@loaders.gl/parquet 3.3.0 → 3.4.0-alpha.1

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 (131) hide show
  1. package/dist/dist.min.js +26 -17
  2. package/dist/dist.min.js.map +3 -3
  3. package/dist/es5/index.js +3 -3
  4. package/dist/es5/index.js.map +1 -1
  5. package/dist/es5/lib/parse-parquet.js +25 -49
  6. package/dist/es5/lib/parse-parquet.js.map +1 -1
  7. package/dist/es5/parquet-loader.js +2 -3
  8. package/dist/es5/parquet-loader.js.map +1 -1
  9. package/dist/es5/parquet-wasm-loader.js +1 -1
  10. package/dist/es5/parquet-wasm-loader.js.map +1 -1
  11. package/dist/es5/parquet-wasm-writer.js +1 -1
  12. package/dist/es5/parquet-wasm-writer.js.map +1 -1
  13. package/dist/es5/parquet-writer.js +1 -1
  14. package/dist/es5/parquet-writer.js.map +1 -1
  15. package/dist/es5/parquetjs/compression.js +5 -15
  16. package/dist/es5/parquetjs/compression.js.map +1 -1
  17. package/dist/es5/parquetjs/encoder/{parquet-encoder.js → writer.js} +158 -70
  18. package/dist/es5/parquetjs/encoder/writer.js.map +1 -0
  19. package/dist/es5/parquetjs/file.js +94 -0
  20. package/dist/es5/parquetjs/file.js.map +1 -0
  21. package/dist/es5/parquetjs/parser/parquet-cursor.js +183 -0
  22. package/dist/es5/parquetjs/parser/parquet-cursor.js.map +1 -0
  23. package/dist/es5/parquetjs/parser/parquet-envelope-reader.js +327 -0
  24. package/dist/es5/parquetjs/parser/parquet-envelope-reader.js.map +1 -0
  25. package/dist/es5/parquetjs/parser/parquet-reader.js +222 -553
  26. package/dist/es5/parquetjs/parser/parquet-reader.js.map +1 -1
  27. package/dist/es5/parquetjs/schema/declare.js +1 -3
  28. package/dist/es5/parquetjs/schema/declare.js.map +1 -1
  29. package/dist/es5/parquetjs/schema/shred.js +33 -39
  30. package/dist/es5/parquetjs/schema/shred.js.map +1 -1
  31. package/dist/es5/parquetjs/schema/types.js.map +1 -1
  32. package/dist/es5/parquetjs/utils/buffer-utils.js +19 -0
  33. package/dist/es5/parquetjs/utils/buffer-utils.js.map +1 -0
  34. package/dist/es5/parquetjs/utils/file-utils.js +3 -2
  35. package/dist/es5/parquetjs/utils/file-utils.js.map +1 -1
  36. package/dist/esm/index.js +1 -1
  37. package/dist/esm/index.js.map +1 -1
  38. package/dist/esm/lib/parse-parquet.js +12 -6
  39. package/dist/esm/lib/parse-parquet.js.map +1 -1
  40. package/dist/esm/parquet-loader.js +2 -3
  41. package/dist/esm/parquet-loader.js.map +1 -1
  42. package/dist/esm/parquet-wasm-loader.js +1 -1
  43. package/dist/esm/parquet-wasm-loader.js.map +1 -1
  44. package/dist/esm/parquet-wasm-writer.js +1 -1
  45. package/dist/esm/parquet-wasm-writer.js.map +1 -1
  46. package/dist/esm/parquet-writer.js +1 -1
  47. package/dist/esm/parquet-writer.js.map +1 -1
  48. package/dist/esm/parquetjs/compression.js +1 -10
  49. package/dist/esm/parquetjs/compression.js.map +1 -1
  50. package/dist/esm/parquetjs/encoder/{parquet-encoder.js → writer.js} +37 -7
  51. package/dist/esm/parquetjs/encoder/writer.js.map +1 -0
  52. package/dist/esm/parquetjs/file.js +81 -0
  53. package/dist/esm/parquetjs/file.js.map +1 -0
  54. package/dist/esm/parquetjs/parser/parquet-cursor.js +78 -0
  55. package/dist/esm/parquetjs/parser/parquet-cursor.js.map +1 -0
  56. package/dist/esm/parquetjs/parser/parquet-envelope-reader.js +129 -0
  57. package/dist/esm/parquetjs/parser/parquet-envelope-reader.js.map +1 -0
  58. package/dist/esm/parquetjs/parser/parquet-reader.js +72 -158
  59. package/dist/esm/parquetjs/parser/parquet-reader.js.map +1 -1
  60. package/dist/esm/parquetjs/schema/declare.js +0 -1
  61. package/dist/esm/parquetjs/schema/declare.js.map +1 -1
  62. package/dist/esm/parquetjs/schema/shred.js +34 -42
  63. package/dist/esm/parquetjs/schema/shred.js.map +1 -1
  64. package/dist/esm/parquetjs/schema/types.js.map +1 -1
  65. package/dist/esm/parquetjs/utils/buffer-utils.js +13 -0
  66. package/dist/esm/parquetjs/utils/buffer-utils.js.map +1 -0
  67. package/dist/esm/parquetjs/utils/file-utils.js +1 -1
  68. package/dist/esm/parquetjs/utils/file-utils.js.map +1 -1
  69. package/dist/index.d.ts +1 -1
  70. package/dist/index.d.ts.map +1 -1
  71. package/dist/index.js +4 -3
  72. package/dist/lib/parse-parquet.d.ts +2 -2
  73. package/dist/lib/parse-parquet.d.ts.map +1 -1
  74. package/dist/lib/parse-parquet.js +12 -24
  75. package/dist/parquet-loader.d.ts +0 -1
  76. package/dist/parquet-loader.d.ts.map +1 -1
  77. package/dist/parquet-loader.js +1 -2
  78. package/dist/parquet-worker.js +24 -15
  79. package/dist/parquet-worker.js.map +3 -3
  80. package/dist/parquetjs/compression.d.ts.map +1 -1
  81. package/dist/parquetjs/compression.js +5 -16
  82. package/dist/parquetjs/encoder/{parquet-encoder.d.ts → writer.d.ts} +19 -10
  83. package/dist/parquetjs/encoder/writer.d.ts.map +1 -0
  84. package/dist/parquetjs/encoder/{parquet-encoder.js → writer.js} +37 -39
  85. package/dist/parquetjs/file.d.ts +10 -0
  86. package/dist/parquetjs/file.d.ts.map +1 -0
  87. package/dist/parquetjs/file.js +99 -0
  88. package/dist/parquetjs/parser/parquet-cursor.d.ts +36 -0
  89. package/dist/parquetjs/parser/parquet-cursor.d.ts.map +1 -0
  90. package/dist/parquetjs/parser/parquet-cursor.js +74 -0
  91. package/dist/parquetjs/parser/parquet-envelope-reader.d.ts +40 -0
  92. package/dist/parquetjs/parser/parquet-envelope-reader.d.ts.map +1 -0
  93. package/dist/parquetjs/parser/parquet-envelope-reader.js +136 -0
  94. package/dist/parquetjs/parser/parquet-reader.d.ts +57 -47
  95. package/dist/parquetjs/parser/parquet-reader.d.ts.map +1 -1
  96. package/dist/parquetjs/parser/parquet-reader.js +102 -168
  97. package/dist/parquetjs/schema/declare.d.ts +7 -14
  98. package/dist/parquetjs/schema/declare.d.ts.map +1 -1
  99. package/dist/parquetjs/schema/declare.js +0 -2
  100. package/dist/parquetjs/schema/shred.d.ts +0 -115
  101. package/dist/parquetjs/schema/shred.d.ts.map +1 -1
  102. package/dist/parquetjs/schema/shred.js +43 -161
  103. package/dist/parquetjs/schema/types.d.ts +2 -2
  104. package/dist/parquetjs/schema/types.d.ts.map +1 -1
  105. package/dist/parquetjs/utils/buffer-utils.d.ts +10 -0
  106. package/dist/parquetjs/utils/buffer-utils.d.ts.map +1 -0
  107. package/dist/parquetjs/utils/buffer-utils.js +22 -0
  108. package/dist/parquetjs/utils/file-utils.d.ts +4 -3
  109. package/dist/parquetjs/utils/file-utils.d.ts.map +1 -1
  110. package/dist/parquetjs/utils/file-utils.js +5 -2
  111. package/package.json +5 -7
  112. package/src/index.ts +2 -2
  113. package/src/lib/parse-parquet.ts +12 -25
  114. package/src/parquet-loader.ts +1 -3
  115. package/src/parquetjs/compression.ts +1 -14
  116. package/src/parquetjs/encoder/{parquet-encoder.ts → writer.ts} +28 -22
  117. package/src/parquetjs/file.ts +90 -0
  118. package/src/parquetjs/parser/parquet-cursor.ts +94 -0
  119. package/src/parquetjs/parser/parquet-envelope-reader.ts +199 -0
  120. package/src/parquetjs/parser/parquet-reader.ts +122 -239
  121. package/src/parquetjs/schema/declare.ts +9 -17
  122. package/src/parquetjs/schema/shred.ts +28 -157
  123. package/src/parquetjs/schema/types.ts +27 -21
  124. package/src/parquetjs/utils/buffer-utils.ts +18 -0
  125. package/src/parquetjs/utils/file-utils.ts +4 -3
  126. package/dist/es5/lib/convert-schema-deep.ts.disabled +0 -910
  127. package/dist/es5/parquetjs/encoder/parquet-encoder.js.map +0 -1
  128. package/dist/esm/lib/convert-schema-deep.ts.disabled +0 -910
  129. package/dist/esm/parquetjs/encoder/parquet-encoder.js.map +0 -1
  130. package/dist/parquetjs/encoder/parquet-encoder.d.ts.map +0 -1
  131. package/src/lib/convert-schema-deep.ts.disabled +0 -910
@@ -10,630 +10,299 @@ var _asyncToGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/
10
10
  var _classCallCheck2 = _interopRequireDefault(require("@babel/runtime/helpers/classCallCheck"));
11
11
  var _createClass2 = _interopRequireDefault(require("@babel/runtime/helpers/createClass"));
12
12
  var _defineProperty2 = _interopRequireDefault(require("@babel/runtime/helpers/defineProperty"));
13
- var _awaitAsyncGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/awaitAsyncGenerator"));
14
- var _wrapAsyncGenerator2 = _interopRequireDefault(require("@babel/runtime/helpers/wrapAsyncGenerator"));
13
+ var _parquetEnvelopeReader = require("./parquet-envelope-reader");
15
14
  var _schema = require("../schema/schema");
16
- var _decoders = require("./decoders");
17
- var _shred = require("../schema/shred");
15
+ var _parquetCursor = require("./parquet-cursor");
18
16
  var _constants = require("../../constants");
19
- var _parquetThrift = require("../parquet-thrift");
20
- var _readUtils = require("../utils/read-utils");
17
+ var _decoders = require("./decoders");
18
+ var _Symbol$asyncIterator;
21
19
  function _createForOfIteratorHelper(o, allowArrayLike) { var it = typeof Symbol !== "undefined" && o[Symbol.iterator] || o["@@iterator"]; if (!it) { if (Array.isArray(o) || (it = _unsupportedIterableToArray(o)) || allowArrayLike && o && typeof o.length === "number") { if (it) o = it; var i = 0; var F = function F() {}; return { s: F, n: function n() { if (i >= o.length) return { done: true }; return { done: false, value: o[i++] }; }, e: function e(_e) { throw _e; }, f: F }; } throw new TypeError("Invalid attempt to iterate non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method."); } var normalCompletion = true, didErr = false, err; return { s: function s() { it = it.call(o); }, n: function n() { var step = it.next(); normalCompletion = step.done; return step; }, e: function e(_e2) { didErr = true; err = _e2; }, f: function f() { try { if (!normalCompletion && it.return != null) it.return(); } finally { if (didErr) throw err; } } }; }
22
20
  function _unsupportedIterableToArray(o, minLen) { if (!o) return; if (typeof o === "string") return _arrayLikeToArray(o, minLen); var n = Object.prototype.toString.call(o).slice(8, -1); if (n === "Object" && o.constructor) n = o.constructor.name; if (n === "Map" || n === "Set") return Array.from(o); if (n === "Arguments" || /^(?:Ui|I)nt(?:8|16|32)(?:Clamped)?Array$/.test(n)) return _arrayLikeToArray(o, minLen); }
23
21
  function _arrayLikeToArray(arr, len) { if (len == null || len > arr.length) len = arr.length; for (var i = 0, arr2 = new Array(len); i < len; i++) { arr2[i] = arr[i]; } return arr2; }
24
- function ownKeys(object, enumerableOnly) { var keys = Object.keys(object); if (Object.getOwnPropertySymbols) { var symbols = Object.getOwnPropertySymbols(object); enumerableOnly && (symbols = symbols.filter(function (sym) { return Object.getOwnPropertyDescriptor(object, sym).enumerable; })), keys.push.apply(keys, symbols); } return keys; }
25
- function _objectSpread(target) { for (var i = 1; i < arguments.length; i++) { var source = null != arguments[i] ? arguments[i] : {}; i % 2 ? ownKeys(Object(source), !0).forEach(function (key) { (0, _defineProperty2.default)(target, key, source[key]); }) : Object.getOwnPropertyDescriptors ? Object.defineProperties(target, Object.getOwnPropertyDescriptors(source)) : ownKeys(Object(source)).forEach(function (key) { Object.defineProperty(target, key, Object.getOwnPropertyDescriptor(source, key)); }); } return target; }
26
- function _asyncIterator(iterable) { var method, async, sync, retry = 2; for ("undefined" != typeof Symbol && (async = Symbol.asyncIterator, sync = Symbol.iterator); retry--;) { if (async && null != (method = iterable[async])) return method.call(iterable); if (sync && null != (method = iterable[sync])) return new AsyncFromSyncIterator(method.call(iterable)); async = "@@asyncIterator", sync = "@@iterator"; } throw new TypeError("Object is not async iterable"); }
27
- function AsyncFromSyncIterator(s) { function AsyncFromSyncIteratorContinuation(r) { if (Object(r) !== r) return Promise.reject(new TypeError(r + " is not an object.")); var done = r.done; return Promise.resolve(r.value).then(function (value) { return { value: value, done: done }; }); } return AsyncFromSyncIterator = function AsyncFromSyncIterator(s) { this.s = s, this.n = s.next; }, AsyncFromSyncIterator.prototype = { s: null, n: null, next: function next() { return AsyncFromSyncIteratorContinuation(this.n.apply(this.s, arguments)); }, return: function _return(value) { var ret = this.s.return; return void 0 === ret ? Promise.resolve({ value: value, done: !0 }) : AsyncFromSyncIteratorContinuation(ret.apply(this.s, arguments)); }, throw: function _throw(value) { var thr = this.s.return; return void 0 === thr ? Promise.reject(value) : AsyncFromSyncIteratorContinuation(thr.apply(this.s, arguments)); } }, new AsyncFromSyncIterator(s); }
28
- var DEFAULT_PROPS = {
29
- defaultDictionarySize: 1e6
30
- };
31
-
22
+ _Symbol$asyncIterator = Symbol.asyncIterator;
32
23
  var ParquetReader = function () {
33
- function ParquetReader(file, props) {
24
+ function ParquetReader(metadata, envelopeReader) {
34
25
  (0, _classCallCheck2.default)(this, ParquetReader);
35
- (0, _defineProperty2.default)(this, "props", void 0);
36
- (0, _defineProperty2.default)(this, "file", void 0);
37
- (0, _defineProperty2.default)(this, "metadata", null);
38
- this.file = file;
39
- this.props = _objectSpread(_objectSpread({}, DEFAULT_PROPS), props);
26
+ (0, _defineProperty2.default)(this, "metadata", void 0);
27
+ (0, _defineProperty2.default)(this, "envelopeReader", void 0);
28
+ (0, _defineProperty2.default)(this, "schema", void 0);
29
+ if (metadata.version !== _constants.PARQUET_VERSION) {
30
+ throw new Error('invalid parquet version');
31
+ }
32
+ this.metadata = metadata;
33
+ this.envelopeReader = envelopeReader;
34
+ var root = this.metadata.schema[0];
35
+ var _decodeSchema = (0, _decoders.decodeSchema)(this.metadata.schema, 1, root.num_children),
36
+ schema = _decodeSchema.schema;
37
+ this.schema = new _schema.ParquetSchema(schema);
40
38
  }
39
+
41
40
  (0, _createClass2.default)(ParquetReader, [{
42
41
  key: "close",
43
- value: function close() {
44
- this.file.close();
45
- }
46
-
47
- }, {
48
- key: "rowIterator",
49
- value:
50
-
51
- function rowIterator(props) {
52
- var _this = this;
53
- return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee() {
54
- var _iteratorAbruptCompletion, _didIteratorError, _iteratorError, _iterator, _step, rows, _iterator3, _step3, row;
42
+ value: function () {
43
+ var _close = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee() {
55
44
  return _regenerator.default.wrap(function _callee$(_context) {
56
45
  while (1) {
57
46
  switch (_context.prev = _context.next) {
58
47
  case 0:
59
- _iteratorAbruptCompletion = false;
60
- _didIteratorError = false;
61
- _context.prev = 2;
62
- _iterator = _asyncIterator(_this.rowBatchIterator(props));
63
- case 4:
64
- _context.next = 6;
65
- return (0, _awaitAsyncGenerator2.default)(_iterator.next());
66
- case 6:
67
- if (!(_iteratorAbruptCompletion = !(_step = _context.sent).done)) {
68
- _context.next = 28;
69
- break;
70
- }
71
- rows = _step.value;
72
- _iterator3 = _createForOfIteratorHelper(rows);
73
- _context.prev = 9;
74
- _iterator3.s();
75
- case 11:
76
- if ((_step3 = _iterator3.n()).done) {
77
- _context.next = 17;
78
- break;
79
- }
80
- row = _step3.value;
81
- _context.next = 15;
82
- return row;
83
- case 15:
84
- _context.next = 11;
85
- break;
86
- case 17:
87
- _context.next = 22;
88
- break;
89
- case 19:
90
- _context.prev = 19;
91
- _context.t0 = _context["catch"](9);
92
- _iterator3.e(_context.t0);
93
- case 22:
94
- _context.prev = 22;
95
- _iterator3.f();
96
- return _context.finish(22);
97
- case 25:
98
- _iteratorAbruptCompletion = false;
99
- _context.next = 4;
100
- break;
101
- case 28:
102
- _context.next = 34;
103
- break;
104
- case 30:
105
- _context.prev = 30;
106
- _context.t1 = _context["catch"](2);
107
- _didIteratorError = true;
108
- _iteratorError = _context.t1;
109
- case 34:
110
- _context.prev = 34;
111
- _context.prev = 35;
112
- if (!(_iteratorAbruptCompletion && _iterator.return != null)) {
113
- _context.next = 39;
114
- break;
115
- }
116
- _context.next = 39;
117
- return (0, _awaitAsyncGenerator2.default)(_iterator.return());
118
- case 39:
119
- _context.prev = 39;
120
- if (!_didIteratorError) {
121
- _context.next = 42;
122
- break;
123
- }
124
- throw _iteratorError;
125
- case 42:
126
- return _context.finish(39);
127
- case 43:
128
- return _context.finish(34);
129
- case 44:
48
+ _context.next = 2;
49
+ return this.envelopeReader.close();
50
+ case 2:
130
51
  case "end":
131
52
  return _context.stop();
132
53
  }
133
54
  }
134
- }, _callee, null, [[2, 30, 34, 44], [9, 19, 22, 25], [35,, 39, 43]]);
135
- }))();
55
+ }, _callee, this);
56
+ }));
57
+ function close() {
58
+ return _close.apply(this, arguments);
59
+ }
60
+ return close;
61
+ }()
62
+ }, {
63
+ key: "getCursor",
64
+ value: function getCursor(columnList) {
65
+ if (!columnList) {
66
+ columnList = [];
67
+ }
68
+
69
+ columnList = columnList.map(function (x) {
70
+ return Array.isArray(x) ? x : [x];
71
+ });
72
+ return new _parquetCursor.ParquetCursor(this.metadata, this.envelopeReader, this.schema, columnList);
136
73
  }
137
74
 
138
75
  }, {
139
- key: "rowBatchIterator",
76
+ key: "getRowCount",
140
77
  value:
141
- function rowBatchIterator(props) {
142
- var _this2 = this;
143
- return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee2() {
144
- var schema, _iteratorAbruptCompletion2, _didIteratorError2, _iteratorError2, _iterator2, _step2, rowGroup;
145
- return _regenerator.default.wrap(function _callee2$(_context2) {
146
- while (1) {
147
- switch (_context2.prev = _context2.next) {
148
- case 0:
149
- _context2.next = 2;
150
- return (0, _awaitAsyncGenerator2.default)(_this2.getSchema());
151
- case 2:
152
- schema = _context2.sent;
153
- _iteratorAbruptCompletion2 = false;
154
- _didIteratorError2 = false;
155
- _context2.prev = 5;
156
- _iterator2 = _asyncIterator(_this2.rowGroupIterator(props));
157
- case 7:
158
- _context2.next = 9;
159
- return (0, _awaitAsyncGenerator2.default)(_iterator2.next());
160
- case 9:
161
- if (!(_iteratorAbruptCompletion2 = !(_step2 = _context2.sent).done)) {
162
- _context2.next = 16;
163
- break;
164
- }
165
- rowGroup = _step2.value;
166
- _context2.next = 13;
167
- return (0, _shred.materializeRecords)(schema, rowGroup);
168
- case 13:
169
- _iteratorAbruptCompletion2 = false;
170
- _context2.next = 7;
171
- break;
172
- case 16:
173
- _context2.next = 22;
174
- break;
175
- case 18:
176
- _context2.prev = 18;
177
- _context2.t0 = _context2["catch"](5);
178
- _didIteratorError2 = true;
179
- _iteratorError2 = _context2.t0;
180
- case 22:
181
- _context2.prev = 22;
182
- _context2.prev = 23;
183
- if (!(_iteratorAbruptCompletion2 && _iterator2.return != null)) {
184
- _context2.next = 27;
185
- break;
186
- }
187
- _context2.next = 27;
188
- return (0, _awaitAsyncGenerator2.default)(_iterator2.return());
189
- case 27:
190
- _context2.prev = 27;
191
- if (!_didIteratorError2) {
192
- _context2.next = 30;
193
- break;
194
- }
195
- throw _iteratorError2;
196
- case 30:
197
- return _context2.finish(27);
198
- case 31:
199
- return _context2.finish(22);
200
- case 32:
201
- case "end":
202
- return _context2.stop();
203
- }
204
- }
205
- }, _callee2, null, [[5, 18, 22, 32], [23,, 27, 31]]);
206
- }))();
78
+ function getRowCount() {
79
+ return Number(this.metadata.num_rows);
207
80
  }
208
81
 
209
82
  }, {
210
- key: "rowGroupIterator",
83
+ key: "getSchema",
211
84
  value:
212
- function rowGroupIterator(props) {
213
- var _this3 = this;
214
- return (0, _wrapAsyncGenerator2.default)(_regenerator.default.mark(function _callee3() {
215
- var columnList, metadata, schema, rowGroupCount, rowGroupIndex, rowGroup;
216
- return _regenerator.default.wrap(function _callee3$(_context3) {
217
- while (1) {
218
- switch (_context3.prev = _context3.next) {
219
- case 0:
220
- columnList = ((props === null || props === void 0 ? void 0 : props.columnList) || []).map(function (x) {
221
- return Array.isArray(x) ? x : [x];
222
- });
223
- _context3.next = 3;
224
- return (0, _awaitAsyncGenerator2.default)(_this3.getFileMetadata());
225
- case 3:
226
- metadata = _context3.sent;
227
- _context3.next = 6;
228
- return (0, _awaitAsyncGenerator2.default)(_this3.getSchema());
229
- case 6:
230
- schema = _context3.sent;
231
- rowGroupCount = (metadata === null || metadata === void 0 ? void 0 : metadata.row_groups.length) || 0;
232
- rowGroupIndex = 0;
233
- case 9:
234
- if (!(rowGroupIndex < rowGroupCount)) {
235
- _context3.next = 18;
236
- break;
237
- }
238
- _context3.next = 12;
239
- return (0, _awaitAsyncGenerator2.default)(_this3.readRowGroup(schema, metadata.row_groups[rowGroupIndex], columnList));
240
- case 12:
241
- rowGroup = _context3.sent;
242
- _context3.next = 15;
243
- return rowGroup;
244
- case 15:
245
- rowGroupIndex++;
246
- _context3.next = 9;
247
- break;
248
- case 18:
249
- case "end":
250
- return _context3.stop();
251
- }
252
- }
253
- }, _callee3);
254
- }))();
85
+ function getSchema() {
86
+ return this.schema;
255
87
  }
88
+
256
89
  }, {
257
- key: "getRowCount",
90
+ key: "getMetadata",
91
+ value:
92
+ function getMetadata() {
93
+ var md = {};
94
+ var _iterator = _createForOfIteratorHelper(this.metadata.key_value_metadata),
95
+ _step;
96
+ try {
97
+ for (_iterator.s(); !(_step = _iterator.n()).done;) {
98
+ var kv = _step.value;
99
+ md[kv.key] = kv.value;
100
+ }
101
+ } catch (err) {
102
+ _iterator.e(err);
103
+ } finally {
104
+ _iterator.f();
105
+ }
106
+ return md;
107
+ }
108
+
109
+ }, {
110
+ key: _Symbol$asyncIterator,
111
+ value:
112
+ function value() {
113
+ return this.getCursor()[Symbol.asyncIterator]();
114
+ }
115
+ }], [{
116
+ key: "openBlob",
258
117
  value: function () {
259
- var _getRowCount = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee4() {
260
- var metadata;
118
+ var _openBlob = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee4(blob) {
119
+ var readFn, closeFn, size, envelopeReader, metadata;
261
120
  return _regenerator.default.wrap(function _callee4$(_context4) {
262
121
  while (1) {
263
122
  switch (_context4.prev = _context4.next) {
264
123
  case 0:
265
- _context4.next = 2;
266
- return this.getFileMetadata();
267
- case 2:
124
+ readFn = function () {
125
+ var _ref = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee2(start, length) {
126
+ var arrayBuffer;
127
+ return _regenerator.default.wrap(function _callee2$(_context2) {
128
+ while (1) {
129
+ switch (_context2.prev = _context2.next) {
130
+ case 0:
131
+ _context2.next = 2;
132
+ return blob.slice(start, start + length).arrayBuffer();
133
+ case 2:
134
+ arrayBuffer = _context2.sent;
135
+ return _context2.abrupt("return", Buffer.from(arrayBuffer));
136
+ case 4:
137
+ case "end":
138
+ return _context2.stop();
139
+ }
140
+ }
141
+ }, _callee2);
142
+ }));
143
+ return function readFn(_x2, _x3) {
144
+ return _ref.apply(this, arguments);
145
+ };
146
+ }();
147
+ closeFn = function () {
148
+ var _ref2 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee3() {
149
+ return _regenerator.default.wrap(function _callee3$(_context3) {
150
+ while (1) {
151
+ switch (_context3.prev = _context3.next) {
152
+ case 0:
153
+ case "end":
154
+ return _context3.stop();
155
+ }
156
+ }
157
+ }, _callee3);
158
+ }));
159
+ return function closeFn() {
160
+ return _ref2.apply(this, arguments);
161
+ };
162
+ }();
163
+ size = blob.size;
164
+ envelopeReader = new _parquetEnvelopeReader.ParquetEnvelopeReader(readFn, closeFn, size);
165
+ _context4.prev = 4;
166
+ _context4.next = 7;
167
+ return envelopeReader.readHeader();
168
+ case 7:
169
+ _context4.next = 9;
170
+ return envelopeReader.readFooter();
171
+ case 9:
268
172
  metadata = _context4.sent;
269
- return _context4.abrupt("return", Number(metadata.num_rows));
270
- case 4:
173
+ return _context4.abrupt("return", new ParquetReader(metadata, envelopeReader));
174
+ case 13:
175
+ _context4.prev = 13;
176
+ _context4.t0 = _context4["catch"](4);
177
+ _context4.next = 17;
178
+ return envelopeReader.close();
179
+ case 17:
180
+ throw _context4.t0;
181
+ case 18:
271
182
  case "end":
272
183
  return _context4.stop();
273
184
  }
274
185
  }
275
- }, _callee4, this);
186
+ }, _callee4, null, [[4, 13]]);
276
187
  }));
277
- function getRowCount() {
278
- return _getRowCount.apply(this, arguments);
188
+ function openBlob(_x) {
189
+ return _openBlob.apply(this, arguments);
279
190
  }
280
- return getRowCount;
191
+ return openBlob;
281
192
  }()
282
193
  }, {
283
- key: "getSchema",
194
+ key: "openArrayBuffer",
284
195
  value: function () {
285
- var _getSchema = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee5() {
286
- var metadata, root, _decodeSchema, schemaDefinition, schema;
287
- return _regenerator.default.wrap(function _callee5$(_context5) {
288
- while (1) {
289
- switch (_context5.prev = _context5.next) {
290
- case 0:
291
- _context5.next = 2;
292
- return this.getFileMetadata();
293
- case 2:
294
- metadata = _context5.sent;
295
- root = metadata.schema[0];
296
- _decodeSchema = (0, _decoders.decodeSchema)(metadata.schema, 1, root.num_children), schemaDefinition = _decodeSchema.schema;
297
- schema = new _schema.ParquetSchema(schemaDefinition);
298
- return _context5.abrupt("return", schema);
299
- case 7:
300
- case "end":
301
- return _context5.stop();
302
- }
303
- }
304
- }, _callee5, this);
305
- }));
306
- function getSchema() {
307
- return _getSchema.apply(this, arguments);
308
- }
309
- return getSchema;
310
- }()
311
- }, {
312
- key: "getSchemaMetadata",
313
- value: function () {
314
- var _getSchemaMetadata = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee6() {
315
- var metadata, md, _iterator4, _step4, kv;
316
- return _regenerator.default.wrap(function _callee6$(_context6) {
317
- while (1) {
318
- switch (_context6.prev = _context6.next) {
319
- case 0:
320
- _context6.next = 2;
321
- return this.getFileMetadata();
322
- case 2:
323
- metadata = _context6.sent;
324
- md = {};
325
- _iterator4 = _createForOfIteratorHelper(metadata.key_value_metadata);
326
- try {
327
- for (_iterator4.s(); !(_step4 = _iterator4.n()).done;) {
328
- kv = _step4.value;
329
- md[kv.key] = kv.value;
330
- }
331
- } catch (err) {
332
- _iterator4.e(err);
333
- } finally {
334
- _iterator4.f();
335
- }
336
- return _context6.abrupt("return", md);
337
- case 7:
338
- case "end":
339
- return _context6.stop();
340
- }
341
- }
342
- }, _callee6, this);
343
- }));
344
- function getSchemaMetadata() {
345
- return _getSchemaMetadata.apply(this, arguments);
346
- }
347
- return getSchemaMetadata;
348
- }()
349
- }, {
350
- key: "getFileMetadata",
351
- value: function () {
352
- var _getFileMetadata = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee7() {
196
+ var _openArrayBuffer = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee7(arrayBuffer) {
197
+ var readFn, closeFn, size, envelopeReader, metadata;
353
198
  return _regenerator.default.wrap(function _callee7$(_context7) {
354
199
  while (1) {
355
200
  switch (_context7.prev = _context7.next) {
356
201
  case 0:
357
- if (this.metadata) {
358
- _context7.next = 4;
359
- break;
360
- }
361
- _context7.next = 3;
362
- return this.readHeader();
363
- case 3:
364
- this.metadata = this.readFooter();
365
- case 4:
366
- return _context7.abrupt("return", this.metadata);
367
- case 5:
202
+ readFn = function () {
203
+ var _ref3 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee5(start, length) {
204
+ return _regenerator.default.wrap(function _callee5$(_context5) {
205
+ while (1) {
206
+ switch (_context5.prev = _context5.next) {
207
+ case 0:
208
+ return _context5.abrupt("return", Buffer.from(arrayBuffer, start, length));
209
+ case 1:
210
+ case "end":
211
+ return _context5.stop();
212
+ }
213
+ }
214
+ }, _callee5);
215
+ }));
216
+ return function readFn(_x5, _x6) {
217
+ return _ref3.apply(this, arguments);
218
+ };
219
+ }();
220
+ closeFn = function () {
221
+ var _ref4 = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee6() {
222
+ return _regenerator.default.wrap(function _callee6$(_context6) {
223
+ while (1) {
224
+ switch (_context6.prev = _context6.next) {
225
+ case 0:
226
+ case "end":
227
+ return _context6.stop();
228
+ }
229
+ }
230
+ }, _callee6);
231
+ }));
232
+ return function closeFn() {
233
+ return _ref4.apply(this, arguments);
234
+ };
235
+ }();
236
+ size = arrayBuffer.byteLength;
237
+ envelopeReader = new _parquetEnvelopeReader.ParquetEnvelopeReader(readFn, closeFn, size);
238
+ _context7.prev = 4;
239
+ _context7.next = 7;
240
+ return envelopeReader.readHeader();
241
+ case 7:
242
+ _context7.next = 9;
243
+ return envelopeReader.readFooter();
244
+ case 9:
245
+ metadata = _context7.sent;
246
+ return _context7.abrupt("return", new ParquetReader(metadata, envelopeReader));
247
+ case 13:
248
+ _context7.prev = 13;
249
+ _context7.t0 = _context7["catch"](4);
250
+ _context7.next = 17;
251
+ return envelopeReader.close();
252
+ case 17:
253
+ throw _context7.t0;
254
+ case 18:
368
255
  case "end":
369
256
  return _context7.stop();
370
257
  }
371
258
  }
372
- }, _callee7, this);
259
+ }, _callee7, null, [[4, 13]]);
373
260
  }));
374
- function getFileMetadata() {
375
- return _getFileMetadata.apply(this, arguments);
261
+ function openArrayBuffer(_x4) {
262
+ return _openArrayBuffer.apply(this, arguments);
376
263
  }
377
- return getFileMetadata;
264
+ return openArrayBuffer;
378
265
  }()
379
266
  }, {
380
- key: "readHeader",
267
+ key: "openBuffer",
381
268
  value: function () {
382
- var _readHeader = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee8() {
383
- var buffer, magic;
269
+ var _openBuffer = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee8(buffer) {
270
+ var envelopeReader, metadata;
384
271
  return _regenerator.default.wrap(function _callee8$(_context8) {
385
272
  while (1) {
386
273
  switch (_context8.prev = _context8.next) {
387
274
  case 0:
388
275
  _context8.next = 2;
389
- return this.file.read(0, _constants.PARQUET_MAGIC.length);
276
+ return _parquetEnvelopeReader.ParquetEnvelopeReader.openBuffer(buffer);
390
277
  case 2:
391
- buffer = _context8.sent;
392
- magic = buffer.toString();
393
- _context8.t0 = magic;
394
- _context8.next = _context8.t0 === _constants.PARQUET_MAGIC ? 7 : _context8.t0 === _constants.PARQUET_MAGIC_ENCRYPTED ? 8 : 9;
395
- break;
396
- case 7:
397
- return _context8.abrupt("break", 10);
278
+ envelopeReader = _context8.sent;
279
+ _context8.prev = 3;
280
+ _context8.next = 6;
281
+ return envelopeReader.readHeader();
282
+ case 6:
283
+ _context8.next = 8;
284
+ return envelopeReader.readFooter();
398
285
  case 8:
399
- throw new Error('Encrypted parquet file not supported');
400
- case 9:
401
- throw new Error("Invalid parquet file (magic=".concat(magic, ")"));
402
- case 10:
403
- case "end":
404
- return _context8.stop();
405
- }
406
- }
407
- }, _callee8, this);
408
- }));
409
- function readHeader() {
410
- return _readHeader.apply(this, arguments);
411
- }
412
- return readHeader;
413
- }()
414
- }, {
415
- key: "readFooter",
416
- value: function () {
417
- var _readFooter = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee9() {
418
- var trailerLen, trailerBuf, magic, metadataSize, metadataOffset, metadataBuf, _decodeFileMetadata, metadata;
419
- return _regenerator.default.wrap(function _callee9$(_context9) {
420
- while (1) {
421
- switch (_context9.prev = _context9.next) {
422
- case 0:
423
- trailerLen = _constants.PARQUET_MAGIC.length + 4;
424
- _context9.next = 3;
425
- return this.file.read(this.file.size - trailerLen, trailerLen);
426
- case 3:
427
- trailerBuf = _context9.sent;
428
- magic = trailerBuf.slice(4).toString();
429
- if (!(magic !== _constants.PARQUET_MAGIC)) {
430
- _context9.next = 7;
431
- break;
432
- }
433
- throw new Error("Not a valid parquet file (magic=\"".concat(magic, ")"));
434
- case 7:
435
- metadataSize = trailerBuf.readUInt32LE(0);
436
- metadataOffset = this.file.size - metadataSize - trailerLen;
437
- if (!(metadataOffset < _constants.PARQUET_MAGIC.length)) {
438
- _context9.next = 11;
439
- break;
440
- }
441
- throw new Error("Invalid metadata size ".concat(metadataOffset));
442
- case 11:
443
- _context9.next = 13;
444
- return this.file.read(metadataOffset, metadataSize);
445
- case 13:
446
- metadataBuf = _context9.sent;
447
- _decodeFileMetadata = (0, _readUtils.decodeFileMetadata)(metadataBuf), metadata = _decodeFileMetadata.metadata;
448
- return _context9.abrupt("return", metadata);
449
- case 16:
450
- case "end":
451
- return _context9.stop();
452
- }
453
- }
454
- }, _callee9, this);
455
- }));
456
- function readFooter() {
457
- return _readFooter.apply(this, arguments);
458
- }
459
- return readFooter;
460
- }()
461
- }, {
462
- key: "readRowGroup",
463
- value: function () {
464
- var _readRowGroup = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee10(schema, rowGroup, columnList) {
465
- var buffer, _iterator5, _step5, colChunk, colMetadata, colKey;
466
- return _regenerator.default.wrap(function _callee10$(_context10) {
467
- while (1) {
468
- switch (_context10.prev = _context10.next) {
469
- case 0:
470
- buffer = {
471
- rowCount: Number(rowGroup.num_rows),
472
- columnData: {}
473
- };
474
- _iterator5 = _createForOfIteratorHelper(rowGroup.columns);
475
- _context10.prev = 2;
476
- _iterator5.s();
477
- case 4:
478
- if ((_step5 = _iterator5.n()).done) {
479
- _context10.next = 15;
480
- break;
481
- }
482
- colChunk = _step5.value;
483
- colMetadata = colChunk.meta_data;
484
- colKey = colMetadata === null || colMetadata === void 0 ? void 0 : colMetadata.path_in_schema;
485
- if (!(columnList.length > 0 && (0, _readUtils.fieldIndexOf)(columnList, colKey) < 0)) {
486
- _context10.next = 10;
487
- break;
488
- }
489
- return _context10.abrupt("continue", 13);
490
- case 10:
491
- _context10.next = 12;
492
- return this.readColumnChunk(schema, colChunk);
286
+ metadata = _context8.sent;
287
+ return _context8.abrupt("return", new ParquetReader(metadata, envelopeReader));
493
288
  case 12:
494
- buffer.columnData[colKey.join()] = _context10.sent;
495
- case 13:
496
- _context10.next = 4;
497
- break;
498
- case 15:
499
- _context10.next = 20;
500
- break;
501
- case 17:
502
- _context10.prev = 17;
503
- _context10.t0 = _context10["catch"](2);
504
- _iterator5.e(_context10.t0);
505
- case 20:
506
- _context10.prev = 20;
507
- _iterator5.f();
508
- return _context10.finish(20);
509
- case 23:
510
- return _context10.abrupt("return", buffer);
511
- case 24:
512
- case "end":
513
- return _context10.stop();
514
- }
515
- }
516
- }, _callee10, this, [[2, 17, 20, 23]]);
517
- }));
518
- function readRowGroup(_x, _x2, _x3) {
519
- return _readRowGroup.apply(this, arguments);
520
- }
521
- return readRowGroup;
522
- }()
523
- }, {
524
- key: "readColumnChunk",
525
- value: function () {
526
- var _readColumnChunk = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee11(schema, colChunk) {
527
- var _colChunk$meta_data, _colChunk$meta_data2, _colChunk$meta_data3, _colChunk$meta_data4, _colChunk$meta_data5, _colChunk$meta_data7, _colChunk$meta_data8, _options$dictionary;
528
- var field, type, compression, pagesOffset, pagesSize, _colChunk$meta_data6, options, dictionary, dictionaryPageOffset, dictionaryOffset, pagesBuf;
529
- return _regenerator.default.wrap(function _callee11$(_context11) {
530
- while (1) {
531
- switch (_context11.prev = _context11.next) {
532
- case 0:
533
- if (!(colChunk.file_path !== undefined && colChunk.file_path !== null)) {
534
- _context11.next = 2;
535
- break;
536
- }
537
- throw new Error('external references are not supported');
538
- case 2:
539
- field = schema.findField((_colChunk$meta_data = colChunk.meta_data) === null || _colChunk$meta_data === void 0 ? void 0 : _colChunk$meta_data.path_in_schema);
540
- type = (0, _readUtils.getThriftEnum)(_parquetThrift.Type, (_colChunk$meta_data2 = colChunk.meta_data) === null || _colChunk$meta_data2 === void 0 ? void 0 : _colChunk$meta_data2.type);
541
- if (!(type !== field.primitiveType)) {
542
- _context11.next = 6;
543
- break;
544
- }
545
- throw new Error("chunk type not matching schema: ".concat(type));
546
- case 6:
547
- compression = (0, _readUtils.getThriftEnum)(_parquetThrift.CompressionCodec, (_colChunk$meta_data3 = colChunk.meta_data) === null || _colChunk$meta_data3 === void 0 ? void 0 : _colChunk$meta_data3.codec);
548
- pagesOffset = Number((_colChunk$meta_data4 = colChunk.meta_data) === null || _colChunk$meta_data4 === void 0 ? void 0 : _colChunk$meta_data4.data_page_offset);
549
- pagesSize = Number((_colChunk$meta_data5 = colChunk.meta_data) === null || _colChunk$meta_data5 === void 0 ? void 0 : _colChunk$meta_data5.total_compressed_size);
550
- if (!colChunk.file_path) {
551
- pagesSize = Math.min(this.file.size - pagesOffset, Number((_colChunk$meta_data6 = colChunk.meta_data) === null || _colChunk$meta_data6 === void 0 ? void 0 : _colChunk$meta_data6.total_compressed_size));
552
- }
553
- options = {
554
- type: type,
555
- rLevelMax: field.rLevelMax,
556
- dLevelMax: field.dLevelMax,
557
- compression: compression,
558
- column: field,
559
- numValues: (_colChunk$meta_data7 = colChunk.meta_data) === null || _colChunk$meta_data7 === void 0 ? void 0 : _colChunk$meta_data7.num_values,
560
- dictionary: []
561
- };
562
- dictionaryPageOffset = colChunk === null || colChunk === void 0 ? void 0 : (_colChunk$meta_data8 = colChunk.meta_data) === null || _colChunk$meta_data8 === void 0 ? void 0 : _colChunk$meta_data8.dictionary_page_offset;
563
- if (!dictionaryPageOffset) {
564
- _context11.next = 17;
565
- break;
566
- }
567
- dictionaryOffset = Number(dictionaryPageOffset);
568
- _context11.next = 16;
569
- return this.getDictionary(dictionaryOffset, options, pagesOffset);
289
+ _context8.prev = 12;
290
+ _context8.t0 = _context8["catch"](3);
291
+ _context8.next = 16;
292
+ return envelopeReader.close();
570
293
  case 16:
571
- dictionary = _context11.sent;
294
+ throw _context8.t0;
572
295
  case 17:
573
- dictionary = (_options$dictionary = options.dictionary) !== null && _options$dictionary !== void 0 && _options$dictionary.length ? options.dictionary : dictionary;
574
- _context11.next = 20;
575
- return this.file.read(pagesOffset, pagesSize);
576
- case 20:
577
- pagesBuf = _context11.sent;
578
- _context11.next = 23;
579
- return (0, _decoders.decodeDataPages)(pagesBuf, _objectSpread(_objectSpread({}, options), {}, {
580
- dictionary: dictionary
581
- }));
582
- case 23:
583
- return _context11.abrupt("return", _context11.sent);
584
- case 24:
585
296
  case "end":
586
- return _context11.stop();
587
- }
588
- }
589
- }, _callee11, this);
590
- }));
591
- function readColumnChunk(_x4, _x5) {
592
- return _readColumnChunk.apply(this, arguments);
593
- }
594
- return readColumnChunk;
595
- }()
596
- }, {
597
- key: "getDictionary",
598
- value: function () {
599
- var _getDictionary = (0, _asyncToGenerator2.default)(_regenerator.default.mark(function _callee12(dictionaryPageOffset, options, pagesOffset) {
600
- var dictionarySize, pagesBuf, cursor, decodedPage;
601
- return _regenerator.default.wrap(function _callee12$(_context12) {
602
- while (1) {
603
- switch (_context12.prev = _context12.next) {
604
- case 0:
605
- if (!(dictionaryPageOffset === 0)) {
606
- _context12.next = 2;
607
- break;
608
- }
609
- return _context12.abrupt("return", []);
610
- case 2:
611
- dictionarySize = Math.min(this.file.size - dictionaryPageOffset, this.props.defaultDictionarySize);
612
- _context12.next = 5;
613
- return this.file.read(dictionaryPageOffset, dictionarySize);
614
- case 5:
615
- pagesBuf = _context12.sent;
616
- cursor = {
617
- buffer: pagesBuf,
618
- offset: 0,
619
- size: pagesBuf.length
620
- };
621
- _context12.next = 9;
622
- return (0, _decoders.decodePage)(cursor, options);
623
- case 9:
624
- decodedPage = _context12.sent;
625
- return _context12.abrupt("return", decodedPage.dictionary);
626
- case 11:
627
- case "end":
628
- return _context12.stop();
297
+ return _context8.stop();
629
298
  }
630
299
  }
631
- }, _callee12, this);
300
+ }, _callee8, null, [[3, 12]]);
632
301
  }));
633
- function getDictionary(_x6, _x7, _x8) {
634
- return _getDictionary.apply(this, arguments);
302
+ function openBuffer(_x7) {
303
+ return _openBuffer.apply(this, arguments);
635
304
  }
636
- return getDictionary;
305
+ return openBuffer;
637
306
  }()
638
307
  }]);
639
308
  return ParquetReader;