@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.
Files changed (497) hide show
  1. package/README.md +2 -2
  2. package/data/PROVENANCE.md +214 -0
  3. package/data/sub-venue-lexicon.json +18269 -0
  4. package/out/src/adapters/ban/adapter.d.ts +5 -4
  5. package/out/src/adapters/ban/adapter.d.ts.map +1 -1
  6. package/out/src/adapters/ban/adapter.js +62 -69
  7. package/out/src/adapters/ban/adapter.js.map +1 -1
  8. package/out/src/adapters/ban/street-decompose.d.ts.map +1 -1
  9. package/out/src/adapters/ban/street-decompose.js +18 -25
  10. package/out/src/adapters/ban/street-decompose.js.map +1 -1
  11. package/out/src/adapters/fcc-bdc/adapter.d.ts +1 -7
  12. package/out/src/adapters/fcc-bdc/adapter.d.ts.map +1 -1
  13. package/out/src/adapters/fcc-bdc/adapter.js +4 -26
  14. package/out/src/adapters/fcc-bdc/adapter.js.map +1 -1
  15. package/out/src/adapters/geonames/adapter.d.ts +1 -1
  16. package/out/src/adapters/geonames/adapter.d.ts.map +1 -1
  17. package/out/src/adapters/geonames/adapter.js +72 -78
  18. package/out/src/adapters/geonames/adapter.js.map +1 -1
  19. package/out/src/adapters/geonames-postal/adapter.d.ts +1 -1
  20. package/out/src/adapters/geonames-postal/adapter.d.ts.map +1 -1
  21. package/out/src/adapters/geonames-postal/adapter.js +46 -51
  22. package/out/src/adapters/geonames-postal/adapter.js.map +1 -1
  23. package/out/src/adapters/gnaf/adapter.d.ts +1 -1
  24. package/out/src/adapters/gnaf/adapter.d.ts.map +1 -1
  25. package/out/src/adapters/gnaf/adapter.js +6 -9
  26. package/out/src/adapters/gnaf/adapter.js.map +1 -1
  27. package/out/src/adapters/gnaf/assemble.d.ts.map +1 -1
  28. package/out/src/adapters/gnaf/assemble.js +7 -12
  29. package/out/src/adapters/gnaf/assemble.js.map +1 -1
  30. package/out/src/adapters/index.d.ts +1 -1
  31. package/out/src/adapters/index.d.ts.map +1 -1
  32. package/out/src/adapters/index.js +1 -1
  33. package/out/src/adapters/index.js.map +1 -1
  34. package/out/src/adapters/openaddresses/adapter.d.ts +1 -1
  35. package/out/src/adapters/openaddresses/adapter.d.ts.map +1 -1
  36. package/out/src/adapters/openaddresses/adapter.js +5 -10
  37. package/out/src/adapters/openaddresses/adapter.js.map +1 -1
  38. package/out/src/adapters/overture/adapter.d.ts +1 -1
  39. package/out/src/adapters/overture/adapter.d.ts.map +1 -1
  40. package/out/src/adapters/overture/adapter.js +5 -9
  41. package/out/src/adapters/overture/adapter.js.map +1 -1
  42. package/out/src/adapters/state-hi-schools/adapter.d.ts +1 -1
  43. package/out/src/adapters/state-hi-schools/adapter.d.ts.map +1 -1
  44. package/out/src/adapters/state-hi-schools/adapter.js +57 -75
  45. package/out/src/adapters/state-hi-schools/adapter.js.map +1 -1
  46. package/out/src/adapters/state-ia-contractors/adapter.d.ts +1 -1
  47. package/out/src/adapters/state-ia-contractors/adapter.d.ts.map +1 -1
  48. package/out/src/adapters/state-ia-contractors/adapter.js +60 -78
  49. package/out/src/adapters/state-ia-contractors/adapter.js.map +1 -1
  50. package/out/src/adapters/state-ny-notaries/adapter.d.ts +1 -1
  51. package/out/src/adapters/state-ny-notaries/adapter.d.ts.map +1 -1
  52. package/out/src/adapters/state-ny-notaries/adapter.js +69 -87
  53. package/out/src/adapters/state-ny-notaries/adapter.js.map +1 -1
  54. package/out/src/adapters/state-tx-notaries/adapter.d.ts +1 -1
  55. package/out/src/adapters/state-tx-notaries/adapter.d.ts.map +1 -1
  56. package/out/src/adapters/state-tx-notaries/adapter.js +71 -89
  57. package/out/src/adapters/state-tx-notaries/adapter.js.map +1 -1
  58. package/out/src/adapters/synth-po-box/adapter.d.ts +3 -3
  59. package/out/src/adapters/synth-po-box/adapter.d.ts.map +1 -1
  60. package/out/src/adapters/synth-po-box/adapter.js +10 -18
  61. package/out/src/adapters/synth-po-box/adapter.js.map +1 -1
  62. package/out/src/adapters/tiger/adapter.d.ts +1 -1
  63. package/out/src/adapters/tiger/adapter.d.ts.map +1 -1
  64. package/out/src/adapters/tiger/adapter.js +2 -2
  65. package/out/src/adapters/tiger/adapter.js.map +1 -1
  66. package/out/src/adapters/tiger/street-decompose.d.ts.map +1 -1
  67. package/out/src/adapters/tiger/street-decompose.js +18 -27
  68. package/out/src/adapters/tiger/street-decompose.js.map +1 -1
  69. package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts +4 -4
  70. package/out/src/adapters/usgov-hrsa-fqhc/adapter.d.ts.map +1 -1
  71. package/out/src/adapters/usgov-hrsa-fqhc/adapter.js +64 -82
  72. package/out/src/adapters/usgov-hrsa-fqhc/adapter.js.map +1 -1
  73. package/out/src/adapters/usgov-imls-pls/adapter.d.ts +1 -1
  74. package/out/src/adapters/usgov-imls-pls/adapter.d.ts.map +1 -1
  75. package/out/src/adapters/usgov-imls-pls/adapter.js +62 -84
  76. package/out/src/adapters/usgov-imls-pls/adapter.js.map +1 -1
  77. package/out/src/adapters/usgov-irs-bmf/adapter.d.ts +1 -1
  78. package/out/src/adapters/usgov-irs-bmf/adapter.d.ts.map +1 -1
  79. package/out/src/adapters/usgov-irs-bmf/adapter.js +59 -66
  80. package/out/src/adapters/usgov-irs-bmf/adapter.js.map +1 -1
  81. package/out/src/adapters/usgov-nad/adapter.d.ts +1 -1
  82. package/out/src/adapters/usgov-nad/adapter.d.ts.map +1 -1
  83. package/out/src/adapters/usgov-nad/adapter.js +6 -9
  84. package/out/src/adapters/usgov-nad/adapter.js.map +1 -1
  85. package/out/src/adapters/usgov-nppes/adapter.d.ts +1 -1
  86. package/out/src/adapters/usgov-nppes/adapter.d.ts.map +1 -1
  87. package/out/src/adapters/usgov-nppes/adapter.js +60 -78
  88. package/out/src/adapters/usgov-nppes/adapter.js.map +1 -1
  89. package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.d.ts +1 -1
  90. package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.d.ts.map +1 -1
  91. package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js +53 -73
  92. package/out/src/adapters/usgov-samhsa-treatment-locator/adapter.js.map +1 -1
  93. package/out/src/{adapter.d.ts → adapters/utils/index.d.ts} +43 -2
  94. package/out/src/adapters/utils/index.d.ts.map +1 -0
  95. package/out/src/{adapter.js → adapters/utils/index.js} +47 -2
  96. package/out/src/adapters/utils/index.js.map +1 -0
  97. package/out/src/adapters/wof-admin-jp/adapter.d.ts +1 -1
  98. package/out/src/adapters/wof-admin-jp/adapter.d.ts.map +1 -1
  99. package/out/src/adapters/wof-admin-jp/adapter.js +24 -6
  100. package/out/src/adapters/wof-admin-jp/adapter.js.map +1 -1
  101. package/out/src/adapters/wof-admin-json/adapter.d.ts +2 -2
  102. package/out/src/adapters/wof-admin-json/adapter.d.ts.map +1 -1
  103. package/out/src/adapters/wof-admin-json/adapter.js +2 -2
  104. package/out/src/adapters/wof-admin-json/adapter.js.map +1 -1
  105. package/out/src/adapters/wof-postalcode-json/adapter.d.ts +2 -2
  106. package/out/src/adapters/wof-postalcode-json/adapter.d.ts.map +1 -1
  107. package/out/src/adapters/wof-postalcode-json/adapter.js +2 -2
  108. package/out/src/adapters/wof-postalcode-json/adapter.js.map +1 -1
  109. package/out/src/build.d.ts +3 -4
  110. package/out/src/build.d.ts.map +1 -1
  111. package/out/src/build.js +6 -8
  112. package/out/src/build.js.map +1 -1
  113. package/out/src/index.d.ts +3 -17
  114. package/out/src/index.d.ts.map +1 -1
  115. package/out/src/index.js +3 -17
  116. package/out/src/index.js.map +1 -1
  117. package/out/src/name-prone-us-suffixes.d.ts +12 -0
  118. package/out/src/name-prone-us-suffixes.d.ts.map +1 -0
  119. package/out/src/name-prone-us-suffixes.js +12 -0
  120. package/out/src/name-prone-us-suffixes.js.map +1 -0
  121. package/out/src/parquet-wrapper/reader.d.ts +8 -0
  122. package/out/src/parquet-wrapper/reader.d.ts.map +1 -1
  123. package/out/src/parquet-wrapper/reader.js +16 -0
  124. package/out/src/parquet-wrapper/reader.js.map +1 -1
  125. package/out/src/parquet-wrapper/schema.d.ts +4 -0
  126. package/out/src/parquet-wrapper/schema.d.ts.map +1 -1
  127. package/out/src/parquet-wrapper/schema.js.map +1 -1
  128. package/out/src/runner.d.ts +2 -2
  129. package/out/src/runner.d.ts.map +1 -1
  130. package/out/src/runner.js +1 -1
  131. package/out/src/runner.js.map +1 -1
  132. package/out/src/shard-recipes/anchor-absorption.d.ts.map +1 -1
  133. package/out/src/shard-recipes/anchor-absorption.js +5 -4
  134. package/out/src/shard-recipes/anchor-absorption.js.map +1 -1
  135. package/out/src/shard-recipes/boundary-stress.d.ts.map +1 -1
  136. package/out/src/shard-recipes/boundary-stress.js +2 -2
  137. package/out/src/shard-recipes/boundary-stress.js.map +1 -1
  138. package/out/src/shard-recipes/country-balanced.d.ts.map +1 -1
  139. package/out/src/shard-recipes/country-balanced.js +75 -75
  140. package/out/src/shard-recipes/country-balanced.js.map +1 -1
  141. package/out/src/shard-recipes/fr-admin-split.js +2 -2
  142. package/out/src/shard-recipes/fr-admin-split.js.map +1 -1
  143. package/out/src/shard-recipes/fr-fragment.d.ts.map +1 -1
  144. package/out/src/shard-recipes/fr-fragment.js +8 -5
  145. package/out/src/shard-recipes/fr-fragment.js.map +1 -1
  146. package/out/src/shard-recipes/fr-lieudit.js +2 -2
  147. package/out/src/shard-recipes/fr-lieudit.js.map +1 -1
  148. package/out/src/shard-recipes/fr-order.d.ts.map +1 -1
  149. package/out/src/shard-recipes/fr-order.js +18 -72
  150. package/out/src/shard-recipes/fr-order.js.map +1 -1
  151. package/out/src/shard-recipes/german.d.ts.map +1 -1
  152. package/out/src/shard-recipes/german.js +17 -70
  153. package/out/src/shard-recipes/german.js.map +1 -1
  154. package/out/src/shard-recipes/house-venue.d.ts.map +1 -1
  155. package/out/src/shard-recipes/house-venue.js +1 -1
  156. package/out/src/shard-recipes/house-venue.js.map +1 -1
  157. package/out/src/shard-recipes/index.d.ts.map +1 -1
  158. package/out/src/shard-recipes/index.js +4 -1
  159. package/out/src/shard-recipes/index.js.map +1 -1
  160. package/out/src/shard-recipes/intersection.d.ts +2 -2
  161. package/out/src/shard-recipes/intersection.d.ts.map +1 -1
  162. package/out/src/shard-recipes/intersection.js +28 -72
  163. package/out/src/shard-recipes/intersection.js.map +1 -1
  164. package/out/src/shard-recipes/locale.d.ts +5 -5
  165. package/out/src/shard-recipes/locale.d.ts.map +1 -1
  166. package/out/src/shard-recipes/locale.js +25 -35
  167. package/out/src/shard-recipes/locale.js.map +1 -1
  168. package/out/src/shard-recipes/no-fragment.d.ts.map +1 -1
  169. package/out/src/shard-recipes/no-fragment.js +8 -5
  170. package/out/src/shard-recipes/no-fragment.js.map +1 -1
  171. package/out/src/shard-recipes/no-street-led.d.ts.map +1 -1
  172. package/out/src/shard-recipes/no-street-led.js +8 -5
  173. package/out/src/shard-recipes/no-street-led.js.map +1 -1
  174. package/out/src/shard-recipes/no-street.d.ts.map +1 -1
  175. package/out/src/shard-recipes/no-street.js +1 -1
  176. package/out/src/shard-recipes/no-street.js.map +1 -1
  177. package/out/src/shard-recipes/po-box-cedex.d.ts +3 -3
  178. package/out/src/shard-recipes/po-box-cedex.d.ts.map +1 -1
  179. package/out/src/shard-recipes/po-box-cedex.js +101 -127
  180. package/out/src/shard-recipes/po-box-cedex.js.map +1 -1
  181. package/out/src/shard-recipes/po-box.d.ts.map +1 -1
  182. package/out/src/shard-recipes/po-box.js +1 -1
  183. package/out/src/shard-recipes/po-box.js.map +1 -1
  184. package/out/src/shard-recipes/scaffold.d.ts +58 -9
  185. package/out/src/shard-recipes/scaffold.d.ts.map +1 -1
  186. package/out/src/shard-recipes/scaffold.js +87 -29
  187. package/out/src/shard-recipes/scaffold.js.map +1 -1
  188. package/out/src/shard-recipes/street-affix.d.ts +54 -0
  189. package/out/src/shard-recipes/street-affix.d.ts.map +1 -1
  190. package/out/src/shard-recipes/street-affix.js +253 -111
  191. package/out/src/shard-recipes/street-affix.js.map +1 -1
  192. package/out/src/shard-recipes/street-bare.d.ts.map +1 -1
  193. package/out/src/shard-recipes/street-bare.js +3 -3
  194. package/out/src/shard-recipes/street-bare.js.map +1 -1
  195. package/out/src/shard-recipes/street.d.ts.map +1 -1
  196. package/out/src/shard-recipes/street.js +2 -2
  197. package/out/src/shard-recipes/street.js.map +1 -1
  198. package/out/src/shard-recipes/sub-venue-sources.d.ts +228 -0
  199. package/out/src/shard-recipes/sub-venue-sources.d.ts.map +1 -0
  200. package/out/src/shard-recipes/sub-venue-sources.js +664 -0
  201. package/out/src/shard-recipes/sub-venue-sources.js.map +1 -0
  202. package/out/src/shard-recipes/sub-venue.d.ts +269 -0
  203. package/out/src/shard-recipes/sub-venue.d.ts.map +1 -0
  204. package/out/src/shard-recipes/sub-venue.js +709 -0
  205. package/out/src/shard-recipes/sub-venue.js.map +1 -0
  206. package/out/src/shard-recipes/unit.d.ts +1 -1
  207. package/out/src/shard-recipes/unit.d.ts.map +1 -1
  208. package/out/src/shard-recipes/unit.js +26 -74
  209. package/out/src/shard-recipes/unit.js.map +1 -1
  210. package/out/src/{synthesize-anchor-absorption.d.ts → synthesizers/anchor-absorption.d.ts} +1 -1
  211. package/out/src/synthesizers/anchor-absorption.d.ts.map +1 -0
  212. package/out/src/{synthesize-anchor-absorption.js → synthesizers/anchor-absorption.js} +1 -1
  213. package/out/src/synthesizers/anchor-absorption.js.map +1 -0
  214. package/out/src/{synthesize-boundary-stress.d.ts → synthesizers/boundary-stress.d.ts} +3 -3
  215. package/out/src/synthesizers/boundary-stress.d.ts.map +1 -0
  216. package/out/src/{synthesize-boundary-stress.js → synthesizers/boundary-stress.js} +2 -2
  217. package/out/src/synthesizers/boundary-stress.js.map +1 -0
  218. package/out/src/{synthesize-german.d.ts → synthesizers/german.d.ts} +2 -2
  219. package/out/src/synthesizers/german.d.ts.map +1 -0
  220. package/out/src/{synthesize-german.js → synthesizers/german.js} +2 -2
  221. package/out/src/synthesizers/german.js.map +1 -0
  222. package/out/src/{synthesize-house-venue.d.ts → synthesizers/house-venue.d.ts} +4 -4
  223. package/out/src/synthesizers/house-venue.d.ts.map +1 -0
  224. package/out/src/{synthesize-house-venue.js → synthesizers/house-venue.js} +4 -4
  225. package/out/src/synthesizers/house-venue.js.map +1 -0
  226. package/out/src/{synthesize-intersection.d.ts → synthesizers/intersection.d.ts} +2 -2
  227. package/out/src/synthesizers/intersection.d.ts.map +1 -0
  228. package/out/src/{synthesize-intersection.js → synthesizers/intersection.js} +1 -1
  229. package/out/src/synthesizers/intersection.js.map +1 -0
  230. package/out/src/{synthesize-no-street.d.ts → synthesizers/no-street.d.ts} +3 -3
  231. package/out/src/synthesizers/no-street.d.ts.map +1 -0
  232. package/out/src/{synthesize-no-street.js → synthesizers/no-street.js} +2 -2
  233. package/out/src/synthesizers/no-street.js.map +1 -0
  234. package/out/src/{synthesize-po-box.d.ts → synthesizers/po-box.d.ts} +2 -2
  235. package/out/src/synthesizers/po-box.d.ts.map +1 -0
  236. package/out/src/{synthesize-po-box.js → synthesizers/po-box.js} +1 -1
  237. package/out/src/synthesizers/po-box.js.map +1 -0
  238. package/out/src/{synthesize-street.d.ts → synthesizers/street.d.ts} +2 -2
  239. package/out/src/synthesizers/street.d.ts.map +1 -0
  240. package/out/src/{synthesize-street.js → synthesizers/street.js} +2 -2
  241. package/out/src/synthesizers/street.js.map +1 -0
  242. package/out/src/{synthesize.d.ts → synthesizers/utils.d.ts} +3 -3
  243. package/out/src/synthesizers/utils.d.ts.map +1 -0
  244. package/out/src/{synthesize.js → synthesizers/utils.js} +4 -15
  245. package/out/src/synthesizers/utils.js.map +1 -0
  246. package/out/src/tools/align-shard.d.ts.map +1 -1
  247. package/out/src/tools/align-shard.js +4 -8
  248. package/out/src/tools/align-shard.js.map +1 -1
  249. package/out/src/tools/audit.d.ts.map +1 -1
  250. package/out/src/tools/audit.js +4 -4
  251. package/out/src/tools/audit.js.map +1 -1
  252. package/out/src/tools/corpus-stats.d.ts +2 -2
  253. package/out/src/tools/corpus-stats.d.ts.map +1 -1
  254. package/out/src/tools/corpus-stats.js +18 -31
  255. package/out/src/tools/corpus-stats.js.map +1 -1
  256. package/out/src/tools/fetch/download.d.ts.map +1 -1
  257. package/out/src/tools/fetch/download.js +5 -6
  258. package/out/src/tools/fetch/download.js.map +1 -1
  259. package/out/src/tools/fetch/imls-pls.d.ts.map +1 -1
  260. package/out/src/tools/fetch/imls-pls.js +3 -15
  261. package/out/src/tools/fetch/imls-pls.js.map +1 -1
  262. package/out/src/tools/fetch/index.d.ts +14 -1
  263. package/out/src/tools/fetch/index.d.ts.map +1 -1
  264. package/out/src/tools/fetch/index.js +14 -1
  265. package/out/src/tools/fetch/index.js.map +1 -1
  266. package/out/src/tools/fetch/nad.d.ts +1 -1
  267. package/out/src/tools/fetch/nad.js +1 -1
  268. package/out/src/tools/fetch/nppes.d.ts.map +1 -1
  269. package/out/src/tools/fetch/nppes.js +8 -14
  270. package/out/src/tools/fetch/nppes.js.map +1 -1
  271. package/out/src/tools/fetch/openaddresses.d.ts +1 -1
  272. package/out/src/tools/fetch/openaddresses.js +1 -1
  273. package/out/src/tools/fetch/ourairports.d.ts +45 -0
  274. package/out/src/tools/fetch/ourairports.d.ts.map +1 -0
  275. package/out/src/tools/fetch/ourairports.js +123 -0
  276. package/out/src/tools/fetch/ourairports.js.map +1 -0
  277. package/out/src/tools/fetch/wikidata-subvenue.d.ts +171 -0
  278. package/out/src/tools/fetch/wikidata-subvenue.d.ts.map +1 -0
  279. package/out/src/tools/fetch/wikidata-subvenue.js +275 -0
  280. package/out/src/tools/fetch/wikidata-subvenue.js.map +1 -0
  281. package/out/src/tools/golden-expand.d.ts.map +1 -1
  282. package/out/src/tools/golden-expand.js +18 -16
  283. package/out/src/tools/golden-expand.js.map +1 -1
  284. package/out/src/tools/golden-promote.d.ts.map +1 -1
  285. package/out/src/tools/golden-promote.js +12 -6
  286. package/out/src/tools/golden-promote.js.map +1 -1
  287. package/out/src/tools/golden-relabel-street.d.ts +196 -0
  288. package/out/src/tools/golden-relabel-street.d.ts.map +1 -0
  289. package/out/src/tools/golden-relabel-street.js +440 -0
  290. package/out/src/tools/golden-relabel-street.js.map +1 -0
  291. package/out/src/tools/index.d.ts +4 -0
  292. package/out/src/tools/index.d.ts.map +1 -1
  293. package/out/src/tools/index.js +4 -0
  294. package/out/src/tools/index.js.map +1 -1
  295. package/out/src/tools/jsonl-to-parquet.d.ts.map +1 -1
  296. package/out/src/tools/jsonl-to-parquet.js +3 -2
  297. package/out/src/tools/jsonl-to-parquet.js.map +1 -1
  298. package/out/src/tools/lint-shard-vocab.d.ts.map +1 -1
  299. package/out/src/tools/lint-shard-vocab.js +3 -2
  300. package/out/src/tools/lint-shard-vocab.js.map +1 -1
  301. package/out/src/tools/lint-shard.d.ts +3 -3
  302. package/out/src/tools/lint-shard.d.ts.map +1 -1
  303. package/out/src/tools/lint-shard.js +23 -32
  304. package/out/src/tools/lint-shard.js.map +1 -1
  305. package/out/src/tools/overlay-manifest.d.ts.map +1 -1
  306. package/out/src/tools/overlay-manifest.js +2 -1
  307. package/out/src/tools/overlay-manifest.js.map +1 -1
  308. package/out/src/tools/overture-subvenue.d.ts +112 -0
  309. package/out/src/tools/overture-subvenue.d.ts.map +1 -0
  310. package/out/src/tools/overture-subvenue.js +144 -0
  311. package/out/src/tools/overture-subvenue.js.map +1 -0
  312. package/out/src/tools/shard-kryptonite.d.ts +1 -1
  313. package/out/src/tools/shard-kryptonite.d.ts.map +1 -1
  314. package/out/src/tools/shard-kryptonite.js +6 -6
  315. package/out/src/tools/shard-kryptonite.js.map +1 -1
  316. package/out/src/tools/shard-translit.d.ts +6 -3
  317. package/out/src/tools/shard-translit.d.ts.map +1 -1
  318. package/out/src/tools/shard-translit.js +9 -8
  319. package/out/src/tools/shard-translit.js.map +1 -1
  320. package/out/src/tools/sub-venue-lexicon.d.ts +507 -0
  321. package/out/src/tools/sub-venue-lexicon.d.ts.map +1 -0
  322. package/out/src/tools/sub-venue-lexicon.js +817 -0
  323. package/out/src/tools/sub-venue-lexicon.js.map +1 -0
  324. package/out/src/tools/sub-venue-promotions.d.ts +94 -0
  325. package/out/src/tools/sub-venue-promotions.d.ts.map +1 -0
  326. package/out/src/tools/sub-venue-promotions.js +266 -0
  327. package/out/src/tools/sub-venue-promotions.js.map +1 -0
  328. package/out/src/{align.d.ts → utils/align.d.ts} +1 -1
  329. package/out/src/utils/align.d.ts.map +1 -0
  330. package/out/src/utils/align.js.map +1 -0
  331. package/out/src/{golden.d.ts → utils/golden.d.ts} +5 -0
  332. package/out/src/utils/golden.d.ts.map +1 -0
  333. package/out/src/{golden.js → utils/golden.js} +24 -10
  334. package/out/src/utils/golden.js.map +1 -0
  335. package/out/src/utils/index.d.ts +14 -0
  336. package/out/src/utils/index.d.ts.map +1 -0
  337. package/out/src/utils/index.js +14 -0
  338. package/out/src/utils/index.js.map +1 -0
  339. package/out/src/utils/license.d.ts.map +1 -0
  340. package/out/src/utils/license.js.map +1 -0
  341. package/out/src/{parquet.d.ts → utils/parquet.d.ts} +2 -2
  342. package/out/src/utils/parquet.d.ts.map +1 -0
  343. package/out/src/{parquet.js → utils/parquet.js} +3 -15
  344. package/out/src/utils/parquet.js.map +1 -0
  345. package/out/src/{split.d.ts → utils/split.d.ts} +8 -1
  346. package/out/src/utils/split.d.ts.map +1 -0
  347. package/out/src/{split.js → utils/split.js} +7 -0
  348. package/out/src/utils/split.js.map +1 -0
  349. package/out/src/utils/tokenize.d.ts.map +1 -0
  350. package/out/src/utils/tokenize.js.map +1 -0
  351. package/out/src/{wof-json.d.ts → utils/wof-json.d.ts} +2 -2
  352. package/out/src/utils/wof-json.d.ts.map +1 -0
  353. package/out/src/{wof-json.js → utils/wof-json.js} +5 -8
  354. package/out/src/utils/wof-json.js.map +1 -0
  355. package/package.json +66 -20
  356. package/src/adapters/ban/adapter.ts +58 -68
  357. package/src/adapters/ban/street-decompose.ts +17 -25
  358. package/src/adapters/fcc-bdc/adapter.ts +5 -36
  359. package/src/adapters/geonames/adapter.ts +67 -76
  360. package/src/adapters/geonames-postal/adapter.ts +45 -54
  361. package/src/adapters/gnaf/adapter.ts +7 -11
  362. package/src/adapters/gnaf/assemble.ts +7 -11
  363. package/src/adapters/index.ts +3 -2
  364. package/src/adapters/openaddresses/adapter.ts +6 -12
  365. package/src/adapters/overture/adapter.ts +6 -10
  366. package/src/adapters/state-hi-schools/adapter.ts +53 -77
  367. package/src/adapters/state-ia-contractors/adapter.ts +55 -80
  368. package/src/adapters/state-ny-notaries/adapter.ts +64 -89
  369. package/src/adapters/state-tx-notaries/adapter.ts +63 -88
  370. package/src/adapters/synth-po-box/adapter.ts +12 -23
  371. package/src/adapters/tiger/adapter.ts +4 -3
  372. package/src/adapters/tiger/street-decompose.ts +21 -31
  373. package/src/adapters/usgov-hrsa-fqhc/adapter.ts +64 -88
  374. package/src/adapters/usgov-imls-pls/adapter.ts +57 -86
  375. package/src/adapters/usgov-irs-bmf/adapter.ts +55 -66
  376. package/src/adapters/usgov-nad/adapter.ts +7 -11
  377. package/src/adapters/usgov-nppes/adapter.ts +54 -79
  378. package/src/adapters/usgov-samhsa-treatment-locator/adapter.ts +48 -75
  379. package/src/{adapter.ts → adapters/utils/index.ts} +59 -3
  380. package/src/adapters/wof-admin-jp/adapter.ts +35 -9
  381. package/src/adapters/wof-admin-json/adapter.ts +3 -4
  382. package/src/adapters/wof-postalcode-json/adapter.ts +3 -4
  383. package/src/build.ts +12 -11
  384. package/src/index.ts +3 -17
  385. package/src/name-prone-us-suffixes.ts +12 -0
  386. package/src/parquet-wrapper/reader.ts +19 -0
  387. package/src/parquet-wrapper/schema.ts +4 -0
  388. package/src/runner.ts +7 -2
  389. package/src/shard-recipes/anchor-absorption.ts +5 -4
  390. package/src/shard-recipes/boundary-stress.ts +6 -2
  391. package/src/shard-recipes/country-balanced.ts +77 -91
  392. package/src/shard-recipes/fr-admin-split.ts +3 -3
  393. package/src/shard-recipes/fr-fragment.ts +10 -7
  394. package/src/shard-recipes/fr-lieudit.ts +3 -3
  395. package/src/shard-recipes/fr-order.ts +18 -83
  396. package/src/shard-recipes/german.ts +17 -80
  397. package/src/shard-recipes/house-venue.ts +2 -1
  398. package/src/shard-recipes/index.ts +4 -1
  399. package/src/shard-recipes/intersection.ts +37 -74
  400. package/src/shard-recipes/locale.ts +31 -39
  401. package/src/shard-recipes/no-fragment.ts +10 -7
  402. package/src/shard-recipes/no-street-led.ts +10 -7
  403. package/src/shard-recipes/no-street.ts +2 -1
  404. package/src/shard-recipes/po-box-cedex.ts +127 -161
  405. package/src/shard-recipes/po-box.ts +6 -1
  406. package/src/shard-recipes/scaffold.ts +125 -27
  407. package/src/shard-recipes/street-affix.ts +335 -134
  408. package/src/shard-recipes/street-bare.ts +4 -3
  409. package/src/shard-recipes/street.ts +3 -2
  410. package/src/shard-recipes/sub-venue-sources.ts +895 -0
  411. package/src/shard-recipes/sub-venue.ts +982 -0
  412. package/src/shard-recipes/unit.ts +27 -85
  413. package/src/{synthesize-boundary-stress.ts → synthesizers/boundary-stress.ts} +2 -2
  414. package/src/{synthesize-german.ts → synthesizers/german.ts} +2 -2
  415. package/src/{synthesize-house-venue.ts → synthesizers/house-venue.ts} +4 -4
  416. package/src/{synthesize-intersection.ts → synthesizers/intersection.ts} +1 -1
  417. package/src/{synthesize-no-street.ts → synthesizers/no-street.ts} +2 -2
  418. package/src/{synthesize-po-box.ts → synthesizers/po-box.ts} +1 -1
  419. package/src/{synthesize-street.ts → synthesizers/street.ts} +2 -2
  420. package/src/{synthesize.ts → synthesizers/utils.ts} +4 -18
  421. package/src/tools/align-shard.ts +4 -7
  422. package/src/tools/audit.ts +6 -5
  423. package/src/tools/corpus-stats.ts +25 -33
  424. package/src/tools/fetch/download.ts +7 -5
  425. package/src/tools/fetch/imls-pls.ts +4 -18
  426. package/src/tools/fetch/index.ts +14 -1
  427. package/src/tools/fetch/nad.ts +1 -1
  428. package/src/tools/fetch/nppes.ts +9 -16
  429. package/src/tools/fetch/openaddresses.ts +1 -1
  430. package/src/tools/fetch/ourairports.ts +166 -0
  431. package/src/tools/fetch/wikidata-subvenue.ts +386 -0
  432. package/src/tools/golden-expand.ts +18 -13
  433. package/src/tools/golden-promote.ts +15 -6
  434. package/src/tools/golden-relabel-street.ts +685 -0
  435. package/src/tools/index.ts +4 -0
  436. package/src/tools/jsonl-to-parquet.ts +3 -2
  437. package/src/tools/lint-shard-vocab.ts +3 -2
  438. package/src/tools/lint-shard.ts +33 -36
  439. package/src/tools/overlay-manifest.ts +2 -1
  440. package/src/tools/overture-subvenue.ts +215 -0
  441. package/src/tools/shard-kryptonite.ts +8 -9
  442. package/src/tools/shard-translit.ts +22 -11
  443. package/src/tools/sub-venue-lexicon.ts +1250 -0
  444. package/src/tools/sub-venue-promotions.ts +330 -0
  445. package/src/{align.ts → utils/align.ts} +1 -1
  446. package/src/{golden.ts → utils/golden.ts} +25 -11
  447. package/src/utils/index.ts +14 -0
  448. package/src/{parquet.ts → utils/parquet.ts} +5 -19
  449. package/src/{split.ts → utils/split.ts} +8 -2
  450. package/src/{wof-json.ts → utils/wof-json.ts} +5 -8
  451. package/out/src/adapter.d.ts.map +0 -1
  452. package/out/src/adapter.js.map +0 -1
  453. package/out/src/align.d.ts.map +0 -1
  454. package/out/src/align.js.map +0 -1
  455. package/out/src/format.d.ts +0 -14
  456. package/out/src/format.d.ts.map +0 -1
  457. package/out/src/format.js +0 -14
  458. package/out/src/format.js.map +0 -1
  459. package/out/src/golden.d.ts.map +0 -1
  460. package/out/src/golden.js.map +0 -1
  461. package/out/src/license.d.ts.map +0 -1
  462. package/out/src/license.js.map +0 -1
  463. package/out/src/parquet.d.ts.map +0 -1
  464. package/out/src/parquet.js.map +0 -1
  465. package/out/src/split.d.ts.map +0 -1
  466. package/out/src/split.js.map +0 -1
  467. package/out/src/synthesize-anchor-absorption.d.ts.map +0 -1
  468. package/out/src/synthesize-anchor-absorption.js.map +0 -1
  469. package/out/src/synthesize-boundary-stress.d.ts.map +0 -1
  470. package/out/src/synthesize-boundary-stress.js.map +0 -1
  471. package/out/src/synthesize-german.d.ts.map +0 -1
  472. package/out/src/synthesize-german.js.map +0 -1
  473. package/out/src/synthesize-house-venue.d.ts.map +0 -1
  474. package/out/src/synthesize-house-venue.js.map +0 -1
  475. package/out/src/synthesize-intersection.d.ts.map +0 -1
  476. package/out/src/synthesize-intersection.js.map +0 -1
  477. package/out/src/synthesize-no-street.d.ts.map +0 -1
  478. package/out/src/synthesize-no-street.js.map +0 -1
  479. package/out/src/synthesize-po-box.d.ts.map +0 -1
  480. package/out/src/synthesize-po-box.js.map +0 -1
  481. package/out/src/synthesize-street.d.ts.map +0 -1
  482. package/out/src/synthesize-street.js.map +0 -1
  483. package/out/src/synthesize.d.ts.map +0 -1
  484. package/out/src/synthesize.js.map +0 -1
  485. package/out/src/tokenize.d.ts.map +0 -1
  486. package/out/src/tokenize.js.map +0 -1
  487. package/out/src/wof-json.d.ts.map +0 -1
  488. package/out/src/wof-json.js.map +0 -1
  489. package/src/format.ts +0 -14
  490. /package/out/src/{align.js → utils/align.js} +0 -0
  491. /package/out/src/{license.d.ts → utils/license.d.ts} +0 -0
  492. /package/out/src/{license.js → utils/license.js} +0 -0
  493. /package/out/src/{tokenize.d.ts → utils/tokenize.d.ts} +0 -0
  494. /package/out/src/{tokenize.js → utils/tokenize.js} +0 -0
  495. /package/src/{synthesize-anchor-absorption.ts → synthesizers/anchor-absorption.ts} +0 -0
  496. /package/src/{license.ts → utils/license.ts} +0 -0
  497. /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. Phase 1.6 §1.2 (#22) selects this source for its adversarial-value-per-row: every
12
- * facility name is a human-typed venue string and the addresses pass through enough hands to
13
- * accumulate the abbreviation drift + suite designator chaos that pure gazetteer data does not.
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 { createReadStream } from "node:fs";
32
- import { parse as csvParse } from "csv-parse";
33
- import { stableSourceID } from "../../adapter.js";
34
- import { lookupStateAbbreviation } from "../../codex/us-fips-state.js";
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` for Phase 1 — Mailwoman's `unit` component exists but the
51
- * address-formatter does not have a clean slot for it, and HRSA addresses do not separate the suite into its own
52
- * column. Leaving the surface form intact in `street` preserves the adversarial training signal (the model learns that
53
- * a trailing "Suite 4" is part of the road line in this distribution).
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 stream = createReadStream(opts.inputPath, { encoding: "utf8" });
88
- const parser = stream.pipe(csvParse({
89
- columns: true,
90
- skip_empty_lines: true,
91
- relax_quotes: true,
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
- try {
96
- for await (const record of parser) {
97
- if (opts.signal?.aborted)
98
- break;
99
- if (opts.limit !== undefined && emitted >= opts.limit)
100
- break;
101
- const venue = (record["Site Name"] ?? "").trim();
102
- const split = splitAddress(record["Site Address"] ?? "");
103
- const city = (record["Site City"] ?? "").trim();
104
- const stateAbbr = (record["Site State Abbreviation"] ?? "").trim();
105
- const postcode = (record["Site Postal Code"] ?? "").trim();
106
- if (!venue || !split || !city || !postcode)
107
- continue;
108
- const state = lookupStateAbbreviation(stateAbbr);
109
- if (!state)
110
- continue;
111
- // Insertion order matters here. `venue` first so alignment claims its span
112
- // (which may contain a token like "Buffalo") before `locality` runs its
113
- // search — the kryptonite case `Buffalo Health Clinic, Buffalo NY`
114
- // otherwise mis-labels the venue's "Buffalo" as locality.
115
- const components = {
116
- venue,
117
- ...(split.house_number ? { house_number: split.house_number } : {}),
118
- street: split.street,
119
- locality: city,
120
- region: state.abbreviation,
121
- postcode,
122
- };
123
- const raw = composeRaw(venue, split.house_number, split.street, city, state.abbreviation, postcode);
124
- if (!raw)
125
- continue;
126
- const aligned = reconcileComponents(components, raw);
127
- if (!Object.keys(aligned).length)
128
- continue;
129
- const siteID = (record["Site ID"] ?? "").trim();
130
- const sourceID = siteID
131
- ? `${USGOV_HRSA_FQHC_ADAPTER_ID}-${siteID}`
132
- : stableSourceID(USGOV_HRSA_FQHC_ADAPTER_ID, aligned);
133
- yield {
134
- raw,
135
- components: aligned,
136
- country: "US",
137
- locale: "en-US",
138
- source: USGOV_HRSA_FQHC_ADAPTER_ID,
139
- source_id: sourceID,
140
- corpus_version: "",
141
- license: USGOV_HRSA_FQHC_DEFAULT_LICENSE,
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,gBAAgB,EAAE,MAAM,SAAS,CAAA;AAE1C,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,WAAW,CAAA;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAA;AACjD,OAAO,EAAE,uBAAuB,EAAE,MAAM,8BAA8B,CAAA;AACtE,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAGrD;;;GAGG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAG,iBAAiB,CAAA;AAC3D;;;GAGG;AACH,MAAM,CAAC,MAAM,+BAA+B,GAAG,eAAe,CAAA;AAmB9D;;;;;;;;GAQG;AACH,MAAM,mBAAmB,GAAG,kCAAkC,CAAA;AAE9D,SAAS,YAAY,CAAC,OAAe;IACpC,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IAE9B,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IACzB,MAAM,CAAC,GAAG,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAE3C,IAAI,CAAC;QAAE,OAAO,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,EAAE,CAAA;IAE1D,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;AAC3B,CAAC;AAED;;;;;;;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,MAAM,GAAG,gBAAgB,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;YAErE,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CACzB,QAAQ,CAAC;gBACR,OAAO,EAAE,IAAI;gBACb,gBAAgB,EAAE,IAAI;gBACtB,YAAY,EAAE,IAAI;gBAClB,kBAAkB,EAAE,IAAI;aACxB,CAAC,CACF,CAAA;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,CAAC;gBACJ,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,MAAoC,EAAE,CAAC;oBACjE,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAE5D,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAChD,MAAM,KAAK,GAAG,YAAY,CAAC,MAAM,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAA;oBACxD,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAC/C,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,yBAAyB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAClE,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,kBAAkB,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAE1D,IAAI,CAAC,KAAK,IAAI,CAAC,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,QAAQ;wBAAE,SAAQ;oBACpD,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;oBAEhD,IAAI,CAAC,KAAK;wBAAE,SAAQ;oBAEpB,2EAA2E;oBAC3E,wEAAwE;oBACxE,mEAAmE;oBACnE,0DAA0D;oBAC1D,MAAM,UAAU,GAA+B;wBAC9C,KAAK;wBACL,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;wBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;wBACpB,QAAQ,EAAE,IAAI;wBACd,MAAM,EAAE,KAAK,CAAC,YAAY;wBAC1B,QAAQ;qBACR,CAAA;oBAED,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,EAAE,KAAK,CAAC,YAAY,EAAE,KAAK,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,CAAC,YAAY,EAAE,QAAQ,CAAC,CAAA;oBAEnG,IAAI,CAAC,GAAG;wBAAE,SAAQ;oBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;oBAEpD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM;wBAAE,SAAQ;oBAE1C,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAE/C,MAAM,QAAQ,GAAG,MAAM;wBACtB,CAAC,CAAC,GAAG,0BAA0B,IAAI,MAAM,EAAE;wBAC3C,CAAC,CAAC,cAAc,CAAC,0BAA0B,EAAE,OAAO,CAAC,CAAA;oBAEtD,MAAM;wBACL,GAAG;wBACH,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,IAAI;wBACb,MAAM,EAAE,OAAO;wBACf,MAAM,EAAE,0BAA0B;wBAClC,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,+BAA+B;qBACxC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;oBAAS,CAAC;gBACV,MAAM,CAAC,OAAO,EAAE,CAAA;YACjB,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG,0BAA0B,EAAE,CAAA"}
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 "../../types.ts";
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;AASH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,gBAAgB,CAAA;AAEjF;;;GAGG;AACH,eAAO,MAAM,yBAAyB,mBAAmB,CAAA;AACzD;;;GAGG;AACH,eAAO,MAAM,8BAA8B,kBAAkB,CAAA;AAyB7D,wBAAgB,yBAAyB,IAAI,aAAa,CAkGzD;AAED;;GAEG;AACH,eAAO,MAAM,mBAAmB,eAA8B,CAAA"}
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 { createReadStream } from "node:fs";
21
- import { parse as csvParse } from "csv-parse";
22
- import { stableSourceID } from "../../adapter.js";
23
- import { lookupStateAbbreviation } from "../../codex/us-fips-state.js";
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 stream = createReadStream(opts.inputPath, { encoding: "utf8" });
55
- const parser = stream.pipe(csvParse({
56
- columns: true,
57
- skip_empty_lines: true,
58
- relax_quotes: true,
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
- try {
63
- for await (const record of parser) {
64
- if (opts.signal?.aborted)
65
- break;
66
- if (opts.limit !== undefined && emitted >= opts.limit)
67
- break;
68
- const libName = (record.LIBNAME ?? "").trim();
69
- const address = (record.ADDRESS ?? "").trim();
70
- const city = (record.CITY ?? "").trim();
71
- const zip = (record.ZIP ?? "").trim();
72
- const stateAbbr = (record.STABR ?? "").trim();
73
- const county = (record.CNTY ?? "").trim();
74
- if (!libName || !city || !zip)
75
- continue;
76
- const state = lookupStateAbbreviation(stateAbbr);
77
- if (!state)
78
- continue;
79
- const split = splitAddress(address);
80
- if (!split)
81
- continue;
82
- const components = {
83
- venue: libName,
84
- ...(split.house_number ? { house_number: split.house_number } : {}),
85
- street: split.street,
86
- locality: city,
87
- region: state.abbreviation,
88
- postcode: zip,
89
- // #552: no subregion — US postal addresses don't surface the county, so emitting
90
- // subregion creates a phantom component with no raw-span to align to, quarantining
91
- // ~21% of rows. The county is still available in the source CSV; it just isn't
92
- // a postal-surface component here.
93
- };
94
- const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim();
95
- const raw = [
96
- libName,
97
- streetPart,
98
- [city, [stateAbbr, zip].filter(Boolean).join(" ")].filter(Boolean).join(", "),
99
- ]
100
- .filter(Boolean)
101
- .join(", ");
102
- const aligned = reconcileComponents(components, raw);
103
- if (Object.keys(aligned).length <= 2)
104
- continue;
105
- const fscsKey = (record.FSCSKEY ?? "").trim();
106
- const sourceID = fscsKey
107
- ? `${USGOV_IMLS_PLS_ADAPTER_ID}-${fscsKey}`
108
- : stableSourceID(USGOV_IMLS_PLS_ADAPTER_ID, aligned);
109
- yield {
110
- raw,
111
- components: aligned,
112
- country: "US",
113
- locale: "en-US",
114
- source: USGOV_IMLS_PLS_ADAPTER_ID,
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,gBAAgB,EAAE,MAAM,SAAS,CAAA;AAE1C,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,WAAW,CAAA;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAA;AACjD,OAAO,EAAE,uBAAuB,EAAE,MAAM,8BAA8B,CAAA;AACtE,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAGrD;;;GAGG;AACH,MAAM,CAAC,MAAM,yBAAyB,GAAG,gBAAgB,CAAA;AACzD;;;GAGG;AACH,MAAM,CAAC,MAAM,8BAA8B,GAAG,eAAe,CAAA;AAE7D,MAAM,mBAAmB,GAAG,kCAAkC,CAAA;AAY9D,SAAS,YAAY,CAAC,OAAe;IACpC,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,EAAE,CAAA;IAE9B,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IACzB,MAAM,CAAC,GAAG,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAE3C,IAAI,CAAC;QAAE,OAAO,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,EAAE,CAAA;IAE1D,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;AAC3B,CAAC;AAED,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,MAAM,GAAG,gBAAgB,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;YAErE,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CACzB,QAAQ,CAAC;gBACR,OAAO,EAAE,IAAI;gBACb,gBAAgB,EAAE,IAAI;gBACtB,YAAY,EAAE,IAAI;gBAClB,kBAAkB,EAAE,IAAI;aACxB,CAAC,CACF,CAAA;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,CAAC;gBACJ,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,MAAsC,EAAE,CAAC;oBACnE,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAE5D,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAC7C,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAC7C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBACvC,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBACrC,MAAM,SAAS,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAC7C,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAEzC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,IAAI,CAAC,GAAG;wBAAE,SAAQ;oBAEvC,MAAM,KAAK,GAAG,uBAAuB,CAAC,SAAS,CAAC,CAAA;oBAEhD,IAAI,CAAC,KAAK;wBAAE,SAAQ;oBAEpB,MAAM,KAAK,GAAG,YAAY,CAAC,OAAO,CAAC,CAAA;oBAEnC,IAAI,CAAC,KAAK;wBAAE,SAAQ;oBAEpB,MAAM,UAAU,GAA+B;wBAC9C,KAAK,EAAE,OAAO;wBACd,GAAG,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC,CAAC,EAAE,YAAY,EAAE,KAAK,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;wBACnE,MAAM,EAAE,KAAK,CAAC,MAAM;wBACpB,QAAQ,EAAE,IAAI;wBACd,MAAM,EAAE,KAAK,CAAC,YAAY;wBAC1B,QAAQ,EAAE,GAAG;wBACb,iFAAiF;wBACjF,mFAAmF;wBACnF,+EAA+E;wBAC/E,mCAAmC;qBACnC,CAAA;oBAED,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;oBAEtF,MAAM,GAAG,GAAG;wBACX,OAAO;wBACP,UAAU;wBACV,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;qBAC7E;yBACC,MAAM,CAAC,OAAO,CAAC;yBACf,IAAI,CAAC,IAAI,CAAC,CAAA;oBAEZ,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;oBAEpD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,IAAI,CAAC;wBAAE,SAAQ;oBAE9C,MAAM,OAAO,GAAG,CAAC,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAE7C,MAAM,QAAQ,GAAG,OAAO;wBACvB,CAAC,CAAC,GAAG,yBAAyB,IAAI,OAAO,EAAE;wBAC3C,CAAC,CAAC,cAAc,CAAC,yBAAyB,EAAE,OAAO,CAAC,CAAA;oBAErD,MAAM;wBACL,GAAG;wBACH,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,IAAI;wBACb,MAAM,EAAE,OAAO;wBACf,MAAM,EAAE,yBAAyB;wBACjC,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,8BAA8B;qBACvC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;oBAAS,CAAC;gBACV,MAAM,CAAC,OAAO,EAAE,CAAA;YACjB,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,EAAE,CAAA"}
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 "../../types.ts";
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;AAQH,OAAO,KAAK,EAAgC,aAAa,EAAE,MAAM,gBAAgB,CAAA;AAEjF;;;GAGG;AACH,eAAO,MAAM,wBAAwB,kBAAkB,CAAA;AACvD;;;GAGG;AACH,eAAO,MAAM,6BAA6B,kBAAkB,CAAA;AA6C5D,wBAAgB,wBAAwB,IAAI,aAAa,CAmFxD;AAED;;GAEG;AACH,eAAO,MAAM,kBAAkB,eAA6B,CAAA"}
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 { createReadStream } from "node:fs";
22
- import { parse as csvParse } from "csv-parse";
23
- import { stableSourceID } from "../../adapter.js";
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+(.+)$/;
36
34
  /**
37
35
  * PO box in its many written forms: "PO BOX 12", "P.O. BOX 12", "P O BOX 12", "POB 12", "BOX 12".
38
36
  */
39
37
  const PO_BOX = /^\s*(?:P\.?\s?O\.?\s*BOX|POB|BOX)\s+\w/i;
40
38
  /**
41
39
  * Classify the street line into a `po_box` or a `{house_number?, street}` split.
40
+ *
41
+ * BMF mixes street addresses and PO boxes in one `STREET` column, so the PO-box shapes have to be claimed BEFORE the
42
+ * shared house-number split runs — otherwise `splitStreetLine` would hand back `"PO Box 1234"` as a plain street, which
43
+ * is correct for every other US adapter and wrong here.
42
44
  */
43
- function splitStreetLine(street) {
45
+ function splitStreetLineOrPOBox(street) {
44
46
  const trimmed = street.trim();
45
- if (!trimmed)
46
- return null;
47
47
  if (PO_BOX.test(trimmed))
48
48
  return { po_box: trimmed };
49
- const m = HOUSE_NUMBER_PREFIX.exec(trimmed);
50
- if (m)
51
- return { house_number: m[1], street: m[2].trim() };
52
- return { street: trimmed };
49
+ return splitStreetLine(trimmed);
53
50
  }
54
51
  function composeRaw(venue, streetPart, city, state, postcode) {
55
52
  const cityPart = [city.trim(), [state, postcode].filter(Boolean).join(" ").trim()].filter(Boolean).join(", ");
@@ -64,61 +61,57 @@ export function createUsgovIrsBmfAdapter() {
64
61
  if (opts.country && opts.country !== "US") {
65
62
  throw new Error(`usgov-irs-bmf adapter: only US supported, got country=${opts.country}`);
66
63
  }
67
- const stream = createReadStream(opts.inputPath, { encoding: "utf8" });
68
- const parser = stream.pipe(csvParse({ columns: true, skip_empty_lines: true, relax_quotes: true, relax_column_count: true, trim: true }));
64
+ const rows = CSVSpliterator.fromAsync(opts.inputPath, {
65
+ mode: "object",
66
+ normalizeKeys: false,
67
+ enableQuoteHandling: true,
68
+ });
69
69
  let emitted = 0;
70
- try {
71
- for await (const record of parser) {
72
- if (opts.signal?.aborted)
73
- break;
74
- if (opts.limit !== undefined && emitted >= opts.limit)
75
- break;
76
- const ein = (record.EIN ?? "").trim();
77
- const venue = (record.NAME ?? "").trim() || undefined;
78
- const street = (record.STREET ?? "").trim();
79
- const city = (record.CITY ?? "").trim();
80
- const state = (record.STATE ?? "").trim();
81
- const zipRaw = (record.ZIP ?? "").trim();
82
- if (!city || !zipRaw)
83
- continue;
84
- const postcode = zipRaw.split("-")[0].trim(); // 5-digit; drop the optional +4
85
- const split = splitStreetLine(street);
86
- if (!split)
87
- continue;
88
- const streetPart = "po_box" in split ? split.po_box : [split.house_number, split.street].filter(Boolean).join(" ");
89
- const components = {
90
- ...(venue ? { venue } : {}),
91
- ...("po_box" in split
92
- ? { po_box: split.po_box }
93
- : { ...(split.house_number ? { house_number: split.house_number } : {}), street: split.street }),
94
- locality: city,
95
- ...(state ? { region: state } : {}),
96
- postcode,
97
- };
98
- const raw = composeRaw(venue, streetPart, city, state, postcode);
99
- if (!raw)
100
- continue;
101
- const aligned = reconcileComponents(components, raw);
102
- if (Object.keys(aligned).length <= 2)
103
- continue;
104
- const sourceID = ein
105
- ? `${USGOV_IRS_BMF_ADAPTER_ID}-${ein}`
106
- : stableSourceID(USGOV_IRS_BMF_ADAPTER_ID, aligned);
107
- yield {
108
- raw,
109
- components: aligned,
110
- country: "US",
111
- locale: "en-US",
112
- source: USGOV_IRS_BMF_ADAPTER_ID,
113
- source_id: sourceID,
114
- corpus_version: "",
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,gBAAgB,EAAE,MAAM,SAAS,CAAA;AAE1C,OAAO,EAAE,KAAK,IAAI,QAAQ,EAAE,MAAM,WAAW,CAAA;AAE7C,OAAO,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAA;AACjD,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAGrD;;;GAGG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,eAAe,CAAA;AACvD;;;GAGG;AACH,MAAM,CAAC,MAAM,6BAA6B,GAAG,eAAe,CAAA;AAE5D,MAAM,mBAAmB,GAAG,kCAAkC,CAAA;AAC9D;;GAEG;AACH,MAAM,MAAM,GAAG,yCAAyC,CAAA;AAWxD;;GAEG;AACH,SAAS,eAAe,CAAC,MAAc;IACtC,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,EAAE,CAAA;IAE7B,IAAI,CAAC,OAAO;QAAE,OAAO,IAAI,CAAA;IAEzB,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;IACpD,MAAM,CAAC,GAAG,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAE3C,IAAI,CAAC;QAAE,OAAO,EAAE,YAAY,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,EAAE,CAAA;IAE1D,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,CAAA;AAC3B,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,MAAM,GAAG,gBAAgB,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAA;YAErE,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CACzB,QAAQ,CAAC,EAAE,OAAO,EAAE,IAAI,EAAE,gBAAgB,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,kBAAkB,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAC7G,CAAA;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,CAAC;gBACJ,IAAI,KAAK,EAAE,MAAM,MAAM,IAAI,MAAkC,EAAE,CAAC;oBAC/D,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAE5D,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBACrC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAA;oBACrD,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAC3C,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBACvC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBACzC,MAAM,MAAM,GAAG,CAAC,MAAM,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;oBAExC,IAAI,CAAC,IAAI,IAAI,CAAC,MAAM;wBAAE,SAAQ;oBAC9B,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAE,CAAC,IAAI,EAAE,CAAA,CAAC,gCAAgC;oBAE9E,MAAM,KAAK,GAAG,eAAe,CAAC,MAAM,CAAC,CAAA;oBAErC,IAAI,CAAC,KAAK;wBAAE,SAAQ;oBAEpB,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;oBAEhG,MAAM,UAAU,GAA+B;wBAC9C,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;wBAC3B,GAAG,CAAC,QAAQ,IAAI,KAAK;4BACpB,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE;4BAC1B,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;wBACjG,QAAQ,EAAE,IAAI;wBACd,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;wBACnC,QAAQ;qBACR,CAAA;oBAED,MAAM,GAAG,GAAG,UAAU,CAAC,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,EAAE,QAAQ,CAAC,CAAA;oBAEhE,IAAI,CAAC,GAAG;wBAAE,SAAQ;oBAElB,MAAM,OAAO,GAAG,mBAAmB,CAAC,UAAU,EAAE,GAAG,CAAC,CAAA;oBAEpD,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,IAAI,CAAC;wBAAE,SAAQ;oBAE9C,MAAM,QAAQ,GAAG,GAAG;wBACnB,CAAC,CAAC,GAAG,wBAAwB,IAAI,GAAG,EAAE;wBACtC,CAAC,CAAC,cAAc,CAAC,wBAAwB,EAAE,OAAO,CAAC,CAAA;oBAEpD,MAAM;wBACL,GAAG;wBACH,UAAU,EAAE,OAAO;wBACnB,OAAO,EAAE,IAAI;wBACb,MAAM,EAAE,OAAO;wBACf,MAAM,EAAE,wBAAwB;wBAChC,SAAS,EAAE,QAAQ;wBACnB,cAAc,EAAE,EAAE;wBAClB,OAAO,EAAE,6BAA6B;qBACtC,CAAA;oBAED,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;oBAAS,CAAC;gBACV,MAAM,CAAC,OAAO,EAAE,CAAA;YACjB,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,wBAAwB,EAAE,CAAA"}
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"}