ts.map +1 -1
- package/out/synthesizers/anchor-absorption.js.map +1 -1
- package/out/synthesizers/boundary-stress.js.map +1 -1
- package/out/synthesizers/house-venue.js.map +1 -1
- package/out/synthesizers/intersection.js.map +1 -1
- package/out/synthesizers/locale.d.ts +116 -0
- package/out/synthesizers/locale.d.ts.map +1 -0
- package/out/synthesizers/locale.js +150 -0
- package/out/synthesizers/locale.js.map +1 -0
- package/out/synthesizers/no-street.d.ts.map +1 -1
- package/out/synthesizers/no-street.js.map +1 -1
- package/out/synthesizers/po-box.js.map +1 -1
- package/out/synthesizers/street.d.ts.map +1 -1
- package/out/synthesizers/street.js +1 -1
- package/out/synthesizers/street.js.map +1 -1
- package/out/synthesizers/utils.js +1 -1
- package/out/synthesizers/utils.js.map +1 -1
- package/out/test-kit/corpus-recipe.d.ts.map +1 -1
- package/out/test-kit/corpus-recipe.js.map +1 -1
- package/out/tools/audit.d.ts.map +1 -1
- package/out/tools/audit.js +3 -2
- package/out/tools/audit.js.map +1 -1
- package/out/tools/corpus-stats.js +3 -3
- package/out/tools/corpus-stats.js.map +1 -1
- package/out/tools/fetch/download/index.d.ts +10 -0
- package/out/tools/fetch/download/index.d.ts.map +1 -0
- package/out/tools/fetch/download/index.js +10 -0
- package/out/tools/fetch/download/index.js.map +1 -0
- package/out/tools/fetch/download/manifest.d.ts +47 -0
- package/out/tools/fetch/download/manifest.d.ts.map +1 -0
- package/out/tools/fetch/download/manifest.js +74 -0
- package/out/tools/fetch/download/manifest.js.map +1 -0
- package/out/tools/fetch/download/network.d.ts +105 -0
- package/out/tools/fetch/download/network.d.ts.map +1 -0
- package/out/tools/fetch/download/network.js +130 -0
- package/out/tools/fetch/download/network.js.map +1 -0
- package/out/tools/fetch/download/retry.d.ts +45 -0
- package/out/tools/fetch/download/retry.d.ts.map +1 -0
- package/out/tools/fetch/download/retry.js +136 -0
- package/out/tools/fetch/download/retry.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/index.d.ts +30 -30
- package/out/tools/fetch/index.d.ts.map +1 -1
- package/out/tools/fetch/index.js +30 -30
- package/out/tools/fetch/index.js.map +1 -1
- package/out/tools/fetch/openaddresses.d.ts +1 -1
- package/out/tools/fetch/openaddresses.d.ts.map +1 -1
- package/out/tools/fetch/openaddresses.js +1 -1
- package/out/tools/fetch/openaddresses.js.map +1 -1
- package/out/tools/fetch/ourairports.d.ts +1 -1
- package/out/tools/fetch/ourairports.d.ts.map +1 -1
- package/out/tools/fetch/ourairports.js +1 -1
- package/out/tools/fetch/ourairports.js.map +1 -1
- package/out/tools/fetch/ppd.js +1 -3
- package/out/tools/fetch/ppd.js.map +1 -1
- package/out/tools/fetch/wikidata-subvenue.d.ts +1 -1
- package/out/tools/fetch/wikidata-subvenue.d.ts.map +1 -1
- package/out/tools/fetch/wikidata-subvenue.js +1 -1
- package/out/tools/fetch/wikidata-subvenue.js.map +1 -1
- 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 +72 -0
- package/out/tools/golden/promote.d.ts.map +1 -0
- package/out/tools/golden/promote.js +243 -0
- package/out/tools/golden/promote.js.map +1 -0
- package/out/tools/golden/relabel-street/directory.d.ts +83 -0
- package/out/tools/golden/relabel-street/directory.d.ts.map +1 -0
- package/out/tools/golden/relabel-street/directory.js +216 -0
- package/out/tools/golden/relabel-street/directory.js.map +1 -0
- package/out/tools/golden/relabel-street/index.d.ts +9 -0
- package/out/tools/golden/relabel-street/index.d.ts.map +1 -0
- package/out/tools/golden/relabel-street/index.js +9 -0
- package/out/tools/golden/relabel-street/index.js.map +1 -0
- package/out/tools/golden/relabel-street/row.d.ts +121 -0
- package/out/tools/golden/relabel-street/row.d.ts.map +1 -0
- package/out/tools/golden/relabel-street/row.js +232 -0
- package/out/tools/golden/relabel-street/row.js.map +1 -0
- package/out/tools/index.d.ts +10 -10
- package/out/tools/index.d.ts.map +1 -1
- package/out/tools/index.js +10 -10
- package/out/tools/index.js.map +1 -1
- package/out/tools/ingest-csv.d.ts.map +1 -1
- package/out/tools/ingest-csv.js +5 -7
- package/out/tools/ingest-csv.js.map +1 -1
- package/out/tools/lint/slice/index.d.ts +93 -0
- package/out/tools/lint/slice/index.d.ts.map +1 -0
- package/out/tools/lint/slice/index.js +322 -0
- package/out/tools/lint/slice/index.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 +254 -0
- package/out/tools/lint/slice/vocab.js.map +1 -0
- package/out/tools/overlay-manifest.d.ts +56 -6
- package/out/tools/overlay-manifest.d.ts.map +1 -1
- package/out/tools/overlay-manifest.js +56 -15
- package/out/tools/overlay-manifest.js.map +1 -1
- package/out/tools/overture-subvenue.d.ts +1 -1
- package/out/tools/postcode-triples.d.ts +36 -0
- package/out/tools/postcode-triples.d.ts.map +1 -1
- package/out/tools/postcode-triples.js +91 -12
- package/out/tools/postcode-triples.js.map +1 -1
- 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.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/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.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.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/tw/tools/fetch/gcis.d.ts +23 -0
- package/out/tw/tools/fetch/gcis.d.ts.map +1 -0
- package/out/tw/tools/fetch/gcis.js +135 -0
- package/out/tw/tools/fetch/gcis.js.map +1 -0
- package/out/types.d.ts +1 -1
- package/out/types.d.ts.map +1 -1
- package/out/us/adapters/fcc-bdc/adapter.d.ts.map +1 -0
- package/out/us/adapters/fcc-bdc/adapter.js +142 -0
- package/out/us/adapters/fcc-bdc/adapter.js.map +1 -0
- package/out/us/adapters/state/hi-schools/adapter.d.ts.map +1 -0
- package/out/us/adapters/state/hi-schools/adapter.js +136 -0
- package/out/us/adapters/state/hi-schools/adapter.js.map +1 -0
- package/out/us/adapters/state/hi-schools/workbook.d.ts.map +1 -0
- package/out/us/adapters/state/hi-schools/workbook.js.map +1 -0
- package/out/us/adapters/state/ia-contractors/adapter.d.ts.map +1 -0
- package/out/us/adapters/state/ia-contractors/adapter.js +104 -0
- package/out/us/adapters/state/ia-contractors/adapter.js.map +1 -0
- package/out/us/adapters/state/ny-notaries/adapter.d.ts.map +1 -0
- package/out/us/adapters/state/ny-notaries/adapter.js +112 -0
- package/out/us/adapters/state/ny-notaries/adapter.js.map +1 -0
- package/out/us/adapters/state/tx-notaries/adapter.d.ts.map +1 -0
- package/out/us/adapters/state/tx-notaries/adapter.js +118 -0
- package/out/us/adapters/state/tx-notaries/adapter.js.map +1 -0
- package/out/us/adapters/tiger/adapter.d.ts +55 -0
- package/out/us/adapters/tiger/adapter.d.ts.map +1 -0
- package/out/us/adapters/tiger/adapter.js +197 -0
- package/out/us/adapters/tiger/adapter.js.map +1 -0
- package/out/us/adapters/tiger/street-decompose.d.ts.map +1 -0
- package/out/us/adapters/tiger/street-decompose.js.map +1 -0
- package/out/us/adapters/usgov/hrsa-fqhc/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/hrsa-fqhc/adapter.js +121 -0
- package/out/us/adapters/usgov/hrsa-fqhc/adapter.js.map +1 -0
- package/out/us/adapters/usgov/imls-pls/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/imls-pls/adapter.js +106 -0
- package/out/us/adapters/usgov/imls-pls/adapter.js.map +1 -0
- package/out/us/adapters/usgov/irs-bmf/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/irs-bmf/adapter.js +116 -0
- package/out/us/adapters/usgov/irs-bmf/adapter.js.map +1 -0
- package/out/us/adapters/usgov/nad/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/nad/adapter.js +239 -0
- package/out/us/adapters/usgov/nad/adapter.js.map +1 -0
- package/out/us/adapters/usgov/nppes/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/nppes/adapter.js +110 -0
- package/out/us/adapters/usgov/nppes/adapter.js.map +1 -0
- package/out/us/adapters/usgov/samhsa-treatment-locator/adapter.d.ts.map +1 -0
- package/out/us/adapters/usgov/samhsa-treatment-locator/adapter.js +143 -0
- package/out/us/adapters/usgov/samhsa-treatment-locator/adapter.js.map +1 -0
- package/out/us/fips-state.d.ts +41 -0
- package/out/us/fips-state.d.ts.map +1 -0
- package/out/us/fips-state.js +100 -0
- package/out/us/fips-state.js.map +1 -0
- package/out/us/tools/fetch/hrsa.d.ts +18 -0
- package/out/us/tools/fetch/hrsa.d.ts.map +1 -0
- package/out/us/tools/fetch/hrsa.js +46 -0
- package/out/us/tools/fetch/hrsa.js.map +1 -0
- package/out/us/tools/fetch/imls-pls.d.ts +25 -0
- package/out/us/tools/fetch/imls-pls.d.ts.map +1 -0
- package/out/us/tools/fetch/imls-pls.js +108 -0
- package/out/us/tools/fetch/imls-pls.js.map +1 -0
- package/out/us/tools/fetch/nad.d.ts +70 -0
- package/out/us/tools/fetch/nad.d.ts.map +1 -0
- package/out/us/tools/fetch/nad.js +234 -0
- package/out/us/tools/fetch/nad.js.map +1 -0
- package/out/us/tools/fetch/nppes.d.ts +26 -0
- package/out/us/tools/fetch/nppes.d.ts.map +1 -0
- package/out/us/tools/fetch/nppes.js +126 -0
- package/out/us/tools/fetch/nppes.js.map +1 -0
- package/out/us/tools/fetch/state/hi-schools.d.ts +26 -0
- package/out/us/tools/fetch/state/hi-schools.d.ts.map +1 -0
- package/out/us/tools/fetch/state/hi-schools.js +112 -0
- package/out/us/tools/fetch/state/hi-schools.js.map +1 -0
- package/out/us/tools/fetch/state/sources.d.ts +26 -0
- package/out/us/tools/fetch/state/sources.d.ts.map +1 -0
- package/out/us/tools/fetch/state/sources.js +111 -0
- package/out/us/tools/fetch/state/sources.js.map +1 -0
- package/out/us/tools/fetch/tiger-full.d.ts +22 -0
- package/out/us/tools/fetch/tiger-full.d.ts.map +1 -0
- package/out/us/tools/fetch/tiger-full.js +217 -0
- package/out/us/tools/fetch/tiger-full.js.map +1 -0
- package/out/utils/align.d.ts +1 -1
- package/out/utils/align.d.ts.map +1 -1
- package/out/utils/align.js.map +1 -1
- package/out/utils/golden.d.ts +7 -3
- package/out/utils/golden.d.ts.map +1 -1
- package/out/utils/golden.js +12 -8
- package/out/utils/golden.js.map +1 -1
- package/out/utils/parquet.d.ts.map +1 -1
- package/out/utils/split.js.map +1 -1
- package/out/utils/wof-json.d.ts.map +1 -1
- package/out/utils/wof-json.js +7 -12
- package/out/utils/wof-json.js.map +1 -1
- package/out/ve/recipes/reviewed-postcode-tail.d.ts.map +1 -0
- package/out/ve/recipes/reviewed-postcode-tail.js.map +1 -0
- package/package.json +196 -46
- package/lib/adapters/ban/adapter.ts +0 -190
- package/lib/adapters/fcc-bdc/adapter.ts +0 -174
- package/lib/adapters/geonames-postal/adapter.ts +0 -122
- package/lib/adapters/gnaf/adapter.ts +0 -149
- package/lib/adapters/gnaf/assemble.ts +0 -225
- package/lib/adapters/state-hi-schools/adapter.ts +0 -170
- package/lib/adapters/state-ia-contractors/adapter.ts +0 -144
- package/lib/adapters/state-ny-notaries/adapter.ts +0 -141
- package/lib/adapters/state-tx-notaries/adapter.ts +0 -157
- package/lib/adapters/tiger/adapter.ts +0 -252
- package/lib/adapters/usgov-hrsa-fqhc/adapter.ts +0 -167
- package/lib/adapters/usgov-imls-pls/adapter.ts +0 -143
- package/lib/adapters/usgov-irs-bmf/adapter.ts +0 -151
- package/lib/adapters/usgov-nad/adapter.ts +0 -334
- package/lib/adapters/usgov-nppes/adapter.ts +0 -151
- package/lib/adapters/usgov-samhsa-treatment-locator/adapter.ts +0 -193
- package/lib/adapters/wof-admin-json/adapter.ts +0 -231
- package/lib/adapters/wof-json-rows.ts +0 -129
- package/lib/adapters/wof-postalcode-json/adapter.ts +0 -176
- package/lib/recipes/fr-admin-split.ts +0 -243
- package/lib/recipes/fr-bare-street.ts +0 -171
- package/lib/recipes/fr-fragment.ts +0 -334
- package/lib/recipes/fr-lieudit.ts +0 -301
- package/lib/recipes/fr-order.ts +0 -246
- package/lib/recipes/german.ts +0 -360
- package/lib/recipes/locale.ts +0 -600
- package/lib/recipes/po-box-cedex.ts +0 -976
- package/lib/recipes/po-box.ts +0 -124
- package/lib/recipes/street-affix.ts +0 -805
- package/lib/recipes/sub-venue-sources.ts +0 -878
- package/lib/recipes/sub-venue.ts +0 -998
- package/lib/synthesizers/german.ts +0 -260
- package/lib/tools/fetch/acra-sg.ts +0 -174
- package/lib/tools/fetch/ban.ts +0 -289
- package/lib/tools/fetch/download.ts +0 -477
- package/lib/tools/fetch/gcis-tw.ts +0 -175
- package/lib/tools/fetch/geonames-dump.ts +0 -257
- package/lib/tools/fetch/geonames-postal.ts +0 -180
- package/lib/tools/fetch/houjin-jp.ts +0 -135
- package/lib/tools/fetch/hrsa.ts +0 -59
- package/lib/tools/fetch/imls-pls.ts +0 -149
- package/lib/tools/fetch/juso-kr.ts +0 -165
- package/lib/tools/fetch/localdata-kr.ts +0 -194
- package/lib/tools/fetch/nad.ts +0 -341
- package/lib/tools/fetch/nppes.ts +0 -165
- package/lib/tools/fetch/state-hi-schools.ts +0 -161
- package/lib/tools/fetch/state-sources.ts +0 -146
- package/lib/tools/fetch/tiger-full.ts +0 -308
- package/lib/tools/golden-promote.ts +0 -337
- package/lib/tools/golden-relabel-street.ts +0 -686
- package/lib/tools/lint-slice-vocab.ts +0 -388
- package/lib/tools/sub-venue/harvest.ts +0 -234
- package/lib/tools/sub-venue/head-nouns.ts +0 -245
- package/lib/tools/sub-venue/surfaces.ts +0 -95
- package/lib/tools/sub-venue/table.ts +0 -281
- package/lib/tools/sub-venue/wikidata.ts +0 -85
- package/lib/tools/sub-venue-lexicon.ts +0 -429
- package/out/adapters/ban/adapter.d.ts.map +0 -1
- package/out/adapters/ban/adapter.js +0 -149
- package/out/adapters/ban/adapter.js.map +0 -1
- package/out/adapters/ban/street-decompose.d.ts.map +0 -1
- package/out/adapters/ban/street-decompose.js.map +0 -1
- package/out/adapters/fcc-bdc/adapter.d.ts.map +0 -1
- package/out/adapters/fcc-bdc/adapter.js +0 -144
- package/out/adapters/fcc-bdc/adapter.js.map +0 -1
- package/out/adapters/geonames-postal/adapter.d.ts.map +0 -1
- package/out/adapters/geonames-postal/adapter.js +0 -107
- package/out/adapters/geonames-postal/adapter.js.map +0 -1
- package/out/adapters/gnaf/adapter.d.ts.map +0 -1
- package/out/adapters/gnaf/adapter.js +0 -124
- package/out/adapters/gnaf/adapter.js.map +0 -1
- package/out/adapters/gnaf/assemble.d.ts +0 -65
- package/out/adapters/gnaf/assemble.d.ts.map +0 -1
- package/out/adapters/gnaf/assemble.js +0 -154
- package/out/adapters/gnaf/assemble.js.map +0 -1
- package/out/adapters/state-hi-schools/adapter.d.ts.map +0 -1
- package/out/adapters/state-hi-schools/adapter.js +0 -145
- package/out/adapters/state-hi-schools/adapter.js.map +0 -1
- package/out/adapters/state-hi-schools/workbook.d.ts.map +0 -1
- package/out/adapters/state-hi-schools/workbook.js.map +0 -1
- package/out/adapters/state-ia-contractors/adapter.d.ts.map +0 -1
- package/out/adapters/state-ia-contractors/adapter.js +0 -111
- package/out/adapters/state-ia-contractors/adapter.js.map +0 -1
- package/out/adapters/state-ny-notaries/adapter.d.ts.map +0 -1
- package/out/adapters/state-ny-notaries/adapter.js +0 -119
- package/out/adapters/state-ny-notaries/adapter.js.map +0 -1
- package/out/adapters/state-tx-notaries/adapter.d.ts.map +0 -1
- package/out/adapters/state-tx-notaries/adapter.js +0 -125
- package/out/adapters/state-tx-notaries/adapter.js.map +0 -1
- package/out/adapters/tiger/adapter.d.ts +0 -55
- package/out/adapters/tiger/adapter.d.ts.map +0 -1
- package/out/adapters/tiger/adapter.js +0 -201
- package/out/adapters/tiger/adapter.js.map +0 -1
- package/out/adapters/tiger/street-decompose.d.ts.map +0 -1
- package/out/adapters/tiger/street-decompose.js.map +0 -1
- package/out/adapters/usgov-hrsa-fqhc/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-hrsa-fqhc/adapter.js +0 -132
- package/out/adapters/usgov-hrsa-fqhc/adapter.js.map +0 -1
- package/out/adapters/usgov-imls-pls/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-imls-pls/adapter.js +0 -114
- package/out/adapters/usgov-imls-pls/adapter.js.map +0 -1
- package/out/adapters/usgov-irs-bmf/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-irs-bmf/adapter.js +0 -120
- package/out/adapters/usgov-irs-bmf/adapter.js.map +0 -1
- package/out/adapters/usgov-nad/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-nad/adapter.js +0 -249
- package/out/adapters/usgov-nad/adapter.js.map +0 -1
- package/out/adapters/usgov-nppes/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-nppes/adapter.js +0 -118
- package/out/adapters/usgov-nppes/adapter.js.map +0 -1
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map +0 -1
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.js +0 -154
- package/out/adapters/usgov-samhsa-treatment-locator/adapter.js.map +0 -1
- package/out/adapters/wof-admin-jp/adapter.d.ts.map +0 -1
- package/out/adapters/wof-admin-jp/adapter.js.map +0 -1
- package/out/adapters/wof-admin-json/adapter.d.ts +0 -79
- package/out/adapters/wof-admin-json/adapter.d.ts.map +0 -1
- package/out/adapters/wof-admin-json/adapter.js +0 -197
- package/out/adapters/wof-admin-json/adapter.js.map +0 -1
- package/out/adapters/wof-json-rows.d.ts +0 -62
- package/out/adapters/wof-json-rows.d.ts.map +0 -1
- package/out/adapters/wof-json-rows.js +0 -83
- package/out/adapters/wof-json-rows.js.map +0 -1
- package/out/adapters/wof-postalcode-json/adapter.d.ts +0 -65
- package/out/adapters/wof-postalcode-json/adapter.d.ts.map +0 -1
- package/out/adapters/wof-postalcode-json/adapter.js +0 -151
- package/out/adapters/wof-postalcode-json/adapter.js.map +0 -1
- package/out/codex/us-fips-state.d.ts +0 -41
- package/out/codex/us-fips-state.d.ts.map +0 -1
- package/out/codex/us-fips-state.js +0 -100
- package/out/codex/us-fips-state.js.map +0 -1
- package/out/recipes/bare-country.d.ts +0 -25
- package/out/recipes/bare-country.d.ts.map +0 -1
- package/out/recipes/bare-country.js +0 -88
- package/out/recipes/bare-country.js.map +0 -1
- package/out/recipes/cn-organizational-units.d.ts +0 -36
- package/out/recipes/cn-organizational-units.d.ts.map +0 -1
- package/out/recipes/cn-organizational-units.js +0 -159
- package/out/recipes/cn-organizational-units.js.map +0 -1
- package/out/recipes/cz-pcfirst-preposition.d.ts +0 -22
- package/out/recipes/cz-pcfirst-preposition.d.ts.map +0 -1
- package/out/recipes/cz-pcfirst-preposition.js +0 -101
- package/out/recipes/cz-pcfirst-preposition.js.map +0 -1
- package/out/recipes/fr-admin-split.d.ts +0 -38
- package/out/recipes/fr-admin-split.d.ts.map +0 -1
- package/out/recipes/fr-admin-split.js +0 -192
- package/out/recipes/fr-admin-split.js.map +0 -1
- package/out/recipes/fr-bare-street.d.ts +0 -28
- package/out/recipes/fr-bare-street.d.ts.map +0 -1
- package/out/recipes/fr-bare-street.js +0 -144
- package/out/recipes/fr-bare-street.js.map +0 -1
- package/out/recipes/fr-fragment.d.ts +0 -77
- package/out/recipes/fr-fragment.d.ts.map +0 -1
- package/out/recipes/fr-fragment.js +0 -277
- package/out/recipes/fr-fragment.js.map +0 -1
- package/out/recipes/fr-lieudit.d.ts +0 -32
- package/out/recipes/fr-lieudit.d.ts.map +0 -1
- package/out/recipes/fr-lieudit.js +0 -227
- package/out/recipes/fr-lieudit.js.map +0 -1
- package/out/recipes/fr-order.d.ts +0 -38
- package/out/recipes/fr-order.d.ts.map +0 -1
- package/out/recipes/fr-order.js +0 -198
- package/out/recipes/fr-order.js.map +0 -1
- package/out/recipes/german.d.ts +0 -44
- package/out/recipes/german.d.ts.map +0 -1
- package/out/recipes/german.js +0 -289
- package/out/recipes/german.js.map +0 -1
- package/out/recipes/locale.d.ts +0 -136
- package/out/recipes/locale.d.ts.map +0 -1
- package/out/recipes/locale.js +0 -460
- package/out/recipes/locale.js.map +0 -1
- package/out/recipes/nl-postcode.d.ts +0 -26
- package/out/recipes/nl-postcode.d.ts.map +0 -1
- package/out/recipes/nl-postcode.js +0 -103
- package/out/recipes/nl-postcode.js.map +0 -1
- package/out/recipes/no-fragment.d.ts +0 -43
- package/out/recipes/no-fragment.d.ts.map +0 -1
- package/out/recipes/no-fragment.js +0 -183
- package/out/recipes/no-fragment.js.map +0 -1
- package/out/recipes/no-street-led.d.ts +0 -24
- package/out/recipes/no-street-led.d.ts.map +0 -1
- package/out/recipes/no-street-led.js +0 -114
- package/out/recipes/no-street-led.js.map +0 -1
- package/out/recipes/no-street.d.ts +0 -18
- package/out/recipes/no-street.d.ts.map +0 -1
- package/out/recipes/no-street.js +0 -72
- package/out/recipes/no-street.js.map +0 -1
- package/out/recipes/po-box-cedex.d.ts +0 -38
- package/out/recipes/po-box-cedex.d.ts.map +0 -1
- package/out/recipes/po-box-cedex.js +0 -768
- package/out/recipes/po-box-cedex.js.map +0 -1
- package/out/recipes/po-box.d.ts +0 -17
- package/out/recipes/po-box.d.ts.map +0 -1
- package/out/recipes/po-box.js +0 -101
- package/out/recipes/po-box.js.map +0 -1
- package/out/recipes/reviewed-postcode-tail.d.ts.map +0 -1
- package/out/recipes/reviewed-postcode-tail.js.map +0 -1
- package/out/recipes/si-bare-village.d.ts +0 -32
- package/out/recipes/si-bare-village.d.ts.map +0 -1
- package/out/recipes/si-bare-village.js +0 -89
- package/out/recipes/si-bare-village.js.map +0 -1
- package/out/recipes/street-affix.d.ts +0 -89
- package/out/recipes/street-affix.d.ts.map +0 -1
- package/out/recipes/street-affix.js +0 -563
- package/out/recipes/street-affix.js.map +0 -1
- package/out/recipes/street-bare.d.ts +0 -25
- package/out/recipes/street-bare.d.ts.map +0 -1
- package/out/recipes/street-bare.js +0 -87
- package/out/recipes/street-bare.js.map +0 -1
- package/out/recipes/sub-venue-sources.d.ts +0 -229
- package/out/recipes/sub-venue-sources.d.ts.map +0 -1
- package/out/recipes/sub-venue-sources.js +0 -654
- package/out/recipes/sub-venue-sources.js.map +0 -1
- package/out/recipes/sub-venue.d.ts +0 -269
- package/out/recipes/sub-venue.d.ts.map +0 -1
- package/out/recipes/sub-venue.js +0 -726
- package/out/recipes/sub-venue.js.map +0 -1
- package/out/synthesizers/german.d.ts +0 -115
- package/out/synthesizers/german.d.ts.map +0 -1
- package/out/synthesizers/german.js +0 -149
- package/out/synthesizers/german.js.map +0 -1
- package/out/tools/fetch/acra-sg.d.ts +0 -23
- package/out/tools/fetch/acra-sg.d.ts.map +0 -1
- package/out/tools/fetch/acra-sg.js +0 -126
- package/out/tools/fetch/acra-sg.js.map +0 -1
- package/out/tools/fetch/ban.d.ts +0 -31
- package/out/tools/fetch/ban.d.ts.map +0 -1
- package/out/tools/fetch/ban.js +0 -239
- package/out/tools/fetch/ban.js.map +0 -1
- package/out/tools/fetch/download.d.ts +0 -182
- package/out/tools/fetch/download.d.ts.map +0 -1
- package/out/tools/fetch/download.js +0 -318
- package/out/tools/fetch/download.js.map +0 -1
- package/out/tools/fetch/gcis-tw.d.ts +0 -23
- package/out/tools/fetch/gcis-tw.d.ts.map +0 -1
- package/out/tools/fetch/gcis-tw.js +0 -135
- package/out/tools/fetch/gcis-tw.js.map +0 -1
- package/out/tools/fetch/geonames-dump.d.ts +0 -91
- package/out/tools/fetch/geonames-dump.d.ts.map +0 -1
- package/out/tools/fetch/geonames-dump.js +0 -171
- package/out/tools/fetch/geonames-dump.js.map +0 -1
- package/out/tools/fetch/geonames-postal.d.ts +0 -76
- package/out/tools/fetch/geonames-postal.d.ts.map +0 -1
- package/out/tools/fetch/geonames-postal.js +0 -126
- package/out/tools/fetch/geonames-postal.js.map +0 -1
- package/out/tools/fetch/houjin-jp.d.ts +0 -24
- package/out/tools/fetch/houjin-jp.d.ts.map +0 -1
- package/out/tools/fetch/houjin-jp.js +0 -102
- package/out/tools/fetch/houjin-jp.js.map +0 -1
- package/out/tools/fetch/hrsa.d.ts +0 -18
- package/out/tools/fetch/hrsa.d.ts.map +0 -1
- package/out/tools/fetch/hrsa.js +0 -46
- package/out/tools/fetch/hrsa.js.map +0 -1
- package/out/tools/fetch/imls-pls.d.ts +0 -25
- package/out/tools/fetch/imls-pls.d.ts.map +0 -1
- package/out/tools/fetch/imls-pls.js +0 -108
- package/out/tools/fetch/imls-pls.js.map +0 -1
- package/out/tools/fetch/juso-kr.d.ts +0 -30
- package/out/tools/fetch/juso-kr.d.ts.map +0 -1
- package/out/tools/fetch/juso-kr.js +0 -120
- package/out/tools/fetch/juso-kr.js.map +0 -1
- package/out/tools/fetch/localdata-kr.d.ts +0 -30
- package/out/tools/fetch/localdata-kr.d.ts.map +0 -1
- package/out/tools/fetch/localdata-kr.js +0 -139
- package/out/tools/fetch/localdata-kr.js.map +0 -1
- package/out/tools/fetch/nad.d.ts +0 -70
- package/out/tools/fetch/nad.d.ts.map +0 -1
- package/out/tools/fetch/nad.js +0 -234
- package/out/tools/fetch/nad.js.map +0 -1
- package/out/tools/fetch/nppes.d.ts +0 -26
- package/out/tools/fetch/nppes.d.ts.map +0 -1
- package/out/tools/fetch/nppes.js +0 -126
- package/out/tools/fetch/nppes.js.map +0 -1
- package/out/tools/fetch/state-hi-schools.d.ts +0 -26
- package/out/tools/fetch/state-hi-schools.d.ts.map +0 -1
- package/out/tools/fetch/state-hi-schools.js +0 -112
- package/out/tools/fetch/state-hi-schools.js.map +0 -1
- package/out/tools/fetch/state-sources.d.ts +0 -26
- package/out/tools/fetch/state-sources.d.ts.map +0 -1
- package/out/tools/fetch/state-sources.js +0 -111
- package/out/tools/fetch/state-sources.js.map +0 -1
- package/out/tools/fetch/tiger-full.d.ts +0 -22
- package/out/tools/fetch/tiger-full.d.ts.map +0 -1
- package/out/tools/fetch/tiger-full.js +0 -217
- package/out/tools/fetch/tiger-full.js.map +0 -1
- package/out/tools/golden-expand.d.ts +0 -90
- package/out/tools/golden-expand.d.ts.map +0 -1
- package/out/tools/golden-expand.js +0 -406
- package/out/tools/golden-expand.js.map +0 -1
- package/out/tools/golden-promote.d.ts +0 -72
- package/out/tools/golden-promote.d.ts.map +0 -1
- package/out/tools/golden-promote.js +0 -240
- package/out/tools/golden-promote.js.map +0 -1
- package/out/tools/golden-relabel-street.d.ts +0 -196
- package/out/tools/golden-relabel-street.d.ts.map +0 -1
- package/out/tools/golden-relabel-street.js +0 -439
- package/out/tools/golden-relabel-street.js.map +0 -1
- package/out/tools/lint-slice-vocab.d.ts +0 -90
- package/out/tools/lint-slice-vocab.d.ts.map +0 -1
- package/out/tools/lint-slice-vocab.js +0 -267
- package/out/tools/lint-slice-vocab.js.map +0 -1
- package/out/tools/lint-slice.d.ts +0 -93
- package/out/tools/lint-slice.d.ts.map +0 -1
- package/out/tools/lint-slice.js +0 -322
- package/out/tools/lint-slice.js.map +0 -1
- package/out/tools/slice-kryptonite.d.ts +0 -32
- package/out/tools/slice-kryptonite.d.ts.map +0 -1
- package/out/tools/slice-kryptonite.js +0 -99
- package/out/tools/slice-kryptonite.js.map +0 -1
- package/out/tools/slice-translit.d.ts +0 -51
- package/out/tools/slice-translit.d.ts.map +0 -1
- package/out/tools/slice-translit.js +0 -166
- package/out/tools/slice-translit.js.map +0 -1
- package/out/tools/sub-venue/harvest.d.ts +0 -100
- package/out/tools/sub-venue/harvest.d.ts.map +0 -1
- package/out/tools/sub-venue/harvest.js +0 -168
- package/out/tools/sub-venue/harvest.js.map +0 -1
- package/out/tools/sub-venue/head-nouns.d.ts +0 -50
- package/out/tools/sub-venue/head-nouns.d.ts.map +0 -1
- package/out/tools/sub-venue/head-nouns.js.map +0 -1
- package/out/tools/sub-venue/surfaces.d.ts +0 -45
- package/out/tools/sub-venue/surfaces.d.ts.map +0 -1
- package/out/tools/sub-venue/surfaces.js.map +0 -1
- package/out/tools/sub-venue/table.d.ts +0 -229
- package/out/tools/sub-venue/table.d.ts.map +0 -1
- package/out/tools/sub-venue/table.js.map +0 -1
- package/out/tools/sub-venue/wikidata.d.ts +0 -26
- package/out/tools/sub-venue/wikidata.d.ts.map +0 -1
- package/out/tools/sub-venue/wikidata.js +0 -62
- package/out/tools/sub-venue/wikidata.js.map +0 -1
- package/out/tools/sub-venue-lexicon.d.ts +0 -156
- package/out/tools/sub-venue-lexicon.d.ts.map +0 -1
- package/out/tools/sub-venue-lexicon.js +0 -283
- package/out/tools/sub-venue-lexicon.js.map +0 -1
- package/out/tools/sub-venue-promotions.d.ts +0 -94
- package/out/tools/sub-venue-promotions.d.ts.map +0 -1
- package/out/tools/sub-venue-promotions.js +0 -266
- package/out/tools/sub-venue-promotions.js.map +0 -1
- /package/lib/{recipes/cn-organizational-units.ts → cn/recipes/organizational-units.ts} +0 -0
- /package/lib/{recipes/cz-pcfirst-preposition.ts → cz/recipes/pcfirst-preposition.ts} +0 -0
- /package/lib/{adapters → fr/adapters}/ban/street-decompose.ts +0 -0
- /package/lib/{adapters/wof-admin-jp → jp/adapters/wof-admin}/adapter.ts +0 -0
- /package/lib/{recipes/nl-postcode.ts → nl/recipes/postcode.ts} +0 -0
- /package/lib/{recipes/no-fragment.ts → no/recipes/fragment.ts} +0 -0
- /package/lib/{recipes/no-street.ts → no/recipes/street/index.ts} +0 -0
- /package/lib/{recipes/no-street-led.ts → no/recipes/street/led.ts} +0 -0
- /package/lib/recipes/{bare-country.ts → bare/country.ts} +0 -0
- /package/lib/recipes/{street-bare.ts → street/bare.ts} +0 -0
- /package/lib/{recipes/si-bare-village.ts → si/recipes/bare-village.ts} +0 -0
- /package/lib/tools/{golden-expand.ts → golden/expand.ts} +0 -0
- /package/lib/tools/{lint-slice.ts → lint/slice/index.ts} +0 -0
- /package/lib/tools/{slice-kryptonite.ts → slice/kryptonite.ts} +0 -0
- /package/lib/tools/{slice-translit.ts → slice/translit.ts} +0 -0
- /package/lib/tools/{sub-venue-promotions.ts → sub/venue/promotions.ts} +0 -0
- /package/lib/{adapters/state-hi-schools → us/adapters/state/hi-schools}/workbook.ts +0 -0
- /package/lib/{adapters → us/adapters}/tiger/street-decompose.ts +0 -0
- /package/lib/{codex/us-fips-state.ts → us/fips-state.ts} +0 -0
- /package/lib/{recipes → ve/recipes}/reviewed-postcode-tail.ts +0 -0
- /package/out/adapters/{geonames-postal → geonames/postal}/adapter.d.ts +0 -0
- /package/out/{adapters → au/adapters}/gnaf/adapter.d.ts +0 -0
- /package/out/{adapters → fr/adapters}/ban/adapter.d.ts +0 -0
- /package/out/{adapters → fr/adapters}/ban/street-decompose.d.ts +0 -0
- /package/out/{adapters → fr/adapters}/ban/street-decompose.js +0 -0
- /package/out/{adapters/wof-admin-jp → jp/adapters/wof-admin}/adapter.d.ts +0 -0
- /package/out/{adapters/wof-admin-jp → jp/adapters/wof-admin}/adapter.js +0 -0
- /package/out/tools/{sub-venue → sub/venue}/head-nouns.js +0 -0
- /package/out/tools/{sub-venue → sub/venue}/surfaces.js +0 -0
- /package/out/tools/{sub-venue → sub/venue}/table.js +0 -0
- /package/out/{adapters → us/adapters}/fcc-bdc/adapter.d.ts +0 -0
- /package/out/{adapters/state-hi-schools → us/adapters/state/hi-schools}/adapter.d.ts +0 -0
- /package/out/{adapters/state-hi-schools → us/adapters/state/hi-schools}/workbook.d.ts +0 -0
- /package/out/{adapters/state-hi-schools → us/adapters/state/hi-schools}/workbook.js +0 -0
- /package/out/{adapters/state-ia-contractors → us/adapters/state/ia-contractors}/adapter.d.ts +0 -0
- /package/out/{adapters/state-ny-notaries → us/adapters/state/ny-notaries}/adapter.d.ts +0 -0
- /package/out/{adapters/state-tx-notaries → us/adapters/state/tx-notaries}/adapter.d.ts +0 -0
- /package/out/{adapters → us/adapters}/tiger/street-decompose.d.ts +0 -0
- /package/out/{adapters → us/adapters}/tiger/street-decompose.js +0 -0
- /package/out/{adapters/usgov-hrsa-fqhc → us/adapters/usgov/hrsa-fqhc}/adapter.d.ts +0 -0
- /package/out/{adapters/usgov-imls-pls → us/adapters/usgov/imls-pls}/adapter.d.ts +0 -0
- /package/out/{adapters/usgov-irs-bmf → us/adapters/usgov/irs-bmf}/adapter.d.ts +0 -0
- /package/out/{adapters/usgov-nad → us/adapters/usgov/nad}/adapter.d.ts +0 -0
- /package/out/{adapters/usgov-nppes → us/adapters/usgov/nppes}/adapter.d.ts +0 -0
- /package/out/{adapters/usgov-samhsa-treatment-locator → us/adapters/usgov/samhsa-treatment-locator}/adapter.d.ts +0 -0
- /package/out/{recipes → ve/recipes}/reviewed-postcode-tail.d.ts +0 -0
- /package/out/{recipes → ve/recipes}/reviewed-postcode-tail.js +0 -0
|
@@ -0,0 +1,146 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
* @license AGPL-3.0
|
|
4
|
+
* @author Teffen Ellis, et al.
|
|
5
|
+
*
|
|
6
|
+
* Re-fetch the state-level open-data sources (NY/TX/DE/OR notaries, IA contractors, WA health
|
|
7
|
+
* providers, HI lobbyists). Reproducible recovery if `$MAILWOMAN_DATA_ROOT` is lost.
|
|
8
|
+
*
|
|
9
|
+
* HI public schools is fetched separately by `mailwoman corpus fetch state-hi-schools` — its
|
|
10
|
+
* upstream is an XLSX workbook that requires an openpyxl-driven sheet-concatenation pre-step
|
|
11
|
+
* before the adapter can consume it.
|
|
12
|
+
*
|
|
13
|
+
* Each source lands in its own subdirectory of `<outRoot>/<slug>/` along with a `MANIFEST.json`
|
|
14
|
+
* recording origin URL + download timestamp + sha256 so downstream adapters can verify provenance.
|
|
15
|
+
*
|
|
16
|
+
* Invoke via `mailwoman corpus fetch state-sources --out-root <path>`. Uses Node's built-in fetch
|
|
17
|
+
* (gzip/brotli) and streaming sha256 instead of curl + sha256sum.
|
|
18
|
+
*/
|
|
19
|
+
|
|
20
|
+
import { BYTES_PER_KIB } from "@mailwoman/core/fs/formatters"
|
|
21
|
+
import { makeDirectories } from "@mailwoman/core/fs/writers"
|
|
22
|
+
import { sha256File } from "@mailwoman/core/hash"
|
|
23
|
+
import { join } from "path-ts"
|
|
24
|
+
|
|
25
|
+
import type { BaseFetchOptions, FetchSummary, SourceManifest } from "#tools/fetch/download/index"
|
|
26
|
+
import { downloadToFile, writeManifest } from "#tools/fetch/download/index"
|
|
27
|
+
|
|
28
|
+
/**
|
|
29
|
+
* Bytes per KiB — the divisor for human-readable sizes, and the floor below which a "download" is an error page rather
|
|
30
|
+
* than data.
|
|
31
|
+
*/
|
|
32
|
+
|
|
33
|
+
export type FetchStateSourcesOptions = BaseFetchOptions
|
|
34
|
+
|
|
35
|
+
interface Source {
|
|
36
|
+
slug: string
|
|
37
|
+
filename: string
|
|
38
|
+
url: string
|
|
39
|
+
}
|
|
40
|
+
|
|
41
|
+
const SOURCES: readonly Source[] = [
|
|
42
|
+
{
|
|
43
|
+
slug: "state-ny-notaries",
|
|
44
|
+
filename: "NY_Commissioned_Notaries.csv",
|
|
45
|
+
url: "https://data.ny.gov/api/views/rwbv-mz6z/rows.csv?accessType=DOWNLOAD",
|
|
46
|
+
},
|
|
47
|
+
{
|
|
48
|
+
slug: "state-tx-notaries",
|
|
49
|
+
filename: "TX_Notary_Public_Commissions.csv",
|
|
50
|
+
url: "https://data.texas.gov/api/views/gmd3-bnrd/rows.csv?accessType=DOWNLOAD",
|
|
51
|
+
},
|
|
52
|
+
{
|
|
53
|
+
slug: "state-de-notaries",
|
|
54
|
+
filename: "DE_Notaries_Commissioned.csv",
|
|
55
|
+
url: "https://data.delaware.gov/api/views/q8dr-mj6p/rows.csv?accessType=DOWNLOAD",
|
|
56
|
+
},
|
|
57
|
+
{
|
|
58
|
+
slug: "state-or-notaries",
|
|
59
|
+
filename: "OR_Active_Notaries.csv",
|
|
60
|
+
url: "https://data.oregon.gov/api/views/j2pk-zk6z/rows.csv?accessType=DOWNLOAD",
|
|
61
|
+
},
|
|
62
|
+
{
|
|
63
|
+
slug: "state-ia-contractors",
|
|
64
|
+
filename: "IA_Active_Construction_Contractor_Registrations.csv",
|
|
65
|
+
url: "https://data.iowa.gov/api/views/dpf3-iz94/rows.csv?accessType=DOWNLOAD",
|
|
66
|
+
},
|
|
67
|
+
{
|
|
68
|
+
slug: "state-wa-health-providers",
|
|
69
|
+
filename: "WA_Health_Care_Provider_Credential_Data.csv",
|
|
70
|
+
url: "https://data.wa.gov/api/views/qxh8-f4bd/rows.csv?accessType=DOWNLOAD",
|
|
71
|
+
},
|
|
72
|
+
{
|
|
73
|
+
slug: "state-hi-lobbyists",
|
|
74
|
+
filename: "HI_Lobbyist_Registration_Statements.csv",
|
|
75
|
+
url: "https://data.hawaii.gov/api/views/cm7c-skav/rows.csv?accessType=DOWNLOAD",
|
|
76
|
+
},
|
|
77
|
+
]
|
|
78
|
+
|
|
79
|
+
export async function fetchStateSources(
|
|
80
|
+
options: FetchStateSourcesOptions,
|
|
81
|
+
report?: (line: string) => void
|
|
82
|
+
): Promise<FetchSummary> {
|
|
83
|
+
await makeDirectories(options.outRoot)
|
|
84
|
+
|
|
85
|
+
let fetched = 0
|
|
86
|
+
let failed = 0
|
|
87
|
+
const failedCodes: string[] = []
|
|
88
|
+
|
|
89
|
+
for (const { slug, filename, url } of SOURCES) {
|
|
90
|
+
const destDir = join(options.outRoot, slug)
|
|
91
|
+
await makeDirectories(destDir)
|
|
92
|
+
const dest = join(destDir, filename)
|
|
93
|
+
|
|
94
|
+
report?.(`=== ${slug} / ${filename}`)
|
|
95
|
+
|
|
96
|
+
let bytes: number
|
|
97
|
+
|
|
98
|
+
try {
|
|
99
|
+
;({ bytes } = await downloadToFile({
|
|
100
|
+
url,
|
|
101
|
+
dest,
|
|
102
|
+
timeoutMs: 600_000,
|
|
103
|
+
headers: { "Accept-Encoding": "gzip, br" },
|
|
104
|
+
report,
|
|
105
|
+
}))
|
|
106
|
+
} catch (error) {
|
|
107
|
+
report?.(` ✗ download failed for ${url}: ${(error as Error).message}`)
|
|
108
|
+
|
|
109
|
+
failed++
|
|
110
|
+
failedCodes.push(slug)
|
|
111
|
+
|
|
112
|
+
continue
|
|
113
|
+
}
|
|
114
|
+
|
|
115
|
+
if (bytes < BYTES_PER_KIB) {
|
|
116
|
+
report?.(` ✗ response too small (${bytes} bytes) — probable 404 / error page`)
|
|
117
|
+
|
|
118
|
+
failed++
|
|
119
|
+
failedCodes.push(slug)
|
|
120
|
+
|
|
121
|
+
continue
|
|
122
|
+
}
|
|
123
|
+
|
|
124
|
+
const sha = await sha256File(dest)
|
|
125
|
+
|
|
126
|
+
const manifest: SourceManifest = {
|
|
127
|
+
source_url: url,
|
|
128
|
+
downloaded_at: new Date().toISOString(),
|
|
129
|
+
filename,
|
|
130
|
+
sha256: sha,
|
|
131
|
+
bytes,
|
|
132
|
+
}
|
|
133
|
+
|
|
134
|
+
await writeManifest(join(destDir, "MANIFEST.json"), manifest)
|
|
135
|
+
|
|
136
|
+
report?.(` ✓ ${(bytes / 1024 / 1024).toFixed(1)} MB sha256=${sha}`)
|
|
137
|
+
|
|
138
|
+
fetched++
|
|
139
|
+
}
|
|
140
|
+
|
|
141
|
+
report?.(`=== summary ===`)
|
|
142
|
+
report?.(`fetched: ${fetched}`)
|
|
143
|
+
report?.(`failed: ${failed}`)
|
|
144
|
+
|
|
145
|
+
return { fetched, skipped: 0, failed, failedCodes }
|
|
146
|
+
}
|
|
@@ -0,0 +1,308 @@
|
|
|
1
|
+
import { APIClient, pluckResponseData } from "@mailwoman/core/api"
|
|
2
|
+
/**
|
|
3
|
+
* @copyright Sister Software
|
|
4
|
+
* @license AGPL-3.0
|
|
5
|
+
* @author Teffen Ellis, et al.
|
|
6
|
+
*
|
|
7
|
+
* Fetch the full TIGER 2024 ADDRFEAT dataset — all US counties.
|
|
8
|
+
*
|
|
9
|
+
* TIGER ADDRFEAT 2024 source:
|
|
10
|
+
*
|
|
11
|
+
* - https://www2.census.gov/geo/tiger/TIGER2024/ADDRFEAT/
|
|
12
|
+
* - Files: `tl_2024_<statefips><countyfips>_addrfeat.zip`
|
|
13
|
+
*
|
|
14
|
+
* Each state's ZIPs land in `<outRoot>/tiger/addrfeat/state-<statefips>/` with a per-state
|
|
15
|
+
* `MANIFEST.json` recording filename, sha256, and bytes for every county ZIP so re-runs can skip
|
|
16
|
+
* already-verified files. (Extraction + ogr2ogr ingestion happen later, in the `tiger` adapter;
|
|
17
|
+
* this module is download + provenance only.)
|
|
18
|
+
*
|
|
19
|
+
* Invoke via `mailwoman corpus fetch tiger-full --out-root <path>`. Native `fetch` streams each
|
|
20
|
+
* county ZIP to disk (no curl subprocess).
|
|
21
|
+
*/
|
|
22
|
+
/* oxlint-disable sister-software/prefer-region-over-marks -- these markers label steps inside one
|
|
23
|
+
procedure, not sections of declarations. A region there folds nothing a reader wants folded. */
|
|
24
|
+
import { BYTES_PER_KIB, ByteFormatter } from "@mailwoman/core/fs/formatters"
|
|
25
|
+
import { statPath, pathExists } from "@mailwoman/core/fs/readers"
|
|
26
|
+
import { makeDirectories, removePathIfPresent } from "@mailwoman/core/fs/writers"
|
|
27
|
+
import { sha256File } from "@mailwoman/core/hash"
|
|
28
|
+
import { isoSeconds } from "@mailwoman/core/utils"
|
|
29
|
+
import { sleep } from "@mailwoman/core/utils/sleep"
|
|
30
|
+
import { basename, join } from "path-ts"
|
|
31
|
+
|
|
32
|
+
import type { BaseFetchOptions, FetchSummary } from "#tools/fetch/download/index"
|
|
33
|
+
import { readManifest, streamDownload, writeManifest } from "#tools/fetch/download/index"
|
|
34
|
+
|
|
35
|
+
/**
|
|
36
|
+
* Bytes per KiB — the divisor for human-readable sizes, and the floor below which a "download" is an error page rather
|
|
37
|
+
* than data.
|
|
38
|
+
*/
|
|
39
|
+
/**
|
|
40
|
+
* Lowest 2xx status; anything below is informational.
|
|
41
|
+
*/
|
|
42
|
+
const HTTP_OK = 200
|
|
43
|
+
|
|
44
|
+
/**
|
|
45
|
+
* Lowest 3xx status; at or above it the response is a redirect or an error, not a body.
|
|
46
|
+
*/
|
|
47
|
+
const HTTP_REDIRECT = 300
|
|
48
|
+
|
|
49
|
+
const TIGER_BASE_URL = "https://www2.census.gov/geo/tiger/TIGER2024/ADDRFEAT"
|
|
50
|
+
|
|
51
|
+
export interface FetchTigerFullOptions extends BaseFetchOptions {
|
|
52
|
+
/**
|
|
53
|
+
* Space-separated list of 2-digit state FIPS codes to skip entirely. Default `"50"` — Vermont, already fetched in
|
|
54
|
+
* v0.1.1.
|
|
55
|
+
*/
|
|
56
|
+
skipStateFips?: string
|
|
57
|
+
/**
|
|
58
|
+
* Seconds to sleep between downloads. Default `0.2`.
|
|
59
|
+
*/
|
|
60
|
+
rateSleep?: number
|
|
61
|
+
/**
|
|
62
|
+
* Max concurrent download workers per state. Default `4`.
|
|
63
|
+
*/
|
|
64
|
+
maxParallel?: number
|
|
65
|
+
/**
|
|
66
|
+
* Print planned downloads without fetching. Default `false`.
|
|
67
|
+
*/
|
|
68
|
+
dryRun?: boolean
|
|
69
|
+
}
|
|
70
|
+
|
|
71
|
+
interface CountyEntry {
|
|
72
|
+
filename: string
|
|
73
|
+
sha256: string
|
|
74
|
+
bytes: number
|
|
75
|
+
}
|
|
76
|
+
|
|
77
|
+
/**
|
|
78
|
+
* Read a per-state MANIFEST.json into a filename → entry map.
|
|
79
|
+
*/
|
|
80
|
+
async function readCountyManifest(manifestPath: string): Promise<Map<string, CountyEntry>> {
|
|
81
|
+
const map = new Map<string, CountyEntry>()
|
|
82
|
+
const parsed = await readManifest<{ counties?: CountyEntry[] }>(manifestPath)
|
|
83
|
+
|
|
84
|
+
for (const c of parsed?.counties ?? []) {
|
|
85
|
+
if (c.filename) {
|
|
86
|
+
map.set(c.filename, { filename: c.filename, sha256: c.sha256, bytes: c.bytes })
|
|
87
|
+
}
|
|
88
|
+
}
|
|
89
|
+
|
|
90
|
+
return map
|
|
91
|
+
}
|
|
92
|
+
|
|
93
|
+
/**
|
|
94
|
+
* Check whether a file already matches a recorded sha256 and byte count.
|
|
95
|
+
*/
|
|
96
|
+
async function fileMatchesSha(path: string, expectedSha: string, expectedBytes: number): Promise<boolean> {
|
|
97
|
+
if (!(await pathExists(path))) return false
|
|
98
|
+
|
|
99
|
+
if ((await statPath(path)).size !== expectedBytes) return false
|
|
100
|
+
|
|
101
|
+
return (await sha256File(path)) === expectedSha
|
|
102
|
+
}
|
|
103
|
+
|
|
104
|
+
type CountyResult =
|
|
105
|
+
| { ok: true; filename: string; sha256: string; bytes: number }
|
|
106
|
+
| { ok: false; filename: string; reason: string }
|
|
107
|
+
|
|
108
|
+
/**
|
|
109
|
+
* Download one county ZIP (size sanity check + sha256).
|
|
110
|
+
*/
|
|
111
|
+
async function downloadCounty(url: string, dest: string): Promise<CountyResult> {
|
|
112
|
+
const filename = basename(dest)
|
|
113
|
+
const status = await streamDownload(url, dest, { timeoutMs: 600_000, retries: 3, retryDelayMs: 5000 })
|
|
114
|
+
|
|
115
|
+
if (status < HTTP_OK || status >= HTTP_REDIRECT) {
|
|
116
|
+
return { ok: false, filename, reason: `HTTP ${status}` }
|
|
117
|
+
}
|
|
118
|
+
|
|
119
|
+
const bytes = (await statPath(dest)).size
|
|
120
|
+
|
|
121
|
+
if (bytes < BYTES_PER_KIB) {
|
|
122
|
+
await removePathIfPresent(dest)
|
|
123
|
+
|
|
124
|
+
return { ok: false, filename, reason: `too small (${bytes} bytes)` }
|
|
125
|
+
}
|
|
126
|
+
|
|
127
|
+
const sha256 = await sha256File(dest)
|
|
128
|
+
|
|
129
|
+
return { ok: true, filename, sha256, bytes }
|
|
130
|
+
}
|
|
131
|
+
|
|
132
|
+
export async function fetchTigerFull(
|
|
133
|
+
options: FetchTigerFullOptions,
|
|
134
|
+
report?: (line: string) => void
|
|
135
|
+
): Promise<FetchSummary> {
|
|
136
|
+
const skipStateFips = (options.skipStateFips ?? "50").split(/\s+/).filter((value) => value.length > 0)
|
|
137
|
+
const rateSleepMs = Math.round((options.rateSleep ?? 0.2) * 1000)
|
|
138
|
+
const maxParallel = options.maxParallel ?? 4
|
|
139
|
+
const dryRun = options.dryRun ?? false
|
|
140
|
+
|
|
141
|
+
const addrfeatDir = join(options.outRoot, "tiger", "addrfeat")
|
|
142
|
+
await makeDirectories(addrfeatDir)
|
|
143
|
+
|
|
144
|
+
// MARK: Step 1 — discover the county file list
|
|
145
|
+
|
|
146
|
+
report?.(`=== Fetching TIGER 2024 ADDRFEAT directory listing...`)
|
|
147
|
+
|
|
148
|
+
// `responseType: "text"` — an Apache directory index, scraped below. The per-county ARCHIVE downloads
|
|
149
|
+
// stay on raw `fetch` (they stream to disk; see `downloadOne`).
|
|
150
|
+
const listingRes = await new APIClient({
|
|
151
|
+
displayName: "tiger-listing",
|
|
152
|
+
retry: true,
|
|
153
|
+
axios: { headers: { "Accept-Encoding": "gzip, br" } },
|
|
154
|
+
})
|
|
155
|
+
.fetch<string>({ url: `${TIGER_BASE_URL}/`, responseType: "text", timeout: 60_000 })
|
|
156
|
+
.then(pluckResponseData)
|
|
157
|
+
|
|
158
|
+
const html = listingRes
|
|
159
|
+
const allZips = [...new Set(html.match(/tl_2024_[0-9]{5}_addrfeat\.zip/g))].toSorted()
|
|
160
|
+
const totalCounties = allZips.length
|
|
161
|
+
report?.(` Found ${totalCounties} county ZIPs in the TIGER 2024 ADDRFEAT index.`)
|
|
162
|
+
|
|
163
|
+
// Build a map: state_fips -> list of filenames.
|
|
164
|
+
// tl_2024_SSCCC_addrfeat.zip — SS = 2-digit state FIPS (chars 8-9), CCC = county FIPS.
|
|
165
|
+
const stateFiles = new Map<string, string[]>()
|
|
166
|
+
|
|
167
|
+
for (const fname of allZips) {
|
|
168
|
+
const stateFips = fname.slice(8, 10)
|
|
169
|
+
const list = stateFiles.get(stateFips) ?? []
|
|
170
|
+
list.push(fname)
|
|
171
|
+
stateFiles.set(stateFips, list)
|
|
172
|
+
}
|
|
173
|
+
|
|
174
|
+
report?.(` Spans ${stateFiles.size} state/territory FIPS codes.`)
|
|
175
|
+
|
|
176
|
+
// MARK: Step 2 — download the county ZIPs
|
|
177
|
+
|
|
178
|
+
let totalFetched = 0
|
|
179
|
+
let totalSkipped = 0
|
|
180
|
+
let totalSkippedState = 0
|
|
181
|
+
let totalFailed = 0
|
|
182
|
+
let totalBytesFetched = 0
|
|
183
|
+
const failedCodes: string[] = []
|
|
184
|
+
|
|
185
|
+
// Process states in sorted FIPS order for predictable output.
|
|
186
|
+
const sortedStates = [...stateFiles.keys()].toSorted()
|
|
187
|
+
|
|
188
|
+
for (const stateFips of sortedStates) {
|
|
189
|
+
const countyFiles = stateFiles.get(stateFips) ?? []
|
|
190
|
+
|
|
191
|
+
// Respect an explicit request to skip this state before scheduling its counties.
|
|
192
|
+
if (skipStateFips.includes(stateFips)) {
|
|
193
|
+
report?.(`--- State ${stateFips} — SKIPPED (in --skip-state-fips, ${countyFiles.length} counties)`)
|
|
194
|
+
totalSkippedState += countyFiles.length
|
|
195
|
+
|
|
196
|
+
continue
|
|
197
|
+
}
|
|
198
|
+
|
|
199
|
+
const stateDir = join(addrfeatDir, `state-${stateFips}`)
|
|
200
|
+
await makeDirectories(stateDir)
|
|
201
|
+
const manifestPath = join(stateDir, "MANIFEST.json")
|
|
202
|
+
|
|
203
|
+
// Load existing manifest for O(1) verified-skip lookup.
|
|
204
|
+
const manifest = await readCountyManifest(manifestPath)
|
|
205
|
+
|
|
206
|
+
report?.(`--- State ${stateFips} — ${countyFiles.length} counties`)
|
|
207
|
+
|
|
208
|
+
// Build a list of URLs+dests that need fetching.
|
|
209
|
+
const pending: Array<{ url: string; dest: string }> = []
|
|
210
|
+
|
|
211
|
+
for (const fname of countyFiles) {
|
|
212
|
+
const dest = join(stateDir, fname)
|
|
213
|
+
const url = `${TIGER_BASE_URL}/${fname}`
|
|
214
|
+
const known = manifest.get(fname)
|
|
215
|
+
|
|
216
|
+
// Skip if already verified via MANIFEST.
|
|
217
|
+
if (known && (await fileMatchesSha(dest, known.sha256, known.bytes))) {
|
|
218
|
+
report?.(` skip (verified) ${fname}`)
|
|
219
|
+
|
|
220
|
+
totalSkipped++
|
|
221
|
+
|
|
222
|
+
continue
|
|
223
|
+
}
|
|
224
|
+
|
|
225
|
+
if (dryRun) {
|
|
226
|
+
report?.(` would fetch: ${url}`)
|
|
227
|
+
|
|
228
|
+
totalFetched++
|
|
229
|
+
|
|
230
|
+
continue
|
|
231
|
+
}
|
|
232
|
+
|
|
233
|
+
pending.push({ url, dest })
|
|
234
|
+
}
|
|
235
|
+
|
|
236
|
+
if (dryRun) continue
|
|
237
|
+
|
|
238
|
+
if (!pending.length) continue
|
|
239
|
+
|
|
240
|
+
// Download pending counties with bounded parallelism and rate-limit spacing.
|
|
241
|
+
const results: CountyResult[] = Array.from({ length: pending.length })
|
|
242
|
+
let cursor = 0
|
|
243
|
+
|
|
244
|
+
const workers = Array.from({ length: Math.min(maxParallel, pending.length) }, async () => {
|
|
245
|
+
while (true) {
|
|
246
|
+
const i = cursor++
|
|
247
|
+
|
|
248
|
+
if (i >= pending.length) return
|
|
249
|
+
const item = pending[i]!
|
|
250
|
+
// Rate-limit: polite spacing before each fetch.
|
|
251
|
+
await sleep(rateSleepMs)
|
|
252
|
+
results[i] = await downloadCounty(item.url, item.dest)
|
|
253
|
+
}
|
|
254
|
+
})
|
|
255
|
+
|
|
256
|
+
await Promise.all(workers)
|
|
257
|
+
|
|
258
|
+
// Collect results from this state.
|
|
259
|
+
for (const result of results) {
|
|
260
|
+
if (result.ok) {
|
|
261
|
+
report?.(
|
|
262
|
+
` ok ${result.filename} ${ByteFormatter.formatIEC(result.bytes)} sha256=${result.sha256.slice(0, 12)}...`
|
|
263
|
+
)
|
|
264
|
+
|
|
265
|
+
manifest.set(result.filename, { filename: result.filename, sha256: result.sha256, bytes: result.bytes })
|
|
266
|
+
|
|
267
|
+
totalFetched++
|
|
268
|
+
totalBytesFetched += result.bytes
|
|
269
|
+
} else {
|
|
270
|
+
report?.(` FAIL ${result.filename} -- ${result.reason}`)
|
|
271
|
+
|
|
272
|
+
totalFailed++
|
|
273
|
+
failedCodes.push(result.filename)
|
|
274
|
+
}
|
|
275
|
+
}
|
|
276
|
+
|
|
277
|
+
// Rewrite per-state MANIFEST.json with all known-good counties (sorted for determinism).
|
|
278
|
+
const counties = [...manifest.values()].toSorted((a, b) => a.filename.localeCompare(b.filename))
|
|
279
|
+
|
|
280
|
+
const manifestDoc = {
|
|
281
|
+
state_fips: stateFips,
|
|
282
|
+
updated_at: isoSeconds(),
|
|
283
|
+
tiger_base_url: TIGER_BASE_URL,
|
|
284
|
+
counties,
|
|
285
|
+
}
|
|
286
|
+
|
|
287
|
+
await writeManifest(manifestPath, manifestDoc)
|
|
288
|
+
}
|
|
289
|
+
|
|
290
|
+
// MARK: Summary
|
|
291
|
+
|
|
292
|
+
report?.(`=== Summary ===`)
|
|
293
|
+
report?.(` Total counties in index : ${totalCounties}`)
|
|
294
|
+
report?.(` State(s) fully skipped : ${totalSkippedState} (--skip-state-fips "${skipStateFips.join(" ")}")`)
|
|
295
|
+
report?.(` Counties already present : ${totalSkipped}`)
|
|
296
|
+
report?.(` Counties fetched this run : ${totalFetched}`)
|
|
297
|
+
report?.(` Counties failed : ${totalFailed}`)
|
|
298
|
+
|
|
299
|
+
if (totalBytesFetched > 0) {
|
|
300
|
+
report?.(` Bytes fetched this run : ${ByteFormatter.formatIEC(totalBytesFetched)} (${totalBytesFetched})`)
|
|
301
|
+
}
|
|
302
|
+
|
|
303
|
+
if (totalFailed > 0) {
|
|
304
|
+
report?.(`WARNING: ${totalFailed} download(s) failed. Re-run to retry.`)
|
|
305
|
+
}
|
|
306
|
+
|
|
307
|
+
return { fetched: totalFetched, skipped: totalSkipped + totalSkippedState, failed: totalFailed, failedCodes }
|
|
308
|
+
}
|
package/lib/utils/align.ts
CHANGED
|
@@ -33,7 +33,7 @@
|
|
|
33
33
|
* — NFD `é` occupies two code units where NFC `é` occupies one — and silently so).
|
|
34
34
|
*/
|
|
35
35
|
|
|
36
|
-
import type { BIOLabel, ComponentTag } from "@mailwoman/
|
|
36
|
+
import type { BIOLabel, ComponentTag } from "@mailwoman/codex/component"
|
|
37
37
|
import { distance as levenshteinDistance } from "fastest-levenshtein"
|
|
38
38
|
|
|
39
39
|
import type { CanonicalRow, LabeledRow, QuarantinedRow } from "#types"
|
package/lib/utils/golden.ts
CHANGED
|
@@ -12,19 +12,19 @@
|
|
|
12
12
|
* - Defines `GoldenEntry` (schema check).
|
|
13
13
|
* - Loads `.jsonl` files (one entry per line).
|
|
14
14
|
* - Validates every entry: schema shape, ComponentTag membership, reachability of each component in
|
|
15
|
-
* `raw` via
|
|
15
|
+
* `raw` via `componentsPresentIn`.
|
|
16
16
|
* - Returns a structured report of per-entry errors so the CLI / CI surface can act on it.
|
|
17
17
|
*
|
|
18
18
|
* The 1000-entry target (500 US + 500 FR) is a human task. This module catches the regressions that
|
|
19
19
|
* creep in over time as new entries land.
|
|
20
20
|
*/
|
|
21
21
|
|
|
22
|
-
import {
|
|
22
|
+
import { componentsPresentIn } from "@mailwoman/codex/address-format"
|
|
23
|
+
import { COMPONENT_TAGS, type ComponentTag } from "@mailwoman/codex/component"
|
|
23
24
|
import { parseJSONStrict } from "@mailwoman/core/json"
|
|
24
|
-
import { COMPONENT_TAGS, type ComponentTag } from "@mailwoman/core/types"
|
|
25
|
-
import { reconcileComponents } from "@mailwoman/formatter"
|
|
26
25
|
import { join } from "path-ts"
|
|
27
26
|
import { TextSpliterator } from "spliterator"
|
|
27
|
+
import { Globerator } from "spliterator/node/fs"
|
|
28
28
|
|
|
29
29
|
const TAG_SET = new Set<string>(COMPONENT_TAGS as readonly string[])
|
|
30
30
|
|
|
@@ -111,14 +111,18 @@ export function parseGoldenLine(line: string): GoldenEntry {
|
|
|
111
111
|
}
|
|
112
112
|
|
|
113
113
|
/**
|
|
114
|
-
* Check that every component in `entry` appears in `entry.raw
|
|
114
|
+
* Check that every component in `entry` appears in `entry.raw`.
|
|
115
|
+
*
|
|
116
|
+
* A golden entry's `raw` is hand-written ground truth, not a render, so the question here really is containment: does
|
|
117
|
+
* this labeled span occur in the string a person typed. That is the weaker of the two reconciliations, and the right
|
|
118
|
+
* one for a string no layout produced.
|
|
115
119
|
*/
|
|
116
120
|
export function unreachableComponents(entry: GoldenEntry): ComponentTag[] {
|
|
117
|
-
const
|
|
121
|
+
const present = componentsPresentIn(entry.components, entry.raw)
|
|
118
122
|
const missing: ComponentTag[] = []
|
|
119
123
|
|
|
120
124
|
for (const tag of Object.keys(entry.components) as ComponentTag[]) {
|
|
121
|
-
if (!(tag in
|
|
125
|
+
if (!(tag in present)) {
|
|
122
126
|
missing.push(tag)
|
|
123
127
|
}
|
|
124
128
|
}
|
|
@@ -169,7 +173,7 @@ export async function validateGoldenFile(path: string): Promise<GoldenIssue[]> {
|
|
|
169
173
|
* Validate every `.jsonl` in a golden directory.
|
|
170
174
|
*/
|
|
171
175
|
export async function validateGoldenDir(dir: string): Promise<GoldenReport> {
|
|
172
|
-
const files =
|
|
176
|
+
const files = await Globerator.files("jsonl", { cwd: dir, absolute: false }).toSorted()
|
|
173
177
|
const issues: GoldenIssue[] = []
|
|
174
178
|
let entries = 0
|
|
175
179
|
|
package/lib/utils/wof-json.ts
CHANGED
|
@@ -31,7 +31,7 @@
|
|
|
31
31
|
|
|
32
32
|
import { readLocalTextFile } from "@mailwoman/core/fs/readers"
|
|
33
33
|
import { tryParsingJSON } from "@mailwoman/core/json"
|
|
34
|
-
import
|
|
34
|
+
import { Globerator } from "spliterator/node/fs"
|
|
35
35
|
|
|
36
36
|
/**
|
|
37
37
|
* A WOF GeoJSON feature, as published by the per-record bundles.
|
|
@@ -174,19 +174,13 @@ function recordFromFeature(feature: WOFFeature): WOFRecord | null {
|
|
|
174
174
|
* walk. Adapters can add stricter validation downstream if they need it.
|
|
175
175
|
*/
|
|
176
176
|
export async function* walkFeatures(repoDir: string, opts: { signal?: AbortSignal } = {}): AsyncIterable<WOFRecord> {
|
|
177
|
-
const
|
|
177
|
+
for await (const filePath of Globerator.from("**/*.geojson", {
|
|
178
178
|
cwd: repoDir,
|
|
179
|
-
|
|
180
|
-
|
|
181
|
-
|
|
182
|
-
|
|
183
|
-
|
|
184
|
-
for await (const entry of stream) {
|
|
185
|
-
if (opts.signal?.aborted) return
|
|
186
|
-
const filePath = String(entry)
|
|
187
|
-
|
|
188
|
-
if (filePath.includes("-alt-")) continue
|
|
189
|
-
|
|
179
|
+
exclude: ["**/*-alt-*.geojson"],
|
|
180
|
+
// Preserve the recursive fast-glob walk this replaces: bundle repositories can expose data through a link.
|
|
181
|
+
followSymlinks: true,
|
|
182
|
+
signal: opts.signal,
|
|
183
|
+
})) {
|
|
190
184
|
let text: string
|
|
191
185
|
|
|
192
186
|
try {
|
|
@@ -22,7 +22,7 @@
|
|
|
22
22
|
* balance so the model sees both domestic and international order) —
|
|
23
23
|
*
|
|
24
24
|
* 1. `{ locality, region }` → "City, Region"
|
|
25
|
-
* 2. `{ locality, region, country }` → "City, Region, Country" `
|
|
25
|
+
* 2. `{ locality, region, country }` → "City, Region, Country" `componentsPresentIn` drops any
|
|
26
26
|
* component that didn't survive into the rendered `raw`.
|
|
27
27
|
*
|
|
28
28
|
* License: stamped `"CC-BY-4.0"` per row (GeoNames' terms); provenance is the `geonames-<id>` key.
|
|
@@ -22,13 +22,13 @@
|
|
|
22
22
|
* balance so the model sees both domestic and international order) —
|
|
23
23
|
*
|
|
24
24
|
* 1. `{ locality, region }` → "City, Region"
|
|
25
|
-
* 2. `{ locality, region, country }` → "City, Region, Country" `
|
|
25
|
+
* 2. `{ locality, region, country }` → "City, Region, Country" `componentsPresentIn` drops any
|
|
26
26
|
* component that didn't survive into the rendered `raw`.
|
|
27
27
|
*
|
|
28
28
|
* License: stamped `"CC-BY-4.0"` per row (GeoNames' terms); provenance is the `geonames-<id>` key.
|
|
29
29
|
*/
|
|
30
|
+
import { componentsPresentIn } from "@mailwoman/codex/address-format";
|
|
30
31
|
import { pathExists } from "@mailwoman/core/fs/readers";
|
|
31
|
-
import { reconcileComponents } from "@mailwoman/formatter";
|
|
32
32
|
import { dirname, join } from "path-ts";
|
|
33
33
|
import { TSVSpliterator } from "spliterator";
|
|
34
34
|
import { stableSourceID } from "#adapters/utils";
|
|
@@ -148,7 +148,7 @@ export function createGeonamesAdapter() {
|
|
|
148
148
|
for (const v of variants) {
|
|
149
149
|
if (opts.limit !== undefined && emitted >= opts.limit)
|
|
150
150
|
break;
|
|
151
|
-
const aligned =
|
|
151
|
+
const aligned = componentsPresentIn(v.comp, v.raw);
|
|
152
152
|
if (!Object.keys(aligned).length)
|
|
153
153
|
continue;
|
|
154
154
|
const sourceID = geonameid
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../lib/adapters/geonames/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EAAE,mBAAmB,EAAE,MAAM,iCAAiC,CAAA;AACrE,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAA;AACvD,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AACvC,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAGhD;;;GAGG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,UAAU,CAAA;AAC7C;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,WAAW,CAAA;AAEnD;;GAEG;AACH,MAAM,GAAG,GAAG;IACX,SAAS,EAAE,CAAC;IACZ,IAAI,EAAE,CAAC;IACP,cAAc,EAAE,CAAC;IACjB,YAAY,EAAE,CAAC;IACf,WAAW,EAAE,CAAC;IACd,OAAO,EAAE,CAAC;IACV,MAAM,EAAE,EAAE;CACD,CAAA;AAEV;;GAEG;AACH,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAA;AAElE;;GAEG;AACH,KAAK,UAAU,UAAU,CAAC,GAAW;IACpC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,sBAAsB,CAAC,CAAA;IAE5C,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,kGAAkG;IAClG,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,cAAc,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;QAC1E,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,aAAa,CAAC,GAAW;IACvC,MAAM,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAA;IACrC,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,EAAE,iBAAiB,CAAC,CAAA;IAEvC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,EAAE,CAAC,CAAC;QAAE,OAAO,GAAG,CAAA;IAEvC,gGAAgG;IAChG,8CAA8C;IAC9C,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,cAAc,CAAC,SAAS,CAAC,EAAE,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC;QAC1E,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC,GAAG,CAAC;YAAE,SAAQ;QAEtC,4DAA4D;QAC5D,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;YACxB,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;QAC1B,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED,MAAM,UAAU,qBAAqB;IACpC,OAAO;QACN,EAAE,EAAE,mBAAmB;QACvB,cAAc,EAAE,wBAAwB;QACxC,WAAW,EACV,0JAA0J;QAE3J,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YACnC,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,GAAG,CAAC,CAAA;YACpC,MAAM,SAAS,GAAG,MAAM,aAAa,CAAC,GAAG,CAAC,CAAA;YAE1C,wDAAwD;YACxD,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAA;YAExE,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;gBAC9B,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,IAAI,GAAG,CAAC,GAAG,CAAC,YAAY,CAAC,KAAK,GAAG;oBAAE,SAAQ;gBAE3C,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;oBAAE,SAAQ;gBAE7D,MAAM,EAAE,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1C,IAAI,CAAC,EAAE;oBAAE,SAAQ;gBAEjB,IAAI,IAAI,CAAC,OAAO,IAAI,EAAE,KAAK,IAAI,CAAC,OAAO;oBAAE,SAAQ;gBAEjD,MAAM,QAAQ,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACvB,MAAM,SAAS,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACnD,MAAM,MAAM,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA;gBACpE,MAAM,OAAO,GAAG,SAAS,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;gBAEjC,8EAA8E;gBAC9E,6CAA6C;gBAC7C,MAAM,QAAQ,GAA2E,EAAE,CAAA;gBAE3F,IAAI,MAAM,EAAE,CAAC;oBACZ,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,EAAE,EAAE,CAAC,CAAA;oBAExF,IAAI,OAAO,EAAE,CAAC;wBACb,QAAQ,CAAC,IAAI,CAAC;4BACb,IAAI,EAAE,KAAK;4BACX,IAAI,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE;4BACnC,GAAG,EAAE,GAAG,QAAQ,KAAK,MAAM,KAAK,OAAO,EAAE;yBACzC,CAAC,CAAA;oBACH,CAAC;gBACF,CAAC;qBAAM,IAAI,OAAO,EAAE,CAAC;oBACpB,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,OAAO,EAAE,EAAE,GAAG,EAAE,GAAG,QAAQ,KAAK,OAAO,EAAE,EAAE,CAAC,CAAA;gBAC3F,CAAC;qBAAM,CAAC;oBACP,QAAQ,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAC,CAAA;gBAChE,CAAC;gBAED,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAC5D,MAAM,OAAO,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAA;oBAElD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;wBAAE,SAAQ;oBAE1C,MAAM,QAAQ,GAAG,SAAS;wBACzB,CAAC,CAAC,GAAG,mBAAmB,IAAI,SAAS,IAAI,CAAC,CAAC,IAAI,EAAE;wBACjD,CAAC,CAAC,cAAc,CAAC,mBAAmB,EAAE,OAAO,CAAC,CAAA;oBAE/C,MAAM;wBACL,GAAG,EAAE,CAAC,CAAC,GAAG;wBACV,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,EAAE;wBACX,MAAM,EAAE,mBAAmB;wBAC3B,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,wBAAwB;qBACjC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,qBAAqB,EAAE,CAAA"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/adapters/geonames/postal/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AAMH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,QAAQ,CAAA;AAEzE;;;GAGG;AACH,eAAO,MAAM,0BAA0B,oBAAoB,CAAA;AAC3D;;;GAGG;AACH,eAAO,MAAM,+BAA+B,cAAc,CAAA;AAE1D;;;;;GAKG;AACH,eAAO,MAAM,uBAAuB;aAAK,OAAO,EAAE,CAAC;aAAE,QAAQ,EAAE,CAAC;aAAE,KAAK,EAAE,CAAC;aAAE,UAAU,EAAE,CAAC;aAAE,UAAU,EAAE,CAAC;CAAW,CAAA;AAEnH,wBAAgB,2BAA2B,IAAI,aAAa,CAkE3D;AAED;;GAEG;AACH,eAAO,MAAM,qBAAqB,eAAgC,CAAA"}
|