@loaders.gl/parquet 4.5.3 → 5.0.0-alpha.10

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 (790) hide show
  1. package/LICENSE +14 -3
  2. package/dist/bundled.d.ts +9 -0
  3. package/dist/bundled.d.ts.map +1 -0
  4. package/dist/bundled.js +10 -0
  5. package/dist/bundled.js.map +1 -0
  6. package/dist/delta-format.d.ts +9 -0
  7. package/dist/delta-format.d.ts.map +1 -0
  8. package/dist/delta-format.js +12 -0
  9. package/dist/delta-format.js.map +1 -0
  10. package/dist/delta-source-loader-types.d.ts +27 -0
  11. package/dist/delta-source-loader-types.d.ts.map +1 -0
  12. package/dist/delta-source-loader-types.js +28 -0
  13. package/dist/delta-source-loader-types.js.map +1 -0
  14. package/dist/delta-source.d.ts +48 -0
  15. package/dist/delta-source.d.ts.map +1 -0
  16. package/dist/delta-source.js +222 -0
  17. package/dist/delta-source.js.map +1 -0
  18. package/dist/delta-types.d.ts +38 -0
  19. package/dist/delta-types.d.ts.map +1 -0
  20. package/dist/delta-types.js +5 -0
  21. package/dist/delta-types.js.map +1 -0
  22. package/dist/geoparquet-loader-with-parser.d.ts +33 -0
  23. package/dist/geoparquet-loader-with-parser.d.ts.map +1 -0
  24. package/dist/geoparquet-loader-with-parser.js +52 -0
  25. package/dist/geoparquet-loader-with-parser.js.map +1 -0
  26. package/dist/geoparquet-loader.d.ts +59 -0
  27. package/dist/geoparquet-loader.d.ts.map +1 -0
  28. package/dist/geoparquet-loader.js +30 -0
  29. package/dist/geoparquet-loader.js.map +1 -0
  30. package/dist/iceberg-rest-catalog.d.ts +49 -0
  31. package/dist/iceberg-rest-catalog.d.ts.map +1 -0
  32. package/dist/iceberg-rest-catalog.js +78 -0
  33. package/dist/iceberg-rest-catalog.js.map +1 -0
  34. package/dist/iceberg-table-source.d.ts +63 -0
  35. package/dist/iceberg-table-source.d.ts.map +1 -0
  36. package/dist/iceberg-table-source.js +797 -0
  37. package/dist/iceberg-table-source.js.map +1 -0
  38. package/dist/iceberg-types.d.ts +150 -0
  39. package/dist/iceberg-types.d.ts.map +1 -0
  40. package/dist/iceberg-types.js +5 -0
  41. package/dist/iceberg-types.js.map +1 -0
  42. package/dist/index.cjs +41568 -6781
  43. package/dist/index.cjs.map +3 -4
  44. package/dist/index.d.ts +15 -9
  45. package/dist/index.d.ts.map +1 -1
  46. package/dist/index.js +11 -12
  47. package/dist/index.js.map +1 -1
  48. package/dist/lib/arrow/convert-columns-to-row-group.js +3 -0
  49. package/dist/lib/arrow/convert-columns-to-row-group.js.map +1 -1
  50. package/dist/lib/arrow/convert-schema-from-parquet.d.ts +3 -1
  51. package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -1
  52. package/dist/lib/arrow/convert-schema-from-parquet.js +164 -29
  53. package/dist/lib/arrow/convert-schema-from-parquet.js.map +1 -1
  54. package/dist/lib/arrow/convert-schema-to-parquet.d.ts.map +1 -1
  55. package/dist/lib/arrow/convert-schema-to-parquet.js +16 -7
  56. package/dist/lib/arrow/convert-schema-to-parquet.js.map +1 -1
  57. package/dist/lib/arrow/create-nested-arrow-vector.d.ts +8 -0
  58. package/dist/lib/arrow/create-nested-arrow-vector.d.ts.map +1 -0
  59. package/dist/lib/arrow/create-nested-arrow-vector.js +485 -0
  60. package/dist/lib/arrow/create-nested-arrow-vector.js.map +1 -0
  61. package/dist/lib/constants.d.ts +0 -1
  62. package/dist/lib/constants.d.ts.map +1 -1
  63. package/dist/lib/constants.js +1 -2
  64. package/dist/lib/constants.js.map +1 -1
  65. package/dist/lib/encoders/encode-arrow-to-parquet.d.ts +2 -2
  66. package/dist/lib/encoders/encode-arrow-to-parquet.d.ts.map +1 -1
  67. package/dist/lib/encoders/encode-arrow-to-parquet.js +3 -10
  68. package/dist/lib/encoders/encode-arrow-to-parquet.js.map +1 -1
  69. package/dist/lib/encoders/encode-table-to-parquet-js.d.ts +5 -0
  70. package/dist/lib/encoders/encode-table-to-parquet-js.d.ts.map +1 -0
  71. package/dist/lib/encoders/encode-table-to-parquet-js.js +394 -0
  72. package/dist/lib/encoders/encode-table-to-parquet-js.js.map +1 -0
  73. package/dist/lib/geo/geoparquet-covering.d.ts +19 -0
  74. package/dist/lib/geo/geoparquet-covering.d.ts.map +1 -0
  75. package/dist/lib/geo/geoparquet-covering.js +137 -0
  76. package/dist/lib/geo/geoparquet-covering.js.map +1 -0
  77. package/dist/lib/geo/geospatial-metadata.d.ts +67 -0
  78. package/dist/lib/geo/geospatial-metadata.d.ts.map +1 -0
  79. package/dist/lib/geo/geospatial-metadata.js +412 -0
  80. package/dist/lib/geo/geospatial-metadata.js.map +1 -0
  81. package/dist/lib/parquet-bloom-filter-planner.d.ts +16 -0
  82. package/dist/lib/parquet-bloom-filter-planner.d.ts.map +1 -0
  83. package/dist/lib/parquet-bloom-filter-planner.js +57 -0
  84. package/dist/lib/parquet-bloom-filter-planner.js.map +1 -0
  85. package/dist/lib/parquet-bloom-filter.d.ts +30 -0
  86. package/dist/lib/parquet-bloom-filter.d.ts.map +1 -0
  87. package/dist/lib/parquet-bloom-filter.js +289 -0
  88. package/dist/lib/parquet-bloom-filter.js.map +1 -0
  89. package/dist/lib/parquet-encryption.d.ts +80 -0
  90. package/dist/lib/parquet-encryption.d.ts.map +1 -0
  91. package/dist/lib/parquet-encryption.js +190 -0
  92. package/dist/lib/parquet-encryption.js.map +1 -0
  93. package/dist/lib/parquet-page-index.d.ts +91 -0
  94. package/dist/lib/parquet-page-index.d.ts.map +1 -0
  95. package/dist/lib/parquet-page-index.js +463 -0
  96. package/dist/lib/parquet-page-index.js.map +1 -0
  97. package/dist/lib/parquet-predicate.d.ts +21 -0
  98. package/dist/lib/parquet-predicate.d.ts.map +1 -0
  99. package/dist/lib/parquet-predicate.js +328 -0
  100. package/dist/lib/parquet-predicate.js.map +1 -0
  101. package/dist/lib/parquet-source-worker-client.d.ts +6 -0
  102. package/dist/lib/parquet-source-worker-client.d.ts.map +1 -0
  103. package/dist/lib/parquet-source-worker-client.js +24 -0
  104. package/dist/lib/parquet-source-worker-client.js.map +1 -0
  105. package/dist/lib/parquet-source-worker-decoder.d.ts +7 -0
  106. package/dist/lib/parquet-source-worker-decoder.d.ts.map +1 -0
  107. package/dist/lib/parquet-source-worker-decoder.js +181 -0
  108. package/dist/lib/parquet-source-worker-decoder.js.map +1 -0
  109. package/dist/lib/parquet-source-worker-file.d.ts +24 -0
  110. package/dist/lib/parquet-source-worker-file.d.ts.map +1 -0
  111. package/dist/lib/parquet-source-worker-file.js +40 -0
  112. package/dist/lib/parquet-source-worker-file.js.map +1 -0
  113. package/dist/lib/parquet-source-worker-types.d.ts +121 -0
  114. package/dist/lib/parquet-source-worker-types.d.ts.map +1 -0
  115. package/dist/lib/parquet-source-worker-types.js +5 -0
  116. package/dist/lib/parquet-source-worker-types.js.map +1 -0
  117. package/dist/lib/parquet-worker-transport.d.ts +8 -0
  118. package/dist/lib/parquet-worker-transport.d.ts.map +1 -0
  119. package/dist/lib/parquet-worker-transport.js +40 -0
  120. package/dist/lib/parquet-worker-transport.js.map +1 -0
  121. package/dist/lib/parsers/convert-parquet-tables.d.ts +16 -0
  122. package/dist/lib/parsers/convert-parquet-tables.d.ts.map +1 -0
  123. package/dist/lib/parsers/convert-parquet-tables.js +30 -0
  124. package/dist/lib/parsers/convert-parquet-tables.js.map +1 -0
  125. package/dist/lib/parsers/get-parquet-schema.d.ts.map +1 -1
  126. package/dist/lib/parsers/get-parquet-schema.js +6 -2
  127. package/dist/lib/parsers/get-parquet-schema.js.map +1 -1
  128. package/dist/lib/parsers/parse-geoparquet-to-geojson.d.ts +4 -4
  129. package/dist/lib/parsers/parse-geoparquet-to-geojson.d.ts.map +1 -1
  130. package/dist/lib/parsers/parse-geoparquet-to-geojson.js +19 -6
  131. package/dist/lib/parsers/parse-geoparquet-to-geojson.js.map +1 -1
  132. package/dist/lib/parsers/parse-parquet-tables.d.ts +37 -0
  133. package/dist/lib/parsers/parse-parquet-tables.d.ts.map +1 -0
  134. package/dist/lib/parsers/parse-parquet-tables.js +50 -0
  135. package/dist/lib/parsers/parse-parquet-tables.js.map +1 -0
  136. package/dist/lib/parsers/parse-parquet-to-arrow-js.d.ts +19 -0
  137. package/dist/lib/parsers/parse-parquet-to-arrow-js.d.ts.map +1 -0
  138. package/dist/lib/parsers/parse-parquet-to-arrow-js.js +921 -0
  139. package/dist/lib/parsers/parse-parquet-to-arrow-js.js.map +1 -0
  140. package/dist/lib/parsers/parse-parquet-to-arrow.d.ts +7 -7
  141. package/dist/lib/parsers/parse-parquet-to-arrow.d.ts.map +1 -1
  142. package/dist/lib/parsers/parse-parquet-to-arrow.js +45 -32
  143. package/dist/lib/parsers/parse-parquet-to-arrow.js.map +1 -1
  144. package/dist/lib/parsers/parse-parquet-to-columns.d.ts +3 -3
  145. package/dist/lib/parsers/parse-parquet-to-columns.d.ts.map +1 -1
  146. package/dist/lib/parsers/parse-parquet-to-columns.js +4 -5
  147. package/dist/lib/parsers/parse-parquet-to-columns.js.map +1 -1
  148. package/dist/lib/parsers/parse-parquet-to-json.d.ts +3 -3
  149. package/dist/lib/parsers/parse-parquet-to-json.d.ts.map +1 -1
  150. package/dist/lib/parsers/parse-parquet-to-json.js +95 -28
  151. package/dist/lib/parsers/parse-parquet-to-json.js.map +1 -1
  152. package/dist/lib/slice-parquet-batch.d.ts +9 -0
  153. package/dist/lib/slice-parquet-batch.d.ts.map +1 -0
  154. package/dist/lib/slice-parquet-batch.js +28 -0
  155. package/dist/lib/slice-parquet-batch.js.map +1 -0
  156. package/dist/lib/sources/parquet-range-file.d.ts +80 -0
  157. package/dist/lib/sources/parquet-range-file.d.ts.map +1 -0
  158. package/dist/lib/sources/parquet-range-file.js +267 -0
  159. package/dist/lib/sources/parquet-range-file.js.map +1 -0
  160. package/dist/lib/sources/single-batch-queue.d.ts +24 -0
  161. package/dist/lib/sources/single-batch-queue.d.ts.map +1 -0
  162. package/dist/lib/sources/single-batch-queue.js +91 -0
  163. package/dist/lib/sources/single-batch-queue.js.map +1 -0
  164. package/dist/lib/utils/load-wasm-node.d.ts +3 -0
  165. package/dist/lib/utils/load-wasm-node.d.ts.map +1 -0
  166. package/dist/lib/utils/load-wasm-node.js +23 -0
  167. package/dist/lib/utils/load-wasm-node.js.map +1 -0
  168. package/dist/lib/utils/load-wasm.d.ts +2 -2
  169. package/dist/lib/utils/load-wasm.d.ts.map +1 -1
  170. package/dist/lib/utils/load-wasm.js +28 -12
  171. package/dist/lib/utils/load-wasm.js.map +1 -1
  172. package/dist/lib/utils/make-stream-iterator.d.ts +2 -0
  173. package/dist/lib/utils/make-stream-iterator.d.ts.map +1 -1
  174. package/dist/lib/utils/make-stream-iterator.js +123 -14
  175. package/dist/lib/utils/make-stream-iterator.js.map +1 -1
  176. package/dist/lib/utils/normalize-parquet-options.d.ts +5 -0
  177. package/dist/lib/utils/normalize-parquet-options.d.ts.map +1 -0
  178. package/dist/lib/utils/normalize-parquet-options.js +25 -0
  179. package/dist/lib/utils/normalize-parquet-options.js.map +1 -0
  180. package/dist/parquet-dataset-source.cjs +38974 -0
  181. package/dist/parquet-dataset-source.cjs.map +6 -0
  182. package/dist/parquet-dataset-source.d.ts +81 -0
  183. package/dist/parquet-dataset-source.d.ts.map +1 -0
  184. package/dist/parquet-dataset-source.js +602 -0
  185. package/dist/parquet-dataset-source.js.map +1 -0
  186. package/dist/parquet-encoded-page-types.d.ts +108 -0
  187. package/dist/parquet-encoded-page-types.d.ts.map +1 -0
  188. package/dist/parquet-encoded-page-types.js +5 -0
  189. package/dist/parquet-encoded-page-types.js.map +1 -0
  190. package/dist/parquet-format.d.ts +2 -0
  191. package/dist/parquet-format.d.ts.map +1 -1
  192. package/dist/parquet-format.js +2 -0
  193. package/dist/parquet-format.js.map +1 -1
  194. package/dist/parquet-js-loader-types.d.ts +59 -0
  195. package/dist/parquet-js-loader-types.d.ts.map +1 -0
  196. package/dist/parquet-js-loader-types.js +34 -0
  197. package/dist/parquet-js-loader-types.js.map +1 -0
  198. package/dist/parquet-js-loader.d.ts +18 -22
  199. package/dist/parquet-js-loader.d.ts.map +1 -1
  200. package/dist/parquet-js-loader.js +53 -14
  201. package/dist/parquet-js-loader.js.map +1 -1
  202. package/dist/parquet-js-writer.d.ts +65 -0
  203. package/dist/parquet-js-writer.d.ts.map +1 -0
  204. package/dist/parquet-js-writer.js +28 -0
  205. package/dist/parquet-js-writer.js.map +1 -0
  206. package/dist/parquet-loader-base.d.ts +25 -0
  207. package/dist/parquet-loader-base.d.ts.map +1 -0
  208. package/dist/parquet-loader-base.js +23 -0
  209. package/dist/parquet-loader-base.js.map +1 -0
  210. package/dist/parquet-loader-options.d.ts +55 -0
  211. package/dist/parquet-loader-options.d.ts.map +1 -0
  212. package/dist/parquet-loader-options.js +10 -0
  213. package/dist/parquet-loader-options.js.map +1 -0
  214. package/dist/parquet-loader-types.d.ts +61 -0
  215. package/dist/parquet-loader-types.d.ts.map +1 -0
  216. package/dist/parquet-loader-types.js +17 -0
  217. package/dist/parquet-loader-types.js.map +1 -0
  218. package/dist/parquet-loader.d.ts +42 -0
  219. package/dist/parquet-loader.d.ts.map +1 -0
  220. package/dist/parquet-loader.js +68 -0
  221. package/dist/parquet-loader.js.map +1 -0
  222. package/dist/parquet-source-capabilities.d.ts +43 -0
  223. package/dist/parquet-source-capabilities.d.ts.map +1 -0
  224. package/dist/parquet-source-capabilities.js +32 -0
  225. package/dist/parquet-source-capabilities.js.map +1 -0
  226. package/dist/parquet-source-loader-types.d.ts +82 -0
  227. package/dist/parquet-source-loader-types.d.ts.map +1 -0
  228. package/dist/parquet-source-loader-types.js +84 -0
  229. package/dist/parquet-source-loader-types.js.map +1 -0
  230. package/dist/parquet-source-loader.cjs +38261 -0
  231. package/dist/parquet-source-loader.cjs.map +6 -0
  232. package/dist/parquet-source-loader.d.ts +167 -0
  233. package/dist/parquet-source-loader.d.ts.map +1 -0
  234. package/dist/parquet-source-loader.js +1438 -0
  235. package/dist/parquet-source-loader.js.map +1 -0
  236. package/dist/parquet-source-types.d.ts +534 -0
  237. package/dist/parquet-source-types.d.ts.map +1 -0
  238. package/dist/parquet-source-types.js +5 -0
  239. package/dist/parquet-source-types.js.map +1 -0
  240. package/dist/parquet-source-worker-factory.d.ts +4 -0
  241. package/dist/parquet-source-worker-factory.d.ts.map +1 -0
  242. package/dist/parquet-source-worker-factory.js +6 -0
  243. package/dist/parquet-source-worker-factory.js.map +1 -0
  244. package/dist/parquet-source-worker-url.d.ts +3 -0
  245. package/dist/parquet-source-worker-url.d.ts.map +1 -0
  246. package/dist/parquet-source-worker-url.js +22 -0
  247. package/dist/parquet-source-worker-url.js.map +1 -0
  248. package/dist/parquet-source-worker.js +38871 -0
  249. package/dist/parquet-source-worker.js.map +7 -0
  250. package/dist/parquet-worker-node.cjs +78507 -0
  251. package/dist/parquet-worker-node.cjs.map +7 -0
  252. package/dist/parquet-worker-url.d.ts +3 -0
  253. package/dist/parquet-worker-url.d.ts.map +1 -0
  254. package/dist/parquet-worker-url.js +19 -0
  255. package/dist/parquet-worker-url.js.map +1 -0
  256. package/dist/parquet-worker.js +20094 -0
  257. package/dist/parquet-worker.js.map +7 -0
  258. package/dist/parquet-writer.d.ts +10 -0
  259. package/dist/parquet-writer.d.ts.map +1 -0
  260. package/dist/parquet-writer.js +27 -0
  261. package/dist/parquet-writer.js.map +1 -0
  262. package/dist/parquet_wasm_bg.wasm +0 -0
  263. package/dist/parquetjs/codecs/byte-stream-split.d.ts +9 -0
  264. package/dist/parquetjs/codecs/byte-stream-split.d.ts.map +1 -0
  265. package/dist/parquetjs/codecs/byte-stream-split.js +159 -0
  266. package/dist/parquetjs/codecs/byte-stream-split.js.map +1 -0
  267. package/dist/parquetjs/codecs/declare.d.ts +35 -3
  268. package/dist/parquetjs/codecs/declare.d.ts.map +1 -1
  269. package/dist/parquetjs/codecs/declare.js +17 -1
  270. package/dist/parquetjs/codecs/declare.js.map +1 -1
  271. package/dist/parquetjs/codecs/delta.d.ts +17 -0
  272. package/dist/parquetjs/codecs/delta.d.ts.map +1 -0
  273. package/dist/parquetjs/codecs/delta.js +788 -0
  274. package/dist/parquetjs/codecs/delta.js.map +1 -0
  275. package/dist/parquetjs/codecs/dictionary.d.ts +5 -2
  276. package/dist/parquetjs/codecs/dictionary.d.ts.map +1 -1
  277. package/dist/parquetjs/codecs/dictionary.js +16 -4
  278. package/dist/parquetjs/codecs/dictionary.js.map +1 -1
  279. package/dist/parquetjs/codecs/index.d.ts.map +1 -1
  280. package/dist/parquetjs/codecs/index.js +22 -0
  281. package/dist/parquetjs/codecs/index.js.map +1 -1
  282. package/dist/parquetjs/codecs/plain.d.ts +3 -3
  283. package/dist/parquetjs/codecs/plain.d.ts.map +1 -1
  284. package/dist/parquetjs/codecs/plain.js +231 -78
  285. package/dist/parquetjs/codecs/plain.js.map +1 -1
  286. package/dist/parquetjs/codecs/rle.d.ts +7 -3
  287. package/dist/parquetjs/codecs/rle.d.ts.map +1 -1
  288. package/dist/parquetjs/codecs/rle.js +613 -69
  289. package/dist/parquetjs/codecs/rle.js.map +1 -1
  290. package/dist/parquetjs/compression.d.ts +16 -8
  291. package/dist/parquetjs/compression.d.ts.map +1 -1
  292. package/dist/parquetjs/compression.js +114 -141
  293. package/dist/parquetjs/compression.js.map +1 -1
  294. package/dist/parquetjs/encoder/dictionary-planner.d.ts +17 -0
  295. package/dist/parquetjs/encoder/dictionary-planner.d.ts.map +1 -0
  296. package/dist/parquetjs/encoder/dictionary-planner.js +67 -0
  297. package/dist/parquetjs/encoder/dictionary-planner.js.map +1 -0
  298. package/dist/parquetjs/encoder/page-planner.d.ts +11 -0
  299. package/dist/parquetjs/encoder/page-planner.d.ts.map +1 -0
  300. package/dist/parquetjs/encoder/page-planner.js +60 -0
  301. package/dist/parquetjs/encoder/page-planner.js.map +1 -0
  302. package/dist/parquetjs/encoder/parquet-encoder.d.ts +54 -5
  303. package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +1 -1
  304. package/dist/parquetjs/encoder/parquet-encoder.js +930 -66
  305. package/dist/parquetjs/encoder/parquet-encoder.js.map +1 -1
  306. package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts +29 -0
  307. package/dist/parquetjs/parquet-thrift/BoundingBox.d.ts.map +1 -0
  308. package/dist/parquetjs/parquet-thrift/BoundingBox.js +85 -0
  309. package/dist/parquetjs/parquet-thrift/BoundingBox.js.map +1 -0
  310. package/dist/parquetjs/parquet-thrift/BsonType.d.ts +1 -1
  311. package/dist/parquetjs/parquet-thrift/BsonType.d.ts.map +1 -1
  312. package/dist/parquetjs/parquet-thrift/BsonType.js +1 -1
  313. package/dist/parquetjs/parquet-thrift/BsonType.js.map +1 -1
  314. package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts +7 -2
  315. package/dist/parquetjs/parquet-thrift/ColumnChunk.d.ts.map +1 -1
  316. package/dist/parquetjs/parquet-thrift/ColumnChunk.js +37 -2
  317. package/dist/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
  318. package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts +35 -0
  319. package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.d.ts.map +1 -0
  320. package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js +135 -0
  321. package/dist/parquetjs/parquet-thrift/ColumnCryptoMetaData.js.map +1 -0
  322. package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts +6 -6
  323. package/dist/parquetjs/parquet-thrift/ColumnIndex.d.ts.map +1 -1
  324. package/dist/parquetjs/parquet-thrift/ColumnIndex.js +3 -3
  325. package/dist/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -1
  326. package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts +12 -2
  327. package/dist/parquetjs/parquet-thrift/ColumnMetaData.d.ts.map +1 -1
  328. package/dist/parquetjs/parquet-thrift/ColumnMetaData.js +75 -2
  329. package/dist/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
  330. package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts +1 -1
  331. package/dist/parquetjs/parquet-thrift/ColumnOrder.d.ts.map +1 -1
  332. package/dist/parquetjs/parquet-thrift/ColumnOrder.js +1 -1
  333. package/dist/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -1
  334. package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts +1 -1
  335. package/dist/parquetjs/parquet-thrift/DataPageHeader.d.ts.map +1 -1
  336. package/dist/parquetjs/parquet-thrift/DataPageHeader.js +1 -1
  337. package/dist/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -1
  338. package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts +1 -1
  339. package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.d.ts.map +1 -1
  340. package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js +1 -1
  341. package/dist/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -1
  342. package/dist/parquetjs/parquet-thrift/DateType.d.ts +1 -1
  343. package/dist/parquetjs/parquet-thrift/DateType.d.ts.map +1 -1
  344. package/dist/parquetjs/parquet-thrift/DateType.js +1 -1
  345. package/dist/parquetjs/parquet-thrift/DateType.js.map +1 -1
  346. package/dist/parquetjs/parquet-thrift/DecimalType.d.ts +1 -1
  347. package/dist/parquetjs/parquet-thrift/DecimalType.d.ts.map +1 -1
  348. package/dist/parquetjs/parquet-thrift/DecimalType.js +1 -1
  349. package/dist/parquetjs/parquet-thrift/DecimalType.js.map +1 -1
  350. package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts +1 -1
  351. package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.d.ts.map +1 -1
  352. package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js +1 -1
  353. package/dist/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -1
  354. package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.d.ts +8 -0
  355. package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.d.ts.map +1 -0
  356. package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.js +14 -0
  357. package/dist/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.js.map +1 -0
  358. package/dist/parquetjs/parquet-thrift/Encoding.d.ts +2 -1
  359. package/dist/parquetjs/parquet-thrift/Encoding.d.ts.map +1 -1
  360. package/dist/parquetjs/parquet-thrift/Encoding.js +1 -0
  361. package/dist/parquetjs/parquet-thrift/Encoding.js.map +1 -1
  362. package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.d.ts +22 -0
  363. package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.d.ts.map +1 -0
  364. package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js +111 -0
  365. package/dist/parquetjs/parquet-thrift/EncryptionAlgorithm.js.map +1 -0
  366. package/dist/parquetjs/parquet-thrift/EnumType.d.ts +1 -1
  367. package/dist/parquetjs/parquet-thrift/EnumType.d.ts.map +1 -1
  368. package/dist/parquetjs/parquet-thrift/EnumType.js +1 -1
  369. package/dist/parquetjs/parquet-thrift/EnumType.js.map +1 -1
  370. package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.d.ts +16 -0
  371. package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.d.ts.map +1 -0
  372. package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js +57 -0
  373. package/dist/parquetjs/parquet-thrift/FileCryptoMetaData.js.map +1 -0
  374. package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts +7 -2
  375. package/dist/parquetjs/parquet-thrift/FileMetaData.d.ts.map +1 -1
  376. package/dist/parquetjs/parquet-thrift/FileMetaData.js +37 -2
  377. package/dist/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
  378. package/dist/parquetjs/parquet-thrift/Float16Type.d.ts +9 -0
  379. package/dist/parquetjs/parquet-thrift/Float16Type.d.ts.map +1 -0
  380. package/dist/parquetjs/parquet-thrift/Float16Type.js +27 -0
  381. package/dist/parquetjs/parquet-thrift/Float16Type.js.map +1 -0
  382. package/dist/parquetjs/parquet-thrift/GeographyType.d.ts +14 -0
  383. package/dist/parquetjs/parquet-thrift/GeographyType.d.ts.map +1 -0
  384. package/dist/parquetjs/parquet-thrift/GeographyType.js +51 -0
  385. package/dist/parquetjs/parquet-thrift/GeographyType.js.map +1 -0
  386. package/dist/parquetjs/parquet-thrift/GeometryType.d.ts +11 -0
  387. package/dist/parquetjs/parquet-thrift/GeometryType.d.ts.map +1 -0
  388. package/dist/parquetjs/parquet-thrift/GeometryType.js +41 -0
  389. package/dist/parquetjs/parquet-thrift/GeometryType.js.map +1 -0
  390. package/dist/parquetjs/parquet-thrift/GeospatialStatistics.d.ts +18 -0
  391. package/dist/parquetjs/parquet-thrift/GeospatialStatistics.d.ts.map +1 -0
  392. package/dist/parquetjs/parquet-thrift/GeospatialStatistics.js +62 -0
  393. package/dist/parquetjs/parquet-thrift/GeospatialStatistics.js.map +1 -0
  394. package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts +1 -1
  395. package/dist/parquetjs/parquet-thrift/IndexPageHeader.d.ts.map +1 -1
  396. package/dist/parquetjs/parquet-thrift/IndexPageHeader.js +1 -1
  397. package/dist/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -1
  398. package/dist/parquetjs/parquet-thrift/IntType.d.ts +1 -1
  399. package/dist/parquetjs/parquet-thrift/IntType.d.ts.map +1 -1
  400. package/dist/parquetjs/parquet-thrift/IntType.js +1 -1
  401. package/dist/parquetjs/parquet-thrift/IntType.js.map +1 -1
  402. package/dist/parquetjs/parquet-thrift/JsonType.d.ts +1 -1
  403. package/dist/parquetjs/parquet-thrift/JsonType.d.ts.map +1 -1
  404. package/dist/parquetjs/parquet-thrift/JsonType.js +1 -1
  405. package/dist/parquetjs/parquet-thrift/JsonType.js.map +1 -1
  406. package/dist/parquetjs/parquet-thrift/KeyValue.d.ts +1 -1
  407. package/dist/parquetjs/parquet-thrift/KeyValue.d.ts.map +1 -1
  408. package/dist/parquetjs/parquet-thrift/KeyValue.js +1 -1
  409. package/dist/parquetjs/parquet-thrift/KeyValue.js.map +1 -1
  410. package/dist/parquetjs/parquet-thrift/ListType.d.ts +1 -1
  411. package/dist/parquetjs/parquet-thrift/ListType.d.ts.map +1 -1
  412. package/dist/parquetjs/parquet-thrift/ListType.js +1 -1
  413. package/dist/parquetjs/parquet-thrift/ListType.js.map +1 -1
  414. package/dist/parquetjs/parquet-thrift/LogicalType.d.ts +17 -1
  415. package/dist/parquetjs/parquet-thrift/LogicalType.d.ts.map +1 -1
  416. package/dist/parquetjs/parquet-thrift/LogicalType.js +97 -1
  417. package/dist/parquetjs/parquet-thrift/LogicalType.js.map +1 -1
  418. package/dist/parquetjs/parquet-thrift/MapType.d.ts +1 -1
  419. package/dist/parquetjs/parquet-thrift/MapType.d.ts.map +1 -1
  420. package/dist/parquetjs/parquet-thrift/MapType.js +1 -1
  421. package/dist/parquetjs/parquet-thrift/MapType.js.map +1 -1
  422. package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts +1 -1
  423. package/dist/parquetjs/parquet-thrift/MicroSeconds.d.ts.map +1 -1
  424. package/dist/parquetjs/parquet-thrift/MicroSeconds.js +1 -1
  425. package/dist/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -1
  426. package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts +1 -1
  427. package/dist/parquetjs/parquet-thrift/MilliSeconds.d.ts.map +1 -1
  428. package/dist/parquetjs/parquet-thrift/MilliSeconds.js +1 -1
  429. package/dist/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -1
  430. package/dist/parquetjs/parquet-thrift/NanoSeconds.d.ts +9 -0
  431. package/dist/parquetjs/parquet-thrift/NanoSeconds.d.ts.map +1 -0
  432. package/dist/parquetjs/parquet-thrift/NanoSeconds.js +27 -0
  433. package/dist/parquetjs/parquet-thrift/NanoSeconds.js.map +1 -0
  434. package/dist/parquetjs/parquet-thrift/NullType.d.ts +1 -1
  435. package/dist/parquetjs/parquet-thrift/NullType.d.ts.map +1 -1
  436. package/dist/parquetjs/parquet-thrift/NullType.js +1 -1
  437. package/dist/parquetjs/parquet-thrift/NullType.js.map +1 -1
  438. package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts +1 -1
  439. package/dist/parquetjs/parquet-thrift/OffsetIndex.d.ts.map +1 -1
  440. package/dist/parquetjs/parquet-thrift/OffsetIndex.js +1 -1
  441. package/dist/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -1
  442. package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts +1 -1
  443. package/dist/parquetjs/parquet-thrift/PageEncodingStats.d.ts.map +1 -1
  444. package/dist/parquetjs/parquet-thrift/PageEncodingStats.js +1 -1
  445. package/dist/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -1
  446. package/dist/parquetjs/parquet-thrift/PageHeader.d.ts +1 -1
  447. package/dist/parquetjs/parquet-thrift/PageHeader.d.ts.map +1 -1
  448. package/dist/parquetjs/parquet-thrift/PageHeader.js +1 -1
  449. package/dist/parquetjs/parquet-thrift/PageHeader.js.map +1 -1
  450. package/dist/parquetjs/parquet-thrift/PageLocation.d.ts +2 -2
  451. package/dist/parquetjs/parquet-thrift/PageLocation.d.ts.map +1 -1
  452. package/dist/parquetjs/parquet-thrift/PageLocation.js +2 -2
  453. package/dist/parquetjs/parquet-thrift/PageLocation.js.map +1 -1
  454. package/dist/parquetjs/parquet-thrift/RowGroup.d.ts +2 -2
  455. package/dist/parquetjs/parquet-thrift/RowGroup.d.ts.map +1 -1
  456. package/dist/parquetjs/parquet-thrift/RowGroup.js +2 -2
  457. package/dist/parquetjs/parquet-thrift/RowGroup.js.map +1 -1
  458. package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts +1 -1
  459. package/dist/parquetjs/parquet-thrift/SchemaElement.d.ts.map +1 -1
  460. package/dist/parquetjs/parquet-thrift/SchemaElement.js +1 -1
  461. package/dist/parquetjs/parquet-thrift/SchemaElement.js.map +1 -1
  462. package/dist/parquetjs/parquet-thrift/SizeStatistics.d.ts +26 -0
  463. package/dist/parquetjs/parquet-thrift/SizeStatistics.d.ts.map +1 -0
  464. package/dist/parquetjs/parquet-thrift/SizeStatistics.js +85 -0
  465. package/dist/parquetjs/parquet-thrift/SizeStatistics.js.map +1 -0
  466. package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts +1 -1
  467. package/dist/parquetjs/parquet-thrift/SortingColumn.d.ts.map +1 -1
  468. package/dist/parquetjs/parquet-thrift/SortingColumn.js +1 -1
  469. package/dist/parquetjs/parquet-thrift/SortingColumn.js.map +1 -1
  470. package/dist/parquetjs/parquet-thrift/Statistics.d.ts +14 -10
  471. package/dist/parquetjs/parquet-thrift/Statistics.d.ts.map +1 -1
  472. package/dist/parquetjs/parquet-thrift/Statistics.js +38 -2
  473. package/dist/parquetjs/parquet-thrift/Statistics.js.map +1 -1
  474. package/dist/parquetjs/parquet-thrift/StringType.d.ts +1 -1
  475. package/dist/parquetjs/parquet-thrift/StringType.d.ts.map +1 -1
  476. package/dist/parquetjs/parquet-thrift/StringType.js +1 -1
  477. package/dist/parquetjs/parquet-thrift/StringType.js.map +1 -1
  478. package/dist/parquetjs/parquet-thrift/TimeType.d.ts +1 -1
  479. package/dist/parquetjs/parquet-thrift/TimeType.d.ts.map +1 -1
  480. package/dist/parquetjs/parquet-thrift/TimeType.js +1 -1
  481. package/dist/parquetjs/parquet-thrift/TimeType.js.map +1 -1
  482. package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts +5 -1
  483. package/dist/parquetjs/parquet-thrift/TimeUnit.d.ts.map +1 -1
  484. package/dist/parquetjs/parquet-thrift/TimeUnit.js +25 -1
  485. package/dist/parquetjs/parquet-thrift/TimeUnit.js.map +1 -1
  486. package/dist/parquetjs/parquet-thrift/TimestampType.d.ts +1 -1
  487. package/dist/parquetjs/parquet-thrift/TimestampType.d.ts.map +1 -1
  488. package/dist/parquetjs/parquet-thrift/TimestampType.js +1 -1
  489. package/dist/parquetjs/parquet-thrift/TimestampType.js.map +1 -1
  490. package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts +1 -1
  491. package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.d.ts.map +1 -1
  492. package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js +1 -1
  493. package/dist/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -1
  494. package/dist/parquetjs/parquet-thrift/UUIDType.d.ts +1 -1
  495. package/dist/parquetjs/parquet-thrift/UUIDType.d.ts.map +1 -1
  496. package/dist/parquetjs/parquet-thrift/UUIDType.js +1 -1
  497. package/dist/parquetjs/parquet-thrift/UUIDType.js.map +1 -1
  498. package/dist/parquetjs/parquet-thrift/VariantType.d.ts +11 -0
  499. package/dist/parquetjs/parquet-thrift/VariantType.d.ts.map +1 -0
  500. package/dist/parquetjs/parquet-thrift/VariantType.js +41 -0
  501. package/dist/parquetjs/parquet-thrift/VariantType.js.map +1 -0
  502. package/dist/parquetjs/parquet-thrift/index.d.ts +12 -2
  503. package/dist/parquetjs/parquet-thrift/index.d.ts.map +1 -1
  504. package/dist/parquetjs/parquet-thrift/index.js +12 -2
  505. package/dist/parquetjs/parquet-thrift/index.js.map +1 -1
  506. package/dist/parquetjs/parser/decoders.d.ts +23 -3
  507. package/dist/parquetjs/parser/decoders.d.ts.map +1 -1
  508. package/dist/parquetjs/parser/decoders.js +687 -119
  509. package/dist/parquetjs/parser/decoders.js.map +1 -1
  510. package/dist/parquetjs/parser/encoded-pages.d.ts +24 -0
  511. package/dist/parquetjs/parser/encoded-pages.d.ts.map +1 -0
  512. package/dist/parquetjs/parser/encoded-pages.js +206 -0
  513. package/dist/parquetjs/parser/encoded-pages.js.map +1 -0
  514. package/dist/parquetjs/parser/parquet-reader.d.ts +111 -10
  515. package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
  516. package/dist/parquetjs/parser/parquet-reader.js +836 -69
  517. package/dist/parquetjs/parser/parquet-reader.js.map +1 -1
  518. package/dist/parquetjs/schema/declare.d.ts +83 -10
  519. package/dist/parquetjs/schema/declare.d.ts.map +1 -1
  520. package/dist/parquetjs/schema/declare.js.map +1 -1
  521. package/dist/parquetjs/schema/schema.d.ts +3 -0
  522. package/dist/parquetjs/schema/schema.d.ts.map +1 -1
  523. package/dist/parquetjs/schema/schema.js +31 -3
  524. package/dist/parquetjs/schema/schema.js.map +1 -1
  525. package/dist/parquetjs/schema/shred.d.ts +2 -0
  526. package/dist/parquetjs/schema/shred.d.ts.map +1 -1
  527. package/dist/parquetjs/schema/shred.js +285 -29
  528. package/dist/parquetjs/schema/shred.js.map +1 -1
  529. package/dist/parquetjs/schema/types.d.ts.map +1 -1
  530. package/dist/parquetjs/schema/types.js +290 -71
  531. package/dist/parquetjs/schema/types.js.map +1 -1
  532. package/dist/parquetjs/schema/variant.d.ts +3 -0
  533. package/dist/parquetjs/schema/variant.d.ts.map +1 -0
  534. package/dist/parquetjs/schema/variant.js +248 -0
  535. package/dist/parquetjs/schema/variant.js.map +1 -0
  536. package/dist/parquetjs/utils/binary-utils.d.ts +41 -0
  537. package/dist/parquetjs/utils/binary-utils.d.ts.map +1 -0
  538. package/dist/parquetjs/utils/binary-utils.js +106 -0
  539. package/dist/parquetjs/utils/binary-utils.js.map +1 -0
  540. package/dist/parquetjs/utils/crc32.d.ts +3 -0
  541. package/dist/parquetjs/utils/crc32.d.ts.map +1 -0
  542. package/dist/parquetjs/utils/crc32.js +15 -0
  543. package/dist/parquetjs/utils/crc32.js.map +1 -0
  544. package/dist/parquetjs/utils/file-utils.d.ts +1 -1
  545. package/dist/parquetjs/utils/file-utils.d.ts.map +1 -1
  546. package/dist/parquetjs/utils/file-utils.js +3 -3
  547. package/dist/parquetjs/utils/file-utils.js.map +1 -1
  548. package/dist/parquetjs/utils/read-utils.d.ts +25 -7
  549. package/dist/parquetjs/utils/read-utils.d.ts.map +1 -1
  550. package/dist/parquetjs/utils/read-utils.js +66 -39
  551. package/dist/parquetjs/utils/read-utils.js.map +1 -1
  552. package/dist/parquetjs/utils/thrift-runtime.d.ts +106 -0
  553. package/dist/parquetjs/utils/thrift-runtime.d.ts.map +1 -0
  554. package/dist/parquetjs/utils/thrift-runtime.js +50 -0
  555. package/dist/parquetjs/utils/thrift-runtime.js.map +1 -0
  556. package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.d.ts +78 -0
  557. package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.d.ts.map +1 -0
  558. package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.js +213 -0
  559. package/dist/parquetjs/utils/uint8-array-compact-protocol-writer.js.map +1 -0
  560. package/dist/parquetjs/utils/uint8-array-compact-protocol.d.ts +80 -0
  561. package/dist/parquetjs/utils/uint8-array-compact-protocol.d.ts.map +1 -0
  562. package/dist/parquetjs/utils/uint8-array-compact-protocol.js +276 -0
  563. package/dist/parquetjs/utils/uint8-array-compact-protocol.js.map +1 -0
  564. package/dist/parquetjs/utils/uint8-array-transport.d.ts +25 -0
  565. package/dist/parquetjs/utils/uint8-array-transport.d.ts.map +1 -0
  566. package/dist/parquetjs/utils/uint8-array-transport.js +62 -0
  567. package/dist/parquetjs/utils/uint8-array-transport.js.map +1 -0
  568. package/dist/polyfills/process.d.ts +12 -0
  569. package/dist/polyfills/process.d.ts.map +1 -0
  570. package/dist/polyfills/process.js +17 -0
  571. package/dist/polyfills/process.js.map +1 -0
  572. package/dist/unbundled.d.ts +10 -0
  573. package/dist/unbundled.d.ts.map +1 -0
  574. package/dist/unbundled.js +11 -0
  575. package/dist/unbundled.js.map +1 -0
  576. package/dist/wasm.d.ts +3 -0
  577. package/dist/wasm.d.ts.map +1 -0
  578. package/dist/wasm.js +6 -0
  579. package/dist/wasm.js.map +1 -0
  580. package/dist/workers/parquet-source-worker.d.ts +2 -0
  581. package/dist/workers/parquet-source-worker.d.ts.map +1 -0
  582. package/dist/workers/parquet-source-worker.js +17 -0
  583. package/dist/workers/parquet-source-worker.js.map +1 -0
  584. package/dist/workers/parquet-worker-node.d.ts +2 -0
  585. package/dist/workers/parquet-worker-node.d.ts.map +1 -0
  586. package/dist/workers/parquet-worker-node.js +5 -0
  587. package/dist/workers/parquet-worker-node.js.map +1 -0
  588. package/dist/workers/parquet-worker.d.ts +2 -0
  589. package/dist/workers/parquet-worker.d.ts.map +1 -0
  590. package/dist/workers/parquet-worker.js +27 -0
  591. package/dist/workers/parquet-worker.js.map +1 -0
  592. package/package.json +84 -40
  593. package/src/bundled.ts +43 -0
  594. package/src/delta-format.ts +12 -0
  595. package/src/delta-source-loader-types.ts +40 -0
  596. package/src/delta-source.ts +305 -0
  597. package/src/delta-types.ts +40 -0
  598. package/src/geoparquet-loader-with-parser.ts +95 -0
  599. package/src/geoparquet-loader.ts +54 -0
  600. package/src/iceberg-rest-catalog.ts +122 -0
  601. package/src/iceberg-table-source.ts +1037 -0
  602. package/src/iceberg-types.ts +165 -0
  603. package/src/index.ts +105 -26
  604. package/src/lib/arrow/convert-columns-to-row-group.ts +3 -0
  605. package/src/lib/arrow/convert-schema-from-parquet.ts +193 -32
  606. package/src/lib/arrow/convert-schema-to-parquet.ts +16 -7
  607. package/src/lib/arrow/create-nested-arrow-vector.ts +705 -0
  608. package/src/lib/constants.ts +0 -2
  609. package/src/lib/encoders/encode-arrow-to-parquet.ts +10 -16
  610. package/src/lib/encoders/encode-table-to-parquet-js.ts +526 -0
  611. package/src/lib/geo/geoparquet-covering.ts +201 -0
  612. package/src/lib/geo/geospatial-metadata.ts +547 -0
  613. package/src/lib/parquet-bloom-filter-planner.ts +86 -0
  614. package/src/lib/parquet-bloom-filter.ts +345 -0
  615. package/src/lib/parquet-encryption.ts +344 -0
  616. package/src/lib/parquet-page-index.ts +727 -0
  617. package/src/lib/parquet-predicate.ts +410 -0
  618. package/src/lib/parquet-source-worker-client.ts +41 -0
  619. package/src/lib/parquet-source-worker-decoder.ts +252 -0
  620. package/src/lib/parquet-source-worker-file.ts +56 -0
  621. package/src/lib/parquet-source-worker-types.ts +132 -0
  622. package/src/lib/parquet-worker-transport.ts +57 -0
  623. package/src/lib/parsers/convert-parquet-tables.ts +39 -0
  624. package/src/lib/parsers/get-parquet-schema.ts +6 -2
  625. package/src/lib/parsers/parse-geoparquet-to-geojson.ts +27 -14
  626. package/src/lib/parsers/parse-parquet-tables.ts +82 -0
  627. package/src/lib/parsers/parse-parquet-to-arrow-js.ts +1213 -0
  628. package/src/lib/parsers/parse-parquet-to-arrow.ts +62 -36
  629. package/src/lib/parsers/parse-parquet-to-columns.ts +9 -8
  630. package/src/lib/parsers/parse-parquet-to-json.ts +119 -35
  631. package/src/lib/slice-parquet-batch.ts +32 -0
  632. package/src/lib/sources/parquet-range-file.ts +365 -0
  633. package/src/lib/sources/single-batch-queue.ts +95 -0
  634. package/src/lib/utils/load-wasm-node.ts +25 -0
  635. package/src/lib/utils/load-wasm.ts +36 -13
  636. package/src/lib/utils/make-stream-iterator.ts +126 -14
  637. package/src/lib/utils/normalize-parquet-options.ts +30 -0
  638. package/src/parquet-dataset-source.ts +800 -0
  639. package/src/parquet-encoded-page-types.ts +122 -0
  640. package/src/parquet-format.ts +2 -0
  641. package/src/parquet-js-loader-types.ts +53 -0
  642. package/src/parquet-js-loader.ts +76 -19
  643. package/src/parquet-js-writer.ts +98 -0
  644. package/src/parquet-loader-base.ts +34 -0
  645. package/src/parquet-loader-options.ts +62 -0
  646. package/src/parquet-loader-types.ts +35 -0
  647. package/src/parquet-loader.ts +115 -0
  648. package/src/parquet-source-capabilities.ts +73 -0
  649. package/src/parquet-source-loader-types.ts +97 -0
  650. package/src/parquet-source-loader.ts +2254 -0
  651. package/src/parquet-source-types.ts +627 -0
  652. package/src/parquet-source-worker-factory.ts +8 -0
  653. package/src/parquet-source-worker-url.ts +23 -0
  654. package/src/parquet-worker-url.ts +23 -0
  655. package/src/parquet-writer.ts +42 -0
  656. package/src/parquetjs/LICENSE +23 -1
  657. package/src/parquetjs/codecs/byte-stream-split.ts +221 -0
  658. package/src/parquetjs/codecs/declare.ts +62 -3
  659. package/src/parquetjs/codecs/delta.ts +1047 -0
  660. package/src/parquetjs/codecs/dictionary.ts +22 -4
  661. package/src/parquetjs/codecs/index.ts +22 -0
  662. package/src/parquetjs/codecs/plain.ts +317 -89
  663. package/src/parquetjs/codecs/rle.ts +852 -82
  664. package/src/parquetjs/compression.ts +137 -156
  665. package/src/parquetjs/encoder/dictionary-planner.ts +104 -0
  666. package/src/parquetjs/encoder/page-planner.ts +83 -0
  667. package/src/parquetjs/encoder/parquet-encoder.ts +1374 -95
  668. package/src/parquetjs/parquet-thrift/BoundingBox.ts +108 -0
  669. package/src/parquetjs/parquet-thrift/BsonType.ts +1 -1
  670. package/src/parquetjs/parquet-thrift/ColumnChunk.ts +37 -2
  671. package/src/parquetjs/parquet-thrift/ColumnCryptoMetaData.ts +139 -0
  672. package/src/parquetjs/parquet-thrift/ColumnIndex.ts +13 -13
  673. package/src/parquetjs/parquet-thrift/ColumnMetaData.ts +75 -2
  674. package/src/parquetjs/parquet-thrift/ColumnOrder.ts +1 -1
  675. package/src/parquetjs/parquet-thrift/DataPageHeader.ts +1 -1
  676. package/src/parquetjs/parquet-thrift/DataPageHeaderV2.ts +1 -1
  677. package/src/parquetjs/parquet-thrift/DateType.ts +1 -1
  678. package/src/parquetjs/parquet-thrift/DecimalType.ts +1 -1
  679. package/src/parquetjs/parquet-thrift/DictionaryPageHeader.ts +1 -1
  680. package/src/parquetjs/parquet-thrift/EdgeInterpolationAlgorithm.ts +13 -0
  681. package/src/parquetjs/parquet-thrift/Encoding.ts +2 -1
  682. package/src/parquetjs/parquet-thrift/EncryptionAlgorithm.ts +127 -0
  683. package/src/parquetjs/parquet-thrift/EnumType.ts +1 -1
  684. package/src/parquetjs/parquet-thrift/FileCryptoMetaData.ts +60 -0
  685. package/src/parquetjs/parquet-thrift/FileMetaData.ts +37 -2
  686. package/src/parquetjs/parquet-thrift/Float16Type.ts +28 -0
  687. package/src/parquetjs/parquet-thrift/GeographyType.ts +54 -0
  688. package/src/parquetjs/parquet-thrift/GeometryType.ts +43 -0
  689. package/src/parquetjs/parquet-thrift/GeospatialStatistics.ts +68 -0
  690. package/src/parquetjs/parquet-thrift/IndexPageHeader.ts +1 -1
  691. package/src/parquetjs/parquet-thrift/IntType.ts +1 -1
  692. package/src/parquetjs/parquet-thrift/JsonType.ts +1 -1
  693. package/src/parquetjs/parquet-thrift/KeyValue.ts +1 -1
  694. package/src/parquetjs/parquet-thrift/ListType.ts +1 -1
  695. package/src/parquetjs/parquet-thrift/LogicalType.ts +97 -1
  696. package/src/parquetjs/parquet-thrift/MapType.ts +1 -1
  697. package/src/parquetjs/parquet-thrift/MicroSeconds.ts +1 -1
  698. package/src/parquetjs/parquet-thrift/MilliSeconds.ts +1 -1
  699. package/src/parquetjs/parquet-thrift/NanoSeconds.ts +28 -0
  700. package/src/parquetjs/parquet-thrift/NullType.ts +1 -1
  701. package/src/parquetjs/parquet-thrift/OffsetIndex.ts +1 -1
  702. package/src/parquetjs/parquet-thrift/PageEncodingStats.ts +1 -1
  703. package/src/parquetjs/parquet-thrift/PageHeader.ts +1 -1
  704. package/src/parquetjs/parquet-thrift/PageLocation.ts +2 -2
  705. package/src/parquetjs/parquet-thrift/RowGroup.ts +2 -2
  706. package/src/parquetjs/parquet-thrift/SchemaElement.ts +1 -1
  707. package/src/parquetjs/parquet-thrift/SizeStatistics.ts +98 -0
  708. package/src/parquetjs/parquet-thrift/SortingColumn.ts +1 -1
  709. package/src/parquetjs/parquet-thrift/Statistics.ts +50 -14
  710. package/src/parquetjs/parquet-thrift/StringType.ts +1 -1
  711. package/src/parquetjs/parquet-thrift/TimeType.ts +1 -1
  712. package/src/parquetjs/parquet-thrift/TimeUnit.ts +25 -1
  713. package/src/parquetjs/parquet-thrift/TimestampType.ts +1 -1
  714. package/src/parquetjs/parquet-thrift/TypeDefinedOrder.ts +1 -1
  715. package/src/parquetjs/parquet-thrift/UUIDType.ts +1 -1
  716. package/src/parquetjs/parquet-thrift/VariantType.ts +43 -0
  717. package/src/parquetjs/parquet-thrift/index.ts +12 -4
  718. package/src/parquetjs/parser/decoders.ts +922 -143
  719. package/src/parquetjs/parser/encoded-pages.ts +278 -0
  720. package/src/parquetjs/parser/parquet-reader.ts +1334 -83
  721. package/src/parquetjs/schema/declare.ts +121 -9
  722. package/src/parquetjs/schema/schema.ts +33 -3
  723. package/src/parquetjs/schema/shred.ts +345 -41
  724. package/src/parquetjs/schema/types.ts +308 -84
  725. package/src/parquetjs/schema/variant.ts +313 -0
  726. package/src/parquetjs/utils/binary-utils.ts +132 -0
  727. package/src/parquetjs/utils/crc32.ts +15 -0
  728. package/src/parquetjs/utils/file-utils.ts +4 -4
  729. package/src/parquetjs/utils/read-utils.ts +77 -51
  730. package/src/parquetjs/utils/thrift-runtime.ts +109 -0
  731. package/src/parquetjs/utils/uint8-array-compact-protocol-writer.ts +254 -0
  732. package/src/parquetjs/utils/uint8-array-compact-protocol.ts +322 -0
  733. package/src/parquetjs/utils/uint8-array-transport.ts +79 -0
  734. package/src/polyfills/process.ts +17 -0
  735. package/src/unbundled.ts +47 -0
  736. package/src/wasm.ts +6 -0
  737. package/src/workers/parquet-source-worker.ts +21 -0
  738. package/src/workers/parquet-worker-node.ts +5 -0
  739. package/src/workers/parquet-worker.ts +31 -0
  740. package/dist/parquet-arrow-loader.d.ts +0 -43
  741. package/dist/parquet-arrow-loader.d.ts.map +0 -1
  742. package/dist/parquet-arrow-loader.js +0 -49
  743. package/dist/parquet-arrow-loader.js.map +0 -1
  744. package/dist/parquet-arrow-writer.d.ts +0 -9
  745. package/dist/parquet-arrow-writer.d.ts.map +0 -1
  746. package/dist/parquet-arrow-writer.js +0 -23
  747. package/dist/parquet-arrow-writer.js.map +0 -1
  748. package/dist/parquet-json-loader.d.ts +0 -148
  749. package/dist/parquet-json-loader.d.ts.map +0 -1
  750. package/dist/parquet-json-loader.js +0 -106
  751. package/dist/parquet-json-loader.js.map +0 -1
  752. package/dist/parquet-json-writer.d.ts +0 -19
  753. package/dist/parquet-json-writer.d.ts.map +0 -1
  754. package/dist/parquet-json-writer.js +0 -20
  755. package/dist/parquet-json-writer.js.map +0 -1
  756. package/dist/polyfills/buffer/buffer-polyfill.browser.d.ts +0 -3
  757. package/dist/polyfills/buffer/buffer-polyfill.browser.d.ts.map +0 -1
  758. package/dist/polyfills/buffer/buffer-polyfill.browser.js +0 -12
  759. package/dist/polyfills/buffer/buffer-polyfill.browser.js.map +0 -1
  760. package/dist/polyfills/buffer/buffer-polyfill.node.d.ts +0 -3
  761. package/dist/polyfills/buffer/buffer-polyfill.node.d.ts.map +0 -1
  762. package/dist/polyfills/buffer/buffer-polyfill.node.js +0 -17
  763. package/dist/polyfills/buffer/buffer-polyfill.node.js.map +0 -1
  764. package/dist/polyfills/buffer/buffer.d.ts +0 -222
  765. package/dist/polyfills/buffer/buffer.d.ts.map +0 -1
  766. package/dist/polyfills/buffer/buffer.js +0 -1868
  767. package/dist/polyfills/buffer/buffer.js.map +0 -1
  768. package/dist/polyfills/buffer/index.d.ts +0 -4
  769. package/dist/polyfills/buffer/index.d.ts.map +0 -1
  770. package/dist/polyfills/buffer/index.js +0 -9
  771. package/dist/polyfills/buffer/index.js.map +0 -1
  772. package/dist/polyfills/buffer/install-buffer-polyfill.d.ts +0 -2
  773. package/dist/polyfills/buffer/install-buffer-polyfill.d.ts.map +0 -1
  774. package/dist/polyfills/buffer/install-buffer-polyfill.js +0 -10
  775. package/dist/polyfills/buffer/install-buffer-polyfill.js.map +0 -1
  776. package/dist/workers/parquet-json-worker.d.ts +0 -2
  777. package/dist/workers/parquet-json-worker.d.ts.map +0 -1
  778. package/dist/workers/parquet-json-worker.js +0 -7
  779. package/dist/workers/parquet-json-worker.js.map +0 -1
  780. package/src/parquet-arrow-loader.ts +0 -74
  781. package/src/parquet-arrow-writer.ts +0 -33
  782. package/src/parquet-json-loader.ts +0 -160
  783. package/src/parquet-json-writer.ts +0 -27
  784. package/src/parquetjs/modules.d.ts +0 -25
  785. package/src/polyfills/buffer/buffer-polyfill.browser.ts +0 -14
  786. package/src/polyfills/buffer/buffer-polyfill.node.ts +0 -19
  787. package/src/polyfills/buffer/buffer.ts +0 -2205
  788. package/src/polyfills/buffer/index.ts +0 -11
  789. package/src/polyfills/buffer/install-buffer-polyfill.ts +0 -12
  790. package/src/workers/parquet-json-worker.ts +0 -8
@@ -0,0 +1,705 @@
1
+ // loaders.gl
2
+ // SPDX-License-Identifier: MIT
3
+ // Copyright (c) vis.gl contributors
4
+
5
+ import * as arrow from 'apache-arrow';
6
+
7
+ import type {
8
+ ParquetColumnChunk,
9
+ ParquetField,
10
+ ParquetRowGroup
11
+ } from '../../parquetjs/schema/declare';
12
+
13
+ /** Largest byte value copied inline to avoid TypedArray#set call overhead. */
14
+ const MAXIMUM_INLINE_BYTE_COPY_LENGTH = 7;
15
+
16
+ /** Primitive Arrow arrays emitted directly from decoded physical Parquet values. */
17
+ type NestedPrimitiveArrowArray =
18
+ | Float32Array
19
+ | Float64Array
20
+ | Int8Array
21
+ | Int16Array
22
+ | Int32Array
23
+ | Uint8Array
24
+ | Uint16Array
25
+ | Uint32Array
26
+ | BigInt64Array
27
+ | BigUint64Array;
28
+
29
+ /** Selected decoded values and their Arrow validity information. */
30
+ type SelectedLeafValues = {
31
+ length: number;
32
+ firstValueIndex: number;
33
+ valueCount: number;
34
+ nullBitmap?: Uint8Array;
35
+ nullCount: number;
36
+ };
37
+
38
+ /** Repeated-list offsets plus the selected primitive values contained by the lists. */
39
+ type SelectedRepeatedValues = SelectedLeafValues & {
40
+ valueOffsets: Int32Array;
41
+ };
42
+
43
+ /**
44
+ * Creates an Arrow List vector directly from a repeated Parquet primitive or repeated struct.
45
+ * Returns undefined for shapes whose repetition tree is not supported by the direct path.
46
+ */
47
+ export function createNestedArrowVector(
48
+ arrowType: arrow.DataType,
49
+ parquetField: ParquetField,
50
+ rowGroup: ParquetRowGroup,
51
+ start: number,
52
+ end: number
53
+ ): arrow.Vector | undefined {
54
+ if (!(arrowType instanceof arrow.List) || parquetField.repetitionType !== 'REPEATED') {
55
+ return undefined;
56
+ }
57
+
58
+ const data = parquetField.fields
59
+ ? createRepeatedStructData(arrowType, parquetField, rowGroup, start, end)
60
+ : createRepeatedPrimitiveData(arrowType, parquetField, rowGroup, start, end);
61
+ return data ? new arrow.Vector([data]) : undefined;
62
+ }
63
+
64
+ /** Creates a top-level repeated primitive as Arrow List data. */
65
+ function createRepeatedPrimitiveData(
66
+ listType: arrow.List,
67
+ parquetField: ParquetField,
68
+ rowGroup: ParquetRowGroup,
69
+ start: number,
70
+ end: number
71
+ ): arrow.Data | undefined {
72
+ const columnData = rowGroup.columnData[parquetField.key];
73
+ if (!columnData) {
74
+ return undefined;
75
+ }
76
+
77
+ const selected = selectRootRepeatedValues(columnData, parquetField, start, end);
78
+ const childData = createSelectedLeafData(
79
+ listType.children[0].type,
80
+ parquetField,
81
+ columnData,
82
+ selected
83
+ );
84
+ if (!childData) {
85
+ return undefined;
86
+ }
87
+
88
+ return arrow.makeData({
89
+ type: listType,
90
+ valueOffsets: selected.valueOffsets,
91
+ child: childData
92
+ } as any);
93
+ }
94
+
95
+ /** Creates a top-level repeated group as Arrow List<Struct> data. */
96
+ function createRepeatedStructData(
97
+ listType: arrow.List,
98
+ parquetField: ParquetField,
99
+ rowGroup: ParquetRowGroup,
100
+ start: number,
101
+ end: number
102
+ ): arrow.Data | undefined {
103
+ const structType = listType.children[0].type;
104
+ if (!(structType instanceof arrow.Struct)) {
105
+ return undefined;
106
+ }
107
+
108
+ const primitiveLeaves = listPrimitiveLeaves(parquetField);
109
+ const driverField = primitiveLeaves.reduce<ParquetField | undefined>(
110
+ (driver, leaf) => (!driver || leaf.rLevelMax < driver.rLevelMax ? leaf : driver),
111
+ undefined
112
+ );
113
+ const driverColumn = driverField && rowGroup.columnData[driverField.key];
114
+ if (!driverField || !driverColumn) {
115
+ return undefined;
116
+ }
117
+
118
+ const rootLayout = selectRepeatedStructLayout(driverColumn, parquetField, start, end);
119
+ const children: arrow.Data[] = [];
120
+ for (const arrowChild of structType.children) {
121
+ const parquetChild = parquetField.fields?.[arrowChild.name];
122
+ if (!parquetChild || parquetChild.fields) {
123
+ return undefined;
124
+ }
125
+ const columnData = rowGroup.columnData[parquetChild.key];
126
+ if (!columnData) {
127
+ return undefined;
128
+ }
129
+
130
+ let childData: arrow.Data | undefined;
131
+ if (parquetChild.repetitionType === 'REPEATED' && arrowChild.type instanceof arrow.List) {
132
+ childData = createRepeatedStructChildData(
133
+ arrowChild.type,
134
+ parquetField,
135
+ parquetChild,
136
+ columnData,
137
+ start,
138
+ end,
139
+ rootLayout.elementCount
140
+ );
141
+ } else {
142
+ const selected = selectStructScalarValues(
143
+ columnData,
144
+ parquetField,
145
+ parquetChild,
146
+ start,
147
+ end,
148
+ rootLayout.elementCount
149
+ );
150
+ childData = createSelectedLeafData(arrowChild.type, parquetChild, columnData, selected);
151
+ }
152
+ if (!childData) {
153
+ return undefined;
154
+ }
155
+ children.push(childData);
156
+ }
157
+
158
+ const structData = arrow.makeData({
159
+ type: structType,
160
+ length: rootLayout.elementCount,
161
+ nullCount: 0,
162
+ children
163
+ });
164
+ return arrow.makeData({
165
+ type: listType,
166
+ valueOffsets: rootLayout.valueOffsets,
167
+ child: structData
168
+ } as any);
169
+ }
170
+
171
+ /** Creates one repeated primitive child within a repeated struct. */
172
+ function createRepeatedStructChildData(
173
+ listType: arrow.List,
174
+ parentField: ParquetField,
175
+ parquetField: ParquetField,
176
+ columnData: ParquetColumnChunk,
177
+ start: number,
178
+ end: number,
179
+ parentCount: number
180
+ ): arrow.Data | undefined {
181
+ const selected = selectStructRepeatedValues(
182
+ columnData,
183
+ parentField,
184
+ parquetField,
185
+ start,
186
+ end,
187
+ parentCount
188
+ );
189
+ const childData = createSelectedLeafData(
190
+ listType.children[0].type,
191
+ parquetField,
192
+ columnData,
193
+ selected
194
+ );
195
+ if (!childData) {
196
+ return undefined;
197
+ }
198
+ return arrow.makeData({
199
+ type: listType,
200
+ valueOffsets: selected.valueOffsets,
201
+ child: childData
202
+ } as any);
203
+ }
204
+
205
+ /** Selects primitive values and row offsets for a top-level repeated leaf. */
206
+ function selectRootRepeatedValues(
207
+ columnData: ParquetColumnChunk,
208
+ parquetField: ParquetField,
209
+ start: number,
210
+ end: number
211
+ ): SelectedRepeatedValues {
212
+ const rowCount = end - start;
213
+ const valueCounts = new Int32Array(rowCount);
214
+ let rowIndex = -1;
215
+ let sourceValueIndex = 0;
216
+ let firstValueIndex = -1;
217
+ let valueCount = 0;
218
+
219
+ for (let levelIndex = 0; levelIndex < columnData.dlevels.length; levelIndex++) {
220
+ if (columnData.rlevels[levelIndex] === 0) {
221
+ rowIndex++;
222
+ }
223
+ const hasValue = columnData.dlevels[levelIndex] === parquetField.dLevelMax;
224
+ if (rowIndex >= start && rowIndex < end && hasValue) {
225
+ firstValueIndex = firstValueIndex < 0 ? sourceValueIndex : firstValueIndex;
226
+ valueCounts[rowIndex - start]++;
227
+ valueCount++;
228
+ }
229
+ if (hasValue) {
230
+ sourceValueIndex++;
231
+ }
232
+ if (rowIndex >= end) {
233
+ break;
234
+ }
235
+ }
236
+
237
+ return {
238
+ length: valueCount,
239
+ firstValueIndex: Math.max(0, firstValueIndex),
240
+ valueCount,
241
+ nullCount: 0,
242
+ valueOffsets: accumulateOffsets(valueCounts)
243
+ };
244
+ }
245
+
246
+ /** Selects row offsets and element count for a top-level repeated group. */
247
+ function selectRepeatedStructLayout(
248
+ columnData: ParquetColumnChunk,
249
+ parentField: ParquetField,
250
+ start: number,
251
+ end: number
252
+ ): {valueOffsets: Int32Array; elementCount: number} {
253
+ const valueCounts = new Int32Array(end - start);
254
+ let rowIndex = -1;
255
+ let elementCount = 0;
256
+
257
+ for (let levelIndex = 0; levelIndex < columnData.dlevels.length; levelIndex++) {
258
+ const repetitionLevel = columnData.rlevels[levelIndex];
259
+ if (repetitionLevel === 0) {
260
+ rowIndex++;
261
+ }
262
+ if (
263
+ rowIndex >= start &&
264
+ rowIndex < end &&
265
+ repetitionLevel <= parentField.rLevelMax &&
266
+ columnData.dlevels[levelIndex] >= parentField.dLevelMax
267
+ ) {
268
+ valueCounts[rowIndex - start]++;
269
+ elementCount++;
270
+ }
271
+ if (rowIndex >= end) {
272
+ break;
273
+ }
274
+ }
275
+
276
+ return {valueOffsets: accumulateOffsets(valueCounts), elementCount};
277
+ }
278
+
279
+ /** Selects scalar child values aligned to the repeated parent struct elements. */
280
+ function selectStructScalarValues(
281
+ columnData: ParquetColumnChunk,
282
+ parentField: ParquetField,
283
+ parquetField: ParquetField,
284
+ start: number,
285
+ end: number,
286
+ parentCount: number
287
+ ): SelectedLeafValues {
288
+ const nullBitmap = new Uint8Array(Math.ceil(parentCount / 8));
289
+ let rowIndex = -1;
290
+ let sourceValueIndex = 0;
291
+ let firstValueIndex = -1;
292
+ let outputIndex = 0;
293
+ let valueCount = 0;
294
+
295
+ for (let levelIndex = 0; levelIndex < columnData.dlevels.length; levelIndex++) {
296
+ const repetitionLevel = columnData.rlevels[levelIndex];
297
+ if (repetitionLevel === 0) {
298
+ rowIndex++;
299
+ }
300
+ const definitionLevel = columnData.dlevels[levelIndex];
301
+ const hasValue = definitionLevel === parquetField.dLevelMax;
302
+ if (
303
+ rowIndex >= start &&
304
+ rowIndex < end &&
305
+ repetitionLevel <= parentField.rLevelMax &&
306
+ definitionLevel >= parentField.dLevelMax
307
+ ) {
308
+ if (hasValue) {
309
+ firstValueIndex = firstValueIndex < 0 ? sourceValueIndex : firstValueIndex;
310
+ nullBitmap[outputIndex >> 3] |= 1 << (outputIndex & 7);
311
+ valueCount++;
312
+ }
313
+ outputIndex++;
314
+ }
315
+ if (hasValue) {
316
+ sourceValueIndex++;
317
+ }
318
+ if (rowIndex >= end) {
319
+ break;
320
+ }
321
+ }
322
+
323
+ if (outputIndex !== parentCount) {
324
+ throw new Error(
325
+ `Parquet nested column ${parquetField.key} has ${outputIndex} parent values; expected ${parentCount}`
326
+ );
327
+ }
328
+ const nullCount = parentCount - valueCount;
329
+ return {
330
+ length: parentCount,
331
+ firstValueIndex: Math.max(0, firstValueIndex),
332
+ valueCount,
333
+ nullBitmap: nullCount ? nullBitmap : undefined,
334
+ nullCount
335
+ };
336
+ }
337
+
338
+ /** Selects repeated child values and list offsets aligned to repeated parent struct elements. */
339
+ function selectStructRepeatedValues(
340
+ columnData: ParquetColumnChunk,
341
+ parentField: ParquetField,
342
+ parquetField: ParquetField,
343
+ start: number,
344
+ end: number,
345
+ parentCount: number
346
+ ): SelectedRepeatedValues {
347
+ const valueCounts = new Int32Array(parentCount);
348
+ let rowIndex = -1;
349
+ let parentIndex = -1;
350
+ let sourceValueIndex = 0;
351
+ let firstValueIndex = -1;
352
+ let valueCount = 0;
353
+
354
+ for (let levelIndex = 0; levelIndex < columnData.dlevels.length; levelIndex++) {
355
+ const repetitionLevel = columnData.rlevels[levelIndex];
356
+ if (repetitionLevel === 0) {
357
+ rowIndex++;
358
+ }
359
+ const definitionLevel = columnData.dlevels[levelIndex];
360
+ const hasValue = definitionLevel === parquetField.dLevelMax;
361
+ if (rowIndex >= start && rowIndex < end) {
362
+ if (repetitionLevel <= parentField.rLevelMax && definitionLevel >= parentField.dLevelMax) {
363
+ parentIndex++;
364
+ }
365
+ if (hasValue) {
366
+ if (parentIndex < 0 || parentIndex >= parentCount) {
367
+ throw new Error(
368
+ `Parquet nested column ${parquetField.key} has invalid repetition levels`
369
+ );
370
+ }
371
+ firstValueIndex = firstValueIndex < 0 ? sourceValueIndex : firstValueIndex;
372
+ valueCounts[parentIndex]++;
373
+ valueCount++;
374
+ }
375
+ }
376
+ if (hasValue) {
377
+ sourceValueIndex++;
378
+ }
379
+ if (rowIndex >= end) {
380
+ break;
381
+ }
382
+ }
383
+
384
+ if (parentIndex + 1 !== parentCount) {
385
+ throw new Error(
386
+ `Parquet nested column ${parquetField.key} has ${parentIndex + 1} parent values; expected ${parentCount}`
387
+ );
388
+ }
389
+ return {
390
+ length: valueCount,
391
+ firstValueIndex: Math.max(0, firstValueIndex),
392
+ valueCount,
393
+ nullCount: 0,
394
+ valueOffsets: accumulateOffsets(valueCounts)
395
+ };
396
+ }
397
+
398
+ /** Creates Arrow primitive or byte data for selected decoded leaf values. */
399
+ function createSelectedLeafData(
400
+ arrowType: arrow.DataType,
401
+ parquetField: ParquetField,
402
+ columnData: ParquetColumnChunk,
403
+ selected: SelectedLeafValues
404
+ ): arrow.Data | undefined {
405
+ const primitiveData = createSelectedPrimitiveData(arrowType, parquetField, columnData, selected);
406
+ if (primitiveData) {
407
+ return primitiveData;
408
+ }
409
+ return createSelectedByteData(arrowType, parquetField, columnData, selected);
410
+ }
411
+
412
+ /** Creates fixed-width Arrow data for selected physical primitive values. */
413
+ function createSelectedPrimitiveData(
414
+ arrowType: arrow.DataType,
415
+ parquetField: ParquetField,
416
+ columnData: ParquetColumnChunk,
417
+ selected: SelectedLeafValues
418
+ ): arrow.Data | undefined {
419
+ const directData = createSelectedPrimitiveArrayView(
420
+ arrowType,
421
+ parquetField,
422
+ columnData,
423
+ selected
424
+ );
425
+ if (directData) {
426
+ return arrow.makeData({type: arrowType, data: directData, nullCount: 0} as any);
427
+ }
428
+
429
+ const data = createNestedPrimitiveArray(arrowType, parquetField, selected.length);
430
+ if (!data) {
431
+ return undefined;
432
+ }
433
+
434
+ const sourceValues = columnData.values;
435
+ let sourceValueIndex = selected.firstValueIndex;
436
+ for (let outputIndex = 0; outputIndex < selected.length; outputIndex++) {
437
+ if (!selected.nullBitmap || selected.nullBitmap[outputIndex >> 3] & (1 << (outputIndex & 7))) {
438
+ setNestedPrimitiveValue(data, outputIndex, sourceValues[sourceValueIndex++]);
439
+ }
440
+ }
441
+
442
+ return arrow.makeData({
443
+ type: arrowType,
444
+ data,
445
+ nullBitmap: selected.nullBitmap,
446
+ nullCount: selected.nullCount
447
+ } as any);
448
+ }
449
+
450
+ /** Reuses a contiguous decoded primitive range when nested selection needs no null expansion. */
451
+ function createSelectedPrimitiveArrayView(
452
+ arrowType: arrow.DataType,
453
+ parquetField: ParquetField,
454
+ columnData: ParquetColumnChunk,
455
+ selected: SelectedLeafValues
456
+ ): NestedPrimitiveArrowArray | undefined {
457
+ if (selected.nullCount || selected.length !== selected.valueCount) {
458
+ return undefined;
459
+ }
460
+ const values = getMatchingNestedPrimitiveArray(arrowType, parquetField, columnData.values);
461
+ return values?.subarray(
462
+ selected.firstValueIndex,
463
+ selected.firstValueIndex + selected.valueCount
464
+ ) as NestedPrimitiveArrowArray | undefined;
465
+ }
466
+
467
+ /** Returns decoded nested values when their physical array exactly matches the Arrow type. */
468
+ function getMatchingNestedPrimitiveArray(
469
+ arrowType: arrow.DataType,
470
+ parquetField: ParquetField,
471
+ values: ParquetColumnChunk['values']
472
+ ): NestedPrimitiveArrowArray | undefined {
473
+ if (
474
+ parquetField.primitiveType === 'FLOAT' &&
475
+ arrowType instanceof arrow.Float32 &&
476
+ values instanceof Float32Array
477
+ ) {
478
+ return values;
479
+ }
480
+ if (
481
+ parquetField.primitiveType === 'DOUBLE' &&
482
+ arrowType instanceof arrow.Float64 &&
483
+ values instanceof Float64Array
484
+ ) {
485
+ return values;
486
+ }
487
+ if (parquetField.primitiveType === 'INT32') {
488
+ if (arrowType instanceof arrow.Int8 && values instanceof Int8Array) return values;
489
+ if (arrowType instanceof arrow.Int16 && values instanceof Int16Array) return values;
490
+ if (
491
+ (arrowType instanceof arrow.Int32 ||
492
+ arrowType instanceof arrow.DateDay ||
493
+ arrowType instanceof arrow.TimeMillisecond) &&
494
+ values instanceof Int32Array
495
+ ) {
496
+ return values;
497
+ }
498
+ if (arrowType instanceof arrow.Uint8 && values instanceof Uint8Array) return values;
499
+ if (arrowType instanceof arrow.Uint16 && values instanceof Uint16Array) return values;
500
+ if (arrowType instanceof arrow.Uint32 && values instanceof Uint32Array) return values;
501
+ }
502
+ if (parquetField.primitiveType === 'INT64') {
503
+ if (
504
+ (arrowType instanceof arrow.Int64 ||
505
+ arrowType instanceof arrow.TimeMicrosecond ||
506
+ arrowType instanceof arrow.TimeNanosecond ||
507
+ arrowType instanceof arrow.TimestampMillisecond ||
508
+ arrowType instanceof arrow.TimestampMicrosecond ||
509
+ arrowType instanceof arrow.TimestampNanosecond) &&
510
+ values instanceof BigInt64Array
511
+ ) {
512
+ return values;
513
+ }
514
+ if (arrowType instanceof arrow.Uint64 && values instanceof BigUint64Array) return values;
515
+ }
516
+ return undefined;
517
+ }
518
+
519
+ /** Allocates the Arrow array matching an unconverted physical Parquet primitive. */
520
+ function createNestedPrimitiveArray(
521
+ arrowType: arrow.DataType,
522
+ parquetField: ParquetField,
523
+ length: number
524
+ ): NestedPrimitiveArrowArray | undefined {
525
+ if (parquetField.primitiveType === 'FLOAT' && arrowType instanceof arrow.Float32) {
526
+ return new Float32Array(length);
527
+ }
528
+ if (parquetField.primitiveType === 'DOUBLE' && arrowType instanceof arrow.Float64) {
529
+ return new Float64Array(length);
530
+ }
531
+ if (parquetField.primitiveType === 'INT32' && arrowType instanceof arrow.Int8) {
532
+ return new Int8Array(length);
533
+ }
534
+ if (parquetField.primitiveType === 'INT32' && arrowType instanceof arrow.Int16) {
535
+ return new Int16Array(length);
536
+ }
537
+ if (
538
+ parquetField.primitiveType === 'INT32' &&
539
+ (arrowType instanceof arrow.Int32 ||
540
+ arrowType instanceof arrow.DateDay ||
541
+ arrowType instanceof arrow.TimeMillisecond)
542
+ ) {
543
+ return new Int32Array(length);
544
+ }
545
+ if (parquetField.primitiveType === 'INT32' && arrowType instanceof arrow.Uint8) {
546
+ return new Uint8Array(length);
547
+ }
548
+ if (parquetField.primitiveType === 'INT32' && arrowType instanceof arrow.Uint16) {
549
+ return new Uint16Array(length);
550
+ }
551
+ if (parquetField.primitiveType === 'INT32' && arrowType instanceof arrow.Uint32) {
552
+ return new Uint32Array(length);
553
+ }
554
+ if (
555
+ parquetField.primitiveType === 'INT64' &&
556
+ (arrowType instanceof arrow.Int64 ||
557
+ arrowType instanceof arrow.TimeMicrosecond ||
558
+ arrowType instanceof arrow.TimeNanosecond ||
559
+ arrowType instanceof arrow.TimestampMillisecond ||
560
+ arrowType instanceof arrow.TimestampMicrosecond ||
561
+ arrowType instanceof arrow.TimestampNanosecond)
562
+ ) {
563
+ return new BigInt64Array(length);
564
+ }
565
+ if (parquetField.primitiveType === 'INT64' && arrowType instanceof arrow.Uint64) {
566
+ return new BigUint64Array(length);
567
+ }
568
+ return undefined;
569
+ }
570
+
571
+ /** Writes one decoded primitive into the exact array expected by Arrow. */
572
+ function setNestedPrimitiveValue(
573
+ data: NestedPrimitiveArrowArray,
574
+ index: number,
575
+ value: unknown
576
+ ): void {
577
+ if (data instanceof BigInt64Array) {
578
+ data[index] = typeof value === 'bigint' ? value : BigInt(value as number | string);
579
+ } else if (data instanceof BigUint64Array) {
580
+ data[index] = BigInt.asUintN(
581
+ 64,
582
+ typeof value === 'bigint' ? value : BigInt(value as number | string)
583
+ );
584
+ } else {
585
+ data[index] = Number(value);
586
+ }
587
+ }
588
+
589
+ /** Creates Arrow Utf8 or Binary data for selected decoded byte values. */
590
+ function createSelectedByteData(
591
+ arrowType: arrow.DataType,
592
+ parquetField: ParquetField,
593
+ columnData: ParquetColumnChunk,
594
+ selected: SelectedLeafValues
595
+ ): arrow.Data | undefined {
596
+ if (!supportsNestedByteData(arrowType, parquetField)) {
597
+ return undefined;
598
+ }
599
+ if (columnData.byteArrayData) {
600
+ return createSelectedBufferedByteData(arrowType, columnData, selected);
601
+ }
602
+
603
+ const byteValues = columnData.values as Uint8Array[];
604
+ const valueOffsets = new Int32Array(selected.length + 1);
605
+ let sourceValueIndex = selected.firstValueIndex;
606
+ let dataByteLength = 0;
607
+ for (let outputIndex = 0; outputIndex < selected.length; outputIndex++) {
608
+ if (!selected.nullBitmap || selected.nullBitmap[outputIndex >> 3] & (1 << (outputIndex & 7))) {
609
+ dataByteLength += byteValues[sourceValueIndex++].byteLength;
610
+ if (dataByteLength > 0x7fffffff) {
611
+ throw new Error('Arrow Utf8/Binary column exceeds the 32-bit offset range');
612
+ }
613
+ }
614
+ valueOffsets[outputIndex + 1] = dataByteLength;
615
+ }
616
+
617
+ const data = new Uint8Array(dataByteLength);
618
+ let dataOffset = 0;
619
+ const valueEnd = selected.firstValueIndex + selected.valueCount;
620
+ for (let valueIndex = selected.firstValueIndex; valueIndex < valueEnd; valueIndex++) {
621
+ const bytes = byteValues[valueIndex];
622
+ if (bytes.byteLength <= MAXIMUM_INLINE_BYTE_COPY_LENGTH) {
623
+ for (let byteIndex = 0; byteIndex < bytes.byteLength; byteIndex++) {
624
+ data[dataOffset++] = bytes[byteIndex];
625
+ }
626
+ } else {
627
+ data.set(bytes, dataOffset);
628
+ dataOffset += bytes.byteLength;
629
+ }
630
+ }
631
+
632
+ return arrow.makeData({
633
+ type: arrowType,
634
+ valueOffsets,
635
+ data,
636
+ nullBitmap: selected.nullBitmap,
637
+ nullCount: selected.nullCount
638
+ } as any);
639
+ }
640
+
641
+ /** Creates selected nested Arrow bytes from the decoder's compact physical value buffer. */
642
+ function createSelectedBufferedByteData(
643
+ arrowType: arrow.DataType,
644
+ columnData: ParquetColumnChunk,
645
+ selected: SelectedLeafValues
646
+ ): arrow.Data {
647
+ const byteArrayData = columnData.byteArrayData!;
648
+ const valueOffsets = new Int32Array(selected.length + 1);
649
+ let sourceValueIndex = selected.firstValueIndex;
650
+ let dataByteLength = 0;
651
+ for (let outputIndex = 0; outputIndex < selected.length; outputIndex++) {
652
+ if (!selected.nullBitmap || selected.nullBitmap[outputIndex >> 3] & (1 << (outputIndex & 7))) {
653
+ dataByteLength +=
654
+ byteArrayData.valueOffsets[sourceValueIndex + 1] -
655
+ byteArrayData.valueOffsets[sourceValueIndex];
656
+ sourceValueIndex++;
657
+ }
658
+ valueOffsets[outputIndex + 1] = dataByteLength;
659
+ }
660
+ const firstByteOffset = byteArrayData.valueOffsets[selected.firstValueIndex];
661
+ const lastByteOffset = byteArrayData.valueOffsets[selected.firstValueIndex + selected.valueCount];
662
+ return arrow.makeData({
663
+ type: arrowType,
664
+ valueOffsets,
665
+ data: byteArrayData.data.subarray(firstByteOffset, lastByteOffset),
666
+ nullBitmap: selected.nullBitmap,
667
+ nullCount: selected.nullCount
668
+ } as any);
669
+ }
670
+
671
+ /** Returns whether one Parquet byte leaf maps directly to Arrow Utf8 or Binary. */
672
+ function supportsNestedByteData(arrowType: arrow.DataType, parquetField: ParquetField): boolean {
673
+ if (arrowType instanceof arrow.Utf8) {
674
+ return parquetField.originalType === 'UTF8' || parquetField.originalType === 'ENUM';
675
+ }
676
+ return (
677
+ arrowType instanceof arrow.Binary &&
678
+ (!parquetField.originalType ||
679
+ parquetField.originalType === 'JSON' ||
680
+ parquetField.originalType === 'BSON' ||
681
+ parquetField.originalType === 'INTERVAL' ||
682
+ parquetField.originalType === 'VARIANT' ||
683
+ parquetField.originalType === 'GEOMETRY' ||
684
+ parquetField.originalType === 'GEOGRAPHY') &&
685
+ (parquetField.primitiveType === 'BYTE_ARRAY' ||
686
+ parquetField.primitiveType === 'FIXED_LEN_BYTE_ARRAY')
687
+ );
688
+ }
689
+
690
+ /** Lists all primitive descendant fields in schema order. */
691
+ function listPrimitiveLeaves(parquetField: ParquetField): ParquetField[] {
692
+ if (!parquetField.fields) {
693
+ return [parquetField];
694
+ }
695
+ return Object.values(parquetField.fields).flatMap(listPrimitiveLeaves);
696
+ }
697
+
698
+ /** Converts per-parent value counts into Arrow list offsets. */
699
+ function accumulateOffsets(valueCounts: Int32Array): Int32Array {
700
+ const valueOffsets = new Int32Array(valueCounts.length + 1);
701
+ for (let index = 0; index < valueCounts.length; index++) {
702
+ valueOffsets[index + 1] = valueOffsets[index] + valueCounts[index];
703
+ }
704
+ return valueOffsets;
705
+ }
@@ -6,8 +6,6 @@
6
6
  // @ts-ignore TS2304: Cannot find name '__VERSION__'.
7
7
  export const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
8
8
 
9
- export const PARQUET_WASM_URL = 'https://unpkg.com/parquet-wasm@0.6.1/esm/parquet_wasm_bg.wasm';
10
-
11
9
  /**
12
10
  * Parquet File Magic String
13
11
  */