@mailwoman/corpus 8.6.0 → 9.1.0
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 +2 -2
- package/data/PROVENANCE.md +214 -0
- package/data/sub-venue-lexicon.json +18269 -0
- package/out/src/adapters/ban/adapter.d.ts +5 -4
- package/out/src/adapters/ban/adapter.d.ts.map +1 -1
- package/out/src/adapters/ban/adapter.js +62 -69
- package/out/src/adapters/ban/adapter.js.map +1 -1
- package/out/src/adapters/ban/street-decompose.d.ts.map +1 -1
- package/out/src/adapters/ban/street-decompose.js +18 -25
- package/out/src/adapters/ban/street-decompose.js.map +1 -1
- package/out/src/adapters/fcc-bdc/adapter.d.ts +1 -7
- package/out/src/adapters/fcc-bdc/adapter.d.ts.map +1 -1
- package/out/src/adapters/fcc-bdc/adapter.js +4 -26
- package/out/src/adapters/fcc-bdc/adapter.js.map +1 -1
- package/out/src/adapters/geonames/adapter.d.ts +1 -1
- package/out/src/adapters/geonames/adapter.d.ts.map +1 -1
- package/out/src/adapters/geonames/adapter.js +72 -78
- package/out/src/adapters/geonames/adapter.js.map +1 -1
- package/out/src/adapters/geonames-postal/adapter.d.ts +1 -1
- package/out/src/adapters/geonames-postal/adapter.d.ts.map +1 -1
- package/out/src/adapters/geonames-postal/adapter.js +46 -51
- package/out/src/adapters/geonames-postal/adapter.js.map +1 -1
- package/out/src/adapters/gnaf/adapter.d.ts +1 -1
- package/out/src/adapters/gnaf/adapter.d.ts.map +1 -1
- package/out/src/adapters/gnaf/adapter.js +6 -9
- package/out/src/adapters/gnaf/adapter.js.map +1 -1
- package/out/src/adapters/gnaf/assemble.d.ts.map +1 -1
- package/out/src/adapters/gnaf/assemble.js +7 -12
- package/out/src/adapters/gnaf/assemble.js.map +1 -1
- package/out/src/adapters/index.d.ts +1 -1
- package/out/src/adapters/index.d.ts.map +1 -1
- package/out/src/adapters/index.js +1 -1
- package/out/src/adapters/index.js.map +1 -1
- package/out/src/adapters/openaddresses/adapter.d.ts +1 -1
- package/out/src/adapters/openaddresses/adapter.d.ts.map +1 -1
- package/out/src/adapters/openaddresses/adapter.js +5 -10
- package/out/src/adapters/openaddresses/adapter.js.map +1 -1
- package/out/src/adapters/overture/adapter.d.ts +1 -1
- package/out/src/adapters/overture/adapter.d.ts.map +1 -1
- package/out/src/adapters/overture/adapter.js +5 -9
- package/out/src/adapters/overture/adapter.js.map +1 -1
- package/out/src/adapters/state-hi-schools/adapter.d.ts +1 -1
- package/out/src/adapters/state-hi-schools/adapter.d.ts.map +1 -1
- package/out/src/adapters/state-hi-schools/adapter.js +57 -75
- package/out/src/adapters/state-hi-schools/adapter.js.map +1 -1
- package/out/src/adapters/state-ia-contractors/adapter.d.ts +1 -1
- package/out/src/adapters/state-ia-contractors/adapter.d.ts.map +1 -1
- package/out/src/adapters/state-ia-contractors/adapter.js +60 -78
- package/out/src/adapters/state-ia-contractors/adapter.js.map +1 -1
- package/out/src/adapters/state-ny-notaries/adapter.d.ts +1 -1
- package/out/src/adapters/state-ny-notaries/adapter.d.ts.map +1 -1
- package/out/src/adapters/state-ny-notaries/adapter.js +69 -87
- package/out/src/adapters/state-ny-notaries/adapter.js.map +1 -1
- package/out/src/adapters/state-tx-notaries/adapter.d.ts +1 -1
- package/out/src/adapters/state-tx-notaries/adapter.d.ts.map +1 -1
- package/out/src/adapters/state-tx-notaries/adapter.js +71 -89
- package/out/src/adapters/state-tx-notaries/adapter.js.map +1 -1
- package/out/src/adapters/synth-po-box/adapter.d.ts +3 -3
- package/out/src/adapters/synth-po-box/adapter.d.ts.map +1 -1
- package/out/src/adapters/synth-po-box/adapter.js +10 -18
- package/out/src/adapters/synth-po-box/adapter.js.map +1 -1
- package/out/src/adapters/tiger/adapter.d.ts +1 -1
- package/out/src/adapters/tiger/adapter.d.ts.map +1 -1
- package/out/src/adapters/tiger/adapter.js +2 -2
- package/out/src/adapters/tiger/adapter.js.map +1 -1
- package/out/src/adapters/tiger/street-decompose.d.ts.map +1 -1
- package/out/src/adapters/tiger/street-decompose.js +18 -27
- package/out/src/adapters/tiger/street-decompose.js.map +1 -1
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts +4 -4
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.js +64 -82
- package/out/src/adapters/usgov-hrsa-fqhc/adapter.js.map +1 -1
- package/out/src/adapters/usgov-imls-pls/adapter.d.ts +1 -1
- package/out/src/adapters/usgov-imls-pls/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-imls-pls/adapter.js +62 -84
- package/out/src/adapters/usgov-imls-pls/adapter.js.map +1 -1
- package/out/src/adapters/usgov-irs-bmf/adapter.d.ts +1 -1
- package/out/src/adapters/usgov-irs-bmf/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-irs-bmf/adapter.js +59 -66
- package/out/src/adapters/usgov-irs-bmf/adapter.js.map +1 -1
- package/out/src/adapters/usgov-nad/adapter.d.ts +1 -1
- package/out/src/adapters/usgov-nad/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-nad/adapter.js +6 -9
- package/out/src/adapters/usgov-nad/adapter.js.map +1 -1
- package/out/src/adapters/usgov-nppes/adapter.d.ts +1 -1
- package/out/src/adapters/usgov-nppes/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-nppes/adapter.js +60 -78
- package/out/src/adapters/usgov-nppes/adapter.js.map +1 -1
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.d.ts +1 -1
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map +1 -1
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js +53 -73
- package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js.map +1 -1
- package/out/src/{adapter.d.ts → adapters/utils/index.d.ts} +43 -2
- package/out/src/adapters/utils/index.d.ts.map +1 -0
- package/out/src/{adapter.js → adapters/utils/index.js} +47 -2
- package/out/src/adapters/utils/index.js.map +1 -0
- package/out/src/adapters/wof-admin-jp/adapter.d.ts +1 -1
- package/out/src/adapters/wof-admin-jp/adapter.d.ts.map +1 -1
- package/out/src/adapters/wof-admin-jp/adapter.js +24 -6
- package/out/src/adapters/wof-admin-jp/adapter.js.map +1 -1
- package/out/src/adapters/wof-admin-json/adapter.d.ts +2 -2
- package/out/src/adapters/wof-admin-json/adapter.d.ts.map +1 -1
- package/out/src/adapters/wof-admin-json/adapter.js +2 -2
- package/out/src/adapters/wof-admin-json/adapter.js.map +1 -1
- package/out/src/adapters/wof-postalcode-json/adapter.d.ts +2 -2
- package/out/src/adapters/wof-postalcode-json/adapter.d.ts.map +1 -1
- package/out/src/adapters/wof-postalcode-json/adapter.js +2 -2
- package/out/src/adapters/wof-postalcode-json/adapter.js.map +1 -1
- package/out/src/build.d.ts +3 -4
- package/out/src/build.d.ts.map +1 -1
- package/out/src/build.js +6 -8
- package/out/src/build.js.map +1 -1
- package/out/src/index.d.ts +3 -17
- package/out/src/index.d.ts.map +1 -1
- package/out/src/index.js +3 -17
- package/out/src/index.js.map +1 -1
- package/out/src/name-prone-us-suffixes.d.ts +12 -0
- package/out/src/name-prone-us-suffixes.d.ts.map +1 -0
- package/out/src/name-prone-us-suffixes.js +12 -0
- package/out/src/name-prone-us-suffixes.js.map +1 -0
- package/out/src/parquet-wrapper/reader.d.ts +8 -0
- package/out/src/parquet-wrapper/reader.d.ts.map +1 -1
- package/out/src/parquet-wrapper/reader.js +16 -0
- package/out/src/parquet-wrapper/reader.js.map +1 -1
- package/out/src/parquet-wrapper/schema.d.ts +4 -0
- package/out/src/parquet-wrapper/schema.d.ts.map +1 -1
- package/out/src/parquet-wrapper/schema.js.map +1 -1
- package/out/src/runner.d.ts +2 -2
- package/out/src/runner.d.ts.map +1 -1
- package/out/src/runner.js +1 -1
- package/out/src/runner.js.map +1 -1
- package/out/src/shard-recipes/anchor-absorption.d.ts.map +1 -1
- package/out/src/shard-recipes/anchor-absorption.js +5 -4
- package/out/src/shard-recipes/anchor-absorption.js.map +1 -1
- package/out/src/shard-recipes/boundary-stress.d.ts.map +1 -1
- package/out/src/shard-recipes/boundary-stress.js +2 -2
- package/out/src/shard-recipes/boundary-stress.js.map +1 -1
- package/out/src/shard-recipes/country-balanced.d.ts.map +1 -1
- package/out/src/shard-recipes/country-balanced.js +75 -75
- package/out/src/shard-recipes/country-balanced.js.map +1 -1
- package/out/src/shard-recipes/fr-admin-split.js +2 -2
- package/out/src/shard-recipes/fr-admin-split.js.map +1 -1
- package/out/src/shard-recipes/fr-fragment.d.ts.map +1 -1
- package/out/src/shard-recipes/fr-fragment.js +8 -5
- package/out/src/shard-recipes/fr-fragment.js.map +1 -1
- package/out/src/shard-recipes/fr-lieudit.js +2 -2
- package/out/src/shard-recipes/fr-lieudit.js.map +1 -1
- package/out/src/shard-recipes/fr-order.d.ts.map +1 -1
- package/out/src/shard-recipes/fr-order.js +18 -72
- package/out/src/shard-recipes/fr-order.js.map +1 -1
- package/out/src/shard-recipes/german.d.ts.map +1 -1
- package/out/src/shard-recipes/german.js +17 -70
- package/out/src/shard-recipes/german.js.map +1 -1
- package/out/src/shard-recipes/house-venue.d.ts.map +1 -1
- package/out/src/shard-recipes/house-venue.js +1 -1
- package/out/src/shard-recipes/house-venue.js.map +1 -1
- package/out/src/shard-recipes/index.d.ts.map +1 -1
- package/out/src/shard-recipes/index.js +4 -1
- package/out/src/shard-recipes/index.js.map +1 -1
- package/out/src/shard-recipes/intersection.d.ts +2 -2
- package/out/src/shard-recipes/intersection.d.ts.map +1 -1
- package/out/src/shard-recipes/intersection.js +28 -72
- package/out/src/shard-recipes/intersection.js.map +1 -1
- package/out/src/shard-recipes/locale.d.ts +5 -5
- package/out/src/shard-recipes/locale.d.ts.map +1 -1
- package/out/src/shard-recipes/locale.js +25 -35
- package/out/src/shard-recipes/locale.js.map +1 -1
- package/out/src/shard-recipes/no-fragment.d.ts.map +1 -1
- package/out/src/shard-recipes/no-fragment.js +8 -5
- package/out/src/shard-recipes/no-fragment.js.map +1 -1
- package/out/src/shard-recipes/no-street-led.d.ts.map +1 -1
- package/out/src/shard-recipes/no-street-led.js +8 -5
- package/out/src/shard-recipes/no-street-led.js.map +1 -1
- package/out/src/shard-recipes/no-street.d.ts.map +1 -1
- package/out/src/shard-recipes/no-street.js +1 -1
- package/out/src/shard-recipes/no-street.js.map +1 -1
- package/out/src/shard-recipes/po-box-cedex.d.ts +3 -3
- package/out/src/shard-recipes/po-box-cedex.d.ts.map +1 -1
- package/out/src/shard-recipes/po-box-cedex.js +101 -127
- package/out/src/shard-recipes/po-box-cedex.js.map +1 -1
- package/out/src/shard-recipes/po-box.d.ts.map +1 -1
- package/out/src/shard-recipes/po-box.js +1 -1
- package/out/src/shard-recipes/po-box.js.map +1 -1
- package/out/src/shard-recipes/scaffold.d.ts +58 -9
- package/out/src/shard-recipes/scaffold.d.ts.map +1 -1
- package/out/src/shard-recipes/scaffold.js +87 -29
- package/out/src/shard-recipes/scaffold.js.map +1 -1
- package/out/src/shard-recipes/street-affix.d.ts +54 -0
- package/out/src/shard-recipes/street-affix.d.ts.map +1 -1
- package/out/src/shard-recipes/street-affix.js +253 -111
- package/out/src/shard-recipes/street-affix.js.map +1 -1
- package/out/src/shard-recipes/street-bare.d.ts.map +1 -1
- package/out/src/shard-recipes/street-bare.js +3 -3
- package/out/src/shard-recipes/street-bare.js.map +1 -1
- package/out/src/shard-recipes/street.d.ts.map +1 -1
- package/out/src/shard-recipes/street.js +2 -2
- package/out/src/shard-recipes/street.js.map +1 -1
- package/out/src/shard-recipes/sub-venue-sources.d.ts +228 -0
- package/out/src/shard-recipes/sub-venue-sources.d.ts.map +1 -0
- package/out/src/shard-recipes/sub-venue-sources.js +664 -0
- package/out/src/shard-recipes/sub-venue-sources.js.map +1 -0
- package/out/src/shard-recipes/sub-venue.d.ts +269 -0
- package/out/src/shard-recipes/sub-venue.d.ts.map +1 -0
- package/out/src/shard-recipes/sub-venue.js +709 -0
- package/out/src/shard-recipes/sub-venue.js.map +1 -0
- package/out/src/shard-recipes/unit.d.ts +1 -1
- package/out/src/shard-recipes/unit.d.ts.map +1 -1
- package/out/src/shard-recipes/unit.js +26 -74
- package/out/src/shard-recipes/unit.js.map +1 -1
- package/out/src/{synthesize-anchor-absorption.d.ts → synthesizers/anchor-absorption.d.ts} +1 -1
- package/out/src/synthesizers/anchor-absorption.d.ts.map +1 -0
- package/out/src/{synthesize-anchor-absorption.js → synthesizers/anchor-absorption.js} +1 -1
- package/out/src/synthesizers/anchor-absorption.js.map +1 -0
- package/out/src/{synthesize-boundary-stress.d.ts → synthesizers/boundary-stress.d.ts} +3 -3
- package/out/src/synthesizers/boundary-stress.d.ts.map +1 -0
- package/out/src/{synthesize-boundary-stress.js → synthesizers/boundary-stress.js} +2 -2
- package/out/src/synthesizers/boundary-stress.js.map +1 -0
- package/out/src/{synthesize-german.d.ts → synthesizers/german.d.ts} +2 -2
- package/out/src/synthesizers/german.d.ts.map +1 -0
- package/out/src/{synthesize-german.js → synthesizers/german.js} +2 -2
- package/out/src/synthesizers/german.js.map +1 -0
- package/out/src/{synthesize-house-venue.d.ts → synthesizers/house-venue.d.ts} +4 -4
- package/out/src/synthesizers/house-venue.d.ts.map +1 -0
- package/out/src/{synthesize-house-venue.js → synthesizers/house-venue.js} +4 -4
- package/out/src/synthesizers/house-venue.js.map +1 -0
- package/out/src/{synthesize-intersection.d.ts → synthesizers/intersection.d.ts} +2 -2
- package/out/src/synthesizers/intersection.d.ts.map +1 -0
- package/out/src/{synthesize-intersection.js → synthesizers/intersection.js} +1 -1
- package/out/src/synthesizers/intersection.js.map +1 -0
- package/out/src/{synthesize-no-street.d.ts → synthesizers/no-street.d.ts} +3 -3
- package/out/src/synthesizers/no-street.d.ts.map +1 -0
- package/out/src/{synthesize-no-street.js → synthesizers/no-street.js} +2 -2
- package/out/src/synthesizers/no-street.js.map +1 -0
- package/out/src/{synthesize-po-box.d.ts → synthesizers/po-box.d.ts} +2 -2
- package/out/src/synthesizers/po-box.d.ts.map +1 -0
- package/out/src/{synthesize-po-box.js → synthesizers/po-box.js} +1 -1
- package/out/src/synthesizers/po-box.js.map +1 -0
- package/out/src/{synthesize-street.d.ts → synthesizers/street.d.ts} +2 -2
- package/out/src/synthesizers/street.d.ts.map +1 -0
- package/out/src/{synthesize-street.js → synthesizers/street.js} +2 -2
- package/out/src/synthesizers/street.js.map +1 -0
- package/out/src/{synthesize.d.ts → synthesizers/utils.d.ts} +3 -3
- package/out/src/synthesizers/utils.d.ts.map +1 -0
- package/out/src/{synthesize.js → synthesizers/utils.js} +4 -15
- package/out/src/synthesizers/utils.js.map +1 -0
- package/out/src/tools/align-shard.d.ts.map +1 -1
- package/out/src/tools/align-shard.js +4 -8
- package/out/src/tools/align-shard.js.map +1 -1
- package/out/src/tools/audit.d.ts.map +1 -1
- package/out/src/tools/audit.js +4 -4
- package/out/src/tools/audit.js.map +1 -1
- package/out/src/tools/corpus-stats.d.ts +2 -2
- package/out/src/tools/corpus-stats.d.ts.map +1 -1
- package/out/src/tools/corpus-stats.js +18 -31
- package/out/src/tools/corpus-stats.js.map +1 -1
- package/out/src/tools/fetch/download.d.ts.map +1 -1
- package/out/src/tools/fetch/download.js +5 -6
- package/out/src/tools/fetch/download.js.map +1 -1
- package/out/src/tools/fetch/imls-pls.d.ts.map +1 -1
- package/out/src/tools/fetch/imls-pls.js +3 -15
- package/out/src/tools/fetch/imls-pls.js.map +1 -1
- package/out/src/tools/fetch/index.d.ts +14 -1
- package/out/src/tools/fetch/index.d.ts.map +1 -1
- package/out/src/tools/fetch/index.js +14 -1
- package/out/src/tools/fetch/index.js.map +1 -1
- package/out/src/tools/fetch/nad.d.ts +1 -1
- package/out/src/tools/fetch/nad.js +1 -1
- package/out/src/tools/fetch/nppes.d.ts.map +1 -1
- package/out/src/tools/fetch/nppes.js +8 -14
- package/out/src/tools/fetch/nppes.js.map +1 -1
- package/out/src/tools/fetch/openaddresses.d.ts +1 -1
- package/out/src/tools/fetch/openaddresses.js +1 -1
- package/out/src/tools/fetch/ourairports.d.ts +45 -0
- package/out/src/tools/fetch/ourairports.d.ts.map +1 -0
- package/out/src/tools/fetch/ourairports.js +123 -0
- package/out/src/tools/fetch/ourairports.js.map +1 -0
- package/out/src/tools/fetch/wikidata-subvenue.d.ts +171 -0
- package/out/src/tools/fetch/wikidata-subvenue.d.ts.map +1 -0
- package/out/src/tools/fetch/wikidata-subvenue.js +275 -0
- package/out/src/tools/fetch/wikidata-subvenue.js.map +1 -0
- package/out/src/tools/golden-expand.d.ts.map +1 -1
- package/out/src/tools/golden-expand.js +18 -16
- package/out/src/tools/golden-expand.js.map +1 -1
- package/out/src/tools/golden-promote.d.ts.map +1 -1
- package/out/src/tools/golden-promote.js +12 -6
- package/out/src/tools/golden-promote.js.map +1 -1
- package/out/src/tools/golden-relabel-street.d.ts +196 -0
- package/out/src/tools/golden-relabel-street.d.ts.map +1 -0
- package/out/src/tools/golden-relabel-street.js +440 -0
- package/out/src/tools/golden-relabel-street.js.map +1 -0
- package/out/src/tools/index.d.ts +4 -0
- package/out/src/tools/index.d.ts.map +1 -1
- package/out/src/tools/index.js +4 -0
- package/out/src/tools/index.js.map +1 -1
- package/out/src/tools/jsonl-to-parquet.d.ts.map +1 -1
- package/out/src/tools/jsonl-to-parquet.js +3 -2
- package/out/src/tools/jsonl-to-parquet.js.map +1 -1
- package/out/src/tools/lint-shard-vocab.d.ts.map +1 -1
- package/out/src/tools/lint-shard-vocab.js +3 -2
- package/out/src/tools/lint-shard-vocab.js.map +1 -1
- package/out/src/tools/lint-shard.d.ts +3 -3
- package/out/src/tools/lint-shard.d.ts.map +1 -1
- package/out/src/tools/lint-shard.js +23 -32
- package/out/src/tools/lint-shard.js.map +1 -1
- package/out/src/tools/overlay-manifest.d.ts.map +1 -1
- package/out/src/tools/overlay-manifest.js +2 -1
- package/out/src/tools/overlay-manifest.js.map +1 -1
- package/out/src/tools/overture-subvenue.d.ts +112 -0
- package/out/src/tools/overture-subvenue.d.ts.map +1 -0
- package/out/src/tools/overture-subvenue.js +144 -0
- package/out/src/tools/overture-subvenue.js.map +1 -0
- package/out/src/tools/shard-kryptonite.d.ts +1 -1
- package/out/src/tools/shard-kryptonite.d.ts.map +1 -1
- package/out/src/tools/shard-kryptonite.js +6 -6
- package/out/src/tools/shard-kryptonite.js.map +1 -1
- package/out/src/tools/shard-translit.d.ts +6 -3
- package/out/src/tools/shard-translit.d.ts.map +1 -1
- package/out/src/tools/shard-translit.js +9 -8
- package/out/src/tools/shard-translit.js.map +1 -1
- package/out/src/tools/sub-venue-lexicon.d.ts +507 -0
- package/out/src/tools/sub-venue-lexicon.d.ts.map +1 -0
- package/out/src/tools/sub-venue-lexicon.js +817 -0
- package/out/src/tools/sub-venue-lexicon.js.map +1 -0
- package/out/src/tools/sub-venue-promotions.d.ts +94 -0
- package/out/src/tools/sub-venue-promotions.d.ts.map +1 -0
- package/out/src/tools/sub-venue-promotions.js +266 -0
- package/out/src/tools/sub-venue-promotions.js.map +1 -0
- package/out/src/{align.d.ts → utils/align.d.ts} +1 -1
- package/out/src/utils/align.d.ts.map +1 -0
- package/out/src/utils/align.js.map +1 -0
- package/out/src/{golden.d.ts → utils/golden.d.ts} +5 -0
- package/out/src/utils/golden.d.ts.map +1 -0
- package/out/src/{golden.js → utils/golden.js} +24 -10
- package/out/src/utils/golden.js.map +1 -0
- package/out/src/utils/index.d.ts +14 -0
- package/out/src/utils/index.d.ts.map +1 -0
- package/out/src/utils/index.js +14 -0
- package/out/src/utils/index.js.map +1 -0
- package/out/src/utils/license.d.ts.map +1 -0
- package/out/src/utils/license.js.map +1 -0
- package/out/src/{parquet.d.ts → utils/parquet.d.ts} +2 -2
- package/out/src/utils/parquet.d.ts.map +1 -0
- package/out/src/{parquet.js → utils/parquet.js} +3 -15
- package/out/src/utils/parquet.js.map +1 -0
- package/out/src/{split.d.ts → utils/split.d.ts} +8 -1
- package/out/src/utils/split.d.ts.map +1 -0
- package/out/src/{split.js → utils/split.js} +7 -0
- package/out/src/utils/split.js.map +1 -0
- package/out/src/utils/tokenize.d.ts.map +1 -0
- package/out/src/utils/tokenize.js.map +1 -0
- package/out/src/{wof-json.d.ts → utils/wof-json.d.ts} +2 -2
- package/out/src/utils/wof-json.d.ts.map +1 -0
- package/out/src/{wof-json.js → utils/wof-json.js} +5 -8
- package/out/src/utils/wof-json.js.map +1 -0
- package/package.json +66 -20
- package/src/adapters/ban/adapter.ts +58 -68
- package/src/adapters/ban/street-decompose.ts +17 -25
- package/src/adapters/fcc-bdc/adapter.ts +5 -36
- package/src/adapters/geonames/adapter.ts +67 -76
- package/src/adapters/geonames-postal/adapter.ts +45 -54
- package/src/adapters/gnaf/adapter.ts +7 -11
- package/src/adapters/gnaf/assemble.ts +7 -11
- package/src/adapters/index.ts +3 -2
- package/src/adapters/openaddresses/adapter.ts +6 -12
- package/src/adapters/overture/adapter.ts +6 -10
- package/src/adapters/state-hi-schools/adapter.ts +53 -77
- package/src/adapters/state-ia-contractors/adapter.ts +55 -80
- package/src/adapters/state-ny-notaries/adapter.ts +64 -89
- package/src/adapters/state-tx-notaries/adapter.ts +63 -88
- package/src/adapters/synth-po-box/adapter.ts +12 -23
- package/src/adapters/tiger/adapter.ts +4 -3
- package/src/adapters/tiger/street-decompose.ts +21 -31
- package/src/adapters/usgov-hrsa-fqhc/adapter.ts +64 -88
- package/src/adapters/usgov-imls-pls/adapter.ts +57 -86
- package/src/adapters/usgov-irs-bmf/adapter.ts +55 -66
- package/src/adapters/usgov-nad/adapter.ts +7 -11
- package/src/adapters/usgov-nppes/adapter.ts +54 -79
- package/src/adapters/usgov-samhsa-treatment-locator/adapter.ts +48 -75
- package/src/{adapter.ts → adapters/utils/index.ts} +59 -3
- package/src/adapters/wof-admin-jp/adapter.ts +35 -9
- package/src/adapters/wof-admin-json/adapter.ts +3 -4
- package/src/adapters/wof-postalcode-json/adapter.ts +3 -4
- package/src/build.ts +12 -11
- package/src/index.ts +3 -17
- package/src/name-prone-us-suffixes.ts +12 -0
- package/src/parquet-wrapper/reader.ts +19 -0
- package/src/parquet-wrapper/schema.ts +4 -0
- package/src/runner.ts +7 -2
- package/src/shard-recipes/anchor-absorption.ts +5 -4
- package/src/shard-recipes/boundary-stress.ts +6 -2
- package/src/shard-recipes/country-balanced.ts +77 -91
- package/src/shard-recipes/fr-admin-split.ts +3 -3
- package/src/shard-recipes/fr-fragment.ts +10 -7
- package/src/shard-recipes/fr-lieudit.ts +3 -3
- package/src/shard-recipes/fr-order.ts +18 -83
- package/src/shard-recipes/german.ts +17 -80
- package/src/shard-recipes/house-venue.ts +2 -1
- package/src/shard-recipes/index.ts +4 -1
- package/src/shard-recipes/intersection.ts +37 -74
- package/src/shard-recipes/locale.ts +31 -39
- package/src/shard-recipes/no-fragment.ts +10 -7
- package/src/shard-recipes/no-street-led.ts +10 -7
- package/src/shard-recipes/no-street.ts +2 -1
- package/src/shard-recipes/po-box-cedex.ts +127 -161
- package/src/shard-recipes/po-box.ts +6 -1
- package/src/shard-recipes/scaffold.ts +125 -27
- package/src/shard-recipes/street-affix.ts +335 -134
- package/src/shard-recipes/street-bare.ts +4 -3
- package/src/shard-recipes/street.ts +3 -2
- package/src/shard-recipes/sub-venue-sources.ts +895 -0
- package/src/shard-recipes/sub-venue.ts +982 -0
- package/src/shard-recipes/unit.ts +27 -85
- package/src/{synthesize-boundary-stress.ts → synthesizers/boundary-stress.ts} +2 -2
- package/src/{synthesize-german.ts → synthesizers/german.ts} +2 -2
- package/src/{synthesize-house-venue.ts → synthesizers/house-venue.ts} +4 -4
- package/src/{synthesize-intersection.ts → synthesizers/intersection.ts} +1 -1
- package/src/{synthesize-no-street.ts → synthesizers/no-street.ts} +2 -2
- package/src/{synthesize-po-box.ts → synthesizers/po-box.ts} +1 -1
- package/src/{synthesize-street.ts → synthesizers/street.ts} +2 -2
- package/src/{synthesize.ts → synthesizers/utils.ts} +4 -18
- package/src/tools/align-shard.ts +4 -7
- package/src/tools/audit.ts +6 -5
- package/src/tools/corpus-stats.ts +25 -33
- package/src/tools/fetch/download.ts +7 -5
- package/src/tools/fetch/imls-pls.ts +4 -18
- package/src/tools/fetch/index.ts +14 -1
- package/src/tools/fetch/nad.ts +1 -1
- package/src/tools/fetch/nppes.ts +9 -16
- package/src/tools/fetch/openaddresses.ts +1 -1
- package/src/tools/fetch/ourairports.ts +166 -0
- package/src/tools/fetch/wikidata-subvenue.ts +386 -0
- package/src/tools/golden-expand.ts +18 -13
- package/src/tools/golden-promote.ts +15 -6
- package/src/tools/golden-relabel-street.ts +685 -0
- package/src/tools/index.ts +4 -0
- package/src/tools/jsonl-to-parquet.ts +3 -2
- package/src/tools/lint-shard-vocab.ts +3 -2
- package/src/tools/lint-shard.ts +33 -36
- package/src/tools/overlay-manifest.ts +2 -1
- package/src/tools/overture-subvenue.ts +215 -0
- package/src/tools/shard-kryptonite.ts +8 -9
- package/src/tools/shard-translit.ts +22 -11
- package/src/tools/sub-venue-lexicon.ts +1250 -0
- package/src/tools/sub-venue-promotions.ts +330 -0
- package/src/{align.ts → utils/align.ts} +1 -1
- package/src/{golden.ts → utils/golden.ts} +25 -11
- package/src/utils/index.ts +14 -0
- package/src/{parquet.ts → utils/parquet.ts} +5 -19
- package/src/{split.ts → utils/split.ts} +8 -2
- package/src/{wof-json.ts → utils/wof-json.ts} +5 -8
- package/out/src/adapter.d.ts.map +0 -1
- package/out/src/adapter.js.map +0 -1
- package/out/src/align.d.ts.map +0 -1
- package/out/src/align.js.map +0 -1
- package/out/src/format.d.ts +0 -14
- package/out/src/format.d.ts.map +0 -1
- package/out/src/format.js +0 -14
- package/out/src/format.js.map +0 -1
- package/out/src/golden.d.ts.map +0 -1
- package/out/src/golden.js.map +0 -1
- package/out/src/license.d.ts.map +0 -1
- package/out/src/license.js.map +0 -1
- package/out/src/parquet.d.ts.map +0 -1
- package/out/src/parquet.js.map +0 -1
- package/out/src/split.d.ts.map +0 -1
- package/out/src/split.js.map +0 -1
- package/out/src/synthesize-anchor-absorption.d.ts.map +0 -1
- package/out/src/synthesize-anchor-absorption.js.map +0 -1
- package/out/src/synthesize-boundary-stress.d.ts.map +0 -1
- package/out/src/synthesize-boundary-stress.js.map +0 -1
- package/out/src/synthesize-german.d.ts.map +0 -1
- package/out/src/synthesize-german.js.map +0 -1
- package/out/src/synthesize-house-venue.d.ts.map +0 -1
- package/out/src/synthesize-house-venue.js.map +0 -1
- package/out/src/synthesize-intersection.d.ts.map +0 -1
- package/out/src/synthesize-intersection.js.map +0 -1
- package/out/src/synthesize-no-street.d.ts.map +0 -1
- package/out/src/synthesize-no-street.js.map +0 -1
- package/out/src/synthesize-po-box.d.ts.map +0 -1
- package/out/src/synthesize-po-box.js.map +0 -1
- package/out/src/synthesize-street.d.ts.map +0 -1
- package/out/src/synthesize-street.js.map +0 -1
- package/out/src/synthesize.d.ts.map +0 -1
- package/out/src/synthesize.js.map +0 -1
- package/out/src/tokenize.d.ts.map +0 -1
- package/out/src/tokenize.js.map +0 -1
- package/out/src/wof-json.d.ts.map +0 -1
- package/out/src/wof-json.js.map +0 -1
- package/src/format.ts +0 -14
- /package/out/src/{align.js → utils/align.js} +0 -0
- /package/out/src/{license.d.ts → utils/license.d.ts} +0 -0
- /package/out/src/{license.js → utils/license.js} +0 -0
- /package/out/src/{tokenize.d.ts → utils/tokenize.d.ts} +0 -0
- /package/out/src/{tokenize.js → utils/tokenize.js} +0 -0
- /package/src/{synthesize-anchor-absorption.ts → synthesizers/anchor-absorption.ts} +0 -0
- /package/src/{license.ts → utils/license.ts} +0 -0
- /package/src/{tokenize.ts → utils/tokenize.ts} +0 -0
|
@@ -8,9 +8,9 @@
|
|
|
8
8
|
* Federally Qualified Health Centers (FQHCs) are HRSA-funded community health programs that
|
|
9
9
|
* self-report site addresses to the HRSA Data Warehouse. The published CSV (`data.hrsa.gov`)
|
|
10
10
|
* carries the site name, the postal-formatted street address, and the locality/region/postcode
|
|
11
|
-
* quad.
|
|
12
|
-
*
|
|
13
|
-
*
|
|
11
|
+
* quad. The source earns its place on adversarial-value-per-row (#22): every facility name is a
|
|
12
|
+
* human-typed venue string and the addresses pass through enough hands to accumulate the
|
|
13
|
+
* abbreviation drift + suite designator chaos that pure gazetteer data does not.
|
|
14
14
|
*
|
|
15
15
|
* The adapter consumes a CSV file the operator pre-downloads. The HRSA data is published as a
|
|
16
16
|
* single national CSV (~10K rows), small enough that the operator can re-fetch on every corpus
|
|
@@ -28,11 +28,10 @@
|
|
|
28
28
|
* License: stamped `"Public Domain"` per the HRSA Data Warehouse's federal government distribution
|
|
29
29
|
* terms.
|
|
30
30
|
*/
|
|
31
|
-
import {
|
|
32
|
-
import {
|
|
33
|
-
import {
|
|
34
|
-
import {
|
|
35
|
-
import { reconcileComponents } from "../../format.js";
|
|
31
|
+
import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils";
|
|
32
|
+
import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state";
|
|
33
|
+
import { reconcileComponents } from "@mailwoman/formatter";
|
|
34
|
+
import { CSVSpliterator } from "spliterator";
|
|
36
35
|
/**
|
|
37
36
|
* Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
|
|
38
37
|
* it came from.
|
|
@@ -47,21 +46,11 @@ export const USGOV_HRSA_FQHC_DEFAULT_LICENSE = "Public Domain";
|
|
|
47
46
|
* Split a "123 Main St Suite 4" surface form into `(house_number, street)`. The regex tolerates one trailing letter on
|
|
48
47
|
* the number (`"123A Main St"`) and a hyphenated form (`"40-12 Bell Blvd"`); anything else falls back to street-only.
|
|
49
48
|
*
|
|
50
|
-
* Suite / Apt / Unit designators stay on `street`
|
|
51
|
-
*
|
|
52
|
-
*
|
|
53
|
-
*
|
|
49
|
+
* Suite / Apt / Unit designators stay on `street` — Mailwoman's `unit` component exists but the address-formatter does
|
|
50
|
+
* not have a clean slot for it, and HRSA addresses do not separate the suite into its own column. Leaving the surface
|
|
51
|
+
* form intact in `street` preserves the adversarial training signal (the model learns that a trailing "Suite 4" is part
|
|
52
|
+
* of the road line in this distribution).
|
|
54
53
|
*/
|
|
55
|
-
const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/;
|
|
56
|
-
function splitAddress(address) {
|
|
57
|
-
const trimmed = address.trim();
|
|
58
|
-
if (!trimmed)
|
|
59
|
-
return null;
|
|
60
|
-
const m = HOUSE_NUMBER_PREFIX.exec(trimmed);
|
|
61
|
-
if (m)
|
|
62
|
-
return { house_number: m[1], street: m[2].trim() };
|
|
63
|
-
return { street: trimmed };
|
|
64
|
-
}
|
|
65
54
|
/**
|
|
66
55
|
* Compose the raw envelope-style address line. Format:
|
|
67
56
|
*
|
|
@@ -84,67 +73,60 @@ export function createUsgovHrsaFqhcAdapter() {
|
|
|
84
73
|
if (opts.country && opts.country !== "US") {
|
|
85
74
|
throw new Error(`usgov-hrsa-fqhc adapter: only US supported, got country=${opts.country}`);
|
|
86
75
|
}
|
|
87
|
-
const
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
relax_column_count: true,
|
|
93
|
-
}));
|
|
76
|
+
const rows = CSVSpliterator.fromAsync(opts.inputPath, {
|
|
77
|
+
mode: "object",
|
|
78
|
+
normalizeKeys: false,
|
|
79
|
+
enableQuoteHandling: true,
|
|
80
|
+
});
|
|
94
81
|
let emitted = 0;
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
|
|
118
|
-
|
|
119
|
-
|
|
120
|
-
|
|
121
|
-
|
|
122
|
-
|
|
123
|
-
|
|
124
|
-
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
131
|
-
|
|
132
|
-
|
|
133
|
-
|
|
134
|
-
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
|
|
138
|
-
|
|
139
|
-
|
|
140
|
-
|
|
141
|
-
|
|
142
|
-
|
|
143
|
-
emitted++;
|
|
144
|
-
}
|
|
145
|
-
}
|
|
146
|
-
finally {
|
|
147
|
-
stream.destroy();
|
|
82
|
+
for await (const record of rows) {
|
|
83
|
+
if (opts.signal?.aborted)
|
|
84
|
+
break;
|
|
85
|
+
if (opts.limit !== undefined && emitted >= opts.limit)
|
|
86
|
+
break;
|
|
87
|
+
const venue = (record["Site Name"] ?? "").trim();
|
|
88
|
+
const split = splitStreetLine(record["Site Address"] ?? "");
|
|
89
|
+
const city = (record["Site City"] ?? "").trim();
|
|
90
|
+
const stateAbbr = (record["Site State Abbreviation"] ?? "").trim();
|
|
91
|
+
const postcode = (record["Site Postal Code"] ?? "").trim();
|
|
92
|
+
if (!venue || !split || !city || !postcode)
|
|
93
|
+
continue;
|
|
94
|
+
const state = lookupStateAbbreviation(stateAbbr);
|
|
95
|
+
if (!state)
|
|
96
|
+
continue;
|
|
97
|
+
// Insertion order matters here. `venue` first so alignment claims its span
|
|
98
|
+
// (which may contain a token like "Buffalo") before `locality` runs its
|
|
99
|
+
// search — the kryptonite case `Buffalo Health Clinic, Buffalo NY`
|
|
100
|
+
// otherwise mis-labels the venue's "Buffalo" as locality.
|
|
101
|
+
const components = {
|
|
102
|
+
venue,
|
|
103
|
+
...(split.house_number ? { house_number: split.house_number } : {}),
|
|
104
|
+
street: split.street,
|
|
105
|
+
locality: city,
|
|
106
|
+
region: state.abbreviation,
|
|
107
|
+
postcode,
|
|
108
|
+
};
|
|
109
|
+
const raw = composeRaw(venue, split.house_number, split.street, city, state.abbreviation, postcode);
|
|
110
|
+
if (!raw)
|
|
111
|
+
continue;
|
|
112
|
+
const aligned = reconcileComponents(components, raw);
|
|
113
|
+
if (!Object.keys(aligned).length)
|
|
114
|
+
continue;
|
|
115
|
+
const siteID = (record["Site ID"] ?? "").trim();
|
|
116
|
+
const sourceID = siteID
|
|
117
|
+
? `${USGOV_HRSA_FQHC_ADAPTER_ID}-${siteID}`
|
|
118
|
+
: stableSourceID(USGOV_HRSA_FQHC_ADAPTER_ID, aligned);
|
|
119
|
+
yield {
|
|
120
|
+
raw,
|
|
121
|
+
components: aligned,
|
|
122
|
+
country: "US",
|
|
123
|
+
locale: "en-US",
|
|
124
|
+
source: USGOV_HRSA_FQHC_ADAPTER_ID,
|
|
125
|
+
source_id: sourceID,
|
|
126
|
+
corpus_version: "",
|
|
127
|
+
license: USGOV_HRSA_FQHC_DEFAULT_LICENSE,
|
|
128
|
+
};
|
|
129
|
+
emitted++;
|
|
148
130
|
}
|
|
149
131
|
},
|
|
150
132
|
};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-hrsa-fqhc/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAEH,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-hrsa-fqhc/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6BG;AAEH,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,kCAAkC,CAAA;AAClF,OAAO,EAAE,uBAAuB,EAAE,MAAM,uCAAuC,CAAA;AAE/E,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C;;;GAGG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iBAAiB,CAAA;AAC3D;;;GAGG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAG,eAAe,CAAA;AAmB9D;;;;;;;;GAQG;AAEH;;;;;;;GAOG;AACH,SAAS,UAAU,CAClB,KAAa,EACb,KAAyB,EACzB,MAAc,EACd,IAAY,EACZ,KAAa,EACb,QAAgB;IAEhB,MAAM,UAAU,GAAG,CAAC,KAAK,EAAE,MAAM,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;IACnE,MAAM,QAAQ,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAE7G,OAAO,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACvE,CAAC;AAED,MAAM,UAAU,0BAA0B;IACzC,OAAO;QACN,EAAE,EAAE,0BAA0B;QAC9B,cAAc,EAAE,+BAA+B;QAC/C,WAAW,EACV,yIAAyI;QAE1I,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,2DAA2D,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YAC3F,CAAC;YAED,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE;gBACrD,IAAI,EAAE,QAAQ;gBACd,aAAa,EAAE,KAAK;gBACpB,mBAAmB,EAAE,IAAI;aACzB,CAAC,CAAA;YAEF,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAAkC,EAAE,CAAC;gBAC/D,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAChD,MAAM,KAAK,GAAG,eAAe,CAAC,MAAM,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAA;gBAC3D,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC/C,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,yBAAyB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAClE,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,kBAAkB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE1D,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,QAAQ;oBAAE,SAAQ;gBACpD,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;gBAEhD,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,2EAA2E;gBAC3E,wEAAwE;gBACxE,mEAAmE;gBACnE,0DAA0D;gBAC1D,MAAM,UAAU,GAA+B;oBAC9C,KAAK;oBACL,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ;iBACR,CAAA;gBAED,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,EAAE,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;gBAEnG,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;oBAAE,SAAQ;gBAE1C,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE/C,MAAM,QAAQ,GAAG,MAAM;oBACtB,CAAC,CAAC,GAAG,0BAA0B,IAAI,MAAM,EAAE;oBAC3C,CAAC,CAAC,cAAc,CAAC,0BAA0B,EAAE,OAAO,CAAC,CAAA;gBAEtD,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,0BAA0B;oBAClC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,+BAA+B;iBACxC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,0BAA0B,EAAE,CAAA"}
|
|
@@ -17,7 +17,7 @@
|
|
|
17
17
|
*
|
|
18
18
|
* License: stamped `"Public Domain"` per IMLS federal government distribution terms.
|
|
19
19
|
*/
|
|
20
|
-
import type { CorpusAdapter } from "
|
|
20
|
+
import type { CorpusAdapter } from "@mailwoman/corpus/types";
|
|
21
21
|
/**
|
|
22
22
|
* Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
|
|
23
23
|
* it came from.
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;
|
|
1
|
+
{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAIH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,yBAAyB,CAAA;AAI1F;;;GAGG;AACH,eAAO,MAAM,yBAAyB,mBAAmB,CAAA;AACzD;;;GAGG;AACH,eAAO,MAAM,8BAA8B,kBAAkB,CAAA;AAY7D,wBAAgB,yBAAyB,IAAI,aAAa,CAqFzD;AAED;;GAEG;AACH,eAAO,MAAM,mBAAmB,eAA8B,CAAA"}
|
|
@@ -17,11 +17,10 @@
|
|
|
17
17
|
*
|
|
18
18
|
* License: stamped `"Public Domain"` per IMLS federal government distribution terms.
|
|
19
19
|
*/
|
|
20
|
-
import {
|
|
21
|
-
import {
|
|
22
|
-
import {
|
|
23
|
-
import {
|
|
24
|
-
import { reconcileComponents } from "../../format.js";
|
|
20
|
+
import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils";
|
|
21
|
+
import { lookupStateAbbreviation } from "@mailwoman/corpus/codex/us-fips-state";
|
|
22
|
+
import { reconcileComponents } from "@mailwoman/formatter";
|
|
23
|
+
import { CSVSpliterator } from "spliterator";
|
|
25
24
|
/**
|
|
26
25
|
* Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
|
|
27
26
|
* it came from.
|
|
@@ -32,16 +31,6 @@ export const USGOV_IMLS_PLS_ADAPTER_ID = "usgov-imls-pls";
|
|
|
32
31
|
* than having to look them up.
|
|
33
32
|
*/
|
|
34
33
|
export const USGOV_IMLS_PLS_DEFAULT_LICENSE = "Public Domain";
|
|
35
|
-
const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/;
|
|
36
|
-
function splitAddress(address) {
|
|
37
|
-
const trimmed = address.trim();
|
|
38
|
-
if (!trimmed)
|
|
39
|
-
return null;
|
|
40
|
-
const m = HOUSE_NUMBER_PREFIX.exec(trimmed);
|
|
41
|
-
if (m)
|
|
42
|
-
return { house_number: m[1], street: m[2].trim() };
|
|
43
|
-
return { street: trimmed };
|
|
44
|
-
}
|
|
45
34
|
export function createUsgovImlsPlsAdapter() {
|
|
46
35
|
return {
|
|
47
36
|
id: USGOV_IMLS_PLS_ADAPTER_ID,
|
|
@@ -51,76 +40,65 @@ export function createUsgovImlsPlsAdapter() {
|
|
|
51
40
|
if (opts.country && opts.country !== "US") {
|
|
52
41
|
throw new Error(`usgov-imls-pls adapter: only US supported, got country=${opts.country}`);
|
|
53
42
|
}
|
|
54
|
-
const
|
|
55
|
-
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
relax_column_count: true,
|
|
60
|
-
}));
|
|
43
|
+
const rows = CSVSpliterator.fromAsync(opts.inputPath, {
|
|
44
|
+
mode: "object",
|
|
45
|
+
normalizeKeys: false,
|
|
46
|
+
enableQuoteHandling: true,
|
|
47
|
+
});
|
|
61
48
|
let emitted = 0;
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
source_id: sourceID,
|
|
116
|
-
corpus_version: "",
|
|
117
|
-
license: USGOV_IMLS_PLS_DEFAULT_LICENSE,
|
|
118
|
-
};
|
|
119
|
-
emitted++;
|
|
120
|
-
}
|
|
121
|
-
}
|
|
122
|
-
finally {
|
|
123
|
-
stream.destroy();
|
|
49
|
+
for await (const record of rows) {
|
|
50
|
+
if (opts.signal?.aborted)
|
|
51
|
+
break;
|
|
52
|
+
if (opts.limit !== undefined && emitted >= opts.limit)
|
|
53
|
+
break;
|
|
54
|
+
const libName = (record.LIBNAME ?? "").trim();
|
|
55
|
+
const address = (record.ADDRESS ?? "").trim();
|
|
56
|
+
const city = (record.CITY ?? "").trim();
|
|
57
|
+
const zip = (record.ZIP ?? "").trim();
|
|
58
|
+
const stateAbbr = (record.STABR ?? "").trim();
|
|
59
|
+
const county = (record.CNTY ?? "").trim();
|
|
60
|
+
if (!libName || !city || !zip)
|
|
61
|
+
continue;
|
|
62
|
+
const state = lookupStateAbbreviation(stateAbbr);
|
|
63
|
+
if (!state)
|
|
64
|
+
continue;
|
|
65
|
+
const split = splitStreetLine(address);
|
|
66
|
+
if (!split)
|
|
67
|
+
continue;
|
|
68
|
+
const components = {
|
|
69
|
+
venue: libName,
|
|
70
|
+
...(split.house_number ? { house_number: split.house_number } : {}),
|
|
71
|
+
street: split.street,
|
|
72
|
+
locality: city,
|
|
73
|
+
region: state.abbreviation,
|
|
74
|
+
postcode: zip,
|
|
75
|
+
// #552: no subregion — US postal addresses don't surface the county, so emitting
|
|
76
|
+
// subregion creates a phantom component with no raw-span to align to, quarantining
|
|
77
|
+
// ~21% of rows. The county is still available in the source CSV; it just isn't
|
|
78
|
+
// a postal-surface component here.
|
|
79
|
+
};
|
|
80
|
+
const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim();
|
|
81
|
+
const raw = [libName, streetPart, [city, [stateAbbr, zip].filter(Boolean).join(" ")].filter(Boolean).join(", ")]
|
|
82
|
+
.filter(Boolean)
|
|
83
|
+
.join(", ");
|
|
84
|
+
const aligned = reconcileComponents(components, raw);
|
|
85
|
+
if (Object.keys(aligned).length <= 2)
|
|
86
|
+
continue;
|
|
87
|
+
const fscsKey = (record.FSCSKEY ?? "").trim();
|
|
88
|
+
const sourceID = fscsKey
|
|
89
|
+
? `${USGOV_IMLS_PLS_ADAPTER_ID}-${fscsKey}`
|
|
90
|
+
: stableSourceID(USGOV_IMLS_PLS_ADAPTER_ID, aligned);
|
|
91
|
+
yield {
|
|
92
|
+
raw,
|
|
93
|
+
components: aligned,
|
|
94
|
+
country: "US",
|
|
95
|
+
locale: "en-US",
|
|
96
|
+
source: USGOV_IMLS_PLS_ADAPTER_ID,
|
|
97
|
+
source_id: sourceID,
|
|
98
|
+
corpus_version: "",
|
|
99
|
+
license: USGOV_IMLS_PLS_DEFAULT_LICENSE,
|
|
100
|
+
};
|
|
101
|
+
emitted++;
|
|
124
102
|
}
|
|
125
103
|
},
|
|
126
104
|
};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAEH,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-imls-pls/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAEH,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,kCAAkC,CAAA;AAClF,OAAO,EAAE,uBAAuB,EAAE,MAAM,uCAAuC,CAAA;AAE/E,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C;;;GAGG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AACzD;;;GAGG;AACH,MAAM,CAAC,MAAM,8BAA8B,GAAG,eAAe,CAAA;AAY7D,MAAM,UAAU,yBAAyB;IACxC,OAAO;QACN,EAAE,EAAE,yBAAyB;QAC7B,cAAc,EAAE,8BAA8B;QAC9C,WAAW,EAAE,yFAAyF;QAEtG,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,0DAA0D,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YAC1F,CAAC;YAED,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE;gBACrD,IAAI,EAAE,QAAQ;gBACd,aAAa,EAAE,KAAK;gBACpB,mBAAmB,EAAE,IAAI;aACzB,CAAC,CAAA;YAEF,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAAoC,EAAE,CAAC;gBACjE,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACvC,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrC,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC7C,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAEzC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAEvC,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;gBAEhD,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,KAAK,GAAG,eAAe,CAAC,OAAO,CAAC,CAAA;gBAEtC,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,UAAU,GAA+B;oBAC9C,KAAK,EAAE,OAAO;oBACd,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI;oBACd,MAAM,EAAE,KAAK,CAAC,YAAY;oBAC1B,QAAQ,EAAE,GAAG;oBACb,iFAAiF;oBACjF,mFAAmF;oBACnF,+EAA+E;oBAC/E,mCAAmC;iBACnC,CAAA;gBAED,MAAM,UAAU,GAAG,CAAC,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;gBAEtF,MAAM,GAAG,GAAG,CAAC,OAAO,EAAE,UAAU,EAAE,CAAC,IAAI,EAAE,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;qBAC9G,MAAM,CAAC,OAAO,CAAC;qBACf,IAAI,CAAC,IAAI,CAAC,CAAA;gBAEZ,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,IAAI,CAAC;oBAAE,SAAQ;gBAE9C,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAE7C,MAAM,QAAQ,GAAG,OAAO;oBACvB,CAAC,CAAC,GAAG,yBAAyB,IAAI,OAAO,EAAE;oBAC3C,CAAC,CAAC,cAAc,CAAC,yBAAyB,EAAE,OAAO,CAAC,CAAA;gBAErD,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,yBAAyB;oBACjC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,8BAA8B;iBACvC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,EAAE,CAAA"}
|
|
@@ -18,7 +18,7 @@
|
|
|
18
18
|
* line. STATE is already a USPS abbreviation in the source. License: `"Public Domain"` (US
|
|
19
19
|
* federal).
|
|
20
20
|
*/
|
|
21
|
-
import type { CorpusAdapter } from "
|
|
21
|
+
import type { CorpusAdapter } from "@mailwoman/corpus/types";
|
|
22
22
|
/**
|
|
23
23
|
* Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
|
|
24
24
|
* it came from.
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;
|
|
1
|
+
{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAGH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,yBAAyB,CAAA;AAI1F;;;GAGG;AACH,eAAO,MAAM,wBAAwB,kBAAkB,CAAA;AACvD;;;GAGG;AACH,eAAO,MAAM,6BAA6B,kBAAkB,CAAA;AA2C5D,wBAAgB,wBAAwB,IAAI,aAAa,CA6ExD;AAED;;GAEG;AACH,eAAO,MAAM,kBAAkB,eAA6B,CAAA"}
|
|
@@ -18,10 +18,9 @@
|
|
|
18
18
|
* line. STATE is already a USPS abbreviation in the source. License: `"Public Domain"` (US
|
|
19
19
|
* federal).
|
|
20
20
|
*/
|
|
21
|
-
import {
|
|
22
|
-
import {
|
|
23
|
-
import {
|
|
24
|
-
import { reconcileComponents } from "../../format.js";
|
|
21
|
+
import { splitStreetLine, stableSourceID } from "@mailwoman/corpus/adapters/utils";
|
|
22
|
+
import { reconcileComponents } from "@mailwoman/formatter";
|
|
23
|
+
import { CSVSpliterator } from "spliterator";
|
|
25
24
|
/**
|
|
26
25
|
* Registry id for this adapter. Stamped into every row it emits, so a corpus record can be traced back to the dataset
|
|
27
26
|
* it came from.
|
|
@@ -32,24 +31,22 @@ export const USGOV_IRS_BMF_ADAPTER_ID = "usgov-irs-bmf";
|
|
|
32
31
|
* than having to look them up.
|
|
33
32
|
*/
|
|
34
33
|
export const USGOV_IRS_BMF_DEFAULT_LICENSE = "Public Domain";
|
|
35
|
-
const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/;
|
|
36
34
|
/**
|
|
37
35
|
* PO box in its many written forms: "PO BOX 12", "P.O. BOX 12", "P O BOX 12", "POB 12", "BOX 12".
|
|
38
36
|
*/
|
|
39
37
|
const PO_BOX = /^\s*(?:P\.?\s?O\.?\s*BOX|POB|BOX)\s+\w/i;
|
|
40
38
|
/**
|
|
41
39
|
* Classify the street line into a `po_box` or a `{house_number?, street}` split.
|
|
40
|
+
*
|
|
41
|
+
* BMF mixes street addresses and PO boxes in one `STREET` column, so the PO-box shapes have to be claimed BEFORE the
|
|
42
|
+
* shared house-number split runs — otherwise `splitStreetLine` would hand back `"PO Box 1234"` as a plain street, which
|
|
43
|
+
* is correct for every other US adapter and wrong here.
|
|
42
44
|
*/
|
|
43
|
-
function
|
|
45
|
+
function splitStreetLineOrPOBox(street) {
|
|
44
46
|
const trimmed = street.trim();
|
|
45
|
-
if (!trimmed)
|
|
46
|
-
return null;
|
|
47
47
|
if (PO_BOX.test(trimmed))
|
|
48
48
|
return { po_box: trimmed };
|
|
49
|
-
|
|
50
|
-
if (m)
|
|
51
|
-
return { house_number: m[1], street: m[2].trim() };
|
|
52
|
-
return { street: trimmed };
|
|
49
|
+
return splitStreetLine(trimmed);
|
|
53
50
|
}
|
|
54
51
|
function composeRaw(venue, streetPart, city, state, postcode) {
|
|
55
52
|
const cityPart = [city.trim(), [state, postcode].filter(Boolean).join(" ").trim()].filter(Boolean).join(", ");
|
|
@@ -64,61 +61,57 @@ export function createUsgovIrsBmfAdapter() {
|
|
|
64
61
|
if (opts.country && opts.country !== "US") {
|
|
65
62
|
throw new Error(`usgov-irs-bmf adapter: only US supported, got country=${opts.country}`);
|
|
66
63
|
}
|
|
67
|
-
const
|
|
68
|
-
|
|
64
|
+
const rows = CSVSpliterator.fromAsync(opts.inputPath, {
|
|
65
|
+
mode: "object",
|
|
66
|
+
normalizeKeys: false,
|
|
67
|
+
enableQuoteHandling: true,
|
|
68
|
+
});
|
|
69
69
|
let emitted = 0;
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
license: USGOV_IRS_BMF_DEFAULT_LICENSE,
|
|
116
|
-
};
|
|
117
|
-
emitted++;
|
|
118
|
-
}
|
|
119
|
-
}
|
|
120
|
-
finally {
|
|
121
|
-
stream.destroy();
|
|
70
|
+
for await (const record of rows) {
|
|
71
|
+
if (opts.signal?.aborted)
|
|
72
|
+
break;
|
|
73
|
+
if (opts.limit !== undefined && emitted >= opts.limit)
|
|
74
|
+
break;
|
|
75
|
+
const ein = (record.EIN ?? "").trim();
|
|
76
|
+
const venue = (record.NAME ?? "").trim() || undefined;
|
|
77
|
+
const street = (record.STREET ?? "").trim();
|
|
78
|
+
const city = (record.CITY ?? "").trim();
|
|
79
|
+
const state = (record.STATE ?? "").trim();
|
|
80
|
+
const zipRaw = (record.ZIP ?? "").trim();
|
|
81
|
+
if (!city || !zipRaw)
|
|
82
|
+
continue;
|
|
83
|
+
const postcode = zipRaw.split("-")[0].trim(); // 5-digit; drop the optional +4
|
|
84
|
+
const split = splitStreetLineOrPOBox(street);
|
|
85
|
+
if (!split)
|
|
86
|
+
continue;
|
|
87
|
+
const streetPart = "po_box" in split ? split.po_box : [split.house_number, split.street].filter(Boolean).join(" ");
|
|
88
|
+
const components = {
|
|
89
|
+
...(venue ? { venue } : {}),
|
|
90
|
+
...("po_box" in split
|
|
91
|
+
? { po_box: split.po_box }
|
|
92
|
+
: { ...(split.house_number ? { house_number: split.house_number } : {}), street: split.street }),
|
|
93
|
+
locality: city,
|
|
94
|
+
...(state ? { region: state } : {}),
|
|
95
|
+
postcode,
|
|
96
|
+
};
|
|
97
|
+
const raw = composeRaw(venue, streetPart, city, state, postcode);
|
|
98
|
+
if (!raw)
|
|
99
|
+
continue;
|
|
100
|
+
const aligned = reconcileComponents(components, raw);
|
|
101
|
+
if (Object.keys(aligned).length <= 2)
|
|
102
|
+
continue;
|
|
103
|
+
const sourceID = ein ? `${USGOV_IRS_BMF_ADAPTER_ID}-${ein}` : stableSourceID(USGOV_IRS_BMF_ADAPTER_ID, aligned);
|
|
104
|
+
yield {
|
|
105
|
+
raw,
|
|
106
|
+
components: aligned,
|
|
107
|
+
country: "US",
|
|
108
|
+
locale: "en-US",
|
|
109
|
+
source: USGOV_IRS_BMF_ADAPTER_ID,
|
|
110
|
+
source_id: sourceID,
|
|
111
|
+
corpus_version: "",
|
|
112
|
+
license: USGOV_IRS_BMF_DEFAULT_LICENSE,
|
|
113
|
+
};
|
|
114
|
+
emitted++;
|
|
122
115
|
}
|
|
123
116
|
},
|
|
124
117
|
};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAE,
|
|
1
|
+
{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../src/adapters/usgov-irs-bmf/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,kCAAkC,CAAA;AAElF,OAAO,EAAE,mBAAmB,EAAE,MAAM,sBAAsB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,eAAe,CAAA;AACvD;;;GAGG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAG,eAAe,CAAA;AAE5D;;GAEG;AACH,MAAM,MAAM,GAAG,yCAAyC,CAAA;AAWxD;;;;;;GAMG;AACH,SAAS,sBAAsB,CAAC,MAAc;IAC7C,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,EAAE,CAAA;IAE7B,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;IAEpD,OAAO,eAAe,CAAC,OAAO,CAAC,CAAA;AAChC,CAAC;AAED,SAAS,UAAU,CAClB,KAAyB,EACzB,UAAkB,EAClB,IAAY,EACZ,KAAa,EACb,QAAgB;IAEhB,MAAM,QAAQ,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAE7G,OAAO,CAAC,KAAK,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AAChE,CAAC;AAED,MAAM,UAAU,wBAAwB;IACvC,OAAO;QACN,EAAE,EAAE,wBAAwB;QAC5B,cAAc,EAAE,6BAA6B;QAC7C,WAAW,EACV,2HAA2H;QAE5H,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,IAAI,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC3C,MAAM,IAAI,KAAK,CAAC,yDAAyD,IAAI,CAAC,OAAO,EAAE,CAAC,CAAA;YACzF,CAAC;YAED,MAAM,IAAI,GAAG,cAAc,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,EAAE;gBACrD,IAAI,EAAE,QAAQ;gBACd,aAAa,EAAE,KAAK;gBACpB,mBAAmB,EAAE,IAAI;aACzB,CAAC,CAAA;YAEF,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,IAAgC,EAAE,CAAC;gBAC7D,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACrC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAA;gBACrD,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAC3C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACvC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBACzC,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;gBAExC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM;oBAAE,SAAQ;gBAC9B,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAA,CAAC,gCAAgC;gBAE9E,MAAM,KAAK,GAAG,sBAAsB,CAAC,MAAM,CAAC,CAAA;gBAE5C,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,UAAU,GACf,QAAQ,IAAI,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;gBAEhG,MAAM,UAAU,GAA+B;oBAC9C,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC3B,GAAG,CAAC,QAAQ,IAAI,KAAK;wBACpB,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE;wBAC1B,CAAC,CAAC,EAAE,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC;oBACjG,QAAQ,EAAE,IAAI;oBACd,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;oBACnC,QAAQ;iBACR,CAAA;gBAED,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAA;gBAEhE,IAAI,CAAC,GAAG;oBAAE,SAAQ;gBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;gBAEpD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,IAAI,CAAC;oBAAE,SAAQ;gBAE9C,MAAM,QAAQ,GAAG,GAAG,CAAC,CAAC,CAAC,GAAG,wBAAwB,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC,cAAc,CAAC,wBAAwB,EAAE,OAAO,CAAC,CAAA;gBAE/G,MAAM;oBACL,GAAG;oBACH,UAAU,EAAE,OAAO;oBACnB,OAAO,EAAE,IAAI;oBACb,MAAM,EAAE,OAAO;oBACf,MAAM,EAAE,wBAAwB;oBAChC,SAAS,EAAE,QAAQ;oBACnB,cAAc,EAAE,EAAE;oBAClB,OAAO,EAAE,6BAA6B;iBACtC,CAAA;gBAED,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,wBAAwB,EAAE,CAAA"}
|