@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 +1,38 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { makeCaPostcode, makePoBoxPhrase } from "#recipes/po/box/cedex/phrases";
|
|
5
|
+
import { CA_EN_LEADERS, CA_FR_LEADERS, ON_FSA_LETTERS, QC_FSA_LETTERS, VENUES_FR, } from "#recipes/po/box/cedex/vocabulary";
|
|
6
|
+
import { pick } from "#synthesizers/utils";
|
|
7
|
+
const CA_FR_GOLDEN_ORDER_CUTOFF = 0.4, CA_FR_NATIVE_CUTOFF = 0.7, CA_FR_BARE_CUTOFF = 0.85;
|
|
8
|
+
const CA_EN_STANDARD_CUTOFF = 0.5, CA_EN_GOLDEN_ORDER_CUTOFF = 0.8;
|
|
9
|
+
export function renderCaFr(random, locality) {
|
|
10
|
+
const phrase = makePoBoxPhrase(random, CA_FR_LEADERS);
|
|
11
|
+
const postcode = makeCaPostcode(random, QC_FSA_LETTERS);
|
|
12
|
+
const components = { po_box: phrase, postcode, locality, region: "QC" };
|
|
13
|
+
const draw = random();
|
|
14
|
+
if (draw < CA_FR_GOLDEN_ORDER_CUTOFF)
|
|
15
|
+
return { fmt: "ca-fr-golden-order", raw: `${phrase}, ${postcode} ${locality}, QC`, components };
|
|
16
|
+
if (draw < CA_FR_NATIVE_CUTOFF)
|
|
17
|
+
return { fmt: "ca-fr-native", raw: `${phrase}, ${locality} QC ${postcode}`, components };
|
|
18
|
+
if (draw < CA_FR_BARE_CUTOFF)
|
|
19
|
+
return { fmt: "ca-fr-bare", raw: phrase, components: { po_box: phrase } };
|
|
20
|
+
const venue = pick(VENUES_FR, random);
|
|
21
|
+
return {
|
|
22
|
+
fmt: "ca-fr-venue",
|
|
23
|
+
raw: `${venue}, ${phrase}, ${locality} QC ${postcode}`,
|
|
24
|
+
components: { venue, ...components },
|
|
25
|
+
};
|
|
26
|
+
}
|
|
27
|
+
export function renderCaEn(random, locality) {
|
|
28
|
+
const phrase = makePoBoxPhrase(random, CA_EN_LEADERS);
|
|
29
|
+
const postcode = makeCaPostcode(random, ON_FSA_LETTERS);
|
|
30
|
+
const components = { po_box: phrase, locality, region: "ON", postcode };
|
|
31
|
+
const draw = random();
|
|
32
|
+
if (draw < CA_EN_STANDARD_CUTOFF)
|
|
33
|
+
return { fmt: "ca-en-standard", raw: `${phrase}, ${locality}, ON ${postcode}`, components };
|
|
34
|
+
if (draw < CA_EN_GOLDEN_ORDER_CUTOFF)
|
|
35
|
+
return { fmt: "ca-en-golden-order", raw: `${phrase}, ${postcode} ${locality}, ON`, components };
|
|
36
|
+
return { fmt: "ca-en-bare", raw: phrase, components: { po_box: phrase } };
|
|
37
|
+
}
|
|
38
|
+
//# sourceMappingURL=ca.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"ca.js","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/ca.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,+BAA+B,CAAA;AAE/E,OAAO,EACN,aAAa,EACb,aAAa,EACb,cAAc,EACd,cAAc,EACd,SAAS,GACT,MAAM,kCAAkC,CAAA;AACzC,OAAO,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAA;AAE1C,MAAM,yBAAyB,GAAG,GAAG,EACpC,mBAAmB,GAAG,GAAG,EACzB,iBAAiB,GAAG,IAAI,CAAA;AAEzB,MAAM,qBAAqB,GAAG,GAAG,EAChC,yBAAyB,GAAG,GAAG,CAAA;AAEhC,MAAM,UAAU,UAAU,CAAC,MAAoB,EAAE,QAAgB;IAChE,MAAM,MAAM,GAAG,eAAe,CAAC,MAAM,EAAE,aAAa,CAAC,CAAA;IACrD,MAAM,QAAQ,GAAG,cAAc,CAAC,MAAM,EAAE,cAAc,CAAC,CAAA;IACvD,MAAM,UAAU,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,CAAA;IACvE,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IAErB,IAAI,IAAI,GAAG,yBAAyB;QACnC,OAAO,EAAE,GAAG,EAAE,oBAAoB,EAAE,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,IAAI,QAAQ,MAAM,EAAE,UAAU,EAAE,CAAA;IAEhG,IAAI,IAAI,GAAG,mBAAmB;QAC7B,OAAO,EAAE,GAAG,EAAE,cAAc,EAAE,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,OAAO,QAAQ,EAAE,EAAE,UAAU,EAAE,CAAA;IAEzF,IAAI,IAAI,GAAG,iBAAiB;QAAE,OAAO,EAAE,GAAG,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,CAAA;IACvG,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;IAErC,OAAO;QACN,GAAG,EAAE,aAAa;QAClB,GAAG,EAAE,GAAG,KAAK,KAAK,MAAM,KAAK,QAAQ,OAAO,QAAQ,EAAE;QACtD,UAAU,EAAE,EAAE,KAAK,EAAE,GAAG,UAAU,EAAE;KACpC,CAAA;AACF,CAAC;AAED,MAAM,UAAU,UAAU,CAAC,MAAoB,EAAE,QAAgB;IAChE,MAAM,MAAM,GAAG,eAAe,CAAC,MAAM,EAAE,aAAa,CAAC,CAAA;IACrD,MAAM,QAAQ,GAAG,cAAc,CAAC,MAAM,EAAE,cAAc,CAAC,CAAA;IACvD,MAAM,UAAU,GAAG,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAA;IACvE,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IAErB,IAAI,IAAI,GAAG,qBAAqB;QAC/B,OAAO,EAAE,GAAG,EAAE,gBAAgB,EAAE,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,QAAQ,QAAQ,EAAE,EAAE,UAAU,EAAE,CAAA;IAE5F,IAAI,IAAI,GAAG,yBAAyB;QACnC,OAAO,EAAE,GAAG,EAAE,oBAAoB,EAAE,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,IAAI,QAAQ,MAAM,EAAE,UAAU,EAAE,CAAA;IAEhG,OAAO,EAAE,GAAG,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,CAAA;AAC1E,CAAC"}
|
|
@@ -0,0 +1,7 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import type { FRTuple, Rendered } from "#recipes/po/box/cedex/types";
|
|
5
|
+
export declare function renderBpFr(random: () => number, tuple: FRTuple): Rendered;
|
|
6
|
+
export declare function renderCedexFr(random: () => number, tuple: FRTuple): Rendered;
|
|
7
|
+
//# sourceMappingURL=fr.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fr.d.ts","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/fr.ts"],"names":[],"mappings":"AAAA;;GAEG;AAGH,OAAO,KAAK,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,6BAA6B,CAAA;AAapE,wBAAgB,UAAU,CAAC,MAAM,EAAE,MAAM,MAAM,EAAE,KAAK,EAAE,OAAO,GAAG,QAAQ,CAiCzE;AAED,wBAAgB,aAAa,CAAC,MAAM,EAAE,MAAM,MAAM,EAAE,KAAK,EAAE,OAAO,GAAG,QAAQ,CAsB5E"}
|
|
@@ -0,0 +1,57 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { makeCedex, makePoBoxPhrase } from "#recipes/po/box/cedex/phrases";
|
|
5
|
+
import { FR_LEADERS, VENUES_FR } from "#recipes/po/box/cedex/vocabulary";
|
|
6
|
+
import { pick } from "#synthesizers/utils";
|
|
7
|
+
const BP_TAIL_CUTOFF = 0.45, BP_BARE_CUTOFF = 0.6, BP_CEDEX_CUTOFF = 0.8;
|
|
8
|
+
const CEDEX_UPPER_LOCALITY_SHARE = 0.6, CEDEX_LINE_CUTOFF = 0.4, CEDEX_FULL_CUTOFF = 0.75, CEDEX_GOLDEN_ORDER_CUTOFF = 0.85;
|
|
9
|
+
export function renderBpFr(random, tuple) {
|
|
10
|
+
const phrase = makePoBoxPhrase(random, FR_LEADERS);
|
|
11
|
+
const { locality, postcode } = tuple;
|
|
12
|
+
const location = random() < 0.5 ? locality.toUpperCase() : locality;
|
|
13
|
+
const draw = random();
|
|
14
|
+
if (draw < BP_TAIL_CUTOFF)
|
|
15
|
+
return {
|
|
16
|
+
fmt: "bp-tail",
|
|
17
|
+
raw: `${phrase}, ${postcode} ${location}`,
|
|
18
|
+
components: { po_box: phrase, postcode, locality: location },
|
|
19
|
+
};
|
|
20
|
+
if (draw < BP_BARE_CUTOFF)
|
|
21
|
+
return { fmt: "bp-bare", raw: phrase, components: { po_box: phrase } };
|
|
22
|
+
if (draw < BP_CEDEX_CUTOFF) {
|
|
23
|
+
const cedex = makeCedex(random);
|
|
24
|
+
const upperLocality = locality.toUpperCase();
|
|
25
|
+
return {
|
|
26
|
+
fmt: "bp-cedex",
|
|
27
|
+
raw: `${phrase}, ${postcode} ${upperLocality} ${cedex}`,
|
|
28
|
+
components: { po_box: phrase, postcode, locality: upperLocality, cedex },
|
|
29
|
+
};
|
|
30
|
+
}
|
|
31
|
+
const venue = pick(VENUES_FR, random);
|
|
32
|
+
return {
|
|
33
|
+
fmt: "bp-venue",
|
|
34
|
+
raw: `${venue}, ${phrase}, ${postcode} ${location}`,
|
|
35
|
+
components: { venue, po_box: phrase, postcode, locality: location },
|
|
36
|
+
};
|
|
37
|
+
}
|
|
38
|
+
export function renderCedexFr(random, tuple) {
|
|
39
|
+
const cedex = makeCedex(random);
|
|
40
|
+
const { house_number, street, locality, postcode } = tuple;
|
|
41
|
+
const location = random() < CEDEX_UPPER_LOCALITY_SHARE ? locality.toUpperCase() : locality;
|
|
42
|
+
const line = { postcode, locality: location, cedex };
|
|
43
|
+
const draw = random();
|
|
44
|
+
if (draw < CEDEX_LINE_CUTOFF)
|
|
45
|
+
return { fmt: "cedex-line", raw: `${postcode} ${location} ${cedex}`, components: line };
|
|
46
|
+
if (draw < CEDEX_FULL_CUTOFF)
|
|
47
|
+
return {
|
|
48
|
+
fmt: "cedex-full",
|
|
49
|
+
raw: `${house_number} ${street}, ${postcode} ${location} ${cedex}`,
|
|
50
|
+
components: { house_number, street, ...line },
|
|
51
|
+
};
|
|
52
|
+
if (draw < CEDEX_GOLDEN_ORDER_CUTOFF)
|
|
53
|
+
return { fmt: "cedex-golden-order", raw: `${postcode} ${cedex} ${location}`, components: line };
|
|
54
|
+
const venue = pick(VENUES_FR, random);
|
|
55
|
+
return { fmt: "cedex-venue", raw: `${venue}, ${postcode} ${location} ${cedex}`, components: { venue, ...line } };
|
|
56
|
+
}
|
|
57
|
+
//# sourceMappingURL=fr.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"fr.js","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/fr.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,EAAE,SAAS,EAAE,eAAe,EAAE,MAAM,+BAA+B,CAAA;AAE1E,OAAO,EAAE,UAAU,EAAE,SAAS,EAAE,MAAM,kCAAkC,CAAA;AACxE,OAAO,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAA;AAE1C,MAAM,cAAc,GAAG,IAAI,EAC1B,cAAc,GAAG,GAAG,EACpB,eAAe,GAAG,GAAG,CAAA;AAEtB,MAAM,0BAA0B,GAAG,GAAG,EACrC,iBAAiB,GAAG,GAAG,EACvB,iBAAiB,GAAG,IAAI,EACxB,yBAAyB,GAAG,IAAI,CAAA;AAEjC,MAAM,UAAU,UAAU,CAAC,MAAoB,EAAE,KAAc;IAC9D,MAAM,MAAM,GAAG,eAAe,CAAC,MAAM,EAAE,UAAU,CAAC,CAAA;IAClD,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAA;IACpC,MAAM,QAAQ,GAAG,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAA;IACnE,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IAErB,IAAI,IAAI,GAAG,cAAc;QACxB,OAAO;YACN,GAAG,EAAE,SAAS;YACd,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,IAAI,QAAQ,EAAE;YACzC,UAAU,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE;SAC5D,CAAA;IAEF,IAAI,IAAI,GAAG,cAAc;QAAE,OAAO,EAAE,GAAG,EAAE,SAAS,EAAE,GAAG,EAAE,MAAM,EAAE,UAAU,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,EAAE,CAAA;IAEjG,IAAI,IAAI,GAAG,eAAe,EAAE,CAAC;QAC5B,MAAM,KAAK,GAAG,SAAS,CAAC,MAAM,CAAC,CAAA;QAC/B,MAAM,aAAa,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAA;QAE5C,OAAO;YACN,GAAG,EAAE,UAAU;YACf,GAAG,EAAE,GAAG,MAAM,KAAK,QAAQ,IAAI,aAAa,IAAI,KAAK,EAAE;YACvD,UAAU,EAAE,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,aAAa,EAAE,KAAK,EAAE;SACxE,CAAA;IACF,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;IAErC,OAAO;QACN,GAAG,EAAE,UAAU;QACf,GAAG,EAAE,GAAG,KAAK,KAAK,MAAM,KAAK,QAAQ,IAAI,QAAQ,EAAE;QACnD,UAAU,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE;KACnE,CAAA;AACF,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,MAAoB,EAAE,KAAc;IACjE,MAAM,KAAK,GAAG,SAAS,CAAC,MAAM,CAAC,CAAA;IAC/B,MAAM,EAAE,YAAY,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,GAAG,KAAK,CAAA;IAC1D,MAAM,QAAQ,GAAG,MAAM,EAAE,GAAG,0BAA0B,CAAC,CAAC,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAA;IAC1F,MAAM,IAAI,GAAG,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAA;IACpD,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IAErB,IAAI,IAAI,GAAG,iBAAiB;QAAE,OAAO,EAAE,GAAG,EAAE,YAAY,EAAE,GAAG,EAAE,GAAG,QAAQ,IAAI,QAAQ,IAAI,KAAK,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAA;IAErH,IAAI,IAAI,GAAG,iBAAiB;QAC3B,OAAO;YACN,GAAG,EAAE,YAAY;YACjB,GAAG,EAAE,GAAG,YAAY,IAAI,MAAM,KAAK,QAAQ,IAAI,QAAQ,IAAI,KAAK,EAAE;YAClE,UAAU,EAAE,EAAE,YAAY,EAAE,MAAM,EAAE,GAAG,IAAI,EAAE;SAC7C,CAAA;IAEF,IAAI,IAAI,GAAG,yBAAyB;QACnC,OAAO,EAAE,GAAG,EAAE,oBAAoB,EAAE,GAAG,EAAE,GAAG,QAAQ,IAAI,KAAK,IAAI,QAAQ,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAA;IAEhG,MAAM,KAAK,GAAG,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,CAAA;IAErC,OAAO,EAAE,GAAG,EAAE,aAAa,EAAE,GAAG,EAAE,GAAG,KAAK,KAAK,QAAQ,IAAI,QAAQ,IAAI,KAAK,EAAE,EAAE,UAAU,EAAE,EAAE,KAAK,EAAE,GAAG,IAAI,EAAE,EAAE,CAAA;AACjH,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
export declare function makePoBoxPhrase(random: () => number, leaders: ReadonlyArray<string>, rareLeaders?: ReadonlyArray<string>): string;
|
|
5
|
+
export declare function makeCedex(random: () => number): string;
|
|
6
|
+
export declare function makeAuNZPoBoxPhrase(random: () => number, leaders: ReadonlyArray<string>, rareLeaders: ReadonlyArray<string>, validate: (input: unknown) => boolean): string;
|
|
7
|
+
export declare function makeCaPostcode(random: () => number, fsaLetters: string[]): string;
|
|
8
|
+
//# sourceMappingURL=phrases.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"phrases.d.ts","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/phrases.ts"],"names":[],"mappings":"AAAA;;GAEG;AAmCH,wBAAgB,eAAe,CAC9B,MAAM,EAAE,MAAM,MAAM,EACpB,OAAO,EAAE,aAAa,CAAC,MAAM,CAAC,EAC9B,WAAW,CAAC,EAAE,aAAa,CAAC,MAAM,CAAC,GACjC,MAAM,CAcR;AAED,wBAAgB,SAAS,CAAC,MAAM,EAAE,MAAM,MAAM,GAAG,MAAM,CAatD;AAED,wBAAgB,mBAAmB,CAClC,MAAM,EAAE,MAAM,MAAM,EACpB,OAAO,EAAE,aAAa,CAAC,MAAM,CAAC,EAC9B,WAAW,EAAE,aAAa,CAAC,MAAM,CAAC,EAClC,QAAQ,EAAE,CAAC,KAAK,EAAE,OAAO,KAAK,OAAO,GACnC,MAAM,CAoBR;AAED,wBAAgB,cAAc,CAAC,MAAM,EAAE,MAAM,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,GAAG,MAAM,CAQjF"}
|
|
@@ -0,0 +1,76 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { normalizeCaPostalCode } from "@mailwoman/codex/ca";
|
|
5
|
+
import { isCedex } from "@mailwoman/codex/fr";
|
|
6
|
+
import { isNZDeliveryService } from "@mailwoman/codex/nz";
|
|
7
|
+
import { isUSPoBoxDesignator, matchPOBox } from "@mailwoman/codex/us";
|
|
8
|
+
import { CA_INTERIOR_LETTERS } from "#recipes/po/box/cedex/vocabulary";
|
|
9
|
+
import { maybeNoisifyBoxNumber } from "#synthesizers/po-box";
|
|
10
|
+
import { pick, tieredNumber } from "#synthesizers/utils";
|
|
11
|
+
const BOX_TWO_DIGIT_CUTOFF = 0.3;
|
|
12
|
+
const BOX_THREE_DIGIT_CUTOFF = 0.7;
|
|
13
|
+
const BOX_FOUR_DIGIT_CUTOFF = 0.95;
|
|
14
|
+
const TEMPLATE_CASE_CUTOFF = 0.7;
|
|
15
|
+
const UPPER_CASE_CUTOFF = 0.92;
|
|
16
|
+
const RARE_LEADER_SHARE = 0.1;
|
|
17
|
+
const CEDEX_UPPER_CUTOFF = 0.6;
|
|
18
|
+
const CEDEX_TITLE_CUTOFF = 0.9;
|
|
19
|
+
const CEDEX_UNNUMBERED_SHARE = 0.2;
|
|
20
|
+
const pickBoxNumber = (random) => tieredNumber(random, [
|
|
21
|
+
{ cutoff: BOX_TWO_DIGIT_CUTOFF, base: 1, span: 99 },
|
|
22
|
+
{ cutoff: BOX_THREE_DIGIT_CUTOFF, base: 100, span: 900 },
|
|
23
|
+
{ cutoff: BOX_FOUR_DIGIT_CUTOFF, base: 1000, span: 9000 },
|
|
24
|
+
{ base: 10_000, span: 90_000 },
|
|
25
|
+
]);
|
|
26
|
+
const caseDial = (random, value) => {
|
|
27
|
+
const draw = random();
|
|
28
|
+
return draw < TEMPLATE_CASE_CUTOFF ? value : draw < UPPER_CASE_CUTOFF ? value.toUpperCase() : value.toLowerCase();
|
|
29
|
+
};
|
|
30
|
+
export function makePoBoxPhrase(random, leaders, rareLeaders) {
|
|
31
|
+
let leader = pick(leaders, random);
|
|
32
|
+
if (rareLeaders && random() < RARE_LEADER_SHARE) {
|
|
33
|
+
leader = pick(rareLeaders, random);
|
|
34
|
+
}
|
|
35
|
+
const number = maybeNoisifyBoxNumber(pickBoxNumber(random), random);
|
|
36
|
+
const phrase = leader === "#" ? `#${number}` : `${caseDial(random, leader)} ${number}`;
|
|
37
|
+
if (isUSPoBoxDesignator(leader) && /^[\dA-Za-z][\dA-Za-z-]*$/.test(number) && !matchPOBox(phrase))
|
|
38
|
+
throw new Error(`generated a po_box phrase the codex matcher rejects: "${phrase}"`);
|
|
39
|
+
return phrase;
|
|
40
|
+
}
|
|
41
|
+
export function makeCedex(random) {
|
|
42
|
+
const draw = random();
|
|
43
|
+
const word = draw < CEDEX_UPPER_CUTOFF ? "CEDEX" : draw < CEDEX_TITLE_CUTOFF ? "Cedex" : "cedex";
|
|
44
|
+
let phrase = word;
|
|
45
|
+
if (random() >= CEDEX_UNNUMBERED_SHARE) {
|
|
46
|
+
const number = 1 + Math.floor(random() * 20);
|
|
47
|
+
phrase = `${word} ${random() < 0.5 ? String(number).padStart(2, "0") : String(number)}`;
|
|
48
|
+
}
|
|
49
|
+
if (!isCedex(phrase))
|
|
50
|
+
throw new Error(`makeCedex emitted a phrase the codex matcher rejects: "${phrase}"`);
|
|
51
|
+
return phrase;
|
|
52
|
+
}
|
|
53
|
+
export function makeAuNZPoBoxPhrase(random, leaders, rareLeaders, validate) {
|
|
54
|
+
let leader = pick(leaders, random);
|
|
55
|
+
if (random() < RARE_LEADER_SHARE) {
|
|
56
|
+
leader = pick(rareLeaders, random);
|
|
57
|
+
}
|
|
58
|
+
let number = maybeNoisifyBoxNumber(pickBoxNumber(random), random);
|
|
59
|
+
if (leader === "CMB" && validate === isNZDeliveryService) {
|
|
60
|
+
number = `B${number}`;
|
|
61
|
+
}
|
|
62
|
+
const phrase = `${caseDial(random, leader)} ${number}`;
|
|
63
|
+
const cleanID = validate === isNZDeliveryService ? /^[\dA-Za-z]+$/ : /^[\dA-Za-z][\dA-Za-z-]*$/;
|
|
64
|
+
if (cleanID.test(number) && !validate(phrase))
|
|
65
|
+
throw new Error(`generated a phrase the codex matcher rejects: "${phrase}"`);
|
|
66
|
+
return phrase;
|
|
67
|
+
}
|
|
68
|
+
export function makeCaPostcode(random, fsaLetters) {
|
|
69
|
+
const letter = () => CA_INTERIOR_LETTERS[Math.floor(random() * CA_INTERIOR_LETTERS.length)];
|
|
70
|
+
const digit = () => String(Math.floor(random() * 10));
|
|
71
|
+
const postcode = `${pick(fsaLetters, random)}${digit()}${letter()} ${digit()}${letter()}${digit()}`;
|
|
72
|
+
if (!normalizeCaPostalCode(postcode))
|
|
73
|
+
throw new Error(`generated an invalid CA postcode: ${postcode}`);
|
|
74
|
+
return postcode;
|
|
75
|
+
}
|
|
76
|
+
//# sourceMappingURL=phrases.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"phrases.js","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/phrases.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,EAAE,qBAAqB,EAAE,MAAM,qBAAqB,CAAA;AAC3D,OAAO,EAAE,OAAO,EAAE,MAAM,qBAAqB,CAAA;AAC7C,OAAO,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAA;AACzD,OAAO,EAAE,mBAAmB,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAErE,OAAO,EAAE,mBAAmB,EAAE,MAAM,kCAAkC,CAAA;AACtE,OAAO,EAAE,qBAAqB,EAAE,MAAM,sBAAsB,CAAA;AAC5D,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAA;AAExD,MAAM,oBAAoB,GAAG,GAAG,CAAA;AAChC,MAAM,sBAAsB,GAAG,GAAG,CAAA;AAClC,MAAM,qBAAqB,GAAG,IAAI,CAAA;AAClC,MAAM,oBAAoB,GAAG,GAAG,CAAA;AAChC,MAAM,iBAAiB,GAAG,IAAI,CAAA;AAC9B,MAAM,iBAAiB,GAAG,GAAG,CAAA;AAC7B,MAAM,kBAAkB,GAAG,GAAG,CAAA;AAC9B,MAAM,kBAAkB,GAAG,GAAG,CAAA;AAC9B,MAAM,sBAAsB,GAAG,GAAG,CAAA;AAElC,MAAM,aAAa,GAAG,CAAC,MAAoB,EAAU,EAAE,CACtD,YAAY,CAAC,MAAM,EAAE;IACpB,EAAE,MAAM,EAAE,oBAAoB,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE;IACnD,EAAE,MAAM,EAAE,sBAAsB,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE;IACxD,EAAE,MAAM,EAAE,qBAAqB,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE;IACzD,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE;CAC9B,CAAC,CAAA;AAEH,MAAM,QAAQ,GAAG,CAAC,MAAoB,EAAE,KAAa,EAAU,EAAE;IAChE,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IAErB,OAAO,IAAI,GAAG,oBAAoB,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,GAAG,iBAAiB,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,WAAW,EAAE,CAAA;AAClH,CAAC,CAAA;AAED,MAAM,UAAU,eAAe,CAC9B,MAAoB,EACpB,OAA8B,EAC9B,WAAmC;IAEnC,IAAI,MAAM,GAAG,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IAElC,IAAI,WAAW,IAAI,MAAM,EAAE,GAAG,iBAAiB,EAAE,CAAC;QACjD,MAAM,GAAG,IAAI,CAAC,WAAW,EAAE,MAAM,CAAC,CAAA;IACnC,CAAC;IAED,MAAM,MAAM,GAAG,qBAAqB,CAAC,aAAa,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC,CAAA;IACnE,MAAM,MAAM,GAAG,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,EAAE,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC,IAAI,MAAM,EAAE,CAAA;IAEtF,IAAI,mBAAmB,CAAC,MAAM,CAAC,IAAI,0BAA0B,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC;QAChG,MAAM,IAAI,KAAK,CAAC,yDAAyD,MAAM,GAAG,CAAC,CAAA;IAEpF,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,UAAU,SAAS,CAAC,MAAoB;IAC7C,MAAM,IAAI,GAAG,MAAM,EAAE,CAAA;IACrB,MAAM,IAAI,GAAG,IAAI,GAAG,kBAAkB,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,GAAG,kBAAkB,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,OAAO,CAAA;IAChG,IAAI,MAAM,GAAG,IAAI,CAAA;IAEjB,IAAI,MAAM,EAAE,IAAI,sBAAsB,EAAE,CAAC;QACxC,MAAM,MAAM,GAAG,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAA;QAC5C,MAAM,GAAG,GAAG,IAAI,IAAI,MAAM,EAAE,GAAG,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE,CAAA;IACxF,CAAC;IAED,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,0DAA0D,MAAM,GAAG,CAAC,CAAA;IAE1G,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,UAAU,mBAAmB,CAClC,MAAoB,EACpB,OAA8B,EAC9B,WAAkC,EAClC,QAAqC;IAErC,IAAI,MAAM,GAAG,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;IAElC,IAAI,MAAM,EAAE,GAAG,iBAAiB,EAAE,CAAC;QAClC,MAAM,GAAG,IAAI,CAAC,WAAW,EAAE,MAAM,CAAC,CAAA;IACnC,CAAC;IAED,IAAI,MAAM,GAAG,qBAAqB,CAAC,aAAa,CAAC,MAAM,CAAC,EAAE,MAAM,CAAC,CAAA;IAEjE,IAAI,MAAM,KAAK,KAAK,IAAI,QAAQ,KAAK,mBAAmB,EAAE,CAAC;QAC1D,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;IACtB,CAAC;IAED,MAAM,MAAM,GAAG,GAAG,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAC,IAAI,MAAM,EAAE,CAAA;IACtD,MAAM,OAAO,GAAG,QAAQ,KAAK,mBAAmB,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,0BAA0B,CAAA;IAE/F,IAAI,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC;QAC5C,MAAM,IAAI,KAAK,CAAC,kDAAkD,MAAM,GAAG,CAAC,CAAA;IAE7E,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,MAAoB,EAAE,UAAoB;IACxE,MAAM,MAAM,GAAG,GAAG,EAAE,CAAC,mBAAmB,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,mBAAmB,CAAC,MAAM,CAAC,CAAE,CAAA;IAC5F,MAAM,KAAK,GAAG,GAAG,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,CAAC,CAAA;IACrD,MAAM,QAAQ,GAAG,GAAG,IAAI,CAAC,UAAU,EAAE,MAAM,CAAC,GAAG,KAAK,EAAE,GAAG,MAAM,EAAE,IAAI,KAAK,EAAE,GAAG,MAAM,EAAE,GAAG,KAAK,EAAE,EAAE,CAAA;IAEnG,IAAI,CAAC,qBAAqB,CAAC,QAAQ,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,qCAAqC,QAAQ,EAAE,CAAC,CAAA;IAEtG,OAAO,QAAQ,CAAA;AAChB,CAAC"}
|
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { type CorpusRecipe } from "#recipes/scaffold";
|
|
5
|
+
/**
|
|
6
|
+
* Generate PO-box and CEDEX rows from US, French, Canadian, Australian, and New Zealand source pools.
|
|
7
|
+
*/
|
|
8
|
+
export declare const poBoxCedexRecipe: CorpusRecipe;
|
|
9
|
+
//# sourceMappingURL=recipe.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"recipe.d.ts","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/recipe.ts"],"names":[],"mappings":"AAAA;;GAEG;AAwBH,OAAO,EAAyC,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AAsB5F;;GAEG;AACH,eAAO,MAAM,gBAAgB,EAAE,YAuJ9B,CAAA"}
|
|
@@ -0,0 +1,153 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { dataRootPath } from "@mailwoman/core/data-root";
|
|
5
|
+
import { mulberry32 as makeMulberry32 } from "@mailwoman/core/utils";
|
|
6
|
+
import { renderAUPoBox, renderNZPoBox } from "#recipes/po/box/cedex/au-nz";
|
|
7
|
+
import { renderCaEn, renderCaFr } from "#recipes/po/box/cedex/ca";
|
|
8
|
+
import { renderBpFr, renderCedexFr } from "#recipes/po/box/cedex/fr";
|
|
9
|
+
import { FR_SOURCE, GEONAMES_CA, GEONAMES_POSTAL_AU, GEONAMES_POSTAL_NZ, isHoldoutLocality, readCaLocalities, readFrTuples, readPostalTuples, readUsTuples, US_EVAL_SOURCE, US_TRAIN_SOURCES, } from "#recipes/po/box/cedex/sources";
|
|
10
|
+
import { renderPmbUs, renderPoBoxUs } from "#recipes/po/box/cedex/us";
|
|
11
|
+
import { CLASS_MIX } from "#recipes/po/box/cedex/vocabulary";
|
|
12
|
+
import { sliceSourceID } from "#recipes/scaffold";
|
|
13
|
+
import { synthesizeMilitaryPoBoxRow } from "#synthesizers/po-box";
|
|
14
|
+
import { pick } from "#synthesizers/utils";
|
|
15
|
+
import { alignRow } from "#utils";
|
|
16
|
+
const COMPONENT_ORDER = ["house_number", "street", "po_box", "venue", "locality", "postcode", "region", "cedex"];
|
|
17
|
+
const orderComponents = (components) => Object.fromEntries(COMPONENT_ORDER.filter((key) => components[key]).map((key) => [key, components[key]]));
|
|
18
|
+
const pickClass = (draw) => {
|
|
19
|
+
let accumulated = 0;
|
|
20
|
+
for (const [name, weight] of CLASS_MIX) {
|
|
21
|
+
accumulated += weight;
|
|
22
|
+
if (draw < accumulated)
|
|
23
|
+
return name;
|
|
24
|
+
}
|
|
25
|
+
return CLASS_MIX.at(-1)[0];
|
|
26
|
+
};
|
|
27
|
+
/**
|
|
28
|
+
* Generate PO-box and CEDEX rows from US, French, Canadian, Australian, and New Zealand source pools.
|
|
29
|
+
*/
|
|
30
|
+
export const poBoxCedexRecipe = {
|
|
31
|
+
name: "po-box-cedex",
|
|
32
|
+
description: "PO box / CEDEX coverage rows (US/FR/CA/AU/NZ) — self-generated from cached OA + GeoNames pools",
|
|
33
|
+
mode: "generate",
|
|
34
|
+
options: [{ flag: "--golden", description: "Emit the leakage-safe holdout variant ({raw, components, country})" }],
|
|
35
|
+
async run(opts, write) {
|
|
36
|
+
const random = makeMulberry32(opts.seed);
|
|
37
|
+
const count = opts.count ?? 50_000;
|
|
38
|
+
const source = opts.sourceName ?? "synth-po-box-cedex";
|
|
39
|
+
const usPool = [];
|
|
40
|
+
for (const sourceEntry of opts.golden ? [US_EVAL_SOURCE] : US_TRAIN_SOURCES) {
|
|
41
|
+
const tuples = await readUsTuples(sourceEntry);
|
|
42
|
+
console.error(` ${sourceEntry.csv}: ${tuples.length} tuples`);
|
|
43
|
+
for (const tuple of tuples) {
|
|
44
|
+
usPool.push(tuple);
|
|
45
|
+
}
|
|
46
|
+
}
|
|
47
|
+
const frAll = await readFrTuples(80_000);
|
|
48
|
+
const frPool = frAll.filter((tuple) => isHoldoutLocality(tuple.locality) === opts.golden);
|
|
49
|
+
console.error(` ${FR_SOURCE.csv}: ${frAll.length} tuples (${frPool.length} after holdout split)`);
|
|
50
|
+
const qcAll = await readCaLocalities("10"), onAll = await readCaLocalities("08");
|
|
51
|
+
const qcPool = qcAll.filter((locality) => isHoldoutLocality(locality) === opts.golden), onPool = onAll.filter((locality) => isHoldoutLocality(locality) === opts.golden);
|
|
52
|
+
console.error(` GeoNames CA: QC ${qcAll.length}→${qcPool.length}, ON ${onAll.length}→${onPool.length}`);
|
|
53
|
+
const auAll = (await readPostalTuples(GEONAMES_POSTAL_AU, { withState: true })), nzAll = (await readPostalTuples(GEONAMES_POSTAL_NZ, { withState: false }));
|
|
54
|
+
const auPool = auAll.filter((tuple) => isHoldoutLocality(tuple.locality) === opts.golden), nzPool = nzAll.filter((tuple) => isHoldoutLocality(tuple.locality) === opts.golden);
|
|
55
|
+
console.error(` GeoNames postal: AU ${auAll.length}→${auPool.length}, NZ ${nzAll.length}→${nzPool.length}`);
|
|
56
|
+
if (!usPool.length || !frPool.length || !qcPool.length || !onPool.length || !auPool.length || !nzPool.length)
|
|
57
|
+
throw new Error(`A base pool is empty — check ${dataRootPath("oa-cache")}, ${GEONAMES_CA}, ${GEONAMES_POSTAL_AU.zip}, and ${GEONAMES_POSTAL_NZ.zip}.`);
|
|
58
|
+
let emitted = 0, skipped = 0, guard = 0;
|
|
59
|
+
while (emitted < count && guard++ < count * 10) {
|
|
60
|
+
const kind = pickClass(random());
|
|
61
|
+
let rendered;
|
|
62
|
+
let country;
|
|
63
|
+
let locale;
|
|
64
|
+
if (kind === "po-box-us") {
|
|
65
|
+
rendered = renderPoBoxUs(random, pick(usPool, random));
|
|
66
|
+
country = "US";
|
|
67
|
+
locale = "en-US";
|
|
68
|
+
}
|
|
69
|
+
else if (kind === "pmb-us") {
|
|
70
|
+
const tuple = pick(usPool, random);
|
|
71
|
+
if (!tuple.postcode || !tuple.street || !tuple.house_number)
|
|
72
|
+
continue;
|
|
73
|
+
rendered = renderPmbUs(random, tuple);
|
|
74
|
+
country = "US";
|
|
75
|
+
locale = "en-US";
|
|
76
|
+
}
|
|
77
|
+
else if (kind === "bp-fr") {
|
|
78
|
+
rendered = renderBpFr(random, pick(frPool, random));
|
|
79
|
+
country = "FR";
|
|
80
|
+
locale = "fr-FR";
|
|
81
|
+
}
|
|
82
|
+
else if (kind === "cedex-fr") {
|
|
83
|
+
rendered = renderCedexFr(random, pick(frPool, random));
|
|
84
|
+
country = "FR";
|
|
85
|
+
locale = "fr-FR";
|
|
86
|
+
}
|
|
87
|
+
else if (kind === "cp-ca-fr") {
|
|
88
|
+
rendered = renderCaFr(random, pick(qcPool, random));
|
|
89
|
+
country = "CA";
|
|
90
|
+
locale = "fr-CA";
|
|
91
|
+
}
|
|
92
|
+
else if (kind === "po-box-au") {
|
|
93
|
+
rendered = renderAUPoBox(random, pick(auPool, random));
|
|
94
|
+
country = "AU";
|
|
95
|
+
locale = "en-AU";
|
|
96
|
+
}
|
|
97
|
+
else if (kind === "po-box-nz") {
|
|
98
|
+
rendered = renderNZPoBox(random, pick(nzPool, random));
|
|
99
|
+
country = "NZ";
|
|
100
|
+
locale = "en-NZ";
|
|
101
|
+
}
|
|
102
|
+
else if (kind === "po-box-us-military") {
|
|
103
|
+
const military = synthesizeMilitaryPoBoxRow({ random });
|
|
104
|
+
const { country: _country, ...components } = military.components;
|
|
105
|
+
rendered = { fmt: "po-box-military", raw: military.raw, components: components };
|
|
106
|
+
country = "US";
|
|
107
|
+
locale = "en-US";
|
|
108
|
+
}
|
|
109
|
+
else {
|
|
110
|
+
rendered = renderCaEn(random, pick(onPool, random));
|
|
111
|
+
country = "CA";
|
|
112
|
+
locale = "en-CA";
|
|
113
|
+
}
|
|
114
|
+
const { raw, components } = rendered;
|
|
115
|
+
if (!Object.values(components).every((value) => raw.includes(value))) {
|
|
116
|
+
skipped++;
|
|
117
|
+
continue;
|
|
118
|
+
}
|
|
119
|
+
if (opts.golden) {
|
|
120
|
+
write(JSON.stringify({ raw, components: orderComponents(components), country }) + "\n");
|
|
121
|
+
emitted++;
|
|
122
|
+
continue;
|
|
123
|
+
}
|
|
124
|
+
const canonical = {
|
|
125
|
+
raw,
|
|
126
|
+
components: orderComponents(components),
|
|
127
|
+
country,
|
|
128
|
+
locale,
|
|
129
|
+
source,
|
|
130
|
+
source_id: sliceSourceID(source, components),
|
|
131
|
+
corpus_version: "0.4.0",
|
|
132
|
+
license: country === "CA"
|
|
133
|
+
? "GeoNames CA (CC-BY 4.0) locality skeletons + Canada Post box forms (corpus templates); postcodes synthesized to the codex CA pattern"
|
|
134
|
+
: country === "FR"
|
|
135
|
+
? "OpenAddresses FR (BAN-derived) skeletons + La Poste BP/CEDEX forms (corpus templates, NF Z 10-011)"
|
|
136
|
+
: country === "AU"
|
|
137
|
+
? "GeoNames AU postal dump (CC-BY 4.0) locality/state/postcode tails + Australia Post Postal Delivery Type designators (@mailwoman/codex/au)"
|
|
138
|
+
: country === "NZ"
|
|
139
|
+
? "GeoNames NZ postal dump (CC-BY 4.0) locality/postcode tails + NZ Post ADV358 Delivery Service Types (@mailwoman/codex/nz)"
|
|
140
|
+
: "OpenAddresses US (non-VT) skeletons + USPS Pub-28 §29 PO-box designators (codex/corpus templates)",
|
|
141
|
+
};
|
|
142
|
+
const aligned = alignRow(canonical);
|
|
143
|
+
if (aligned.kind !== "labeled" || !aligned.row) {
|
|
144
|
+
skipped++;
|
|
145
|
+
continue;
|
|
146
|
+
}
|
|
147
|
+
write(JSON.stringify({ ...aligned.row, synth_method: kind, synth_base_id: null }) + "\n");
|
|
148
|
+
emitted++;
|
|
149
|
+
}
|
|
150
|
+
return { emitted, skipped };
|
|
151
|
+
},
|
|
152
|
+
};
|
|
153
|
+
//# sourceMappingURL=recipe.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"recipe.js","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/recipe.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,UAAU,IAAI,cAAc,EAAE,MAAM,uBAAuB,CAAA;AAEpE,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAC1E,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAA;AACjE,OAAO,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACpE,OAAO,EACN,SAAS,EACT,WAAW,EACX,kBAAkB,EAClB,kBAAkB,EAClB,iBAAiB,EACjB,gBAAgB,EAChB,YAAY,EACZ,gBAAgB,EAChB,YAAY,EACZ,cAAc,EACd,gBAAgB,GAChB,MAAM,+BAA+B,CAAA;AAEtC,OAAO,EAAE,WAAW,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAA;AACrE,OAAO,EAAE,SAAS,EAAE,MAAM,kCAAkC,CAAA;AAC5D,OAAO,EAAE,aAAa,EAA6C,MAAM,mBAAmB,CAAA;AAC5F,OAAO,EAAE,0BAA0B,EAAE,MAAM,sBAAsB,CAAA;AACjE,OAAO,EAAE,IAAI,EAAE,MAAM,qBAAqB,CAAA;AAC1C,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAEjC,MAAM,eAAe,GAAG,CAAC,cAAc,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,CAAC,CAAA;AAEhH,MAAM,eAAe,GAAG,CAAC,UAAkC,EAA0B,EAAE,CACtF,MAAM,CAAC,WAAW,CAAC,eAAe,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE,UAAU,CAAC,GAAG,CAAE,CAAC,CAAC,CAAC,CAAA;AAE3G,MAAM,SAAS,GAAG,CAAC,IAAY,EAAU,EAAE;IAC1C,IAAI,WAAW,GAAG,CAAC,CAAA;IAEnB,KAAK,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;QACxC,WAAW,IAAI,MAAM,CAAA;QAErB,IAAI,IAAI,GAAG,WAAW;YAAE,OAAO,IAAI,CAAA;IACpC,CAAC;IAED,OAAO,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC,CAAC,CAAC,CAAA;AAC5B,CAAC,CAAA;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAiB;IAC7C,IAAI,EAAE,cAAc;IACpB,WAAW,EAAE,gGAAgG;IAC7G,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,UAAU,EAAE,WAAW,EAAE,oEAAoE,EAAE,CAAC;IAClH,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,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,MAAM,GAAc,EAAE,CAAA;QAE5B,KAAK,MAAM,WAAW,IAAI,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,gBAAgB,EAAE,CAAC;YAC7E,MAAM,MAAM,GAAG,MAAM,YAAY,CAAC,WAAW,CAAC,CAAA;YAE9C,OAAO,CAAC,KAAK,CAAC,KAAK,WAAW,CAAC,GAAG,KAAK,MAAM,CAAC,MAAM,SAAS,CAAC,CAAA;YAE9D,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;gBAC5B,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YACnB,CAAC;QACF,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,YAAY,CAAC,MAAM,CAAC,CAAA;QACxC,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,CAAA;QAEzF,OAAO,CAAC,KAAK,CAAC,KAAK,SAAS,CAAC,GAAG,KAAK,KAAK,CAAC,MAAM,YAAY,MAAM,CAAC,MAAM,uBAAuB,CAAC,CAAA;QAElG,MAAM,KAAK,GAAG,MAAM,gBAAgB,CAAC,IAAI,CAAC,EACzC,KAAK,GAAG,MAAM,gBAAgB,CAAC,IAAI,CAAC,CAAA;QAErC,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,iBAAiB,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,EACrF,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,iBAAiB,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,CAAA;QAEjF,OAAO,CAAC,KAAK,CAAC,qBAAqB,KAAK,CAAC,MAAM,IAAI,MAAM,CAAC,MAAM,QAAQ,KAAK,CAAC,MAAM,IAAI,MAAM,CAAC,MAAM,EAAE,CAAC,CAAA;QAExG,MAAM,KAAK,GAAG,CAAC,MAAM,gBAAgB,CAAC,kBAAkB,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAc,EAC3F,KAAK,GAAG,CAAC,MAAM,gBAAgB,CAAC,kBAAkB,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAc,CAAA;QAExF,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,EACxF,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,MAAM,CAAC,CAAA;QAEpF,OAAO,CAAC,KAAK,CAAC,yBAAyB,KAAK,CAAC,MAAM,IAAI,MAAM,CAAC,MAAM,QAAQ,KAAK,CAAC,MAAM,IAAI,MAAM,CAAC,MAAM,EAAE,CAAC,CAAA;QAE5G,IAAI,CAAC,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,MAAM;YAC3G,MAAM,IAAI,KAAK,CACd,gCAAgC,YAAY,CAAC,UAAU,CAAC,KAAK,WAAW,KAAK,kBAAkB,CAAC,GAAG,SAAS,kBAAkB,CAAC,GAAG,GAAG,CACrI,CAAA;QAEF,IAAI,OAAO,GAAG,CAAC,EACd,OAAO,GAAG,CAAC,EACX,KAAK,GAAG,CAAC,CAAA;QAEV,OAAO,OAAO,GAAG,KAAK,IAAI,KAAK,EAAE,GAAG,KAAK,GAAG,EAAE,EAAE,CAAC;YAChD,MAAM,IAAI,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,CAAA;YAChC,IAAI,QAAkB,CAAA;YACtB,IAAI,OAAe,CAAA;YACnB,IAAI,MAAc,CAAA;YAElB,IAAI,IAAI,KAAK,WAAW,EAAE,CAAC;gBAC1B,QAAQ,GAAG,aAAa,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACtD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,QAAQ,EAAE,CAAC;gBAC9B,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;gBAElC,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,YAAY;oBAAE,SAAQ;gBACrE,QAAQ,GAAG,WAAW,CAAC,MAAM,EAAE,KAAK,CAAC,CAAA;gBACrC,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,OAAO,EAAE,CAAC;gBAC7B,QAAQ,GAAG,UAAU,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACnD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,UAAU,EAAE,CAAC;gBAChC,QAAQ,GAAG,aAAa,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACtD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,UAAU,EAAE,CAAC;gBAChC,QAAQ,GAAG,UAAU,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACnD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,WAAW,EAAE,CAAC;gBACjC,QAAQ,GAAG,aAAa,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACtD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,WAAW,EAAE,CAAC;gBACjC,QAAQ,GAAG,aAAa,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACtD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,IAAI,IAAI,KAAK,oBAAoB,EAAE,CAAC;gBAC1C,MAAM,QAAQ,GAAG,0BAA0B,CAAC,EAAE,MAAM,EAAE,CAAC,CAAA;gBACvD,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,GAAG,UAAU,EAAE,GAAG,QAAQ,CAAC,UAAU,CAAA;gBAChE,QAAQ,GAAG,EAAE,GAAG,EAAE,iBAAiB,EAAE,GAAG,EAAE,QAAQ,CAAC,GAAG,EAAE,UAAU,EAAE,UAAoC,EAAE,CAAA;gBAC1G,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;iBAAM,CAAC;gBACP,QAAQ,GAAG,UAAU,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;gBACnD,OAAO,GAAG,IAAI,CAAA;gBACd,MAAM,GAAG,OAAO,CAAA;YACjB,CAAC;YAED,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,GAAG,QAAQ,CAAA;YAEpC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,UAAU,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;gBACtE,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBACjB,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,UAAU,EAAE,eAAe,CAAC,UAAU,CAAC,EAAE,OAAO,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;gBAEvF,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,SAAS,GAAsB;gBACpC,GAAG;gBACH,UAAU,EAAE,eAAe,CAAC,UAAU,CAAC;gBACvC,OAAO;gBACP,MAAM;gBACN,MAAM;gBACN,SAAS,EAAE,aAAa,CAAC,MAAM,EAAE,UAAU,CAAC;gBAC5C,cAAc,EAAE,OAAO;gBACvB,OAAO,EACN,OAAO,KAAK,IAAI;oBACf,CAAC,CAAC,sIAAsI;oBACxI,CAAC,CAAC,OAAO,KAAK,IAAI;wBACjB,CAAC,CAAC,oGAAoG;wBACtG,CAAC,CAAC,OAAO,KAAK,IAAI;4BACjB,CAAC,CAAC,2IAA2I;4BAC7I,CAAC,CAAC,OAAO,KAAK,IAAI;gCACjB,CAAC,CAAC,2HAA2H;gCAC7H,CAAC,CAAC,mGAAmG;aAC1G,CAAA;YAED,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,CAAC,CAAA;YAErE,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;YAEzF,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}
|
|
@@ -0,0 +1,86 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import type { PathBuilderLike } from "path-ts";
|
|
5
|
+
import type { AUTuple, FRTuple, NZTuple, USTuple } from "#recipes/po/box/cedex/types";
|
|
6
|
+
/**
|
|
7
|
+
* Non-Vermont OpenAddresses files used for training rows.
|
|
8
|
+
*/
|
|
9
|
+
export declare const US_TRAIN_SOURCES: ({
|
|
10
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__ca__berkeley.zip">;
|
|
11
|
+
csv: string;
|
|
12
|
+
region: string;
|
|
13
|
+
} | {
|
|
14
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__ca__marin.zip">;
|
|
15
|
+
csv: string;
|
|
16
|
+
region: string;
|
|
17
|
+
} | {
|
|
18
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__dc__statewide.zip">;
|
|
19
|
+
csv: string;
|
|
20
|
+
region: string;
|
|
21
|
+
} | {
|
|
22
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__ia__statewide.zip">;
|
|
23
|
+
csv: string;
|
|
24
|
+
region: string;
|
|
25
|
+
} | {
|
|
26
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__il__cook.zip">;
|
|
27
|
+
csv: string;
|
|
28
|
+
region: string;
|
|
29
|
+
} | {
|
|
30
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__mt__statewide.zip">;
|
|
31
|
+
csv: string;
|
|
32
|
+
region: string;
|
|
33
|
+
} | {
|
|
34
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__sd__statewide.zip">;
|
|
35
|
+
csv: string;
|
|
36
|
+
region: string;
|
|
37
|
+
})[];
|
|
38
|
+
/**
|
|
39
|
+
* Vermont OpenAddresses file reserved for golden rows.
|
|
40
|
+
*/
|
|
41
|
+
export declare const US_EVAL_SOURCE: {
|
|
42
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/us__vt__statewide.zip">;
|
|
43
|
+
csv: string;
|
|
44
|
+
region: string;
|
|
45
|
+
};
|
|
46
|
+
/**
|
|
47
|
+
* Countrywide French OpenAddresses source.
|
|
48
|
+
*/
|
|
49
|
+
export declare const FR_SOURCE: {
|
|
50
|
+
zip: import("path-ts").PathBuilder<"~/oa-cache/fr__countrywide.zip">;
|
|
51
|
+
csv: string;
|
|
52
|
+
};
|
|
53
|
+
/**
|
|
54
|
+
* GeoNames Canada feature dump.
|
|
55
|
+
*/
|
|
56
|
+
export declare const GEONAMES_CA: import("path-ts").PathBuilder<"~/geonames/CA.zip">;
|
|
57
|
+
/**
|
|
58
|
+
* GeoNames Australian postal-code dump.
|
|
59
|
+
*/
|
|
60
|
+
export declare const GEONAMES_POSTAL_AU: {
|
|
61
|
+
zip: import("path-ts").PathBuilder<"~/geonames-postal/AU.zip">;
|
|
62
|
+
txt: string;
|
|
63
|
+
};
|
|
64
|
+
/**
|
|
65
|
+
* GeoNames New Zealand postal-code dump.
|
|
66
|
+
*/
|
|
67
|
+
export declare const GEONAMES_POSTAL_NZ: {
|
|
68
|
+
zip: import("path-ts").PathBuilder<"~/geonames-postal/NZ.zip">;
|
|
69
|
+
txt: string;
|
|
70
|
+
};
|
|
71
|
+
export declare const localityHash: (name: string) => number;
|
|
72
|
+
export declare const isHoldoutLocality: (name: string) => boolean;
|
|
73
|
+
export declare function readUsTuples(source: {
|
|
74
|
+
zip: PathBuilderLike;
|
|
75
|
+
csv: string;
|
|
76
|
+
region: string;
|
|
77
|
+
}): Promise<USTuple[]>;
|
|
78
|
+
export declare function readFrTuples(limit: number): Promise<FRTuple[]>;
|
|
79
|
+
export declare function readCaLocalities(admin1: string): Promise<string[]>;
|
|
80
|
+
export declare function readPostalTuples(source: {
|
|
81
|
+
zip: PathBuilderLike;
|
|
82
|
+
txt: string;
|
|
83
|
+
}, opts: {
|
|
84
|
+
withState: boolean;
|
|
85
|
+
}): Promise<Array<AUTuple | NZTuple>>;
|
|
86
|
+
//# sourceMappingURL=sources.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"sources.d.ts","sourceRoot":"","sources":["../../../../../lib/recipes/po/box/cedex/sources.ts"],"names":[],"mappings":"AAAA;;GAEG;AAOH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAA;AAG9C,OAAO,KAAK,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,6BAA6B,CAAA;AAGrF;;GAEG;AACH,eAAO,MAAM,gBAAgB;;;;;;;;;;;;;;;;;;;;;;;;;;;;IAQ5B,CAAA;AAED;;GAEG;AACH,eAAO,MAAM,cAAc;IAC1B,GAAG;IACH,GAAG;IACH,MAAM;CACN,CAAA;AAED;;GAEG;AACH,eAAO,MAAM,SAAS;IAAK,GAAG;IAAmD,GAAG;CAAwB,CAAA;AAC5G;;GAEG;AACH,eAAO,MAAM,WAAW,oDAAqC,CAAA;AAC7D;;GAEG;AACH,eAAO,MAAM,kBAAkB;IAAK,GAAG;IAA6C,GAAG;CAAY,CAAA;AACnG;;GAEG;AACH,eAAO,MAAM,kBAAkB;IAAK,GAAG;IAA6C,GAAG;CAAY,CAAA;AAOnG,eAAO,MAAM,YAAY,SAAU,MAAM,KAAG,MAQ3C,CAAA;AAED,eAAO,MAAM,iBAAiB,SAAU,MAAM,KAAG,OAAwC,CAAA;AAKzF,wBAAsB,YAAY,CAAC,MAAM,EAAE;IAAE,GAAG,EAAE,eAAe,CAAC;IAAC,GAAG,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,GAAG,OAAO,CAAC,OAAO,EAAE,CAAC,CAwBpH;AAED,wBAAsB,YAAY,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,EAAE,CAAC,CAgCpE;AAED,wBAAsB,gBAAgB,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CAsBxE;AAED,wBAAsB,gBAAgB,CACrC,MAAM,EAAE;IAAE,GAAG,EAAE,eAAe,CAAC;IAAC,GAAG,EAAE,MAAM,CAAA;CAAE,EAC7C,IAAI,EAAE;IAAE,SAAS,EAAE,OAAO,CAAA;CAAE,GAC1B,OAAO,CAAC,KAAK,CAAC,OAAO,GAAG,OAAO,CAAC,CAAC,CA2BnC"}
|
|
@@ -0,0 +1,143 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* @copyright Sister Software
|
|
3
|
+
*/
|
|
4
|
+
import { isAuPostcode, isAuStateAbbreviation } from "@mailwoman/codex/au";
|
|
5
|
+
import { isNZPostcode } from "@mailwoman/codex/nz";
|
|
6
|
+
import { dataRootPath } from "@mailwoman/core/data-root";
|
|
7
|
+
import { pathExists } from "@mailwoman/core/fs/readers";
|
|
8
|
+
import { readZipEntry } from "@mailwoman/core/fs/zip";
|
|
9
|
+
import { TextSpliterator, TSVSpliterator } from "spliterator";
|
|
10
|
+
import { readCSVRecords, readZippedCSVRecords } from "#recipes/scaffold";
|
|
11
|
+
/**
|
|
12
|
+
* Non-Vermont OpenAddresses files used for training rows.
|
|
13
|
+
*/
|
|
14
|
+
export const US_TRAIN_SOURCES = [
|
|
15
|
+
{ zip: dataRootPath("oa-cache", "us__ca__berkeley.zip"), csv: "us/ca/berkeley.csv", region: "CA" },
|
|
16
|
+
{ zip: dataRootPath("oa-cache", "us__ca__marin.zip"), csv: "us/ca/marin.csv", region: "CA" },
|
|
17
|
+
{ zip: dataRootPath("oa-cache", "us__dc__statewide.zip"), csv: "us/dc/statewide.csv", region: "DC" },
|
|
18
|
+
{ zip: dataRootPath("oa-cache", "us__ia__statewide.zip"), csv: "us/ia/statewide.csv", region: "IA" },
|
|
19
|
+
{ zip: dataRootPath("oa-cache", "us__il__cook.zip"), csv: "us/il/cook.csv", region: "IL" },
|
|
20
|
+
{ zip: dataRootPath("oa-cache", "us__mt__statewide.zip"), csv: "us/mt/statewide.csv", region: "MT" },
|
|
21
|
+
{ zip: dataRootPath("oa-cache", "us__sd__statewide.zip"), csv: "us/sd/statewide.csv", region: "SD" },
|
|
22
|
+
];
|
|
23
|
+
/**
|
|
24
|
+
* Vermont OpenAddresses file reserved for golden rows.
|
|
25
|
+
*/
|
|
26
|
+
export const US_EVAL_SOURCE = {
|
|
27
|
+
zip: dataRootPath("oa-cache", "us__vt__statewide.zip"),
|
|
28
|
+
csv: "us/vt/statewide.csv",
|
|
29
|
+
region: "VT",
|
|
30
|
+
};
|
|
31
|
+
/**
|
|
32
|
+
* Countrywide French OpenAddresses source.
|
|
33
|
+
*/
|
|
34
|
+
export const FR_SOURCE = { zip: dataRootPath("oa-cache", "fr__countrywide.zip"), csv: "fr/countrywide.csv" };
|
|
35
|
+
/**
|
|
36
|
+
* GeoNames Canada feature dump.
|
|
37
|
+
*/
|
|
38
|
+
export const GEONAMES_CA = dataRootPath("geonames", "CA.zip");
|
|
39
|
+
/**
|
|
40
|
+
* GeoNames Australian postal-code dump.
|
|
41
|
+
*/
|
|
42
|
+
export const GEONAMES_POSTAL_AU = { zip: dataRootPath("geonames-postal", "AU.zip"), txt: "AU.txt" };
|
|
43
|
+
/**
|
|
44
|
+
* GeoNames New Zealand postal-code dump.
|
|
45
|
+
*/
|
|
46
|
+
export const GEONAMES_POSTAL_NZ = { zip: dataRootPath("geonames-postal", "NZ.zip"), txt: "NZ.txt" };
|
|
47
|
+
const MAX_LOCALITY_LENGTH = 40;
|
|
48
|
+
const FR_STRIDE = 211;
|
|
49
|
+
const FR_STRIDE_OFFSET = 3;
|
|
50
|
+
const CA_LOCALITY_MIN_POPULATION = 1000;
|
|
51
|
+
export const localityHash = (name) => {
|
|
52
|
+
let hash = 5381;
|
|
53
|
+
for (const character of name.toLowerCase()) {
|
|
54
|
+
hash = ((hash << 5) + hash + character.charCodeAt(0)) >>> 0;
|
|
55
|
+
}
|
|
56
|
+
return hash;
|
|
57
|
+
};
|
|
58
|
+
export const isHoldoutLocality = (name) => localityHash(name) % 10 === 0;
|
|
59
|
+
const cleanLocality = (locality) => locality && locality.length <= MAX_LOCALITY_LENGTH && !/\d|,/.test(locality) && !/cedex/i.test(locality);
|
|
60
|
+
export async function readUsTuples(source) {
|
|
61
|
+
const tuples = [];
|
|
62
|
+
const seen = new Set();
|
|
63
|
+
for await (const row of readZippedCSVRecords(source.zip, source.csv)) {
|
|
64
|
+
const locality = row.city ?? "";
|
|
65
|
+
if (!cleanLocality(locality))
|
|
66
|
+
continue;
|
|
67
|
+
const street = row.street ?? "";
|
|
68
|
+
const pairKey = `${locality}|${street}`.toLowerCase();
|
|
69
|
+
if (seen.has(pairKey))
|
|
70
|
+
continue;
|
|
71
|
+
seen.add(pairKey);
|
|
72
|
+
tuples.push({
|
|
73
|
+
house_number: row.number ?? "",
|
|
74
|
+
street,
|
|
75
|
+
locality,
|
|
76
|
+
region: source.region,
|
|
77
|
+
postcode: row.postcode ?? "",
|
|
78
|
+
});
|
|
79
|
+
}
|
|
80
|
+
return tuples;
|
|
81
|
+
}
|
|
82
|
+
export async function readFrTuples(limit) {
|
|
83
|
+
if (!(await pathExists(FR_SOURCE.zip))) {
|
|
84
|
+
console.error(` WARN: ${FR_SOURCE.zip} is not cached — skipping ${FR_SOURCE.csv}`);
|
|
85
|
+
return [];
|
|
86
|
+
}
|
|
87
|
+
const encoder = new TextEncoder();
|
|
88
|
+
const strided = TextSpliterator.fromAsync(readZipEntry(FR_SOURCE.zip, FR_SOURCE.csv), { skipEmpty: false })
|
|
89
|
+
.filter((_line, index) => index === 0 || (index + 1) % FR_STRIDE === FR_STRIDE_OFFSET)
|
|
90
|
+
.take(limit + 1)
|
|
91
|
+
.map((line) => encoder.encode(line + "\n"));
|
|
92
|
+
const tuples = [];
|
|
93
|
+
const seen = new Set();
|
|
94
|
+
for await (const row of readCSVRecords(strided)) {
|
|
95
|
+
const locality = row.city ?? "", postcode = row.postcode ?? "", street = row.street ?? "", house_number = row.number ?? "";
|
|
96
|
+
if (!cleanLocality(locality) || !/^\d{5}$/.test(postcode) || !street || !house_number)
|
|
97
|
+
continue;
|
|
98
|
+
const key = `${locality}|${street}`.toLowerCase();
|
|
99
|
+
if (seen.has(key))
|
|
100
|
+
continue;
|
|
101
|
+
seen.add(key);
|
|
102
|
+
tuples.push({ house_number, street, locality, postcode });
|
|
103
|
+
}
|
|
104
|
+
return tuples;
|
|
105
|
+
}
|
|
106
|
+
export async function readCaLocalities(admin1) {
|
|
107
|
+
if (!(await pathExists(GEONAMES_CA))) {
|
|
108
|
+
console.error(` WARN: ${GEONAMES_CA} is not cached — skipping CA localities (admin1=${admin1})`);
|
|
109
|
+
return [];
|
|
110
|
+
}
|
|
111
|
+
const localities = new Set();
|
|
112
|
+
for await (const columns of TSVSpliterator.fromAsync(readZipEntry(GEONAMES_CA, "CA.txt"), { header: false })) {
|
|
113
|
+
if (columns[6] === "P" &&
|
|
114
|
+
columns[8] === "CA" &&
|
|
115
|
+
columns[10] === admin1 &&
|
|
116
|
+
Number(columns[14]) > CA_LOCALITY_MIN_POPULATION &&
|
|
117
|
+
cleanLocality(columns[1] ?? "")) {
|
|
118
|
+
localities.add(columns[1]);
|
|
119
|
+
}
|
|
120
|
+
}
|
|
121
|
+
return [...localities];
|
|
122
|
+
}
|
|
123
|
+
export async function readPostalTuples(source, opts) {
|
|
124
|
+
if (!(await pathExists(source.zip))) {
|
|
125
|
+
console.error(` WARN: ${source.zip} is not cached — skipping ${source.txt}`);
|
|
126
|
+
return [];
|
|
127
|
+
}
|
|
128
|
+
const tuples = [];
|
|
129
|
+
const seen = new Set();
|
|
130
|
+
const validPostcode = opts.withState ? isAuPostcode : isNZPostcode;
|
|
131
|
+
for await (const columns of TSVSpliterator.fromAsync(readZipEntry(source.zip, source.txt), { header: false })) {
|
|
132
|
+
const postcode = (columns[1] ?? "").trim(), locality = (columns[2] ?? "").trim(), region = (columns[4] ?? "").trim();
|
|
133
|
+
if (!cleanLocality(locality) || !validPostcode(postcode) || (opts.withState && !isAuStateAbbreviation(region)))
|
|
134
|
+
continue;
|
|
135
|
+
const key = `${locality}|${postcode}`.toLowerCase();
|
|
136
|
+
if (seen.has(key))
|
|
137
|
+
continue;
|
|
138
|
+
seen.add(key);
|
|
139
|
+
tuples.push(opts.withState ? { locality, region, postcode } : { locality, postcode });
|
|
140
|
+
}
|
|
141
|
+
return tuples;
|
|
142
|
+
}
|
|
143
|
+
//# sourceMappingURL=sources.js.map
|