@mailwoman/resolver-wof-sqlite 9.1.0 → 9.3.0
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 +36 -20
- package/lib/address/index.ts +9 -0
- package/{address-point-interpolation.ts → lib/address/point-interpolation.ts} +40 -28
- package/{address-point-schema.ts → lib/address/point-schema.ts} +34 -8
- package/lib/address/point.ts +280 -0
- package/{ancestry-backfill.ts → lib/ancestry/backfill.ts} +14 -19
- package/{ancestry.ts → lib/ancestry/index.ts} +13 -8
- package/{build-candidate.ts → lib/build-candidate.ts} +243 -199
- package/{build-slim.ts → lib/build-slim.ts} +153 -168
- package/lib/candidate/alias-bags.ts +56 -0
- package/lib/candidate/ancestors-sidecar.ts +204 -0
- package/lib/candidate/country-display-names.ts +79 -0
- package/lib/candidate/extract-fold.ts +138 -0
- package/lib/candidate/name-roles.ts +238 -0
- package/lib/candidate/own-name.ts +146 -0
- package/lib/candidate/place-attrs.ts +44 -0
- package/lib/candidate-ancestors-schema.ts +195 -0
- package/{candidate-fts.ts → lib/candidate-fts.ts} +7 -5
- package/{candidate-importance.ts → lib/candidate-importance.ts} +54 -57
- package/lib/candidate-lookup.ts +976 -0
- package/{candidate-schema.ts → lib/candidate-schema.ts} +34 -6
- package/lib/candidate-scoring.ts +261 -0
- package/lib/capital-schema.ts +89 -0
- package/lib/capitals.ts +148 -0
- package/{coincident-roles.ts → lib/coincident-roles.ts} +75 -19
- package/{convention.ts → lib/convention/index.ts} +5 -5
- package/lib/convention/schema.ts +72 -0
- package/{coverage-manifest-schema.ts → lib/coverage-manifest-schema.ts} +15 -16
- package/lib/currency-backfill.ts +334 -0
- package/lib/env.ts +50 -0
- package/lib/exact-match.ts +113 -0
- package/{sharding.ts → lib/extracts.ts} +63 -60
- package/lib/fst/autocomplete.ts +176 -0
- package/{fst-builder.ts → lib/fst/builder.ts} +30 -27
- package/{fst-deserialize-web.ts → lib/fst/deserialize-web.ts} +27 -87
- package/lib/fst/format.ts +106 -0
- package/{fst-freshness.ts → lib/fst/freshness.ts} +42 -75
- package/lib/fst/index.ts +14 -0
- package/{fst-matcher.ts → lib/fst/matcher.ts} +1 -1
- package/{fst-serialize.ts → lib/fst/serialize.ts} +37 -115
- package/{fst-types.ts → lib/fst/types.ts} +4 -2
- package/{fts.ts → lib/fts/index.ts} +28 -23
- package/{fts-query.ts → lib/fts/query.ts} +2 -2
- package/{geonames-aliases.ts → lib/geonames/aliases.ts} +13 -13
- package/lib/geonames/index.ts +8 -0
- package/{geonames-postal.ts → lib/geonames/postal.ts} +16 -15
- package/{index.ts → lib/index.ts} +52 -48
- package/{interpolation.ts → lib/interpolation.ts} +137 -41
- package/lib/lookup.ts +861 -0
- package/{name-score.ts → lib/name-score.ts} +6 -4
- package/lib/nsul/index.ts +8 -0
- package/lib/nsul/lookup.ts +144 -0
- package/lib/nsul/schema.ts +130 -0
- package/{place-importance-schema.ts → lib/place-importance-schema.ts} +70 -22
- package/lib/poi/index.ts +8 -0
- package/{poi-lookup.ts → lib/poi/lookup.ts} +32 -39
- package/{poi-schema.ts → lib/poi/schema.ts} +10 -6
- package/lib/polygon-schema.ts +47 -0
- package/{postal-city-alias-lookup.ts → lib/postal/city-alias-lookup.ts} +15 -21
- package/{postal-city-candidate-schema.ts → lib/postal/city-candidate-schema.ts} +4 -2
- package/lib/postal/index.ts +9 -0
- package/{postcode-point-lookup.ts → lib/postcode-point-lookup.ts} +17 -15
- package/lib/primary-preference.ts +226 -0
- package/lib/proximity-rerank.ts +120 -0
- package/{ranking-weights.ts → lib/ranking-weights.ts} +30 -4
- package/lib/region-keys.ts +144 -0
- package/{reverse.ts → lib/reverse.ts} +45 -56
- package/{schema.ts → lib/schema.ts} +11 -1
- package/lib/search-fetch.ts +257 -0
- package/{sqlite-convention-source.ts → lib/sqlite-convention-source.ts} +9 -10
- package/lib/sqlite-utils.ts +83 -0
- package/{street-centroid-schema.ts → lib/street/centroid-schema.ts} +12 -6
- package/{street-centroid.ts → lib/street/centroid.ts} +26 -28
- package/lib/street/index.ts +13 -0
- package/{street-morphology-fst-builder.ts → lib/street/morphology-fst-builder.ts} +24 -14
- package/{street-morphology-fst-loader.ts → lib/street/morphology-fst-loader.ts} +14 -12
- package/{street-name-lookup.ts → lib/street/name-lookup.ts} +14 -30
- package/lib/street/normalize.ts +572 -0
- package/{street-segment-schema.ts → lib/street/segment-schema.ts} +13 -8
- package/{types.ts → lib/types.ts} +49 -9
- package/{unified-schema.ts → lib/unified-schema.ts} +21 -25
- package/lib/uprn/existence.ts +84 -0
- package/lib/uprn/index.ts +9 -0
- package/lib/uprn/lookup.ts +206 -0
- package/lib/uprn/schema.ts +124 -0
- package/lib/weights-overlay-linker.ts +878 -0
- package/out/address/index.d.ts +9 -0
- package/out/address/index.d.ts.map +1 -0
- package/out/address/index.js +9 -0
- package/out/address/index.js.map +1 -0
- package/out/{address-point-interpolation.d.ts → address/point-interpolation.d.ts} +11 -10
- package/out/address/point-interpolation.d.ts.map +1 -0
- package/out/{address-point-interpolation.js → address/point-interpolation.js} +29 -23
- package/out/address/point-interpolation.js.map +1 -0
- package/out/{address-point-schema.d.ts → address/point-schema.d.ts} +30 -10
- package/out/address/point-schema.d.ts.map +1 -0
- package/out/{address-point-schema.js → address/point-schema.js} +7 -3
- package/out/address/point-schema.js.map +1 -0
- package/out/{address-point.d.ts → address/point.d.ts} +16 -8
- package/out/address/point.d.ts.map +1 -0
- package/out/address/point.js +191 -0
- package/out/address/point.js.map +1 -0
- package/out/{ancestry-backfill.d.ts → ancestry/backfill.d.ts} +7 -5
- package/out/ancestry/backfill.d.ts.map +1 -0
- package/out/{ancestry-backfill.js → ancestry/backfill.js} +12 -16
- package/out/ancestry/backfill.js.map +1 -0
- package/out/{ancestry.d.ts → ancestry/index.d.ts} +6 -5
- package/out/ancestry/index.d.ts.map +1 -0
- package/out/{ancestry.js → ancestry/index.js} +7 -7
- package/out/ancestry/index.js.map +1 -0
- package/out/build-candidate.d.ts +88 -7
- package/out/build-candidate.d.ts.map +1 -1
- package/out/build-candidate.js +145 -149
- package/out/build-candidate.js.map +1 -1
- package/out/build-slim.d.ts +5 -5
- package/out/build-slim.d.ts.map +1 -1
- package/out/build-slim.js +127 -142
- package/out/build-slim.js.map +1 -1
- package/out/candidate/alias-bags.d.ts +19 -0
- package/out/candidate/alias-bags.d.ts.map +1 -0
- package/out/candidate/alias-bags.js +39 -0
- package/out/candidate/alias-bags.js.map +1 -0
- package/out/candidate/ancestors-sidecar.d.ts +32 -0
- package/out/candidate/ancestors-sidecar.d.ts.map +1 -0
- package/out/candidate/ancestors-sidecar.js +140 -0
- package/out/candidate/ancestors-sidecar.js.map +1 -0
- package/out/candidate/country-display-names.d.ts +35 -0
- package/out/candidate/country-display-names.d.ts.map +1 -0
- package/out/candidate/country-display-names.js +59 -0
- package/out/candidate/country-display-names.js.map +1 -0
- package/out/candidate/extract-fold.d.ts +32 -0
- package/out/candidate/extract-fold.d.ts.map +1 -0
- package/out/candidate/extract-fold.js +104 -0
- package/out/candidate/extract-fold.js.map +1 -0
- package/out/candidate/name-roles.d.ts +57 -0
- package/out/candidate/name-roles.d.ts.map +1 -0
- package/out/candidate/name-roles.js +166 -0
- package/out/candidate/name-roles.js.map +1 -0
- package/out/candidate/own-name.d.ts +50 -0
- package/out/candidate/own-name.d.ts.map +1 -0
- package/out/candidate/own-name.js +132 -0
- package/out/candidate/own-name.js.map +1 -0
- package/out/candidate/place-attrs.d.ts +43 -0
- package/out/candidate/place-attrs.d.ts.map +1 -0
- package/out/candidate/place-attrs.js +15 -0
- package/out/candidate/place-attrs.js.map +1 -0
- package/out/candidate-ancestors-schema.d.ts +150 -0
- package/out/candidate-ancestors-schema.d.ts.map +1 -0
- package/out/candidate-ancestors-schema.js +123 -0
- package/out/candidate-ancestors-schema.js.map +1 -0
- package/out/candidate-fts.d.ts +7 -5
- package/out/candidate-fts.d.ts.map +1 -1
- package/out/candidate-fts.js +5 -3
- package/out/candidate-fts.js.map +1 -1
- package/out/candidate-importance.d.ts +16 -16
- package/out/candidate-importance.d.ts.map +1 -1
- package/out/candidate-importance.js +47 -52
- package/out/candidate-importance.js.map +1 -1
- package/out/candidate-lookup.d.ts +31 -52
- package/out/candidate-lookup.d.ts.map +1 -1
- package/out/candidate-lookup.js +380 -172
- package/out/candidate-lookup.js.map +1 -1
- package/out/candidate-schema.d.ts +31 -7
- package/out/candidate-schema.d.ts.map +1 -1
- package/out/candidate-schema.js +3 -0
- package/out/candidate-schema.js.map +1 -1
- package/out/candidate-scoring.d.ts +34 -0
- package/out/candidate-scoring.d.ts.map +1 -0
- package/out/candidate-scoring.js +196 -0
- package/out/candidate-scoring.js.map +1 -0
- package/out/capital-schema.d.ts +51 -0
- package/out/capital-schema.d.ts.map +1 -0
- package/out/capital-schema.js +63 -0
- package/out/capital-schema.js.map +1 -0
- package/out/capitals.d.ts +69 -0
- package/out/capitals.d.ts.map +1 -0
- package/out/capitals.js +98 -0
- package/out/capitals.js.map +1 -0
- package/out/coincident-roles.d.ts +12 -4
- package/out/coincident-roles.d.ts.map +1 -1
- package/out/coincident-roles.js +43 -11
- package/out/coincident-roles.js.map +1 -1
- package/out/{convention.d.ts → convention/index.d.ts} +5 -5
- package/out/convention/index.d.ts.map +1 -0
- package/out/{convention.js → convention/index.js} +4 -4
- package/out/convention/index.js.map +1 -0
- package/out/convention/schema.d.ts +51 -0
- package/out/convention/schema.d.ts.map +1 -0
- package/out/convention/schema.js +34 -0
- package/out/convention/schema.js.map +1 -0
- package/out/coverage-manifest-schema.d.ts +8 -8
- package/out/coverage-manifest-schema.d.ts.map +1 -1
- package/out/coverage-manifest-schema.js +6 -10
- package/out/coverage-manifest-schema.js.map +1 -1
- package/out/currency-backfill.d.ts +90 -0
- package/out/currency-backfill.d.ts.map +1 -0
- package/out/currency-backfill.js +196 -0
- package/out/currency-backfill.js.map +1 -0
- package/out/env.d.ts +43 -0
- package/out/env.d.ts.map +1 -0
- package/out/env.js +48 -0
- package/out/env.js.map +1 -0
- package/out/exact-match.d.ts +25 -0
- package/out/exact-match.d.ts.map +1 -0
- package/out/exact-match.js +88 -0
- package/out/exact-match.js.map +1 -0
- package/out/extracts.d.ts +114 -0
- package/out/extracts.d.ts.map +1 -0
- package/out/{sharding.js → extracts.js} +38 -38
- package/out/extracts.js.map +1 -0
- package/out/{fst-autocomplete.d.ts → fst/autocomplete.d.ts} +13 -13
- package/out/fst/autocomplete.d.ts.map +1 -0
- package/out/fst/autocomplete.js +113 -0
- package/out/fst/autocomplete.js.map +1 -0
- package/out/{fst-builder.d.ts → fst/builder.d.ts} +5 -5
- package/out/fst/builder.d.ts.map +1 -0
- package/out/{fst-builder.js → fst/builder.js} +24 -24
- package/out/fst/builder.js.map +1 -0
- package/out/{fst-deserialize-web.d.ts → fst/deserialize-web.d.ts} +3 -3
- package/out/fst/deserialize-web.d.ts.map +1 -0
- package/out/{fst-deserialize-web.js → fst/deserialize-web.js} +10 -74
- package/out/fst/deserialize-web.js.map +1 -0
- package/out/fst/format.d.ts +80 -0
- package/out/fst/format.d.ts.map +1 -0
- package/out/fst/format.js +91 -0
- package/out/fst/format.js.map +1 -0
- package/out/{fst-freshness.d.ts → fst/freshness.d.ts} +11 -14
- package/out/fst/freshness.d.ts.map +1 -0
- package/out/{fst-freshness.js → fst/freshness.js} +42 -74
- package/out/fst/freshness.js.map +1 -0
- package/out/fst/index.d.ts +14 -0
- package/out/fst/index.d.ts.map +1 -0
- package/out/fst/index.js +14 -0
- package/out/fst/index.js.map +1 -0
- package/out/{fst-matcher.d.ts → fst/matcher.d.ts} +2 -2
- package/out/fst/matcher.d.ts.map +1 -0
- package/out/{fst-matcher.js → fst/matcher.js} +1 -1
- package/out/fst/matcher.js.map +1 -0
- package/out/{fst-serialize.d.ts → fst/serialize.d.ts} +5 -10
- package/out/fst/serialize.d.ts.map +1 -0
- package/out/{fst-serialize.js → fst/serialize.js} +18 -98
- package/out/fst/serialize.js.map +1 -0
- package/out/{fst-types.d.ts → fst/types.d.ts} +4 -3
- package/out/fst/types.d.ts.map +1 -0
- package/out/{fst-types.js → fst/types.js} +1 -1
- package/out/fst/types.js.map +1 -0
- package/out/{fts.d.ts → fts/index.d.ts} +20 -13
- package/out/fts/index.d.ts.map +1 -0
- package/out/{fts.js → fts/index.js} +20 -15
- package/out/fts/index.js.map +1 -0
- package/out/{fts-query.d.ts → fts/query.d.ts} +2 -2
- package/out/fts/query.d.ts.map +1 -0
- package/out/{fts-query.js → fts/query.js} +2 -2
- package/out/fts/query.js.map +1 -0
- package/out/{geonames-aliases.d.ts → geonames/aliases.d.ts} +8 -6
- package/out/geonames/aliases.d.ts.map +1 -0
- package/out/{geonames-aliases.js → geonames/aliases.js} +7 -7
- package/out/geonames/aliases.js.map +1 -0
- package/out/geonames/index.d.ts +8 -0
- package/out/geonames/index.d.ts.map +1 -0
- package/out/geonames/index.js +8 -0
- package/out/geonames/index.js.map +1 -0
- package/out/{geonames-postal.d.ts → geonames/postal.d.ts} +12 -10
- package/out/geonames/postal.d.ts.map +1 -0
- package/out/{geonames-postal.js → geonames/postal.js} +11 -11
- package/out/geonames/postal.js.map +1 -0
- package/out/index.d.ts +32 -31
- package/out/index.d.ts.map +1 -1
- package/out/index.js +24 -24
- package/out/index.js.map +1 -1
- package/out/interpolation.d.ts +18 -9
- package/out/interpolation.d.ts.map +1 -1
- package/out/interpolation.js +108 -36
- package/out/interpolation.js.map +1 -1
- package/out/lookup.d.ts +25 -25
- package/out/lookup.d.ts.map +1 -1
- package/out/lookup.js +174 -558
- package/out/lookup.js.map +1 -1
- package/out/name-score.d.ts +0 -10
- package/out/name-score.d.ts.map +1 -1
- package/out/name-score.js +6 -4
- package/out/name-score.js.map +1 -1
- package/out/nsul/index.d.ts +8 -0
- package/out/nsul/index.d.ts.map +1 -0
- package/out/nsul/index.js +8 -0
- package/out/nsul/index.js.map +1 -0
- package/out/nsul/lookup.d.ts +83 -0
- package/out/nsul/lookup.d.ts.map +1 -0
- package/out/nsul/lookup.js +80 -0
- package/out/nsul/lookup.js.map +1 -0
- package/out/nsul/schema.d.ts +100 -0
- package/out/nsul/schema.d.ts.map +1 -0
- package/out/nsul/schema.js +78 -0
- package/out/nsul/schema.js.map +1 -0
- package/out/place-importance-schema.d.ts +47 -10
- package/out/place-importance-schema.d.ts.map +1 -1
- package/out/place-importance-schema.js +56 -10
- package/out/place-importance-schema.js.map +1 -1
- package/out/poi/index.d.ts +8 -0
- package/out/poi/index.d.ts.map +1 -0
- package/out/poi/index.js +8 -0
- package/out/poi/index.js.map +1 -0
- package/out/{poi-lookup.d.ts → poi/lookup.d.ts} +8 -10
- package/out/poi/lookup.d.ts.map +1 -0
- package/out/{poi-lookup.js → poi/lookup.js} +24 -29
- package/out/poi/lookup.js.map +1 -0
- package/out/{poi-schema.d.ts → poi/schema.d.ts} +10 -6
- package/out/poi/schema.d.ts.map +1 -0
- package/out/{poi-schema.js → poi/schema.js} +1 -1
- package/out/poi/schema.js.map +1 -0
- package/out/polygon-schema.d.ts +37 -0
- package/out/polygon-schema.d.ts.map +1 -0
- package/out/polygon-schema.js +23 -0
- package/out/polygon-schema.js.map +1 -0
- package/out/{postal-city-alias-lookup.d.ts → postal/city-alias-lookup.d.ts} +7 -6
- package/out/postal/city-alias-lookup.d.ts.map +1 -0
- package/out/{postal-city-alias-lookup.js → postal/city-alias-lookup.js} +12 -16
- package/out/postal/city-alias-lookup.js.map +1 -0
- package/out/{postal-city-alias-schema.d.ts → postal/city-alias-schema.d.ts} +1 -1
- package/out/{postal-city-alias-schema.d.ts.map → postal/city-alias-schema.d.ts.map} +1 -1
- package/out/{postal-city-alias-schema.js → postal/city-alias-schema.js} +1 -1
- package/out/{postal-city-alias-schema.js.map → postal/city-alias-schema.js.map} +1 -1
- package/out/{postal-city-candidate-schema.d.ts → postal/city-candidate-schema.d.ts} +4 -3
- package/out/postal/city-candidate-schema.d.ts.map +1 -0
- package/out/{postal-city-candidate-schema.js → postal/city-candidate-schema.js} +2 -2
- package/out/{postal-city-candidate-schema.js.map → postal/city-candidate-schema.js.map} +1 -1
- package/out/postal/index.d.ts +9 -0
- package/out/postal/index.d.ts.map +1 -0
- package/out/postal/index.js +9 -0
- package/out/postal/index.js.map +1 -0
- package/out/postcode-point-lookup.d.ts +10 -10
- package/out/postcode-point-lookup.d.ts.map +1 -1
- package/out/postcode-point-lookup.js +13 -13
- package/out/postcode-point-lookup.js.map +1 -1
- package/out/primary-preference.d.ts +133 -0
- package/out/primary-preference.d.ts.map +1 -0
- package/out/primary-preference.js +146 -0
- package/out/primary-preference.js.map +1 -0
- package/out/proximity-rerank.d.ts +77 -0
- package/out/proximity-rerank.d.ts.map +1 -0
- package/out/proximity-rerank.js +86 -0
- package/out/proximity-rerank.js.map +1 -0
- package/out/ranking-weights.d.ts +15 -3
- package/out/ranking-weights.d.ts.map +1 -1
- package/out/ranking-weights.js +21 -3
- package/out/ranking-weights.js.map +1 -1
- package/out/region-keys.d.ts +47 -0
- package/out/region-keys.d.ts.map +1 -0
- package/out/region-keys.js +121 -0
- package/out/region-keys.js.map +1 -0
- package/out/reverse.d.ts +6 -6
- package/out/reverse.d.ts.map +1 -1
- package/out/reverse.js +23 -35
- package/out/reverse.js.map +1 -1
- package/out/schema.d.ts +10 -1
- package/out/schema.d.ts.map +1 -1
- package/out/schema.js.map +1 -1
- package/out/search-fetch.d.ts +59 -0
- package/out/search-fetch.d.ts.map +1 -0
- package/out/search-fetch.js +184 -0
- package/out/search-fetch.js.map +1 -0
- package/out/sqlite-convention-source.d.ts +7 -7
- package/out/sqlite-convention-source.d.ts.map +1 -1
- package/out/sqlite-convention-source.js +5 -5
- package/out/sqlite-convention-source.js.map +1 -1
- package/out/sqlite-utils.d.ts +24 -6
- package/out/sqlite-utils.d.ts.map +1 -1
- package/out/sqlite-utils.js +24 -7
- package/out/sqlite-utils.js.map +1 -1
- package/out/{street-centroid-schema.d.ts → street/centroid-schema.d.ts} +12 -7
- package/out/street/centroid-schema.d.ts.map +1 -0
- package/out/{street-centroid-schema.js → street/centroid-schema.js} +5 -5
- package/out/{street-centroid-schema.js.map → street/centroid-schema.js.map} +1 -1
- package/out/{street-centroid.d.ts → street/centroid.d.ts} +6 -6
- package/out/street/centroid.d.ts.map +1 -0
- package/out/{street-centroid.js → street/centroid.js} +18 -24
- package/out/street/centroid.js.map +1 -0
- package/out/street/index.d.ts +13 -0
- package/out/street/index.d.ts.map +1 -0
- package/out/street/index.js +13 -0
- package/out/street/index.js.map +1 -0
- package/out/{street-morphology-fst-builder.d.ts → street/morphology-fst-builder.d.ts} +4 -4
- package/out/street/morphology-fst-builder.d.ts.map +1 -0
- package/out/{street-morphology-fst-builder.js → street/morphology-fst-builder.js} +18 -13
- package/out/street/morphology-fst-builder.js.map +1 -0
- package/out/{street-morphology-fst-loader.d.ts → street/morphology-fst-loader.d.ts} +5 -5
- package/out/street/morphology-fst-loader.d.ts.map +1 -0
- package/out/{street-morphology-fst-loader.js → street/morphology-fst-loader.js} +11 -10
- package/out/street/morphology-fst-loader.js.map +1 -0
- package/out/{street-name-lookup.d.ts → street/name-lookup.d.ts} +4 -7
- package/out/street/name-lookup.d.ts.map +1 -0
- package/out/{street-name-lookup.js → street/name-lookup.js} +9 -23
- package/out/street/name-lookup.js.map +1 -0
- package/out/street/normalize.d.ts +179 -0
- package/out/street/normalize.d.ts.map +1 -0
- package/out/street/normalize.js +457 -0
- package/out/street/normalize.js.map +1 -0
- package/out/{street-segment-schema.d.ts → street/segment-schema.d.ts} +13 -9
- package/out/street/segment-schema.d.ts.map +1 -0
- package/out/{street-segment-schema.js → street/segment-schema.js} +4 -4
- package/out/{street-segment-schema.js.map → street/segment-schema.js.map} +1 -1
- package/out/types.d.ts +49 -9
- package/out/types.d.ts.map +1 -1
- package/out/types.js.map +1 -1
- package/out/unified-schema.d.ts +6 -5
- package/out/unified-schema.d.ts.map +1 -1
- package/out/unified-schema.js +16 -19
- package/out/unified-schema.js.map +1 -1
- package/out/uprn/existence.d.ts +53 -0
- package/out/uprn/existence.d.ts.map +1 -0
- package/out/uprn/existence.js +61 -0
- package/out/uprn/existence.js.map +1 -0
- package/out/uprn/index.d.ts +9 -0
- package/out/uprn/index.d.ts.map +1 -0
- package/out/uprn/index.js +9 -0
- package/out/uprn/index.js.map +1 -0
- package/out/uprn/lookup.d.ts +85 -0
- package/out/uprn/lookup.d.ts.map +1 -0
- package/out/uprn/lookup.js +149 -0
- package/out/uprn/lookup.js.map +1 -0
- package/out/uprn/schema.d.ts +93 -0
- package/out/uprn/schema.d.ts.map +1 -0
- package/out/uprn/schema.js +78 -0
- package/out/uprn/schema.js.map +1 -0
- package/out/weights-overlay-linker.d.ts +299 -0
- package/out/weights-overlay-linker.d.ts.map +1 -0
- package/out/weights-overlay-linker.js +511 -0
- package/out/weights-overlay-linker.js.map +1 -0
- package/package.json +622 -154
- package/address-point.ts +0 -125
- package/candidate-lookup.ts +0 -720
- package/fst-autocomplete.ts +0 -204
- package/geo.ts +0 -121
- package/lookup.ts +0 -1354
- package/out/address-point-interpolation.d.ts.map +0 -1
- package/out/address-point-interpolation.js.map +0 -1
- package/out/address-point-schema.d.ts.map +0 -1
- package/out/address-point-schema.js.map +0 -1
- package/out/address-point.d.ts.map +0 -1
- package/out/address-point.js +0 -84
- package/out/address-point.js.map +0 -1
- package/out/ancestry-backfill.d.ts.map +0 -1
- package/out/ancestry-backfill.js.map +0 -1
- package/out/ancestry.d.ts.map +0 -1
- package/out/ancestry.js.map +0 -1
- package/out/convention.d.ts.map +0 -1
- package/out/convention.js.map +0 -1
- package/out/fst-autocomplete.d.ts.map +0 -1
- package/out/fst-autocomplete.js +0 -130
- package/out/fst-autocomplete.js.map +0 -1
- package/out/fst-builder.d.ts.map +0 -1
- package/out/fst-builder.js.map +0 -1
- package/out/fst-deserialize-web.d.ts.map +0 -1
- package/out/fst-deserialize-web.js.map +0 -1
- package/out/fst-freshness.d.ts.map +0 -1
- package/out/fst-freshness.js.map +0 -1
- package/out/fst-matcher.d.ts.map +0 -1
- package/out/fst-matcher.js.map +0 -1
- package/out/fst-serialize.d.ts.map +0 -1
- package/out/fst-serialize.js.map +0 -1
- package/out/fst-types.d.ts.map +0 -1
- package/out/fst-types.js.map +0 -1
- package/out/fts-query.d.ts.map +0 -1
- package/out/fts-query.js.map +0 -1
- package/out/fts.d.ts.map +0 -1
- package/out/fts.js.map +0 -1
- package/out/geo.d.ts +0 -74
- package/out/geo.d.ts.map +0 -1
- package/out/geo.js +0 -71
- package/out/geo.js.map +0 -1
- package/out/geonames-aliases.d.ts.map +0 -1
- package/out/geonames-aliases.js.map +0 -1
- package/out/geonames-postal.d.ts.map +0 -1
- package/out/geonames-postal.js.map +0 -1
- package/out/poi-lookup.d.ts.map +0 -1
- package/out/poi-lookup.js.map +0 -1
- package/out/poi-schema.d.ts.map +0 -1
- package/out/poi-schema.js.map +0 -1
- package/out/postal-city-alias-lookup.d.ts.map +0 -1
- package/out/postal-city-alias-lookup.js.map +0 -1
- package/out/postal-city-candidate-schema.d.ts.map +0 -1
- package/out/sharding.d.ts +0 -114
- package/out/sharding.d.ts.map +0 -1
- package/out/sharding.js.map +0 -1
- package/out/street-centroid-schema.d.ts.map +0 -1
- package/out/street-centroid.d.ts.map +0 -1
- package/out/street-centroid.js.map +0 -1
- package/out/street-morphology-fst-builder.d.ts.map +0 -1
- package/out/street-morphology-fst-builder.js.map +0 -1
- package/out/street-morphology-fst-loader.d.ts.map +0 -1
- package/out/street-morphology-fst-loader.js.map +0 -1
- package/out/street-name-lookup.d.ts.map +0 -1
- package/out/street-name-lookup.js.map +0 -1
- package/out/street-normalize.d.ts +0 -95
- package/out/street-normalize.d.ts.map +0 -1
- package/out/street-normalize.js +0 -272
- package/out/street-normalize.js.map +0 -1
- package/out/street-segment-schema.d.ts.map +0 -1
- package/sqlite-utils.ts +0 -46
- package/street-normalize.ts +0 -318
- /package/{postal-city-alias-schema.ts → lib/postal/city-alias-schema.ts} +0 -0
package/out/candidate-lookup.js
CHANGED
|
@@ -14,7 +14,7 @@
|
|
|
14
14
|
* {@link normalizeLocalityForKey} (build- and query-consistent), each row is denormalized (display
|
|
15
15
|
* `name`, centroid, bbox), and population rank is precomputed into `neg_rank` — so the result is
|
|
16
16
|
* POPULATION-FIRST and COUNTRY-AGNOSTIC (when no `country` filter is given), exactly like the
|
|
17
|
-
* demo. That's the deliberate divergence from {@link
|
|
17
|
+
* demo. That's the deliberate divergence from {@link WOFSQLitePlaceLookup}'s FTS/bm25 ranking: a
|
|
18
18
|
* bare "Moscow" resolves to the 10.4 M-pop Russian city, not whichever same-name US township bm25
|
|
19
19
|
* floats to the top.
|
|
20
20
|
*
|
|
@@ -22,91 +22,50 @@
|
|
|
22
22
|
* stored bbox and the locality query is point-in-bbox-filtered on the candidate centroid (the
|
|
23
23
|
* `bbox` field on {@link FindPlaceQuery}).
|
|
24
24
|
*/
|
|
25
|
-
import {
|
|
26
|
-
import { expandPlacetypeFilter } from "@mailwoman/resolver";
|
|
27
|
-
import {
|
|
28
|
-
import {
|
|
29
|
-
import {
|
|
30
|
-
import {
|
|
31
|
-
import {
|
|
32
|
-
import {
|
|
33
|
-
import {
|
|
34
|
-
import {
|
|
25
|
+
import { jaroWinkler, levenshteinSimilarity } from "@mailwoman/match/comparators";
|
|
26
|
+
import { expandPlacetypeFilter, partitionByContainment, } from "@mailwoman/resolver";
|
|
27
|
+
import { haversineKm } from "@mailwoman/spatial";
|
|
28
|
+
import { DatabaseClient } from "@mailwoman/sqlite/client";
|
|
29
|
+
import { CANDIDATE_ANCESTOR_TABLE, CANDIDATE_INTERVAL_TABLE, intervalContains, } from "#candidate-ancestors-schema";
|
|
30
|
+
import { CANDIDATE_FTS_TABLE } from "#candidate-fts";
|
|
31
|
+
import { readGazetteerCoverageManifest } from "#coverage-manifest-schema";
|
|
32
|
+
import { referentialFromPopulation } from "#place-importance-schema";
|
|
33
|
+
import { POSTAL_CITY_CANDIDATE_TABLE } from "#postal/city-candidate-schema";
|
|
34
|
+
import { rankByPrimaryPreference, RERANK_FETCH } from "#primary-preference";
|
|
35
|
+
import { applyProximityRerank } from "#proximity-rerank";
|
|
36
|
+
import { REGION_CLASS_PLACETYPES, regionQualifierProbeKeys } from "#region-keys";
|
|
37
|
+
import { allRows, hasColumn, hasTable } from "#sqlite-utils";
|
|
38
|
+
import { normalizeLocalityForKey, stripLocalityQualifier } from "#street/normalize";
|
|
39
|
+
export { rankByPrimaryPreference } from "#primary-preference";
|
|
35
40
|
/**
|
|
36
|
-
* FTS5-trigram over-fetch before the
|
|
37
|
-
*
|
|
41
|
+
* FTS5-trigram over-fetch before the WORD-LEVEL re-rank. The trigram index stays the candidate GENERATOR (it is what
|
|
42
|
+
* the artifact carries); the scoring moved off trigram-Jaccard on 2026-08-12 (#1614), which the aucklnad receipts
|
|
43
|
+
* falsified as a typo measure: it scored the true transposition correction 'auckland' at 0.333 — below its own 0.34 bar
|
|
44
|
+
* — while 'auckley' scored 0.375 and the 'gore bay' scrape 0.455, because shared generic suffixes count as trigram
|
|
45
|
+
* evidence and transpositions count against it.
|
|
38
46
|
*/
|
|
39
47
|
const FUZZY_FETCH = 40;
|
|
40
|
-
const FUZZY_MIN = 0.34;
|
|
41
48
|
/**
|
|
42
|
-
*
|
|
43
|
-
*
|
|
44
|
-
*
|
|
45
|
-
*
|
|
49
|
+
* Minimum WORD-LEVEL similarity (max of Jaro-Winkler and normalized edit similarity — the `match/comparators`
|
|
50
|
+
* primitives, deliberately WITHOUT `nameSimilarity`'s token-subset floor, which is a person-name rule that would hand
|
|
51
|
+
* 'stanmore bay' to a place named 'Bay') for a fuzzy correction to count. Measured on the #1614 receipts:
|
|
52
|
+
* 'aucklnad'→'auckland' 0.975 (in), →'auckley' 0.868 (in, but outranked), 'stanmore bay'→'gore bay' ~0.70 (out),
|
|
53
|
+
* 'sacremento'→'sacramento' ~0.97 (in).
|
|
46
54
|
*/
|
|
47
|
-
const
|
|
55
|
+
const WORD_FUZZY_MIN = 0.85;
|
|
48
56
|
/**
|
|
49
|
-
*
|
|
50
|
-
*
|
|
51
|
-
* The raw candidate order is population-first (`neg_rank ASC`) and treats an ALIAS row (a place's alt-name / exonym,
|
|
52
|
-
* `is_primary=0`) and a PRIMARY-name row (`is_primary=1`) on equal footing. So a foreign place whose transliterated
|
|
53
|
-
* exonym coincidentally normalizes to a query — Changchun CN stores the Turkish exonym "Çançun" (`name_key="cancun"`),
|
|
54
|
-
* 4.19 M pop — outranks the PRIMARY-name place the query actually means (Cancún MX, 0.89 M pop). This penalty makes a
|
|
55
|
-
* same-key alias have to clear a population MARGIN over a foreign primary before it wins.
|
|
56
|
-
*
|
|
57
|
-
* It is deliberately NOT a dominant sort key (no `ORDER BY is_primary DESC`, which would make every primary outrank
|
|
58
|
-
* every alias and break the alt-names users depend on — "NYC"→New York, "LA"→Los Angeles, "Frisco"→San Francisco). Two
|
|
59
|
-
* bounds keep it a soft prior:
|
|
60
|
-
*
|
|
61
|
-
* 1. **Cross-country only.** The penalty applies to an alias ONLY when the top-population primary sharing the key is in a
|
|
62
|
-
* DIFFERENT country. A SAME-country nickname contest (San Francisco's alias "Frisco" vs the primary Frisco, TX —
|
|
63
|
-
* both US) is left on pure population, so the legitimate alias still wins.
|
|
64
|
-
* 2. **Population-bounded.** The penalty is {@link PRIMARY_PREFERENCE_LOG10} in log10-population units — an alias must be
|
|
65
|
-
* at least 10x more populous than the foreign primary to still win. So a genuinely dominant alias keeps winning
|
|
66
|
-
* ("Los Angeles" over La, Ghana — gap 1.6; "Las Vegas" over Vegas, Cuba — gap 2.4) while a near-tie coincidental
|
|
67
|
-
* collision defers to the primary (Cancún over Changchun — gap 0.7).
|
|
68
|
-
*/
|
|
69
|
-
const PRIMARY_PREFERENCE_LOG10 = 1;
|
|
70
|
-
/**
|
|
71
|
-
* Over-fetch cap for {@link rankByPrimaryPreference}: the candidate rows for one `name_key` (all same-name places
|
|
72
|
-
* worldwide) are re-ranked in-process, so the probe fetches this many (population-ordered) before the re-rank rather
|
|
73
|
-
* than the caller's small `limit`, ensuring the intended primary isn't cut below the fold by a cluster of more-populous
|
|
74
|
-
* foreign aliases. Bounded and small — a single contiguous B-tree scan.
|
|
57
|
+
* The word-level correction similarity — see {@link WORD_FUZZY_MIN}.
|
|
75
58
|
*/
|
|
76
|
-
|
|
59
|
+
function wordFuzzySimilarity(a, b) {
|
|
60
|
+
return Math.max(jaroWinkler(a, b), levenshteinSimilarity(a, b));
|
|
61
|
+
}
|
|
77
62
|
/**
|
|
78
|
-
*
|
|
79
|
-
* `
|
|
80
|
-
*
|
|
81
|
-
*
|
|
82
|
-
* the primary. Returns the top `limit` after the re-rank, each annotated.
|
|
63
|
+
* Postcode-containment re-rank check radius (km) — the SAME value the resolver's country pass measures at
|
|
64
|
+
* (`POSTCODE_COUNTRY_COHERENCE_THRESHOLD_KM`, resolver/postcode-country-coherence.ts): a locality within this distance
|
|
65
|
+
* of the postcode's own centroid counts as "containing" it. One number, two passes — a divergence here would make the
|
|
66
|
+
* two mechanisms disagree about what is proximal.
|
|
83
67
|
*/
|
|
84
|
-
|
|
85
|
-
// The primary the alias actually competes with for the top slot: highest population (min neg_rank). Undefined
|
|
86
|
-
// when the set has no primary → nothing to prefer, penalty is 0, order stays population-first (today's behavior).
|
|
87
|
-
let topPrimary;
|
|
88
|
-
for (const r of rows) {
|
|
89
|
-
if (r.is_primary === 1 && (topPrimary === undefined || r.neg_rank < topPrimary.neg_rank)) {
|
|
90
|
-
topPrimary = r;
|
|
91
|
-
}
|
|
92
|
-
}
|
|
93
|
-
const topCountry = topPrimary?.country_id;
|
|
94
|
-
// A cross-country alias (different country than the top primary) is penalized; it is DEMOTED when even after — i.e.
|
|
95
|
-
// the penalty leaves its effective rank behind the primary's raw rank (it lost the bounded population contest).
|
|
96
|
-
const isCrossCountryAlias = (r) => topCountry !== undefined && r.is_primary !== 1 && r.country_id !== topCountry;
|
|
97
|
-
const annotate = (r) => {
|
|
98
|
-
const penalized = isCrossCountryAlias(r);
|
|
99
|
-
const effectiveNegRank = r.neg_rank + (penalized ? delta : 0);
|
|
100
|
-
return { ...r, effectiveNegRank, demoted: penalized && effectiveNegRank > topPrimary.neg_rank };
|
|
101
|
-
};
|
|
102
|
-
return (rows
|
|
103
|
-
.map((r, i) => ({ row: annotate(r), i }))
|
|
104
|
-
// Effective rank ASC; ties keep population order, then original index (stable).
|
|
105
|
-
// oxlint-disable-next-line unicorn/no-array-sort -- sorts a freshly-built array; toSorted would double-allocate on a hot path
|
|
106
|
-
.sort((a, b) => a.row.effectiveNegRank - b.row.effectiveNegRank || a.row.neg_rank - b.row.neg_rank || a.i - b.i)
|
|
107
|
-
.slice(0, limit)
|
|
108
|
-
.map((x) => x.row));
|
|
109
|
-
}
|
|
68
|
+
const POSTCODE_CONTAINMENT_THRESHOLD_KM = 25;
|
|
110
69
|
/**
|
|
111
70
|
* Unpadded character-trigrams of `s`, OR'd into an FTS5 trigram MATCH query (each quoted so FTS treats it as a literal
|
|
112
71
|
* term). Returns "" when `s` is shorter than a trigram or yields no clean grams — the caller then skips the fuzzy
|
|
@@ -123,12 +82,16 @@ function ftsTrigramQuery(s) {
|
|
|
123
82
|
return [...grams].map((g) => `"${g}"`).join(" OR ");
|
|
124
83
|
}
|
|
125
84
|
/**
|
|
126
|
-
* Node {@link PlaceLookup} over `candidate.db`. Drop-in for {@link
|
|
85
|
+
* Node {@link PlaceLookup} over `candidate.db`. Drop-in for {@link WOFSQLitePlaceLookup} in `createWOFResolver(backend)`
|
|
127
86
|
* — same `findPlace` contract, population-first ranking.
|
|
128
87
|
*/
|
|
129
88
|
export class WOFCandidateTableLookup {
|
|
130
89
|
#db;
|
|
131
|
-
|
|
90
|
+
/**
|
|
91
|
+
* Resources this instance opened. A connection handed in by a caller is NOT in here, so disposal cannot reach it —
|
|
92
|
+
* ownership is membership rather than a flag a later branch has to check.
|
|
93
|
+
*/
|
|
94
|
+
#resources = new DisposableStack();
|
|
132
95
|
#countryToID = new Map();
|
|
133
96
|
#idToCountry = new Map();
|
|
134
97
|
#placetypeToID = new Map();
|
|
@@ -144,48 +107,86 @@ export class WOFCandidateTableLookup {
|
|
|
144
107
|
*/
|
|
145
108
|
#ftsProbe;
|
|
146
109
|
/**
|
|
147
|
-
* Prepared UNFILTERED existence probe (`name_key` present anywhere, ignoring country/placetype/bbox).
|
|
148
|
-
* fallback: fuzzy is a TYPO corrector, so it engages only when the name doesn't exist in the gazetteer at all.
|
|
149
|
-
* that DOES exist but missed under the active filter is a filter miss (e.g. a placer misroute "Vienna,
|
|
150
|
-
* not a spelling miss — fuzzing it would scrape an unrelated same-country place and defeat the
|
|
151
|
-
* country-agnostic retry. Prepared only alongside `#ftsProbe`.
|
|
110
|
+
* Prepared UNFILTERED existence probe (`name_key` present anywhere, ignoring country/placetype/bbox). Checks the
|
|
111
|
+
* fuzzy fallback: fuzzy is a TYPO corrector, so it engages only when the name doesn't exist in the gazetteer at all.
|
|
112
|
+
* A name that DOES exist but missed under the active filter is a filter miss (e.g. a placer misroute "Vienna,
|
|
113
|
+
* Austria"→IT), not a spelling miss — fuzzing it would scrape an unrelated same-country place and defeat the
|
|
114
|
+
* cascade's country-agnostic retry. Prepared only alongside `#ftsProbe`.
|
|
152
115
|
*/
|
|
153
116
|
#nameKeyExistsProbe;
|
|
154
117
|
/**
|
|
155
118
|
* Facts this candidate DB declares about itself — the coverage manifest (`country_coverage` + `country_bbox`) the
|
|
156
119
|
* gazetteer build emits, read once at open. `undefined` when the artifact predates the manifest, so every consumer
|
|
157
|
-
* (the hard-country coverage
|
|
120
|
+
* (the hard-country coverage check, guard-B plausibility) falls back to its code constants byte-identically.
|
|
158
121
|
*/
|
|
159
122
|
artifactCoverage;
|
|
160
123
|
/**
|
|
161
124
|
* `", importance"` when this artifact carries the #28 fame column, `""` when it does not — spliced into the probe's
|
|
162
|
-
* SELECT list. Existence-
|
|
125
|
+
* SELECT list. Existence-restricted exactly like `#ftsProbe` and `#postalCityProbe` above, and for the same reason: a
|
|
163
126
|
* candidate.db built before the column is a valid artifact, and naming a column it lacks would turn a stale gazetteer
|
|
164
127
|
* into `no such column` on the first keystroke rather than into "no fame signal", which is what it is.
|
|
165
128
|
*/
|
|
166
129
|
#importanceSelect;
|
|
130
|
+
/**
|
|
131
|
+
* Whether the artifact carries the #1730 `name_role` column — absent on pre-role builds, where the `excludeNameRoles`
|
|
132
|
+
* filter degrades to a no-op rather than erroring on a missing column.
|
|
133
|
+
*/
|
|
134
|
+
#hasNameRole;
|
|
135
|
+
#variantAliasExemption;
|
|
136
|
+
/**
|
|
137
|
+
* `", name_role"` when the artifact carries the column — the probe SELECT rides it so the #1882 exemption can read
|
|
138
|
+
* the stamp off the row; empty on a pre-role build.
|
|
139
|
+
*/
|
|
140
|
+
#roleSelect;
|
|
141
|
+
/**
|
|
142
|
+
* Prepared chain probe over the `candidate_ancestor` sidecar — `undefined` when the artifact predates it.
|
|
143
|
+
*/
|
|
144
|
+
#ancestorsProbe;
|
|
145
|
+
#ancestorsCache = new Map();
|
|
146
|
+
/**
|
|
147
|
+
* Prepared interval-label probe over `candidate_interval` — `undefined` when the artifact predates the sidecar, which
|
|
148
|
+
* is what makes the admin-containment re-rank (#1717 stage 2) capability-restricted: without it,
|
|
149
|
+
* `FindPlaceQuery.regionQualifier` is ignored, no candidate carries a `containedByQualifier` stamp, and the resolver
|
|
150
|
+
* walk reports the change `unavailable` instead of silently dead.
|
|
151
|
+
*/
|
|
152
|
+
#intervalProbe;
|
|
153
|
+
#intervalCache = new Map();
|
|
154
|
+
/**
|
|
155
|
+
* Prepared qualifier probe: the region-band rows (plus `country` — the region SLOT can hold a mislabeled country
|
|
156
|
+
* name, "Moscow, Russia" parses region="Russia") for one folded qualifier key. `undefined` when the artifact lacks
|
|
157
|
+
* the sidecar or the placetype dictionary lacks the band entirely.
|
|
158
|
+
*/
|
|
159
|
+
#qualifierProbe;
|
|
160
|
+
/**
|
|
161
|
+
* The ancestor lineage of a resolved place — nearest-first (locality-tier → county → region → … → country), the same
|
|
162
|
+
* order the FTS backend's `ancestorLineage` serves, read from the `candidate_ancestor` sidecar in one clustered
|
|
163
|
+
* probe. Backs `ResolveOpts.includeAncestors` (#404) on this backend, which is what puts region-class ancestry in
|
|
164
|
+
* front of the admin-coherence check (#1717).
|
|
165
|
+
*
|
|
166
|
+
* A PROPERTY, not a method, and assigned only when the artifact carries the sidecar: capability probes (`typeof
|
|
167
|
+
* backend.ancestors === "function"` — the resolver's gap report) then read the ARTIFACT truthfully. A candidate.db
|
|
168
|
+
* built before the sidecar reports the capability absent instead of presenting a method that answers `[]` for every
|
|
169
|
+
* place, which would be an absence dressed as a negative answer.
|
|
170
|
+
*/
|
|
171
|
+
ancestors;
|
|
167
172
|
constructor(opts) {
|
|
168
173
|
if (opts.database) {
|
|
169
174
|
this.#db = opts.database;
|
|
170
|
-
this.#ownsDB = false;
|
|
171
175
|
}
|
|
172
176
|
else if (opts.databasePath) {
|
|
173
|
-
this.#db = new
|
|
174
|
-
this.#ownsDB = true;
|
|
177
|
+
this.#db = this.#resources.use(new DatabaseClient(opts.databasePath, { readOnly: true }));
|
|
175
178
|
}
|
|
176
179
|
else {
|
|
177
180
|
throw new Error("WOFCandidateTableLookup needs `databasePath` or `database`");
|
|
178
181
|
}
|
|
179
182
|
// The code tables are tiny (country/placetype dictionaries) — load them once at construction so
|
|
180
183
|
// `findPlace` is a single B-tree probe with no dictionary round-trip.
|
|
181
|
-
for (const r of this.#db.prepare("SELECT id, code FROM country_codes")
|
|
184
|
+
for (const r of allRows(this.#db.prepare("SELECT id, code FROM country_codes"))) {
|
|
182
185
|
const code = String(r.code).toUpperCase();
|
|
183
186
|
this.#countryToID.set(code, Number(r.id));
|
|
184
187
|
this.#idToCountry.set(Number(r.id), code);
|
|
185
188
|
}
|
|
186
|
-
for (const r of this.#db
|
|
187
|
-
.prepare("SELECT id, placetype FROM placetype_codes")
|
|
188
|
-
.all()) {
|
|
189
|
+
for (const r of allRows(this.#db.prepare("SELECT id, placetype FROM placetype_codes"))) {
|
|
189
190
|
this.#placetypeToID.set(String(r.placetype), Number(r.id));
|
|
190
191
|
this.#idToPlacetype.set(Number(r.id), String(r.placetype));
|
|
191
192
|
}
|
|
@@ -202,11 +203,192 @@ export class WOFCandidateTableLookup {
|
|
|
202
203
|
}
|
|
203
204
|
// #28 fame column: probed ONCE here (it runs a PRAGMA, and `findPlace` is per-keystroke hot).
|
|
204
205
|
this.#importanceSelect = hasColumn(this.#db, "candidate", "importance") ? ", importance" : "";
|
|
205
|
-
|
|
206
|
+
this.#hasNameRole = hasColumn(this.#db, "candidate", "name_role");
|
|
207
|
+
this.#variantAliasExemption = opts.variantAliasExemption === true;
|
|
208
|
+
this.#roleSelect = this.#hasNameRole ? ", name_role" : "";
|
|
209
|
+
// Ancestors sidecar (#1717): existence-restricted like the probes above, and the CAPABILITY checks with
|
|
210
|
+
// it — see the `ancestors` property doc for why an older artifact must read as "no ancestors()"
|
|
211
|
+
// rather than as a method that answers [] everywhere.
|
|
212
|
+
if (hasTable(this.#db, CANDIDATE_ANCESTOR_TABLE)) {
|
|
213
|
+
this.#ancestorsProbe = this.#db.prepare(`SELECT parent_spr_id, parent_placetype_id, parent_name FROM ${CANDIDATE_ANCESTOR_TABLE}` +
|
|
214
|
+
" WHERE spr_id = ? ORDER BY depth ASC");
|
|
215
|
+
this.ancestors = (id) => this.#ancestorLineage(id);
|
|
216
|
+
}
|
|
217
|
+
// Admin-containment re-rank (#1717 stage 2): conditioned on the interval half of the sidecar (built in
|
|
218
|
+
// the same pass as the closure rows; probed separately so a hand-degraded artifact degrades
|
|
219
|
+
// truthfully) AND on the placetype dictionary carrying the qualifier band at all.
|
|
220
|
+
if (this.#ancestorsProbe && hasTable(this.#db, CANDIDATE_INTERVAL_TABLE)) {
|
|
221
|
+
this.#intervalProbe = this.#db.prepare(`SELECT pre, post FROM ${CANDIDATE_INTERVAL_TABLE} WHERE spr_id = ?`);
|
|
222
|
+
const bandIDs = [...REGION_CLASS_PLACETYPES, "country"]
|
|
223
|
+
.map((placetype) => this.#placetypeToID.get(placetype))
|
|
224
|
+
.filter((id) => id !== undefined);
|
|
225
|
+
if (bandIDs.length) {
|
|
226
|
+
this.#qualifierProbe = this.#db.prepare(`SELECT DISTINCT spr_id FROM candidate WHERE name_key = ? AND placetype_id IN (${bandIDs.join(",")}) LIMIT 8`);
|
|
227
|
+
}
|
|
228
|
+
}
|
|
229
|
+
// Coverage manifest (survey candidate #2): the artifact's own coverage facts, existence-restricted like
|
|
206
230
|
// the probes above — a candidate.db built before the manifest reads `undefined` and consumers keep
|
|
207
231
|
// their code-constant fallbacks byte-identically.
|
|
208
232
|
this.artifactCoverage = readGazetteerCoverageManifest(this.#db);
|
|
209
233
|
}
|
|
234
|
+
/**
|
|
235
|
+
* The memoized chain read behind {@link ancestors}. Sync raw `.prepare()` on purpose — the backend contract's
|
|
236
|
+
* `ancestors()` is synchronous (the sync-by-interface resolver-reader rule), and the sidecar row already carries the
|
|
237
|
+
* parent's name and placetype, so this is one clustered probe with no join.
|
|
238
|
+
*/
|
|
239
|
+
#ancestorLineage(id) {
|
|
240
|
+
const pid = typeof id === "number" ? id : Number(id);
|
|
241
|
+
if (!Number.isFinite(pid) || !this.#ancestorsProbe)
|
|
242
|
+
return [];
|
|
243
|
+
const cached = this.#ancestorsCache.get(pid);
|
|
244
|
+
if (cached)
|
|
245
|
+
return cached;
|
|
246
|
+
const rows = allRows(this.#ancestorsProbe, pid);
|
|
247
|
+
const lineage = rows.map((r) => ({
|
|
248
|
+
id: Number(r.parent_spr_id),
|
|
249
|
+
placetype: this.#idToPlacetype.get(Number(r.parent_placetype_id)) ?? "",
|
|
250
|
+
name: String(r.parent_name ?? ""),
|
|
251
|
+
}));
|
|
252
|
+
this.#ancestorsCache.set(pid, lineage);
|
|
253
|
+
return lineage;
|
|
254
|
+
}
|
|
255
|
+
/**
|
|
256
|
+
* The interval label for one place, memoized. `null` is a real answer — the place has no recorded ancestry in the
|
|
257
|
+
* source (absence semantics: UNVERIFIABLE, never a containment verdict) — and is cached as such.
|
|
258
|
+
*/
|
|
259
|
+
#intervalLabel(sprID) {
|
|
260
|
+
if (!this.#intervalProbe)
|
|
261
|
+
return null;
|
|
262
|
+
const cached = this.#intervalCache.get(sprID);
|
|
263
|
+
if (cached !== undefined)
|
|
264
|
+
return cached;
|
|
265
|
+
const row = this.#intervalProbe.get(sprID);
|
|
266
|
+
const label = row ? { pre: Number(row.pre), post: Number(row.post) } : null;
|
|
267
|
+
this.#intervalCache.set(sprID, label);
|
|
268
|
+
return label;
|
|
269
|
+
}
|
|
270
|
+
/**
|
|
271
|
+
* The qualifier's own rows in the candidate table: every region-band (+ country) place whose `name_key` matches one
|
|
272
|
+
* of the qualifier's {@link regionQualifierProbeKeys} expansions. Alias keys participate — `Thüringen` finds the row
|
|
273
|
+
* stored as `Thuringia` through the artifact's own alias keying, which is precisely the variant-form bridge the
|
|
274
|
+
* admin-coherence verdicts' fold-equality bound cannot offer (its stated v1 bound). Empty = the qualifier names
|
|
275
|
+
* nothing the artifact knows; the caller then stamps `false` everywhere and reorders nothing.
|
|
276
|
+
*/
|
|
277
|
+
#qualifierRegionIDs(qualifier, country) {
|
|
278
|
+
const ids = new Set();
|
|
279
|
+
if (!this.#qualifierProbe)
|
|
280
|
+
return ids;
|
|
281
|
+
for (const key of regionQualifierProbeKeys(qualifier, country)) {
|
|
282
|
+
if (!key)
|
|
283
|
+
continue;
|
|
284
|
+
for (const row of allRows(this.#qualifierProbe, key)) {
|
|
285
|
+
ids.add(Number(row.spr_id));
|
|
286
|
+
}
|
|
287
|
+
}
|
|
288
|
+
return ids;
|
|
289
|
+
}
|
|
290
|
+
/**
|
|
291
|
+
* Is `sprID` contained by ANY of the qualifier's rows? Interval first — {@link intervalContains}, O(1), reflexive —
|
|
292
|
+
* then the closure rows where intervals abstain: the interval forest encodes only the CANONICAL parent per place, so
|
|
293
|
+
* a `false` there means "not contained along the canonical hierarchy", and the chain probe (one clustered read of
|
|
294
|
+
* ≤{@link MAX_ANCESTOR_DEPTH} rows) is the complete record that determines the result.
|
|
295
|
+
*/
|
|
296
|
+
#containedByQualifier(sprID, qualifierIDs, qualifierLabels) {
|
|
297
|
+
if (qualifierIDs.has(sprID))
|
|
298
|
+
return true;
|
|
299
|
+
const label = this.#intervalLabel(sprID);
|
|
300
|
+
if (label && qualifierLabels.some((outer) => intervalContains(outer, label)))
|
|
301
|
+
return true;
|
|
302
|
+
return this.#ancestorLineage(sprID).some((ancestor) => qualifierIDs.has(Number(ancestor.id)));
|
|
303
|
+
}
|
|
304
|
+
/**
|
|
305
|
+
* The #1717 stage-2 re-rank over one lookup's final row set. Three steps, each additive:
|
|
306
|
+
*
|
|
307
|
+
* 1. Resolve the qualifier to its region-band rows ({@link #qualifierRegionIDs}) and stamp every existing row's
|
|
308
|
+
* `containedByQualifier` — the stamp is the trace surface, written even when nothing reorders.
|
|
309
|
+
* 2. INJECT contained same-key candidates the country scope hid: the deciding-site measurement (2026-08-18, the #1729
|
|
310
|
+
* lesson re-confirmed) showed `Weimar, Thüringen` under the en-US locale probes `country_id = US`, so the DE row
|
|
311
|
+
* is not IN the list and no reorder of the list can reach it. The injection probe runs the same exact fold (and,
|
|
312
|
+
* on a contained-miss, the qualifier-strip variant restricted to primary keys — the #1626 alias-scrape guard)
|
|
313
|
+
* under the SHAPE conds only, appends contained rows not already present, and never removes anything — recall can
|
|
314
|
+
* only widen. The typo-fuzzy tier is deliberately not probed: a qualifier cannot vouch for a name the gazetteer
|
|
315
|
+
* does not carry.
|
|
316
|
+
* 3. Partition contained-first — the SHARED {@link partitionByContainment} (tier-safe, stable; the resolver walk runs
|
|
317
|
+
* the same function after its fame re-rank, one function at both deciding sites per the #861 rule) — then
|
|
318
|
+
* re-window to `limit`.
|
|
319
|
+
*
|
|
320
|
+
* A qualifier that matches nothing stamps `false` everywhere and reorders nothing — byte-identical answers, and the
|
|
321
|
+
* walk's verdict reads `no_contained_candidate` rather than `unavailable` (the question WAS asked).
|
|
322
|
+
*/
|
|
323
|
+
#applyAdminContainment(rows, qualifier, country, opts) {
|
|
324
|
+
const qualifierIDs = this.#qualifierRegionIDs(qualifier, country);
|
|
325
|
+
if (!qualifierIDs.size) {
|
|
326
|
+
for (const row of rows) {
|
|
327
|
+
row.containedByQualifier = false;
|
|
328
|
+
}
|
|
329
|
+
return rows;
|
|
330
|
+
}
|
|
331
|
+
const qualifierLabels = [...qualifierIDs]
|
|
332
|
+
.map((id) => this.#intervalLabel(id))
|
|
333
|
+
.filter((label) => label !== null);
|
|
334
|
+
const contained = (sprID) => this.#containedByQualifier(sprID, qualifierIDs, qualifierLabels);
|
|
335
|
+
for (const row of rows) {
|
|
336
|
+
row.containedByQualifier = contained(Number(row.spr_id));
|
|
337
|
+
}
|
|
338
|
+
const present = new Set(rows.map((row) => Number(row.spr_id)));
|
|
339
|
+
const injected = [];
|
|
340
|
+
const injectSQL = (primaryOnly) => "SELECT spr_id, name, country_id, placetype_id, latitude, longitude, min_lat, min_lon, max_lat, max_lon, neg_rank, is_primary, population" +
|
|
341
|
+
`${this.#importanceSelect} FROM candidate WHERE ${["name_key = ?", ...opts.shapeFilters, ...(primaryOnly ? ["is_primary = 1"] : [])].join(" AND ")} ` +
|
|
342
|
+
"ORDER BY neg_rank ASC LIMIT ?";
|
|
343
|
+
const injectFrom = (key, primaryOnly) => {
|
|
344
|
+
const fetched = allRows(this.#db.prepare(injectSQL(primaryOnly)), key, ...opts.shapeParams, RERANK_FETCH);
|
|
345
|
+
for (const row of fetched) {
|
|
346
|
+
const sprID = Number(row.spr_id);
|
|
347
|
+
if (present.has(sprID) || !contained(sprID))
|
|
348
|
+
continue;
|
|
349
|
+
present.add(sprID);
|
|
350
|
+
injected.push({ ...row, effectiveNegRank: row.neg_rank, demoted: false, containedByQualifier: true });
|
|
351
|
+
}
|
|
352
|
+
};
|
|
353
|
+
injectFrom(opts.nameKey, false);
|
|
354
|
+
// #1731: the dependent-locality band. A locality query's filter group (locality/borough/localadmin)
|
|
355
|
+
// cannot reach a neighbourhood-tier namesake, so a CONTAINED one is structurally invisible no matter
|
|
356
|
+
// how the list reorders — the Astoria class: Queens' Astoria is a WOF neighbourhood, and the walk
|
|
357
|
+
// answered the Oregon locality under `qualifier="NY"` because nothing in the pool sat under NY. The
|
|
358
|
+
// widening is injection-only and triple-conditioned: the band is explicit (neighbourhood/macrohood/microhood
|
|
359
|
+
// — never region or country tiers), admission still requires the sidecar's containment proof, and only
|
|
360
|
+
// primary-keyed rows enter (an alias-keyed neighbourhood is the #1626 scrape class). Recall can only
|
|
361
|
+
// widen, and only toward rows the qualifier vouches for. `opts.shapeFilters`' bbox clause is
|
|
362
|
+
// deliberately not carried: containment is the stronger constraint, and the two co-occurring is not a
|
|
363
|
+
// measured shape.
|
|
364
|
+
const bandIDs = ["neighbourhood", "macrohood", "microhood"]
|
|
365
|
+
.map((placetype) => this.#placetypeToID.get(placetype))
|
|
366
|
+
.filter((id) => id !== undefined);
|
|
367
|
+
if (bandIDs.length) {
|
|
368
|
+
const bandSQL = "SELECT spr_id, name, country_id, placetype_id, latitude, longitude, min_lat, min_lon, max_lat, max_lon, neg_rank, is_primary, population" +
|
|
369
|
+
`${this.#importanceSelect} FROM candidate WHERE name_key = ? AND placetype_id IN (${bandIDs.map(() => "?").join(",")}) AND is_primary = 1 ` +
|
|
370
|
+
"ORDER BY neg_rank ASC LIMIT ?";
|
|
371
|
+
const fetched = allRows(this.#db.prepare(bandSQL), opts.nameKey, ...bandIDs, RERANK_FETCH);
|
|
372
|
+
for (const row of fetched) {
|
|
373
|
+
const sprID = Number(row.spr_id);
|
|
374
|
+
if (present.has(sprID) || !contained(sprID))
|
|
375
|
+
continue;
|
|
376
|
+
present.add(sprID);
|
|
377
|
+
injected.push({ ...row, effectiveNegRank: row.neg_rank, demoted: false, containedByQualifier: true });
|
|
378
|
+
}
|
|
379
|
+
}
|
|
380
|
+
// The strip variant mirrors the cascade's discipline: tried only when the exact fold vouched for
|
|
381
|
+
// nothing, and primary-keyed only (a stripped surface never named an alias — #1626).
|
|
382
|
+
if (!injected.length &&
|
|
383
|
+
!rows.some((row) => row.containedByQualifier) &&
|
|
384
|
+
opts.strippedKey &&
|
|
385
|
+
opts.strippedKey !== opts.nameKey) {
|
|
386
|
+
injectFrom(opts.strippedKey, true);
|
|
387
|
+
}
|
|
388
|
+
if (!injected.length && !rows.some((row) => row.containedByQualifier))
|
|
389
|
+
return rows;
|
|
390
|
+
return partitionByContainment([...rows, ...injected], (row) => row.containedByQualifier === true, (row) => !row.demoted && !row.fuzzy).slice(0, opts.limit);
|
|
391
|
+
}
|
|
210
392
|
/**
|
|
211
393
|
* Does this query want a locality-tier place? Postal-city aliases (#741) are all localities.
|
|
212
394
|
*/
|
|
@@ -259,7 +441,7 @@ export class WOFCandidateTableLookup {
|
|
|
259
441
|
return [];
|
|
260
442
|
// #741: postcode-keyed postal-city alias. An exact `(name_key, postcode)` hit resolves a
|
|
261
443
|
// user-typed POSTAL city ("Antioch", 37013) to the geographic locality the postcode sits in
|
|
262
|
-
// ("Nashville"), bypassing the population/region ranking that can't see the postcode.
|
|
444
|
+
// ("Nashville"), bypassing the population/region ranking that can't see the postcode. Conditioned on
|
|
263
445
|
// the side-index being present, a postcode in the query, and a locality-tier request — so the
|
|
264
446
|
// common (no-postcode / non-locality) path is untouched. A hit short-circuits: the postcode is
|
|
265
447
|
// an exact, high-confidence disambiguator, so we return the single geographic locality.
|
|
@@ -281,9 +463,15 @@ export class WOFCandidateTableLookup {
|
|
|
281
463
|
}
|
|
282
464
|
}
|
|
283
465
|
const limit = Math.max(1, query.limit ?? 10);
|
|
284
|
-
// Filter conds shared by the exact-key + strip-fallback probes (everything but name_key).
|
|
466
|
+
// Filter conds shared by the exact-key + strip-fallback probes (everything but name_key). The
|
|
467
|
+
// SHAPE subset (placetype/bbox/primary — everything but the country scope) is kept separately
|
|
468
|
+
// because the admin-containment injection probe (#1717 stage 2) runs under the shape conds
|
|
469
|
+
// WITHOUT the country: bypassing a locale-inferred country scope for a qualifier-vouched
|
|
470
|
+
// candidate is the change's whole point and the one filter injection may cross.
|
|
285
471
|
const filters = [];
|
|
286
472
|
const filterParams = [];
|
|
473
|
+
const shapeFilters = [];
|
|
474
|
+
const shapeParams = [];
|
|
287
475
|
if (query.country) {
|
|
288
476
|
const cid = this.#countryToID.get(query.country.toUpperCase());
|
|
289
477
|
if (cid === undefined)
|
|
@@ -300,14 +488,32 @@ export class WOFCandidateTableLookup {
|
|
|
300
488
|
.filter((v) => v !== undefined);
|
|
301
489
|
if (!ids.length)
|
|
302
490
|
return [];
|
|
303
|
-
|
|
304
|
-
|
|
491
|
+
shapeFilters.push(`placetype_id IN (${ids.map(() => "?").join(",")})`);
|
|
492
|
+
shapeParams.push(...ids);
|
|
305
493
|
}
|
|
306
494
|
if (query.bbox) {
|
|
307
495
|
const b = query.bbox;
|
|
308
|
-
|
|
309
|
-
|
|
496
|
+
shapeFilters.push("latitude BETWEEN ? AND ? AND longitude BETWEEN ? AND ?");
|
|
497
|
+
shapeParams.push(b.minLat, b.maxLat, b.minLon, b.maxLon);
|
|
310
498
|
}
|
|
499
|
+
// The re-reading guard (#1632, the #1626 rationale generalized to the caller): a probe whose surface
|
|
500
|
+
// is a token taken out of a longer classified span never NAMED an alias, so alias-keyed rows must not
|
|
501
|
+
// answer it — 'Savile Row''s token 'Row' resolved Rhu, Scotland (585 km) through the village's
|
|
502
|
+
// historical-name alias key. Whole-input bare probes never set this, keeping the exonym recall the
|
|
503
|
+
// #1546 note protects (Москва's alias rows answer 'Moscow').
|
|
504
|
+
if (query.primaryOnly) {
|
|
505
|
+
shapeFilters.push("is_primary = 1");
|
|
506
|
+
}
|
|
507
|
+
// The role guard (#1730): a probe may refuse abbreviation/gloss alias rows while keeping the
|
|
508
|
+
// role-NULL exonym tier open — the distinction `primaryOnly` cannot express. Degrades to a no-op
|
|
509
|
+
// on an artifact without the column.
|
|
510
|
+
if (query.excludeNameRoles?.length && this.#hasNameRole) {
|
|
511
|
+
shapeFilters.push(`(name_role IS NULL OR name_role NOT IN (${query.excludeNameRoles.map(() => "?").join(",")}))`);
|
|
512
|
+
shapeParams.push(...query.excludeNameRoles);
|
|
513
|
+
}
|
|
514
|
+
// The main-probe conds are country-then-shape, exactly the order they have always been.
|
|
515
|
+
filters.push(...shapeFilters);
|
|
516
|
+
filterParams.push(...shapeParams);
|
|
311
517
|
// Region scope: when the cascade resolves a region and passes it down as `parentID` (the walk sets
|
|
312
518
|
// `query.parentID = parentResolved.id`), the candidate build stamps each place's region-tier ancestor
|
|
313
519
|
// id into `region_id` (build-candidate.ts `regionOf`), and that id equals the resolved region's WOF id
|
|
@@ -317,13 +523,18 @@ export class WOFCandidateTableLookup {
|
|
|
317
523
|
// country/non-region parent (no `region_id` match), a `region_id=0` row (place with no region
|
|
318
524
|
// ancestor), or a wrong parent degrades to today's behavior — never worse, recall-safe by construction.
|
|
319
525
|
const regionParentID = query.parentID || undefined;
|
|
320
|
-
const probe = (nk, regionID) => {
|
|
526
|
+
const probe = (nk, regionID, countryID) => {
|
|
321
527
|
const conds = ["name_key = ?", ...filters];
|
|
322
528
|
const params = [nk, ...filterParams];
|
|
323
529
|
if (regionID !== undefined) {
|
|
324
530
|
conds.push("region_id = ?");
|
|
325
531
|
params.push(regionID);
|
|
326
532
|
}
|
|
533
|
+
// #1585: the fuzzy tier's country scope — only the corrected-key probes pass this.
|
|
534
|
+
if (typeof countryID === "number") {
|
|
535
|
+
conds.push("country_id = ?");
|
|
536
|
+
params.push(countryID);
|
|
537
|
+
}
|
|
327
538
|
// Fetch population-ordered (the clustered-key order — a cheap ordered scan), over-fetching to
|
|
328
539
|
// RERANK_FETCH so the bounded cross-country primary-preference re-rank (below) can promote the
|
|
329
540
|
// intended primary even when a cluster of more-populous foreign aliases sits ahead of it. `is_primary`
|
|
@@ -337,9 +548,9 @@ export class WOFCandidateTableLookup {
|
|
|
337
548
|
// whether the query was bare enough to deserve it. A backend that pre-sorted by fame would apply
|
|
338
549
|
// it to every lookup, including the qualified addresses the D-rule guard exists to protect.
|
|
339
550
|
const sql = "SELECT spr_id, name, country_id, placetype_id, latitude, longitude, min_lat, min_lon, max_lat, max_lon, neg_rank, is_primary, population" +
|
|
340
|
-
`${this.#importanceSelect} FROM candidate WHERE ${conds.join(" AND ")} ORDER BY neg_rank ASC LIMIT ?`;
|
|
341
|
-
const fetched = this.#db.prepare(sql)
|
|
342
|
-
return rankByPrimaryPreference(fetched, limit);
|
|
551
|
+
`${this.#importanceSelect}${this.#roleSelect} FROM candidate WHERE ${conds.join(" AND ")} ORDER BY neg_rank ASC LIMIT ?`;
|
|
552
|
+
const fetched = allRows(this.#db.prepare(sql), ...params, Math.max(limit, RERANK_FETCH));
|
|
553
|
+
return rankByPrimaryPreference(fetched, limit, undefined, this.#idToPlacetype, this.#variantAliasExemption);
|
|
343
554
|
};
|
|
344
555
|
// The exact → qualifier-strip → typo-fuzzy probe cascade, run at a fixed region scope. Region scoping
|
|
345
556
|
// only tightens an already-population-first pick, so a region MISS re-runs the whole cascade unscoped
|
|
@@ -352,7 +563,14 @@ export class WOFCandidateTableLookup {
|
|
|
352
563
|
// miss; the cascade's region bbox disambiguates any base-name ambiguity.
|
|
353
564
|
const strippedKey = normalizeLocalityForKey(stripLocalityQualifier(text));
|
|
354
565
|
if (strippedKey && strippedKey !== nameKey) {
|
|
355
|
-
|
|
566
|
+
// #1626: a stripped probe may answer only through a NON-PRIMARY alias key, which is a
|
|
567
|
+
// scrape, not a qualifier match — 'Savile Row' stripped to 'row' resolved Rhu, Scotland
|
|
568
|
+
// (585 km) through the village's historical-name alias. The legitimate qualifier class
|
|
569
|
+
// matches the place's own primary key ('Lenk im Simmental' → the Lenk row keyed 'lenk',
|
|
570
|
+
// is_primary=1), so refusing alias-keyed rows keeps every intended case and kills the
|
|
571
|
+
// scrape. The alias tier remains fully available to EXACT queries — only the stripped
|
|
572
|
+
// RETRY loses it, because the query's own surface never named the alias.
|
|
573
|
+
rows = probe(strippedKey, regionID).filter((r) => r.is_primary === 1);
|
|
356
574
|
}
|
|
357
575
|
}
|
|
358
576
|
// Typo-tolerant fallback (the unified gazetteer's fuzzy mode): an exact + strip miss may be a
|
|
@@ -361,7 +579,7 @@ export class WOFCandidateTableLookup {
|
|
|
361
579
|
// the same on either backend. The country/placetype/bbox/region filters still apply via `probe`.
|
|
362
580
|
// Skipped when the index is absent (byte-stable for an older candidate.db).
|
|
363
581
|
//
|
|
364
|
-
//
|
|
582
|
+
// Condition: only when the name doesn't exist in the gazetteer AT ALL (unfiltered). A name that exists
|
|
365
583
|
// but missed under the active country/placetype/bbox filter is a FILTER miss, not a spelling miss
|
|
366
584
|
// — fuzzing it scrapes an unrelated same-filter place ("Vienna, Austria" misrouted to IT would
|
|
367
585
|
// pull a tiny Italian name_key near Siena) and masks the cascade's country-agnostic retry that
|
|
@@ -371,17 +589,23 @@ export class WOFCandidateTableLookup {
|
|
|
371
589
|
// DIFFERENT postcode. The 2026-08-05 Code-Point swap exposed the trap at scale: Northern Ireland's
|
|
372
590
|
// `BT3 9QQ` (absent — no permissive NI source) trigram-matched Sheffield's `S3 9QQ` (Jaccard 0.4
|
|
373
591
|
// on {39q, 9qq}) and resolved 200+ km wrong with full confidence. An unknown postcode must abstain.
|
|
592
|
+
// #1585: a locale HINT scopes the typo tier to its country. Only when no hard `country`
|
|
593
|
+
// filter is active (that is already narrower); a scope naming a country the table doesn't
|
|
594
|
+
// carry is a SCOPED-EMPTY — the fuzzy tier abstains rather than falling through worldwide.
|
|
595
|
+
const fuzzyCountryID = !query.country && query.fuzzyCountry ? this.#countryToID.get(query.fuzzyCountry.toUpperCase()) : undefined;
|
|
596
|
+
const fuzzyScopedOut = !query.country && !!query.fuzzyCountry && typeof fuzzyCountryID !== "number";
|
|
374
597
|
if (!rows.length &&
|
|
375
598
|
!wantsPostcode &&
|
|
599
|
+
!fuzzyScopedOut &&
|
|
376
600
|
this.#ftsProbe &&
|
|
377
601
|
this.#nameKeyExistsProbe &&
|
|
378
602
|
!this.#nameKeyExistsProbe.get(nameKey)) {
|
|
379
603
|
const match = ftsTrigramQuery(nameKey);
|
|
380
604
|
if (match) {
|
|
381
|
-
const hits = this.#ftsProbe
|
|
605
|
+
const hits = allRows(this.#ftsProbe, match, FUZZY_FETCH);
|
|
382
606
|
const ranked = hits
|
|
383
|
-
.map((h) => ({ nk: String(h.name_key), s:
|
|
384
|
-
.filter((h) => h.s >=
|
|
607
|
+
.map((h) => ({ nk: String(h.name_key), s: wordFuzzySimilarity(nameKey, String(h.name_key)) }))
|
|
608
|
+
.filter((h) => h.s >= WORD_FUZZY_MIN)
|
|
385
609
|
// oxlint-disable-next-line unicorn/no-array-sort -- sorts a freshly-built array; toSorted would double-allocate on a hot path
|
|
386
610
|
.sort((a, b) => b.s - a.s);
|
|
387
611
|
const seen = new Set();
|
|
@@ -391,7 +615,7 @@ export class WOFCandidateTableLookup {
|
|
|
391
615
|
seen.add(h.nk);
|
|
392
616
|
// #17: stamp the tier. These rows answer a name the gazetteer does not carry, so they are
|
|
393
617
|
// fuzzy matches and `exactMatch` below must say so — see `RankedRow.fuzzy`.
|
|
394
|
-
rows.push(...probe(h.nk, regionID).map((r) => ({ ...r, fuzzy: true })));
|
|
618
|
+
rows.push(...probe(h.nk, regionID, fuzzyCountryID).map((r) => ({ ...r, fuzzy: true })));
|
|
395
619
|
if (rows.length >= limit)
|
|
396
620
|
break;
|
|
397
621
|
}
|
|
@@ -401,19 +625,24 @@ export class WOFCandidateTableLookup {
|
|
|
401
625
|
return rows;
|
|
402
626
|
};
|
|
403
627
|
let rows = cascade(regionParentID);
|
|
628
|
+
// #1731: whether the rows the caller receives came from the UNSCOPED fallback below — the backend's
|
|
629
|
+
// interior check the resolver-side trace (#1721) cannot otherwise see. Stamped onto every returned
|
|
630
|
+
// place, because the re-admission path is exactly where a wrong-instance namesake enters.
|
|
631
|
+
let regionScopeMiss = false;
|
|
404
632
|
// Region-scope fallback: if scoping to the parent region found nothing across the whole cascade, retry
|
|
405
633
|
// unscoped so a place with no in-region row (missing ancestry, or a country/non-region parent) still
|
|
406
634
|
// resolves exactly as it does today. Only when a region scope was actually applied.
|
|
407
635
|
if (!rows.length && regionParentID !== undefined) {
|
|
408
636
|
rows = cascade(undefined);
|
|
637
|
+
regionScopeMiss = rows.length > 0;
|
|
409
638
|
}
|
|
410
639
|
// Postcode-containment coherence (#31, Mechanism 2): re-rank the rows by proximity to the postcode's
|
|
411
640
|
// own centroid, so the locality that CONTAINS the postcode wins the name-match tie (the "Paris" that
|
|
412
641
|
// holds 75001, not the one that holds a 75001-free namesake). The resolver sends this flag on locality
|
|
413
642
|
// lookups when `ResolveOpts.postcodeContainmentCoherence` is on. Strictly beneath the #741 postal-city
|
|
414
643
|
// short-circuit above — an exact (name, postcode) hit IS the answer and outranks any re-rank — and after
|
|
415
|
-
// the region-scope fallback, so it sees the final row set. Rows within the
|
|
416
|
-
// the out-
|
|
644
|
+
// the region-scope fallback, so it sees the final row set. Rows within the radius sort by distance first;
|
|
645
|
+
// the out-of-radius tail keeps its original population-first order. No in-radius row, or no postcode row in
|
|
417
646
|
// the candidate table → unchanged (byte-identical to the flag-off path).
|
|
418
647
|
if (query.postcode &&
|
|
419
648
|
query.postcodeContainmentCoherence === true &&
|
|
@@ -421,37 +650,51 @@ export class WOFCandidateTableLookup {
|
|
|
421
650
|
rows.length > 1) {
|
|
422
651
|
const anchor = this.#postcodeAnchor(query.postcode, query.country);
|
|
423
652
|
if (anchor) {
|
|
424
|
-
const
|
|
425
|
-
const
|
|
653
|
+
const insideThreshold = [];
|
|
654
|
+
const outsideThreshold = [];
|
|
426
655
|
for (const row of rows) {
|
|
427
656
|
const distanceKm = haversineKm(anchor.lat, anchor.lon, Number(row.latitude), Number(row.longitude));
|
|
428
|
-
if (distanceKm <=
|
|
429
|
-
|
|
657
|
+
if (distanceKm <= POSTCODE_CONTAINMENT_THRESHOLD_KM) {
|
|
658
|
+
insideThreshold.push({ row, distanceKm });
|
|
430
659
|
}
|
|
431
660
|
else {
|
|
432
|
-
|
|
661
|
+
outsideThreshold.push(row);
|
|
433
662
|
}
|
|
434
663
|
}
|
|
435
|
-
if (
|
|
664
|
+
if (insideThreshold.length) {
|
|
436
665
|
// oxlint-disable-next-line unicorn/no-array-sort -- sorts a freshly-built array; toSorted would double-allocate on a hot path
|
|
437
|
-
|
|
438
|
-
rows = [...
|
|
666
|
+
insideThreshold.sort((a, b) => a.distanceKm - b.distanceKm);
|
|
667
|
+
rows = [...insideThreshold.map(({ row }) => row), ...outsideThreshold];
|
|
439
668
|
}
|
|
440
669
|
}
|
|
441
670
|
}
|
|
671
|
+
// Admin-containment re-rank (#1717 stage 2): LAST, on the final row set — the qualifier is the
|
|
672
|
+
// address's outermost explicit statement, so its partition outranks the postcode-proximity order
|
|
673
|
+
// above (contained rows keep that order among themselves). Capability-conditioned on the sidecar
|
|
674
|
+
// (`#qualifierProbe`); when the artifact predates it, `regionQualifier` is ignored, no stamp is
|
|
675
|
+
// written, and the resolver walk reports the change `unavailable`.
|
|
676
|
+
if (query.regionQualifier?.trim() && this.#qualifierProbe && this.#wantsLocality(query.placetype)) {
|
|
677
|
+
rows = this.#applyAdminContainment(rows, query.regionQualifier.trim(), query.country, {
|
|
678
|
+
nameKey,
|
|
679
|
+
strippedKey: normalizeLocalityForKey(stripLocalityQualifier(text)),
|
|
680
|
+
shapeFilters,
|
|
681
|
+
shapeParams,
|
|
682
|
+
limit,
|
|
683
|
+
});
|
|
684
|
+
}
|
|
442
685
|
const candidates = rows.map((row) => {
|
|
443
686
|
const hasBbox = row.min_lat != null && row.max_lat != null && row.min_lon != null && row.max_lon != null;
|
|
444
687
|
return {
|
|
445
688
|
id: Number(row.spr_id),
|
|
446
689
|
name: String(row.name ?? ""),
|
|
447
690
|
placetype: (this.#idToPlacetype.get(Number(row.placetype_id)) ?? ""),
|
|
448
|
-
// Surfaced so the cascade can country-
|
|
691
|
+
// Surfaced so the cascade can country-restrict a postcode by the resolved locality (an ambiguous
|
|
449
692
|
// international postcode like 10115 = Berlin DE AND New York US must not out-resolve the city).
|
|
450
693
|
country: this.#idToCountry.get(Number(row.country_id)) ?? "",
|
|
451
694
|
lat: Number(row.latitude),
|
|
452
695
|
lon: Number(row.longitude),
|
|
453
696
|
// `score` stays the RAW population rank (`-neg_rank`) — it feeds the resolver walk's absolute
|
|
454
|
-
// `minWinningScore`
|
|
697
|
+
// `minWinningScore` floor (`resolve.ts`), which must see real prominence, never a penalized value.
|
|
455
698
|
score: -Number(row.neg_rank),
|
|
456
699
|
// `prominence` carries the bounded cross-country primary preference (the effective, penalty-adjusted
|
|
457
700
|
// rank). The walk ORDERS candidates by `prominence ?? score` (`resolve.ts`), so this is what makes the
|
|
@@ -466,6 +709,16 @@ export class WOFCandidateTableLookup {
|
|
|
466
709
|
// EXCEPT a row the typo-corrector produced (`fuzzy`), which by definition answers a name the
|
|
467
710
|
// gazetteer does not carry (see `RankedRow.fuzzy`).
|
|
468
711
|
exactMatch: !row.demoted && !row.fuzzy,
|
|
712
|
+
// #1731: emitted ONLY when a region scope was applied, missed, and the unscoped fallback
|
|
713
|
+
// produced this row — the re-admission path. Absence means the question never arose.
|
|
714
|
+
...(regionScopeMiss ? { regionScopeMiss: true } : {}),
|
|
715
|
+
// #1717 stage 2 — the containment stamp, tri-state: emitted ONLY when the question was
|
|
716
|
+
// asked (a `regionQualifier` query over a sidecar-bearing artifact); its absence is what
|
|
717
|
+
// the resolver walk reports as `unavailable` (meaning-of-zero).
|
|
718
|
+
...(row.containedByQualifier === undefined ? {} : { containedByQualifier: row.containedByQualifier }),
|
|
719
|
+
// #1893 — the exemption's firing mark, carried only when the ranker actually spared this row
|
|
720
|
+
// the cross-country penalty (see RankedRow.variantExempted).
|
|
721
|
+
...(row.variantExempted ? { variantAliasExempted: true } : {}),
|
|
469
722
|
// The two-score split's carry (ROAD_TO_V9 §2). `referential` names the prominence this
|
|
470
723
|
// backend has always ordered by — `neg_rank` IS `-log10(population + 1)`, so the score and
|
|
471
724
|
// the sort key are two readings of the same number.
|
|
@@ -500,60 +753,15 @@ export class WOFCandidateTableLookup {
|
|
|
500
753
|
: {}),
|
|
501
754
|
};
|
|
502
755
|
});
|
|
503
|
-
// Proximity re-rank (#938)
|
|
504
|
-
//
|
|
505
|
-
// nearness in one additive scale — so an in-view namesake wins a tie without a hard filter. Byte-
|
|
506
|
-
// identical to the plain population order when no bias is passed. `score` here is -neg_rank =
|
|
507
|
-
// log10(population + 1), so popTerm is the server formula read straight off it. Constants MIRROR
|
|
508
|
-
// lookup.ts's DEFAULT_WEIGHTS (biasBoost 4, populationBoost 4, populationScaleLog10 6,
|
|
509
|
-
// proximityScaleKm 100) — the #861 server↔demo parity contract; keep them in lockstep.
|
|
756
|
+
// Proximity re-rank (#938) — the shared implementation, so the browser byte-range twin runs the same
|
|
757
|
+
// code rather than the same constants. See proximity-rerank.ts for why that distinction mattered.
|
|
510
758
|
if (query.bias && query.bias.length) {
|
|
511
|
-
|
|
512
|
-
const POP_BOOST = 4;
|
|
513
|
-
const POP_SCALE_LOG10 = 6;
|
|
514
|
-
// SHARPER than lookup.ts's 100 km on purpose: this backend's `score` is log-population ALONE
|
|
515
|
-
// (no bm25 document term), so the population signal is weaker relative to the bias and the
|
|
516
|
-
// gentle 100 km decay let a 230 km-distant alias-exact township ("Paris Township", OH) edge
|
|
517
|
-
// out a global city ("Paris", FR) from a nearby view. A ~30 km scale keeps the boost to
|
|
518
|
-
// candidates the user is actually LOOKING at — an in-view namesake still wins (Dublin, OH from
|
|
519
|
-
// an Ohio view), a distant one no longer does (Paris stays FR from a Michigan view).
|
|
520
|
-
const PROX_SCALE_KM = 30;
|
|
521
|
-
const combinedProminence = (c) => {
|
|
522
|
-
// Population base is the PENALIZED `prominence` (set above = -effectiveNegRank), not raw `score`, so
|
|
523
|
-
// the cross-country primary preference carries into the bias-weighted order too — a coincidental
|
|
524
|
-
// foreign alias doesn't ride population back over a primary just because a viewport hint is present.
|
|
525
|
-
const popBase = c.prominence ?? c.score;
|
|
526
|
-
const popTerm = POP_BOOST * Math.min(1, Math.max(0, popBase) / POP_SCALE_LOG10);
|
|
527
|
-
let proxTerm = 0;
|
|
528
|
-
if (!(c.lat === 0 && c.lon === 0)) {
|
|
529
|
-
for (const b of query.bias) {
|
|
530
|
-
const d = haversineKm(b.lat, b.lon, c.lat, c.lon);
|
|
531
|
-
const term = (BIAS_BOOST * (b.weight ?? 1)) / (1 + d / PROX_SCALE_KM);
|
|
532
|
-
if (term > proxTerm) {
|
|
533
|
-
proxTerm = term;
|
|
534
|
-
}
|
|
535
|
-
}
|
|
536
|
-
}
|
|
537
|
-
return popTerm + proxTerm;
|
|
538
|
-
};
|
|
539
|
-
// Persist the combined value into `prominence` so the resolver walk's `prominence ?? score` sort (and any
|
|
540
|
-
// other node consumer) honors the bias order — then sort. Stable within equal prominence (preserves the
|
|
541
|
-
// population order the B-tree already gave).
|
|
542
|
-
candidates
|
|
543
|
-
.map((c, i) => {
|
|
544
|
-
c.prominence = combinedProminence(c);
|
|
545
|
-
return { c, i, p: c.prominence };
|
|
546
|
-
})
|
|
547
|
-
// oxlint-disable-next-line unicorn/no-array-sort -- sorts a freshly-built array; toSorted would double-allocate on a hot path
|
|
548
|
-
.sort((a, b) => b.p - a.p || a.i - b.i)
|
|
549
|
-
.forEach((x, j) => (candidates[j] = x.c));
|
|
759
|
+
applyProximityRerank(candidates, query.bias);
|
|
550
760
|
}
|
|
551
761
|
return candidates;
|
|
552
762
|
}
|
|
553
|
-
|
|
554
|
-
|
|
555
|
-
this.#db.close();
|
|
556
|
-
}
|
|
763
|
+
[Symbol.dispose]() {
|
|
764
|
+
this.#resources[Symbol.dispose]();
|
|
557
765
|
}
|
|
558
766
|
}
|
|
559
767
|
//# sourceMappingURL=candidate-lookup.js.map
|