@loaders.gl/parquet 3.1.0-beta.3 → 3.1.1
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.
- package/dist/dist.min.js +6 -6
- package/dist/dist.min.js.map +2 -2
- package/dist/es5/bundle.js +1 -1
- package/dist/es5/bundle.js.map +1 -1
- package/dist/es5/constants.js +5 -5
- package/dist/es5/constants.js.map +1 -1
- package/dist/es5/index.js +19 -10
- package/dist/es5/index.js.map +1 -1
- package/dist/es5/lib/convert-schema.js +13 -13
- package/dist/es5/lib/convert-schema.js.map +1 -1
- package/dist/es5/lib/parse-parquet.js +154 -19
- package/dist/es5/lib/parse-parquet.js.map +1 -1
- package/dist/es5/lib/read-array-buffer.js +43 -6
- package/dist/es5/lib/read-array-buffer.js.map +1 -1
- package/dist/es5/parquet-loader.js +4 -4
- package/dist/es5/parquet-loader.js.map +1 -1
- package/dist/es5/parquet-writer.js +4 -4
- package/dist/es5/parquet-writer.js.map +1 -1
- package/dist/es5/parquetjs/codecs/dictionary.js +10 -2
- package/dist/es5/parquetjs/codecs/dictionary.js.map +1 -1
- package/dist/es5/parquetjs/codecs/index.js +6 -4
- package/dist/es5/parquetjs/codecs/index.js.map +1 -1
- package/dist/es5/parquetjs/codecs/plain.js +43 -41
- package/dist/es5/parquetjs/codecs/plain.js.map +1 -1
- package/dist/es5/parquetjs/codecs/rle.js +35 -25
- package/dist/es5/parquetjs/codecs/rle.js.map +1 -1
- package/dist/es5/parquetjs/compression.js +110 -27
- package/dist/es5/parquetjs/compression.js.map +1 -1
- package/dist/es5/parquetjs/encoder/writer.js +737 -301
- package/dist/es5/parquetjs/encoder/writer.js.map +1 -1
- package/dist/es5/parquetjs/file.js +15 -15
- package/dist/es5/parquetjs/file.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/BoundaryOrder.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/BsonType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js +152 -141
- package/dist/es5/parquetjs/parquet-thrift/ColumnChunk.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js +160 -147
- package/dist/es5/parquetjs/parquet-thrift/ColumnIndex.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js +259 -248
- package/dist/es5/parquetjs/parquet-thrift/ColumnMetaData.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js +79 -67
- package/dist/es5/parquetjs/parquet-thrift/ColumnOrder.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/CompressionCodec.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ConvertedType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js +124 -113
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js +169 -158
- package/dist/es5/parquetjs/parquet-thrift/DataPageHeaderV2.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DateType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/DateType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js +79 -68
- package/dist/es5/parquetjs/parquet-thrift/DecimalType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js +94 -83
- package/dist/es5/parquetjs/parquet-thrift/DictionaryPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Encoding.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/EnumType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/FieldRepetitionType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js +182 -170
- package/dist/es5/parquetjs/parquet-thrift/FileMetaData.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/IndexPageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/IntType.js +79 -68
- package/dist/es5/parquetjs/parquet-thrift/IntType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/JsonType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js +79 -68
- package/dist/es5/parquetjs/parquet-thrift/KeyValue.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/ListType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/ListType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js +343 -319
- package/dist/es5/parquetjs/parquet-thrift/LogicalType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MapType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/MapType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/MicroSeconds.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/MilliSeconds.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/NullType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/NullType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js +75 -64
- package/dist/es5/parquetjs/parquet-thrift/OffsetIndex.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js +94 -83
- package/dist/es5/parquetjs/parquet-thrift/PageEncodingStats.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js +169 -158
- package/dist/es5/parquetjs/parquet-thrift/PageHeader.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js +94 -83
- package/dist/es5/parquetjs/parquet-thrift/PageLocation.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/PageType.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js +124 -113
- package/dist/es5/parquetjs/parquet-thrift/RowGroup.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js +199 -188
- package/dist/es5/parquetjs/parquet-thrift/SchemaElement.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js +94 -83
- package/dist/es5/parquetjs/parquet-thrift/SortingColumn.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js +135 -124
- package/dist/es5/parquetjs/parquet-thrift/Statistics.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/StringType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/StringType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js +79 -68
- package/dist/es5/parquetjs/parquet-thrift/TimeType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js +101 -88
- package/dist/es5/parquetjs/parquet-thrift/TimeUnit.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js +79 -68
- package/dist/es5/parquetjs/parquet-thrift/TimestampType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/Type.js +1 -1
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/TypeDefinedOrder.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js +45 -31
- package/dist/es5/parquetjs/parquet-thrift/UUIDType.js.map +1 -1
- package/dist/es5/parquetjs/parquet-thrift/index.js +43 -43
- package/dist/es5/parquetjs/parser/decoders.js +391 -218
- package/dist/es5/parquetjs/parser/decoders.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-cursor.js +180 -62
- package/dist/es5/parquetjs/parser/parquet-cursor.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-envelope-reader.js +370 -125
- package/dist/es5/parquetjs/parser/parquet-envelope-reader.js.map +1 -1
- package/dist/es5/parquetjs/parser/parquet-reader.js +320 -91
- package/dist/es5/parquetjs/parser/parquet-reader.js.map +1 -1
- package/dist/es5/parquetjs/schema/declare.js +11 -9
- package/dist/es5/parquetjs/schema/declare.js.map +1 -1
- package/dist/es5/parquetjs/schema/schema.js +87 -73
- package/dist/es5/parquetjs/schema/schema.js.map +1 -1
- package/dist/es5/parquetjs/schema/shred.js +96 -56
- package/dist/es5/parquetjs/schema/shred.js.map +1 -1
- package/dist/es5/parquetjs/schema/types.js +40 -39
- package/dist/es5/parquetjs/schema/types.js.map +1 -1
- package/dist/es5/parquetjs/utils/buffer-utils.js +1 -1
- package/dist/es5/parquetjs/utils/buffer-utils.js.map +1 -1
- package/dist/es5/parquetjs/utils/file-utils.js +12 -8
- package/dist/es5/parquetjs/utils/file-utils.js.map +1 -1
- package/dist/es5/parquetjs/utils/read-utils.js +50 -22
- package/dist/es5/parquetjs/utils/read-utils.js.map +1 -1
- package/dist/esm/parquet-loader.js +1 -1
- package/dist/esm/parquet-loader.js.map +1 -1
- package/dist/esm/parquet-writer.js +1 -1
- package/dist/esm/parquet-writer.js.map +1 -1
- package/package.json +5 -5
|
@@ -1,5 +1,7 @@
|
|
|
1
1
|
"use strict";
|
|
2
2
|
|
|
3
|
+
var _typeof = require("@babel/runtime/helpers/typeof");
|
|
4
|
+
|
|
3
5
|
Object.defineProperty(exports, "__esModule", {
|
|
4
6
|
value: true
|
|
5
7
|
});
|
|
@@ -22,17 +24,17 @@ Object.keys(_declare).forEach(function (key) {
|
|
|
22
24
|
if (key in exports && exports[key] === _declare[key]) return;
|
|
23
25
|
Object.defineProperty(exports, key, {
|
|
24
26
|
enumerable: true,
|
|
25
|
-
get: function () {
|
|
27
|
+
get: function get() {
|
|
26
28
|
return _declare[key];
|
|
27
29
|
}
|
|
28
30
|
});
|
|
29
31
|
});
|
|
30
32
|
|
|
31
|
-
function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function (nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
|
|
33
|
+
function _getRequireWildcardCache(nodeInterop) { if (typeof WeakMap !== "function") return null; var cacheBabelInterop = new WeakMap(); var cacheNodeInterop = new WeakMap(); return (_getRequireWildcardCache = function _getRequireWildcardCache(nodeInterop) { return nodeInterop ? cacheNodeInterop : cacheBabelInterop; })(nodeInterop); }
|
|
32
34
|
|
|
33
|
-
function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null ||
|
|
35
|
+
function _interopRequireWildcard(obj, nodeInterop) { if (!nodeInterop && obj && obj.__esModule) { return obj; } if (obj === null || _typeof(obj) !== "object" && typeof obj !== "function") { return { default: obj }; } var cache = _getRequireWildcardCache(nodeInterop); if (cache && cache.has(obj)) { return cache.get(obj); } var newObj = {}; var hasPropertyDescriptor = Object.defineProperty && Object.getOwnPropertyDescriptor; for (var key in obj) { if (key !== "default" && Object.prototype.hasOwnProperty.call(obj, key)) { var desc = hasPropertyDescriptor ? Object.getOwnPropertyDescriptor(obj, key) : null; if (desc && (desc.get || desc.set)) { Object.defineProperty(newObj, key, desc); } else { newObj[key] = obj[key]; } } } newObj.default = obj; if (cache) { cache.set(obj, newObj); } return newObj; }
|
|
34
36
|
|
|
35
|
-
|
|
37
|
+
var PARQUET_CODECS = {
|
|
36
38
|
PLAIN: {
|
|
37
39
|
encodeValues: PLAIN.encodeValues,
|
|
38
40
|
decodeValues: PLAIN.decodeValues
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../../../../src/parquetjs/codecs/index.ts"],"names":["PARQUET_CODECS","PLAIN","encodeValues","decodeValues","RLE","PLAIN_DICTIONARY","DICTIONARY","RLE_DICTIONARY"],"mappings":"
|
|
1
|
+
{"version":3,"sources":["../../../../src/parquetjs/codecs/index.ts"],"names":["PARQUET_CODECS","PLAIN","encodeValues","decodeValues","RLE","PLAIN_DICTIONARY","DICTIONARY","RLE_DICTIONARY"],"mappings":";;;;;;;;;;;;AAGA;;AACA;;AACA;;AAEA;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;;;;;;AAEO,IAAMA,cAAqD,GAAG;AACnEC,EAAAA,KAAK,EAAE;AACLC,IAAAA,YAAY,EAAED,KAAK,CAACC,YADf;AAELC,IAAAA,YAAY,EAAEF,KAAK,CAACE;AAFf,GAD4D;AAKnEC,EAAAA,GAAG,EAAE;AACHF,IAAAA,YAAY,EAAEE,GAAG,CAACF,YADf;AAEHC,IAAAA,YAAY,EAAEC,GAAG,CAACD;AAFf,GAL8D;AAUnEE,EAAAA,gBAAgB,EAAE;AAEhBH,IAAAA,YAAY,EAAEI,UAAU,CAACJ,YAFT;AAGhBC,IAAAA,YAAY,EAAEG,UAAU,CAACH;AAHT,GAViD;AAgBnEI,EAAAA,cAAc,EAAE;AAEdL,IAAAA,YAAY,EAAEI,UAAU,CAACJ,YAFX;AAGdC,IAAAA,YAAY,EAAEG,UAAU,CAACH;AAHX;AAhBmD,CAA9D","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\nimport type {ParquetCodec} from '../schema/declare';\nimport type {ParquetCodecKit} from './declare';\nimport * as PLAIN from './plain';\nimport * as RLE from './rle';\nimport * as DICTIONARY from './dictionary';\n\nexport * from './declare';\n\nexport const PARQUET_CODECS: Record<ParquetCodec, ParquetCodecKit> = {\n PLAIN: {\n encodeValues: PLAIN.encodeValues,\n decodeValues: PLAIN.decodeValues\n },\n RLE: {\n encodeValues: RLE.encodeValues,\n decodeValues: RLE.decodeValues\n },\n // Using the PLAIN_DICTIONARY enum value is deprecated in the Parquet 2.0 specification.\n PLAIN_DICTIONARY: {\n // @ts-ignore\n encodeValues: DICTIONARY.encodeValues,\n decodeValues: DICTIONARY.decodeValues\n },\n // Prefer using RLE_DICTIONARY in a data page and PLAIN in a dictionary page for Parquet 2.0+ files.\n RLE_DICTIONARY: {\n // @ts-ignore\n encodeValues: DICTIONARY.encodeValues,\n decodeValues: DICTIONARY.decodeValues\n }\n};\n"],"file":"index.js"}
|
|
@@ -73,10 +73,10 @@ function decodeValues(type, cursor, count, opts) {
|
|
|
73
73
|
}
|
|
74
74
|
|
|
75
75
|
function encodeValues_BOOLEAN(values) {
|
|
76
|
-
|
|
76
|
+
var buf = Buffer.alloc(Math.ceil(values.length / 8));
|
|
77
77
|
buf.fill(0);
|
|
78
78
|
|
|
79
|
-
for (
|
|
79
|
+
for (var i = 0; i < values.length; i++) {
|
|
80
80
|
if (values[i]) {
|
|
81
81
|
buf[Math.floor(i / 8)] |= 1 << i % 8;
|
|
82
82
|
}
|
|
@@ -86,10 +86,10 @@ function encodeValues_BOOLEAN(values) {
|
|
|
86
86
|
}
|
|
87
87
|
|
|
88
88
|
function decodeValues_BOOLEAN(cursor, count) {
|
|
89
|
-
|
|
89
|
+
var values = [];
|
|
90
90
|
|
|
91
|
-
for (
|
|
92
|
-
|
|
91
|
+
for (var i = 0; i < count; i++) {
|
|
92
|
+
var b = cursor.buffer[cursor.offset + Math.floor(i / 8)];
|
|
93
93
|
values.push((b & 1 << i % 8) > 0);
|
|
94
94
|
}
|
|
95
95
|
|
|
@@ -98,9 +98,9 @@ function decodeValues_BOOLEAN(cursor, count) {
|
|
|
98
98
|
}
|
|
99
99
|
|
|
100
100
|
function encodeValues_INT32(values) {
|
|
101
|
-
|
|
101
|
+
var buf = Buffer.alloc(4 * values.length);
|
|
102
102
|
|
|
103
|
-
for (
|
|
103
|
+
for (var i = 0; i < values.length; i++) {
|
|
104
104
|
buf.writeInt32LE(values[i], i * 4);
|
|
105
105
|
}
|
|
106
106
|
|
|
@@ -108,9 +108,9 @@ function encodeValues_INT32(values) {
|
|
|
108
108
|
}
|
|
109
109
|
|
|
110
110
|
function decodeValues_INT32(cursor, count) {
|
|
111
|
-
|
|
111
|
+
var values = [];
|
|
112
112
|
|
|
113
|
-
for (
|
|
113
|
+
for (var i = 0; i < count; i++) {
|
|
114
114
|
values.push(cursor.buffer.readInt32LE(cursor.offset));
|
|
115
115
|
cursor.offset += 4;
|
|
116
116
|
}
|
|
@@ -119,9 +119,9 @@ function decodeValues_INT32(cursor, count) {
|
|
|
119
119
|
}
|
|
120
120
|
|
|
121
121
|
function encodeValues_INT64(values) {
|
|
122
|
-
|
|
122
|
+
var buf = Buffer.alloc(8 * values.length);
|
|
123
123
|
|
|
124
|
-
for (
|
|
124
|
+
for (var i = 0; i < values.length; i++) {
|
|
125
125
|
_int.default.writeInt64LE(values[i], buf, i * 8);
|
|
126
126
|
}
|
|
127
127
|
|
|
@@ -129,9 +129,9 @@ function encodeValues_INT64(values) {
|
|
|
129
129
|
}
|
|
130
130
|
|
|
131
131
|
function decodeValues_INT64(cursor, count) {
|
|
132
|
-
|
|
132
|
+
var values = [];
|
|
133
133
|
|
|
134
|
-
for (
|
|
134
|
+
for (var i = 0; i < count; i++) {
|
|
135
135
|
values.push(_int.default.readInt64LE(cursor.buffer, cursor.offset));
|
|
136
136
|
cursor.offset += 8;
|
|
137
137
|
}
|
|
@@ -140,9 +140,9 @@ function decodeValues_INT64(cursor, count) {
|
|
|
140
140
|
}
|
|
141
141
|
|
|
142
142
|
function encodeValues_INT96(values) {
|
|
143
|
-
|
|
143
|
+
var buf = Buffer.alloc(12 * values.length);
|
|
144
144
|
|
|
145
|
-
for (
|
|
145
|
+
for (var i = 0; i < values.length; i++) {
|
|
146
146
|
if (values[i] >= 0) {
|
|
147
147
|
_int.default.writeInt64LE(values[i], buf, i * 12);
|
|
148
148
|
|
|
@@ -158,12 +158,12 @@ function encodeValues_INT96(values) {
|
|
|
158
158
|
}
|
|
159
159
|
|
|
160
160
|
function decodeValues_INT96(cursor, count) {
|
|
161
|
-
|
|
161
|
+
var values = [];
|
|
162
162
|
|
|
163
|
-
for (
|
|
164
|
-
|
|
163
|
+
for (var i = 0; i < count; i++) {
|
|
164
|
+
var low = _int.default.readInt64LE(cursor.buffer, cursor.offset);
|
|
165
165
|
|
|
166
|
-
|
|
166
|
+
var high = cursor.buffer.readUInt32LE(cursor.offset + 8);
|
|
167
167
|
|
|
168
168
|
if (high === 0xffffffff) {
|
|
169
169
|
values.push(~-low + 1);
|
|
@@ -178,9 +178,9 @@ function decodeValues_INT96(cursor, count) {
|
|
|
178
178
|
}
|
|
179
179
|
|
|
180
180
|
function encodeValues_FLOAT(values) {
|
|
181
|
-
|
|
181
|
+
var buf = Buffer.alloc(4 * values.length);
|
|
182
182
|
|
|
183
|
-
for (
|
|
183
|
+
for (var i = 0; i < values.length; i++) {
|
|
184
184
|
buf.writeFloatLE(values[i], i * 4);
|
|
185
185
|
}
|
|
186
186
|
|
|
@@ -188,9 +188,9 @@ function encodeValues_FLOAT(values) {
|
|
|
188
188
|
}
|
|
189
189
|
|
|
190
190
|
function decodeValues_FLOAT(cursor, count) {
|
|
191
|
-
|
|
191
|
+
var values = [];
|
|
192
192
|
|
|
193
|
-
for (
|
|
193
|
+
for (var i = 0; i < count; i++) {
|
|
194
194
|
values.push(cursor.buffer.readFloatLE(cursor.offset));
|
|
195
195
|
cursor.offset += 4;
|
|
196
196
|
}
|
|
@@ -199,9 +199,9 @@ function decodeValues_FLOAT(cursor, count) {
|
|
|
199
199
|
}
|
|
200
200
|
|
|
201
201
|
function encodeValues_DOUBLE(values) {
|
|
202
|
-
|
|
202
|
+
var buf = Buffer.alloc(8 * values.length);
|
|
203
203
|
|
|
204
|
-
for (
|
|
204
|
+
for (var i = 0; i < values.length; i++) {
|
|
205
205
|
buf.writeDoubleLE(values[i], i * 8);
|
|
206
206
|
}
|
|
207
207
|
|
|
@@ -209,9 +209,9 @@ function encodeValues_DOUBLE(values) {
|
|
|
209
209
|
}
|
|
210
210
|
|
|
211
211
|
function decodeValues_DOUBLE(cursor, count) {
|
|
212
|
-
|
|
212
|
+
var values = [];
|
|
213
213
|
|
|
214
|
-
for (
|
|
214
|
+
for (var i = 0; i < count; i++) {
|
|
215
215
|
values.push(cursor.buffer.readDoubleLE(cursor.offset));
|
|
216
216
|
cursor.offset += 8;
|
|
217
217
|
}
|
|
@@ -220,30 +220,32 @@ function decodeValues_DOUBLE(cursor, count) {
|
|
|
220
220
|
}
|
|
221
221
|
|
|
222
222
|
function encodeValues_BYTE_ARRAY(values) {
|
|
223
|
-
|
|
223
|
+
var buf_len = 0;
|
|
224
224
|
|
|
225
|
-
for (
|
|
225
|
+
for (var i = 0; i < values.length; i++) {
|
|
226
226
|
values[i] = Buffer.from(values[i]);
|
|
227
227
|
buf_len += 4 + values[i].length;
|
|
228
228
|
}
|
|
229
229
|
|
|
230
|
-
|
|
231
|
-
|
|
230
|
+
var buf = Buffer.alloc(buf_len);
|
|
231
|
+
var buf_pos = 0;
|
|
232
232
|
|
|
233
|
-
for (
|
|
234
|
-
buf.writeUInt32LE(values[
|
|
235
|
-
|
|
236
|
-
buf_pos
|
|
233
|
+
for (var _i = 0; _i < values.length; _i++) {
|
|
234
|
+
buf.writeUInt32LE(values[_i].length, buf_pos);
|
|
235
|
+
|
|
236
|
+
values[_i].copy(buf, buf_pos + 4);
|
|
237
|
+
|
|
238
|
+
buf_pos += 4 + values[_i].length;
|
|
237
239
|
}
|
|
238
240
|
|
|
239
241
|
return buf;
|
|
240
242
|
}
|
|
241
243
|
|
|
242
244
|
function decodeValues_BYTE_ARRAY(cursor, count) {
|
|
243
|
-
|
|
245
|
+
var values = [];
|
|
244
246
|
|
|
245
|
-
for (
|
|
246
|
-
|
|
247
|
+
for (var i = 0; i < count; i++) {
|
|
248
|
+
var len = cursor.buffer.readUInt32LE(cursor.offset);
|
|
247
249
|
cursor.offset += 4;
|
|
248
250
|
values.push(cursor.buffer.slice(cursor.offset, cursor.offset + len));
|
|
249
251
|
cursor.offset += len;
|
|
@@ -257,7 +259,7 @@ function encodeValues_FIXED_LEN_BYTE_ARRAY(values, opts) {
|
|
|
257
259
|
throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');
|
|
258
260
|
}
|
|
259
261
|
|
|
260
|
-
for (
|
|
262
|
+
for (var i = 0; i < values.length; i++) {
|
|
261
263
|
values[i] = Buffer.from(values[i]);
|
|
262
264
|
|
|
263
265
|
if (values[i].length !== opts.typeLength) {
|
|
@@ -269,13 +271,13 @@ function encodeValues_FIXED_LEN_BYTE_ARRAY(values, opts) {
|
|
|
269
271
|
}
|
|
270
272
|
|
|
271
273
|
function decodeValues_FIXED_LEN_BYTE_ARRAY(cursor, count, opts) {
|
|
272
|
-
|
|
274
|
+
var values = [];
|
|
273
275
|
|
|
274
276
|
if (!opts.typeLength) {
|
|
275
277
|
throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');
|
|
276
278
|
}
|
|
277
279
|
|
|
278
|
-
for (
|
|
280
|
+
for (var i = 0; i < count; i++) {
|
|
279
281
|
values.push(cursor.buffer.slice(cursor.offset, cursor.offset + opts.typeLength));
|
|
280
282
|
cursor.offset += opts.typeLength;
|
|
281
283
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../../../../src/parquetjs/codecs/plain.ts"],"names":["encodeValues","type","values","opts","encodeValues_BOOLEAN","encodeValues_INT32","encodeValues_INT64","encodeValues_INT96","encodeValues_FLOAT","encodeValues_DOUBLE","encodeValues_BYTE_ARRAY","encodeValues_FIXED_LEN_BYTE_ARRAY","Error","decodeValues","cursor","count","decodeValues_BOOLEAN","decodeValues_INT32","decodeValues_INT64","decodeValues_INT96","decodeValues_FLOAT","decodeValues_DOUBLE","decodeValues_BYTE_ARRAY","decodeValues_FIXED_LEN_BYTE_ARRAY","buf","Buffer","alloc","Math","ceil","length","fill","i","floor","b","buffer","offset","push","writeInt32LE","readInt32LE","INT53","writeInt64LE","readInt64LE","writeUInt32LE","low","high","readUInt32LE","writeFloatLE","readFloatLE","writeDoubleLE","readDoubleLE","buf_len","from","buf_pos","copy","len","slice","typeLength","concat"],"mappings":";;;;;;;;;;AAIA;;AAEO,SAASA,YAAT,CACLC,IADK,EAELC,MAFK,EAGLC,IAHK,EAIG;AACR,UAAQF,IAAR;AACE,SAAK,SAAL;AACE,aAAOG,oBAAoB,CAACF,MAAD,CAA3B;;AACF,SAAK,OAAL;AACE,aAAOG,kBAAkB,CAACH,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOI,kBAAkB,CAACJ,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOK,kBAAkB,CAACL,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOM,kBAAkB,CAACN,MAAD,CAAzB;;AACF,SAAK,QAAL;AACE,aAAOO,mBAAmB,CAACP,MAAD,CAA1B;;AACF,SAAK,YAAL;AACE,aAAOQ,uBAAuB,CAACR,MAAD,CAA9B;;AACF,SAAK,sBAAL;AACE,aAAOS,iCAAiC,CAACT,MAAD,EAASC,IAAT,CAAxC;;AACF;AACE,YAAM,IAAIS,KAAJ,6BAA+BX,IAA/B,EAAN;AAlBJ;AAoBD;;AAEM,SAASY,YAAT,CACLZ,IADK,EAELa,MAFK,EAGLC,KAHK,EAILZ,IAJK,EAKE;AACP,UAAQF,IAAR;AACE,SAAK,SAAL;AACE,aAAOe,oBAAoB,CAACF,MAAD,EAASC,KAAT,CAA3B;;AACF,SAAK,OAAL;AACE,aAAOE,kBAAkB,CAACH,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOG,kBAAkB,CAACJ,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOI,kBAAkB,CAACL,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOK,kBAAkB,CAACN,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,QAAL;AACE,aAAOM,mBAAmB,CAACP,MAAD,EAASC,KAAT,CAA1B;;AACF,SAAK,YAAL;AACE,aAAOO,uBAAuB,CAACR,MAAD,EAASC,KAAT,CAA9B;;AACF,SAAK,sBAAL;AACE,aAAOQ,iCAAiC,CAACT,MAAD,EAASC,KAAT,EAAgBZ,IAAhB,CAAxC;;AACF;AACE,YAAM,IAAIS,KAAJ,6BAA+BX,IAA/B,EAAN;AAlBJ;AAoBD;;AAED,SAASG,oBAAT,CAA8BF,MAA9B,EAAyD;AACvD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAaC,IAAI,CAACC,IAAL,CAAU1B,MAAM,CAAC2B,MAAP,GAAgB,CAA1B,CAAb,CAAZ;AACAL,EAAAA,GAAG,CAACM,IAAJ,CAAS,CAAT;;AACA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC,QAAI7B,MAAM,CAAC6B,CAAD,CAAV,EAAe;AACbP,MAAAA,GAAG,CAACG,IAAI,CAACK,KAAL,CAAWD,CAAC,GAAG,CAAf,CAAD,CAAH,IAA0B,KAAKA,CAAC,GAAG,CAAnC;AACD;AACF;;AACD,SAAOP,GAAP;AACD;;AAED,SAASR,oBAAT,CAA8BF,MAA9B,EAAoDC,KAApD,EAA8E;AAC5E,QAAMb,MAAiB,GAAG,EAA1B;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,UAAME,CAAC,GAAGnB,MAAM,CAACoB,MAAP,CAAcpB,MAAM,CAACqB,MAAP,GAAgBR,IAAI,CAACK,KAAL,CAAWD,CAAC,GAAG,CAAf,CAA9B,CAAV;AACA7B,IAAAA,MAAM,CAACkC,IAAP,CAAY,CAACH,CAAC,GAAI,KAAKF,CAAC,GAAG,CAAf,IAAqB,CAAjC;AACD;;AACDjB,EAAAA,MAAM,CAACqB,MAAP,IAAiBR,IAAI,CAACC,IAAL,CAAUb,KAAK,GAAG,CAAlB,CAAjB;AACA,SAAOb,MAAP;AACD;;AAED,SAASG,kBAAT,CAA4BH,MAA5B,EAAsD;AACpD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACa,YAAJ,CAAiBnC,MAAM,CAAC6B,CAAD,CAAvB,EAA4BA,CAAC,GAAG,CAAhC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASP,kBAAT,CAA4BH,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcI,WAAd,CAA0BxB,MAAM,CAACqB,MAAjC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASI,kBAAT,CAA4BJ,MAA5B,EAAsD;AACpD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCQ,iBAAMC,YAAN,CAAmBtC,MAAM,CAAC6B,CAAD,CAAzB,EAA8BP,GAA9B,EAAmCO,CAAC,GAAG,CAAvC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASN,kBAAT,CAA4BJ,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYG,aAAME,WAAN,CAAkB3B,MAAM,CAACoB,MAAzB,EAAiCpB,MAAM,CAACqB,MAAxC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASK,kBAAT,CAA4BL,MAA5B,EAAsD;AACpD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,KAAKxB,MAAM,CAAC2B,MAAzB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC,QAAI7B,MAAM,CAAC6B,CAAD,CAAN,IAAa,CAAjB,EAAoB;AAClBQ,mBAAMC,YAAN,CAAmBtC,MAAM,CAAC6B,CAAD,CAAzB,EAA8BP,GAA9B,EAAmCO,CAAC,GAAG,EAAvC;;AACAP,MAAAA,GAAG,CAACkB,aAAJ,CAAkB,CAAlB,EAAqBX,CAAC,GAAG,EAAJ,GAAS,CAA9B;AACD,KAHD,MAGO;AACLQ,mBAAMC,YAAN,CAAmB,CAAC,CAACtC,MAAM,CAAC6B,CAAD,CAAR,GAAc,CAAjC,EAAoCP,GAApC,EAAyCO,CAAC,GAAG,EAA7C;;AACAP,MAAAA,GAAG,CAACkB,aAAJ,CAAkB,UAAlB,EAA8BX,CAAC,GAAG,EAAJ,GAAS,CAAvC;AACD;AACF;;AACD,SAAOP,GAAP;AACD;;AAED,SAASL,kBAAT,CAA4BL,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,UAAMY,GAAG,GAAGJ,aAAME,WAAN,CAAkB3B,MAAM,CAACoB,MAAzB,EAAiCpB,MAAM,CAACqB,MAAxC,CAAZ;;AACA,UAAMS,IAAI,GAAG9B,MAAM,CAACoB,MAAP,CAAcW,YAAd,CAA2B/B,MAAM,CAACqB,MAAP,GAAgB,CAA3C,CAAb;;AACA,QAAIS,IAAI,KAAK,UAAb,EAAyB;AACvB1C,MAAAA,MAAM,CAACkC,IAAP,CAAY,CAAC,CAACO,GAAF,GAAQ,CAApB;AACD,KAFD,MAEO;AACLzC,MAAAA,MAAM,CAACkC,IAAP,CAAYO,GAAZ;AACD;;AACD7B,IAAAA,MAAM,CAACqB,MAAP,IAAiB,EAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASM,kBAAT,CAA4BN,MAA5B,EAAsD;AACpD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACsB,YAAJ,CAAiB5C,MAAM,CAAC6B,CAAD,CAAvB,EAA4BA,CAAC,GAAG,CAAhC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASJ,kBAAT,CAA4BN,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAca,WAAd,CAA0BjC,MAAM,CAACqB,MAAjC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASO,mBAAT,CAA6BP,MAA7B,EAAuD;AACrD,QAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACwB,aAAJ,CAAkB9C,MAAM,CAAC6B,CAAD,CAAxB,EAA6BA,CAAC,GAAG,CAAjC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASH,mBAAT,CAA6BP,MAA7B,EAAmDC,KAAnD,EAA4E;AAC1E,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAce,YAAd,CAA2BnC,MAAM,CAACqB,MAAlC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASQ,uBAAT,CAAiCR,MAAjC,EAA2D;AAEzD,MAAIgD,OAAO,GAAG,CAAd;;AACA,OAAK,IAAInB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC7B,IAAAA,MAAM,CAAC6B,CAAD,CAAN,GAAYN,MAAM,CAAC0B,IAAP,CAAYjD,MAAM,CAAC6B,CAAD,CAAlB,CAAZ;AACAmB,IAAAA,OAAO,IAAI,IAAIhD,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAAzB;AACD;;AACD,QAAML,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAawB,OAAb,CAAZ;AAEA,MAAIE,OAAO,GAAG,CAAd;;AACA,OAAK,IAAIrB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACkB,aAAJ,CAAkBxC,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAA5B,EAAoCuB,OAApC;AACAlD,IAAAA,MAAM,CAAC6B,CAAD,CAAN,CAAUsB,IAAV,CAAe7B,GAAf,EAAoB4B,OAAO,GAAG,CAA9B;AACAA,IAAAA,OAAO,IAAI,IAAIlD,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAAzB;AACD;;AACD,SAAOL,GAAP;AACD;;AAED,SAASF,uBAAT,CAAiCR,MAAjC,EAAuDC,KAAvD,EAAgF;AAC9E,QAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,UAAMuB,GAAG,GAAGxC,MAAM,CAACoB,MAAP,CAAcW,YAAd,CAA2B/B,MAAM,CAACqB,MAAlC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACAjC,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcqB,KAAd,CAAoBzC,MAAM,CAACqB,MAA3B,EAAmCrB,MAAM,CAACqB,MAAP,GAAgBmB,GAAnD,CAAZ;AACAxC,IAAAA,MAAM,CAACqB,MAAP,IAAiBmB,GAAjB;AACD;;AACD,SAAOpD,MAAP;AACD;;AAED,SAASS,iCAAT,CAA2CT,MAA3C,EAA6DC,IAA7D,EAAgG;AAC9F,MAAI,CAACA,IAAI,CAACqD,UAAV,EAAsB;AACpB,UAAM,IAAI5C,KAAJ,CAAU,gEAAV,CAAN;AACD;;AACD,OAAK,IAAImB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC7B,IAAAA,MAAM,CAAC6B,CAAD,CAAN,GAAYN,MAAM,CAAC0B,IAAP,CAAYjD,MAAM,CAAC6B,CAAD,CAAlB,CAAZ;;AACA,QAAI7B,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAAV,KAAqB1B,IAAI,CAACqD,UAA9B,EAA0C;AACxC,YAAM,IAAI5C,KAAJ,mDAAqDV,MAAM,CAAC6B,CAAD,CAA3D,EAAN;AACD;AACF;;AACD,SAAON,MAAM,CAACgC,MAAP,CAAcvD,MAAd,CAAP;AACD;;AAED,SAASqB,iCAAT,CACET,MADF,EAEEC,KAFF,EAGEZ,IAHF,EAIY;AACV,QAAMD,MAAgB,GAAG,EAAzB;;AACA,MAAI,CAACC,IAAI,CAACqD,UAAV,EAAsB;AACpB,UAAM,IAAI5C,KAAJ,CAAU,gEAAV,CAAN;AACD;;AACD,OAAK,IAAImB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcqB,KAAd,CAAoBzC,MAAM,CAACqB,MAA3B,EAAmCrB,MAAM,CAACqB,MAAP,GAAgBhC,IAAI,CAACqD,UAAxD,CAAZ;AACA1C,IAAAA,MAAM,CAACqB,MAAP,IAAiBhC,IAAI,CAACqD,UAAtB;AACD;;AACD,SAAOtD,MAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n/* eslint-disable camelcase */\nimport type {PrimitiveType} from '../schema/declare';\nimport type {CursorBuffer, ParquetCodecOptions} from './declare';\nimport INT53 from 'int53';\n\nexport function encodeValues(\n type: PrimitiveType,\n values: any[],\n opts: ParquetCodecOptions\n): Buffer {\n switch (type) {\n case 'BOOLEAN':\n return encodeValues_BOOLEAN(values);\n case 'INT32':\n return encodeValues_INT32(values);\n case 'INT64':\n return encodeValues_INT64(values);\n case 'INT96':\n return encodeValues_INT96(values);\n case 'FLOAT':\n return encodeValues_FLOAT(values);\n case 'DOUBLE':\n return encodeValues_DOUBLE(values);\n case 'BYTE_ARRAY':\n return encodeValues_BYTE_ARRAY(values);\n case 'FIXED_LEN_BYTE_ARRAY':\n return encodeValues_FIXED_LEN_BYTE_ARRAY(values, opts);\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n}\n\nexport function decodeValues(\n type: PrimitiveType,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): any[] {\n switch (type) {\n case 'BOOLEAN':\n return decodeValues_BOOLEAN(cursor, count);\n case 'INT32':\n return decodeValues_INT32(cursor, count);\n case 'INT64':\n return decodeValues_INT64(cursor, count);\n case 'INT96':\n return decodeValues_INT96(cursor, count);\n case 'FLOAT':\n return decodeValues_FLOAT(cursor, count);\n case 'DOUBLE':\n return decodeValues_DOUBLE(cursor, count);\n case 'BYTE_ARRAY':\n return decodeValues_BYTE_ARRAY(cursor, count);\n case 'FIXED_LEN_BYTE_ARRAY':\n return decodeValues_FIXED_LEN_BYTE_ARRAY(cursor, count, opts);\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n}\n\nfunction encodeValues_BOOLEAN(values: boolean[]): Buffer {\n const buf = Buffer.alloc(Math.ceil(values.length / 8));\n buf.fill(0);\n for (let i = 0; i < values.length; i++) {\n if (values[i]) {\n buf[Math.floor(i / 8)] |= 1 << i % 8;\n }\n }\n return buf;\n}\n\nfunction decodeValues_BOOLEAN(cursor: CursorBuffer, count: number): boolean[] {\n const values: boolean[] = [];\n for (let i = 0; i < count; i++) {\n const b = cursor.buffer[cursor.offset + Math.floor(i / 8)];\n values.push((b & (1 << i % 8)) > 0);\n }\n cursor.offset += Math.ceil(count / 8);\n return values;\n}\n\nfunction encodeValues_INT32(values: number[]): Buffer {\n const buf = Buffer.alloc(4 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeInt32LE(values[i], i * 4);\n }\n return buf;\n}\n\nfunction decodeValues_INT32(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readInt32LE(cursor.offset));\n cursor.offset += 4;\n }\n return values;\n}\n\nfunction encodeValues_INT64(values: number[]): Buffer {\n const buf = Buffer.alloc(8 * values.length);\n for (let i = 0; i < values.length; i++) {\n INT53.writeInt64LE(values[i], buf, i * 8);\n }\n return buf;\n}\n\nfunction decodeValues_INT64(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(INT53.readInt64LE(cursor.buffer, cursor.offset));\n cursor.offset += 8;\n }\n return values;\n}\n\nfunction encodeValues_INT96(values: number[]): Buffer {\n const buf = Buffer.alloc(12 * values.length);\n for (let i = 0; i < values.length; i++) {\n if (values[i] >= 0) {\n INT53.writeInt64LE(values[i], buf, i * 12);\n buf.writeUInt32LE(0, i * 12 + 8); // truncate to 64 actual precision\n } else {\n INT53.writeInt64LE(~-values[i] + 1, buf, i * 12);\n buf.writeUInt32LE(0xffffffff, i * 12 + 8); // truncate to 64 actual precision\n }\n }\n return buf;\n}\n\nfunction decodeValues_INT96(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n const low = INT53.readInt64LE(cursor.buffer, cursor.offset);\n const high = cursor.buffer.readUInt32LE(cursor.offset + 8);\n if (high === 0xffffffff) {\n values.push(~-low + 1); // truncate to 64 actual precision\n } else {\n values.push(low); // truncate to 64 actual precision\n }\n cursor.offset += 12;\n }\n return values;\n}\n\nfunction encodeValues_FLOAT(values: number[]): Buffer {\n const buf = Buffer.alloc(4 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeFloatLE(values[i], i * 4);\n }\n return buf;\n}\n\nfunction decodeValues_FLOAT(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readFloatLE(cursor.offset));\n cursor.offset += 4;\n }\n return values;\n}\n\nfunction encodeValues_DOUBLE(values: number[]): Buffer {\n const buf = Buffer.alloc(8 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeDoubleLE(values[i], i * 8);\n }\n return buf;\n}\n\nfunction decodeValues_DOUBLE(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readDoubleLE(cursor.offset));\n cursor.offset += 8;\n }\n return values;\n}\n\nfunction encodeValues_BYTE_ARRAY(values: Buffer[]): Buffer {\n // tslint:disable-next-line:variable-name\n let buf_len = 0;\n for (let i = 0; i < values.length; i++) {\n values[i] = Buffer.from(values[i]);\n buf_len += 4 + values[i].length;\n }\n const buf = Buffer.alloc(buf_len);\n // tslint:disable-next-line:variable-name\n let buf_pos = 0;\n for (let i = 0; i < values.length; i++) {\n buf.writeUInt32LE(values[i].length, buf_pos);\n values[i].copy(buf, buf_pos + 4);\n buf_pos += 4 + values[i].length;\n }\n return buf;\n}\n\nfunction decodeValues_BYTE_ARRAY(cursor: CursorBuffer, count: number): Buffer[] {\n const values: Buffer[] = [];\n for (let i = 0; i < count; i++) {\n const len = cursor.buffer.readUInt32LE(cursor.offset);\n cursor.offset += 4;\n values.push(cursor.buffer.slice(cursor.offset, cursor.offset + len));\n cursor.offset += len;\n }\n return values;\n}\n\nfunction encodeValues_FIXED_LEN_BYTE_ARRAY(values: Buffer[], opts: ParquetCodecOptions): Buffer {\n if (!opts.typeLength) {\n throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');\n }\n for (let i = 0; i < values.length; i++) {\n values[i] = Buffer.from(values[i]);\n if (values[i].length !== opts.typeLength) {\n throw new Error(`invalid value for FIXED_LEN_BYTE_ARRAY: ${values[i]}`);\n }\n }\n return Buffer.concat(values);\n}\n\nfunction decodeValues_FIXED_LEN_BYTE_ARRAY(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): Buffer[] {\n const values: Buffer[] = [];\n if (!opts.typeLength) {\n throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');\n }\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.slice(cursor.offset, cursor.offset + opts.typeLength));\n cursor.offset += opts.typeLength;\n }\n return values;\n}\n"],"file":"plain.js"}
|
|
1
|
+
{"version":3,"sources":["../../../../src/parquetjs/codecs/plain.ts"],"names":["encodeValues","type","values","opts","encodeValues_BOOLEAN","encodeValues_INT32","encodeValues_INT64","encodeValues_INT96","encodeValues_FLOAT","encodeValues_DOUBLE","encodeValues_BYTE_ARRAY","encodeValues_FIXED_LEN_BYTE_ARRAY","Error","decodeValues","cursor","count","decodeValues_BOOLEAN","decodeValues_INT32","decodeValues_INT64","decodeValues_INT96","decodeValues_FLOAT","decodeValues_DOUBLE","decodeValues_BYTE_ARRAY","decodeValues_FIXED_LEN_BYTE_ARRAY","buf","Buffer","alloc","Math","ceil","length","fill","i","floor","b","buffer","offset","push","writeInt32LE","readInt32LE","INT53","writeInt64LE","readInt64LE","writeUInt32LE","low","high","readUInt32LE","writeFloatLE","readFloatLE","writeDoubleLE","readDoubleLE","buf_len","from","buf_pos","copy","len","slice","typeLength","concat"],"mappings":";;;;;;;;;;AAIA;;AAEO,SAASA,YAAT,CACLC,IADK,EAELC,MAFK,EAGLC,IAHK,EAIG;AACR,UAAQF,IAAR;AACE,SAAK,SAAL;AACE,aAAOG,oBAAoB,CAACF,MAAD,CAA3B;;AACF,SAAK,OAAL;AACE,aAAOG,kBAAkB,CAACH,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOI,kBAAkB,CAACJ,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOK,kBAAkB,CAACL,MAAD,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOM,kBAAkB,CAACN,MAAD,CAAzB;;AACF,SAAK,QAAL;AACE,aAAOO,mBAAmB,CAACP,MAAD,CAA1B;;AACF,SAAK,YAAL;AACE,aAAOQ,uBAAuB,CAACR,MAAD,CAA9B;;AACF,SAAK,sBAAL;AACE,aAAOS,iCAAiC,CAACT,MAAD,EAASC,IAAT,CAAxC;;AACF;AACE,YAAM,IAAIS,KAAJ,6BAA+BX,IAA/B,EAAN;AAlBJ;AAoBD;;AAEM,SAASY,YAAT,CACLZ,IADK,EAELa,MAFK,EAGLC,KAHK,EAILZ,IAJK,EAKE;AACP,UAAQF,IAAR;AACE,SAAK,SAAL;AACE,aAAOe,oBAAoB,CAACF,MAAD,EAASC,KAAT,CAA3B;;AACF,SAAK,OAAL;AACE,aAAOE,kBAAkB,CAACH,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOG,kBAAkB,CAACJ,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOI,kBAAkB,CAACL,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,OAAL;AACE,aAAOK,kBAAkB,CAACN,MAAD,EAASC,KAAT,CAAzB;;AACF,SAAK,QAAL;AACE,aAAOM,mBAAmB,CAACP,MAAD,EAASC,KAAT,CAA1B;;AACF,SAAK,YAAL;AACE,aAAOO,uBAAuB,CAACR,MAAD,EAASC,KAAT,CAA9B;;AACF,SAAK,sBAAL;AACE,aAAOQ,iCAAiC,CAACT,MAAD,EAASC,KAAT,EAAgBZ,IAAhB,CAAxC;;AACF;AACE,YAAM,IAAIS,KAAJ,6BAA+BX,IAA/B,EAAN;AAlBJ;AAoBD;;AAED,SAASG,oBAAT,CAA8BF,MAA9B,EAAyD;AACvD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAaC,IAAI,CAACC,IAAL,CAAU1B,MAAM,CAAC2B,MAAP,GAAgB,CAA1B,CAAb,CAAZ;AACAL,EAAAA,GAAG,CAACM,IAAJ,CAAS,CAAT;;AACA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC,QAAI7B,MAAM,CAAC6B,CAAD,CAAV,EAAe;AACbP,MAAAA,GAAG,CAACG,IAAI,CAACK,KAAL,CAAWD,CAAC,GAAG,CAAf,CAAD,CAAH,IAA0B,KAAKA,CAAC,GAAG,CAAnC;AACD;AACF;;AACD,SAAOP,GAAP;AACD;;AAED,SAASR,oBAAT,CAA8BF,MAA9B,EAAoDC,KAApD,EAA8E;AAC5E,MAAMb,MAAiB,GAAG,EAA1B;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,QAAME,CAAC,GAAGnB,MAAM,CAACoB,MAAP,CAAcpB,MAAM,CAACqB,MAAP,GAAgBR,IAAI,CAACK,KAAL,CAAWD,CAAC,GAAG,CAAf,CAA9B,CAAV;AACA7B,IAAAA,MAAM,CAACkC,IAAP,CAAY,CAACH,CAAC,GAAI,KAAKF,CAAC,GAAG,CAAf,IAAqB,CAAjC;AACD;;AACDjB,EAAAA,MAAM,CAACqB,MAAP,IAAiBR,IAAI,CAACC,IAAL,CAAUb,KAAK,GAAG,CAAlB,CAAjB;AACA,SAAOb,MAAP;AACD;;AAED,SAASG,kBAAT,CAA4BH,MAA5B,EAAsD;AACpD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACa,YAAJ,CAAiBnC,MAAM,CAAC6B,CAAD,CAAvB,EAA4BA,CAAC,GAAG,CAAhC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASP,kBAAT,CAA4BH,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcI,WAAd,CAA0BxB,MAAM,CAACqB,MAAjC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASI,kBAAT,CAA4BJ,MAA5B,EAAsD;AACpD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCQ,iBAAMC,YAAN,CAAmBtC,MAAM,CAAC6B,CAAD,CAAzB,EAA8BP,GAA9B,EAAmCO,CAAC,GAAG,CAAvC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASN,kBAAT,CAA4BJ,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYG,aAAME,WAAN,CAAkB3B,MAAM,CAACoB,MAAzB,EAAiCpB,MAAM,CAACqB,MAAxC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASK,kBAAT,CAA4BL,MAA5B,EAAsD;AACpD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,KAAKxB,MAAM,CAAC2B,MAAzB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC,QAAI7B,MAAM,CAAC6B,CAAD,CAAN,IAAa,CAAjB,EAAoB;AAClBQ,mBAAMC,YAAN,CAAmBtC,MAAM,CAAC6B,CAAD,CAAzB,EAA8BP,GAA9B,EAAmCO,CAAC,GAAG,EAAvC;;AACAP,MAAAA,GAAG,CAACkB,aAAJ,CAAkB,CAAlB,EAAqBX,CAAC,GAAG,EAAJ,GAAS,CAA9B;AACD,KAHD,MAGO;AACLQ,mBAAMC,YAAN,CAAmB,CAAC,CAACtC,MAAM,CAAC6B,CAAD,CAAR,GAAc,CAAjC,EAAoCP,GAApC,EAAyCO,CAAC,GAAG,EAA7C;;AACAP,MAAAA,GAAG,CAACkB,aAAJ,CAAkB,UAAlB,EAA8BX,CAAC,GAAG,EAAJ,GAAS,CAAvC;AACD;AACF;;AACD,SAAOP,GAAP;AACD;;AAED,SAASL,kBAAT,CAA4BL,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,QAAMY,GAAG,GAAGJ,aAAME,WAAN,CAAkB3B,MAAM,CAACoB,MAAzB,EAAiCpB,MAAM,CAACqB,MAAxC,CAAZ;;AACA,QAAMS,IAAI,GAAG9B,MAAM,CAACoB,MAAP,CAAcW,YAAd,CAA2B/B,MAAM,CAACqB,MAAP,GAAgB,CAA3C,CAAb;;AACA,QAAIS,IAAI,KAAK,UAAb,EAAyB;AACvB1C,MAAAA,MAAM,CAACkC,IAAP,CAAY,CAAC,CAACO,GAAF,GAAQ,CAApB;AACD,KAFD,MAEO;AACLzC,MAAAA,MAAM,CAACkC,IAAP,CAAYO,GAAZ;AACD;;AACD7B,IAAAA,MAAM,CAACqB,MAAP,IAAiB,EAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASM,kBAAT,CAA4BN,MAA5B,EAAsD;AACpD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACsB,YAAJ,CAAiB5C,MAAM,CAAC6B,CAAD,CAAvB,EAA4BA,CAAC,GAAG,CAAhC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASJ,kBAAT,CAA4BN,MAA5B,EAAkDC,KAAlD,EAA2E;AACzE,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAca,WAAd,CAA0BjC,MAAM,CAACqB,MAAjC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASO,mBAAT,CAA6BP,MAA7B,EAAuD;AACrD,MAAMsB,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,IAAIxB,MAAM,CAAC2B,MAAxB,CAAZ;;AACA,OAAK,IAAIE,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACwB,aAAJ,CAAkB9C,MAAM,CAAC6B,CAAD,CAAxB,EAA6BA,CAAC,GAAG,CAAjC;AACD;;AACD,SAAOP,GAAP;AACD;;AAED,SAASH,mBAAT,CAA6BP,MAA7B,EAAmDC,KAAnD,EAA4E;AAC1E,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAce,YAAd,CAA2BnC,MAAM,CAACqB,MAAlC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACD;;AACD,SAAOjC,MAAP;AACD;;AAED,SAASQ,uBAAT,CAAiCR,MAAjC,EAA2D;AAEzD,MAAIgD,OAAO,GAAG,CAAd;;AACA,OAAK,IAAInB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC7B,IAAAA,MAAM,CAAC6B,CAAD,CAAN,GAAYN,MAAM,CAAC0B,IAAP,CAAYjD,MAAM,CAAC6B,CAAD,CAAlB,CAAZ;AACAmB,IAAAA,OAAO,IAAI,IAAIhD,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAAzB;AACD;;AACD,MAAML,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAawB,OAAb,CAAZ;AAEA,MAAIE,OAAO,GAAG,CAAd;;AACA,OAAK,IAAIrB,EAAC,GAAG,CAAb,EAAgBA,EAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,EAAC,EAApC,EAAwC;AACtCP,IAAAA,GAAG,CAACkB,aAAJ,CAAkBxC,MAAM,CAAC6B,EAAD,CAAN,CAAUF,MAA5B,EAAoCuB,OAApC;;AACAlD,IAAAA,MAAM,CAAC6B,EAAD,CAAN,CAAUsB,IAAV,CAAe7B,GAAf,EAAoB4B,OAAO,GAAG,CAA9B;;AACAA,IAAAA,OAAO,IAAI,IAAIlD,MAAM,CAAC6B,EAAD,CAAN,CAAUF,MAAzB;AACD;;AACD,SAAOL,GAAP;AACD;;AAED,SAASF,uBAAT,CAAiCR,MAAjC,EAAuDC,KAAvD,EAAgF;AAC9E,MAAMb,MAAgB,GAAG,EAAzB;;AACA,OAAK,IAAI6B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B,QAAMuB,GAAG,GAAGxC,MAAM,CAACoB,MAAP,CAAcW,YAAd,CAA2B/B,MAAM,CAACqB,MAAlC,CAAZ;AACArB,IAAAA,MAAM,CAACqB,MAAP,IAAiB,CAAjB;AACAjC,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcqB,KAAd,CAAoBzC,MAAM,CAACqB,MAA3B,EAAmCrB,MAAM,CAACqB,MAAP,GAAgBmB,GAAnD,CAAZ;AACAxC,IAAAA,MAAM,CAACqB,MAAP,IAAiBmB,GAAjB;AACD;;AACD,SAAOpD,MAAP;AACD;;AAED,SAASS,iCAAT,CAA2CT,MAA3C,EAA6DC,IAA7D,EAAgG;AAC9F,MAAI,CAACA,IAAI,CAACqD,UAAV,EAAsB;AACpB,UAAM,IAAI5C,KAAJ,CAAU,gEAAV,CAAN;AACD;;AACD,OAAK,IAAImB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG7B,MAAM,CAAC2B,MAA3B,EAAmCE,CAAC,EAApC,EAAwC;AACtC7B,IAAAA,MAAM,CAAC6B,CAAD,CAAN,GAAYN,MAAM,CAAC0B,IAAP,CAAYjD,MAAM,CAAC6B,CAAD,CAAlB,CAAZ;;AACA,QAAI7B,MAAM,CAAC6B,CAAD,CAAN,CAAUF,MAAV,KAAqB1B,IAAI,CAACqD,UAA9B,EAA0C;AACxC,YAAM,IAAI5C,KAAJ,mDAAqDV,MAAM,CAAC6B,CAAD,CAA3D,EAAN;AACD;AACF;;AACD,SAAON,MAAM,CAACgC,MAAP,CAAcvD,MAAd,CAAP;AACD;;AAED,SAASqB,iCAAT,CACET,MADF,EAEEC,KAFF,EAGEZ,IAHF,EAIY;AACV,MAAMD,MAAgB,GAAG,EAAzB;;AACA,MAAI,CAACC,IAAI,CAACqD,UAAV,EAAsB;AACpB,UAAM,IAAI5C,KAAJ,CAAU,gEAAV,CAAN;AACD;;AACD,OAAK,IAAImB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGhB,KAApB,EAA2BgB,CAAC,EAA5B,EAAgC;AAC9B7B,IAAAA,MAAM,CAACkC,IAAP,CAAYtB,MAAM,CAACoB,MAAP,CAAcqB,KAAd,CAAoBzC,MAAM,CAACqB,MAA3B,EAAmCrB,MAAM,CAACqB,MAAP,GAAgBhC,IAAI,CAACqD,UAAxD,CAAZ;AACA1C,IAAAA,MAAM,CAACqB,MAAP,IAAiBhC,IAAI,CAACqD,UAAtB;AACD;;AACD,SAAOtD,MAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n/* eslint-disable camelcase */\nimport type {PrimitiveType} from '../schema/declare';\nimport type {CursorBuffer, ParquetCodecOptions} from './declare';\nimport INT53 from 'int53';\n\nexport function encodeValues(\n type: PrimitiveType,\n values: any[],\n opts: ParquetCodecOptions\n): Buffer {\n switch (type) {\n case 'BOOLEAN':\n return encodeValues_BOOLEAN(values);\n case 'INT32':\n return encodeValues_INT32(values);\n case 'INT64':\n return encodeValues_INT64(values);\n case 'INT96':\n return encodeValues_INT96(values);\n case 'FLOAT':\n return encodeValues_FLOAT(values);\n case 'DOUBLE':\n return encodeValues_DOUBLE(values);\n case 'BYTE_ARRAY':\n return encodeValues_BYTE_ARRAY(values);\n case 'FIXED_LEN_BYTE_ARRAY':\n return encodeValues_FIXED_LEN_BYTE_ARRAY(values, opts);\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n}\n\nexport function decodeValues(\n type: PrimitiveType,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): any[] {\n switch (type) {\n case 'BOOLEAN':\n return decodeValues_BOOLEAN(cursor, count);\n case 'INT32':\n return decodeValues_INT32(cursor, count);\n case 'INT64':\n return decodeValues_INT64(cursor, count);\n case 'INT96':\n return decodeValues_INT96(cursor, count);\n case 'FLOAT':\n return decodeValues_FLOAT(cursor, count);\n case 'DOUBLE':\n return decodeValues_DOUBLE(cursor, count);\n case 'BYTE_ARRAY':\n return decodeValues_BYTE_ARRAY(cursor, count);\n case 'FIXED_LEN_BYTE_ARRAY':\n return decodeValues_FIXED_LEN_BYTE_ARRAY(cursor, count, opts);\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n}\n\nfunction encodeValues_BOOLEAN(values: boolean[]): Buffer {\n const buf = Buffer.alloc(Math.ceil(values.length / 8));\n buf.fill(0);\n for (let i = 0; i < values.length; i++) {\n if (values[i]) {\n buf[Math.floor(i / 8)] |= 1 << i % 8;\n }\n }\n return buf;\n}\n\nfunction decodeValues_BOOLEAN(cursor: CursorBuffer, count: number): boolean[] {\n const values: boolean[] = [];\n for (let i = 0; i < count; i++) {\n const b = cursor.buffer[cursor.offset + Math.floor(i / 8)];\n values.push((b & (1 << i % 8)) > 0);\n }\n cursor.offset += Math.ceil(count / 8);\n return values;\n}\n\nfunction encodeValues_INT32(values: number[]): Buffer {\n const buf = Buffer.alloc(4 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeInt32LE(values[i], i * 4);\n }\n return buf;\n}\n\nfunction decodeValues_INT32(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readInt32LE(cursor.offset));\n cursor.offset += 4;\n }\n return values;\n}\n\nfunction encodeValues_INT64(values: number[]): Buffer {\n const buf = Buffer.alloc(8 * values.length);\n for (let i = 0; i < values.length; i++) {\n INT53.writeInt64LE(values[i], buf, i * 8);\n }\n return buf;\n}\n\nfunction decodeValues_INT64(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(INT53.readInt64LE(cursor.buffer, cursor.offset));\n cursor.offset += 8;\n }\n return values;\n}\n\nfunction encodeValues_INT96(values: number[]): Buffer {\n const buf = Buffer.alloc(12 * values.length);\n for (let i = 0; i < values.length; i++) {\n if (values[i] >= 0) {\n INT53.writeInt64LE(values[i], buf, i * 12);\n buf.writeUInt32LE(0, i * 12 + 8); // truncate to 64 actual precision\n } else {\n INT53.writeInt64LE(~-values[i] + 1, buf, i * 12);\n buf.writeUInt32LE(0xffffffff, i * 12 + 8); // truncate to 64 actual precision\n }\n }\n return buf;\n}\n\nfunction decodeValues_INT96(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n const low = INT53.readInt64LE(cursor.buffer, cursor.offset);\n const high = cursor.buffer.readUInt32LE(cursor.offset + 8);\n if (high === 0xffffffff) {\n values.push(~-low + 1); // truncate to 64 actual precision\n } else {\n values.push(low); // truncate to 64 actual precision\n }\n cursor.offset += 12;\n }\n return values;\n}\n\nfunction encodeValues_FLOAT(values: number[]): Buffer {\n const buf = Buffer.alloc(4 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeFloatLE(values[i], i * 4);\n }\n return buf;\n}\n\nfunction decodeValues_FLOAT(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readFloatLE(cursor.offset));\n cursor.offset += 4;\n }\n return values;\n}\n\nfunction encodeValues_DOUBLE(values: number[]): Buffer {\n const buf = Buffer.alloc(8 * values.length);\n for (let i = 0; i < values.length; i++) {\n buf.writeDoubleLE(values[i], i * 8);\n }\n return buf;\n}\n\nfunction decodeValues_DOUBLE(cursor: CursorBuffer, count: number): number[] {\n const values: number[] = [];\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.readDoubleLE(cursor.offset));\n cursor.offset += 8;\n }\n return values;\n}\n\nfunction encodeValues_BYTE_ARRAY(values: Buffer[]): Buffer {\n // tslint:disable-next-line:variable-name\n let buf_len = 0;\n for (let i = 0; i < values.length; i++) {\n values[i] = Buffer.from(values[i]);\n buf_len += 4 + values[i].length;\n }\n const buf = Buffer.alloc(buf_len);\n // tslint:disable-next-line:variable-name\n let buf_pos = 0;\n for (let i = 0; i < values.length; i++) {\n buf.writeUInt32LE(values[i].length, buf_pos);\n values[i].copy(buf, buf_pos + 4);\n buf_pos += 4 + values[i].length;\n }\n return buf;\n}\n\nfunction decodeValues_BYTE_ARRAY(cursor: CursorBuffer, count: number): Buffer[] {\n const values: Buffer[] = [];\n for (let i = 0; i < count; i++) {\n const len = cursor.buffer.readUInt32LE(cursor.offset);\n cursor.offset += 4;\n values.push(cursor.buffer.slice(cursor.offset, cursor.offset + len));\n cursor.offset += len;\n }\n return values;\n}\n\nfunction encodeValues_FIXED_LEN_BYTE_ARRAY(values: Buffer[], opts: ParquetCodecOptions): Buffer {\n if (!opts.typeLength) {\n throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');\n }\n for (let i = 0; i < values.length; i++) {\n values[i] = Buffer.from(values[i]);\n if (values[i].length !== opts.typeLength) {\n throw new Error(`invalid value for FIXED_LEN_BYTE_ARRAY: ${values[i]}`);\n }\n }\n return Buffer.concat(values);\n}\n\nfunction decodeValues_FIXED_LEN_BYTE_ARRAY(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): Buffer[] {\n const values: Buffer[] = [];\n if (!opts.typeLength) {\n throw new Error('missing option: typeLength (required for FIXED_LEN_BYTE_ARRAY)');\n }\n for (let i = 0; i < count; i++) {\n values.push(cursor.buffer.slice(cursor.offset, cursor.offset + opts.typeLength));\n cursor.offset += opts.typeLength;\n }\n return values;\n}\n"],"file":"plain.js"}
|
|
@@ -8,6 +8,8 @@ Object.defineProperty(exports, "__esModule", {
|
|
|
8
8
|
exports.encodeValues = encodeValues;
|
|
9
9
|
exports.decodeValues = decodeValues;
|
|
10
10
|
|
|
11
|
+
var _toConsumableArray2 = _interopRequireDefault(require("@babel/runtime/helpers/toConsumableArray"));
|
|
12
|
+
|
|
11
13
|
var _varint = _interopRequireDefault(require("varint"));
|
|
12
14
|
|
|
13
15
|
function encodeValues(type, values, opts) {
|
|
@@ -19,18 +21,20 @@ function encodeValues(type, values, opts) {
|
|
|
19
21
|
case 'BOOLEAN':
|
|
20
22
|
case 'INT32':
|
|
21
23
|
case 'INT64':
|
|
22
|
-
values = values.map(
|
|
24
|
+
values = values.map(function (x) {
|
|
25
|
+
return parseInt(x, 10);
|
|
26
|
+
});
|
|
23
27
|
break;
|
|
24
28
|
|
|
25
29
|
default:
|
|
26
30
|
throw new Error("unsupported type: ".concat(type));
|
|
27
31
|
}
|
|
28
32
|
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
33
|
+
var buf = Buffer.alloc(0);
|
|
34
|
+
var run = [];
|
|
35
|
+
var repeats = 0;
|
|
32
36
|
|
|
33
|
-
for (
|
|
37
|
+
for (var i = 0; i < values.length; i++) {
|
|
34
38
|
if (repeats === 0 && run.length % 8 === 0 && values[i] === values[i + 1]) {
|
|
35
39
|
if (run.length) {
|
|
36
40
|
buf = Buffer.concat([buf, encodeRunBitpacked(run, opts)]);
|
|
@@ -60,7 +64,7 @@ function encodeValues(type, values, opts) {
|
|
|
60
64
|
return buf;
|
|
61
65
|
}
|
|
62
66
|
|
|
63
|
-
|
|
67
|
+
var envelope = Buffer.alloc(buf.length + 4);
|
|
64
68
|
envelope.writeUInt32LE(buf.length, undefined);
|
|
65
69
|
buf.copy(envelope, 4);
|
|
66
70
|
return envelope;
|
|
@@ -75,19 +79,25 @@ function decodeValues(type, cursor, count, opts) {
|
|
|
75
79
|
cursor.offset += 4;
|
|
76
80
|
}
|
|
77
81
|
|
|
78
|
-
|
|
82
|
+
var values = [];
|
|
79
83
|
|
|
80
84
|
while (values.length < count) {
|
|
81
|
-
|
|
85
|
+
var header = _varint.default.decode(cursor.buffer, cursor.offset);
|
|
82
86
|
|
|
83
87
|
cursor.offset += _varint.default.encodingLength(header);
|
|
84
88
|
|
|
85
89
|
if (header & 1) {
|
|
86
|
-
|
|
87
|
-
|
|
90
|
+
var _values;
|
|
91
|
+
|
|
92
|
+
var _count = (header >> 1) * 8;
|
|
93
|
+
|
|
94
|
+
(_values = values).push.apply(_values, (0, _toConsumableArray2.default)(decodeRunBitpacked(cursor, _count, opts)));
|
|
88
95
|
} else {
|
|
89
|
-
|
|
90
|
-
|
|
96
|
+
var _values2;
|
|
97
|
+
|
|
98
|
+
var _count2 = header >> 1;
|
|
99
|
+
|
|
100
|
+
(_values2 = values).push.apply(_values2, (0, _toConsumableArray2.default)(decodeRunRepeated(cursor, _count2, opts)));
|
|
91
101
|
}
|
|
92
102
|
}
|
|
93
103
|
|
|
@@ -101,15 +111,15 @@ function decodeValues(type, cursor, count, opts) {
|
|
|
101
111
|
}
|
|
102
112
|
|
|
103
113
|
function decodeRunBitpacked(cursor, count, opts) {
|
|
104
|
-
|
|
114
|
+
var bitWidth = opts.bitWidth;
|
|
105
115
|
|
|
106
116
|
if (count % 8 !== 0) {
|
|
107
117
|
throw new Error('must be a multiple of 8');
|
|
108
118
|
}
|
|
109
119
|
|
|
110
|
-
|
|
120
|
+
var values = new Array(count).fill(0);
|
|
111
121
|
|
|
112
|
-
for (
|
|
122
|
+
for (var b = 0; b < bitWidth * count; b++) {
|
|
113
123
|
if (cursor.buffer[cursor.offset + Math.floor(b / 8)] & 1 << b % 8) {
|
|
114
124
|
values[Math.floor(b / bitWidth)] |= 1 << b % bitWidth;
|
|
115
125
|
}
|
|
@@ -120,10 +130,10 @@ function decodeRunBitpacked(cursor, count, opts) {
|
|
|
120
130
|
}
|
|
121
131
|
|
|
122
132
|
function decodeRunRepeated(cursor, count, opts) {
|
|
123
|
-
|
|
124
|
-
|
|
133
|
+
var bitWidth = opts.bitWidth;
|
|
134
|
+
var value = 0;
|
|
125
135
|
|
|
126
|
-
for (
|
|
136
|
+
for (var i = 0; i < Math.ceil(bitWidth / 8); i++) {
|
|
127
137
|
value << 8;
|
|
128
138
|
value += cursor.buffer[cursor.offset];
|
|
129
139
|
cursor.offset += 1;
|
|
@@ -133,15 +143,15 @@ function decodeRunRepeated(cursor, count, opts) {
|
|
|
133
143
|
}
|
|
134
144
|
|
|
135
145
|
function encodeRunBitpacked(values, opts) {
|
|
136
|
-
|
|
146
|
+
var bitWidth = opts.bitWidth;
|
|
137
147
|
|
|
138
|
-
for (
|
|
148
|
+
for (var i = 0; i < values.length % 8; i++) {
|
|
139
149
|
values.push(0);
|
|
140
150
|
}
|
|
141
151
|
|
|
142
|
-
|
|
152
|
+
var buf = Buffer.alloc(Math.ceil(bitWidth * (values.length / 8)));
|
|
143
153
|
|
|
144
|
-
for (
|
|
154
|
+
for (var b = 0; b < bitWidth * values.length; b++) {
|
|
145
155
|
if ((values[Math.floor(b / bitWidth)] & 1 << b % bitWidth) > 0) {
|
|
146
156
|
buf[Math.floor(b / 8)] |= 1 << b % 8;
|
|
147
157
|
}
|
|
@@ -151,10 +161,10 @@ function encodeRunBitpacked(values, opts) {
|
|
|
151
161
|
}
|
|
152
162
|
|
|
153
163
|
function encodeRunRepeated(value, count, opts) {
|
|
154
|
-
|
|
155
|
-
|
|
164
|
+
var bitWidth = opts.bitWidth;
|
|
165
|
+
var buf = Buffer.alloc(Math.ceil(bitWidth / 8));
|
|
156
166
|
|
|
157
|
-
for (
|
|
167
|
+
for (var i = 0; i < buf.length; i++) {
|
|
158
168
|
buf.writeUInt8(value & 0xff, i);
|
|
159
169
|
value >> 8;
|
|
160
170
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["../../../../src/parquetjs/codecs/rle.ts"],"names":["encodeValues","type","values","opts","Error","map","x","parseInt","buf","Buffer","alloc","run","repeats","i","length","concat","encodeRunBitpacked","encodeRunRepeated","push","disableEnvelope","envelope","writeUInt32LE","undefined","copy","decodeValues","cursor","count","offset","header","varint","decode","buffer","encodingLength","decodeRunBitpacked","decodeRunRepeated","slice","bitWidth","Array","fill","b","Math","floor","value","ceil","from","encode","writeUInt8"],"mappings":";;;;;;;;;;AAIA;;AAGO,SAASA,YAAT,CACLC,IADK,EAELC,MAFK,EAGLC,IAHK,EAIG;AACR,MAAI,EAAE,cAAcA,IAAhB,CAAJ,EAA2B;AACzB,UAAM,IAAIC,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,UAAQH,IAAR;AACE,SAAK,SAAL;AACA,SAAK,OAAL;AACA,SAAK,OAAL;AAEEC,MAAAA,MAAM,GAAGA,MAAM,CAACG,GAAP,CAAYC,CAAD,IAAOC,QAAQ,CAACD,CAAD,EAAI,EAAJ,CAA1B,CAAT;AACA;;AAEF;AACE,YAAM,IAAIF,KAAJ,6BAA+BH,IAA/B,EAAN;AATJ;;AAYA,MAAIO,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,CAAb,CAAV;AACA,MAAIC,GAAU,GAAG,EAAjB;AACA,MAAIC,OAAO,GAAG,CAAd;;AAEA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGX,MAAM,CAACY,MAA3B,EAAmCD,CAAC,EAApC,EAAwC;AAGtC,QAAID,OAAO,KAAK,CAAZ,IAAiBD,GAAG,CAACG,MAAJ,GAAa,CAAb,KAAmB,CAApC,IAAyCZ,MAAM,CAACW,CAAD,CAAN,KAAcX,MAAM,CAACW,CAAC,GAAG,CAAL,CAAjE,EAA0E;AAExE,UAAIF,GAAG,CAACG,MAAR,EAAgB;AACdN,QAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMQ,kBAAkB,CAACL,GAAD,EAAMR,IAAN,CAAxB,CAAd,CAAN;AACAQ,QAAAA,GAAG,GAAG,EAAN;AACD;;AACDC,MAAAA,OAAO,GAAG,CAAV;AACD,KAPD,MAOO,IAAIA,OAAO,GAAG,CAAV,IAAeV,MAAM,CAACW,CAAD,CAAN,KAAcX,MAAM,CAACW,CAAC,GAAG,CAAL,CAAvC,EAAgD;AACrDD,MAAAA,OAAO,IAAI,CAAX;AACD,KAFM,MAEA;AAEL,UAAIA,OAAJ,EAAa;AACXJ,QAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMS,iBAAiB,CAACf,MAAM,CAACW,CAAC,GAAG,CAAL,CAAP,EAAgBD,OAAhB,EAAyBT,IAAzB,CAAvB,CAAd,CAAN;AACAS,QAAAA,OAAO,GAAG,CAAV;AACD;;AACDD,MAAAA,GAAG,CAACO,IAAJ,CAAShB,MAAM,CAACW,CAAD,CAAf;AACD;AACF;;AAED,MAAID,OAAJ,EAAa;AACXJ,IAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMS,iBAAiB,CAACf,MAAM,CAACA,MAAM,CAACY,MAAP,GAAgB,CAAjB,CAAP,EAA4BF,OAA5B,EAAqCT,IAArC,CAAvB,CAAd,CAAN;AACD,GAFD,MAEO,IAAIQ,GAAG,CAACG,MAAR,EAAgB;AACrBN,IAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMQ,kBAAkB,CAACL,GAAD,EAAMR,IAAN,CAAxB,CAAd,CAAN;AACD;;AAED,MAAIA,IAAI,CAACgB,eAAT,EAA0B;AACxB,WAAOX,GAAP;AACD;;AAED,QAAMY,QAAQ,GAAGX,MAAM,CAACC,KAAP,CAAaF,GAAG,CAACM,MAAJ,GAAa,CAA1B,CAAjB;AACAM,EAAAA,QAAQ,CAACC,aAAT,CAAuBb,GAAG,CAACM,MAA3B,EAAmCQ,SAAnC;AACAd,EAAAA,GAAG,CAACe,IAAJ,CAASH,QAAT,EAAmB,CAAnB;AAEA,SAAOA,QAAP;AACD;;AAEM,SAASI,YAAT,CACLvB,IADK,EAELwB,MAFK,EAGLC,KAHK,EAILvB,IAJK,EAKK;AACV,MAAI,EAAE,cAAcA,IAAhB,CAAJ,EAA2B;AACzB,UAAM,IAAIC,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,MAAI,CAACD,IAAI,CAACgB,eAAV,EAA2B;AACzBM,IAAAA,MAAM,CAACE,MAAP,IAAiB,CAAjB;AACD;;AAED,MAAIzB,MAAgB,GAAG,EAAvB;;AACA,SAAOA,MAAM,CAACY,MAAP,GAAgBY,KAAvB,EAA8B;AAC5B,UAAME,MAAM,GAAGC,gBAAOC,MAAP,CAAcL,MAAM,CAACM,MAArB,EAA6BN,MAAM,CAACE,MAApC,CAAf;;AACAF,IAAAA,MAAM,CAACE,MAAP,IAAiBE,gBAAOG,cAAP,CAAsBJ,MAAtB,CAAjB;;AACA,QAAIA,MAAM,GAAG,CAAb,EAAgB;AACd,YAAMF,KAAK,GAAG,CAACE,MAAM,IAAI,CAAX,IAAgB,CAA9B;AACA1B,MAAAA,MAAM,CAACgB,IAAP,CAAY,GAAGe,kBAAkB,CAACR,MAAD,EAASC,KAAT,EAAgBvB,IAAhB,CAAjC;AACD,KAHD,MAGO;AACL,YAAMuB,KAAK,GAAGE,MAAM,IAAI,CAAxB;AACA1B,MAAAA,MAAM,CAACgB,IAAP,CAAY,GAAGgB,iBAAiB,CAACT,MAAD,EAASC,KAAT,EAAgBvB,IAAhB,CAAhC;AACD;AACF;;AACDD,EAAAA,MAAM,GAAGA,MAAM,CAACiC,KAAP,CAAa,CAAb,EAAgBT,KAAhB,CAAT;;AAEA,MAAIxB,MAAM,CAACY,MAAP,KAAkBY,KAAtB,EAA6B;AAC3B,UAAM,IAAItB,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,SAAOF,MAAP;AACD;;AAED,SAAS+B,kBAAT,CACER,MADF,EAEEC,KAFF,EAGEvB,IAHF,EAIY;AAEV,QAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;;AAEA,MAAIV,KAAK,GAAG,CAAR,KAAc,CAAlB,EAAqB;AACnB,UAAM,IAAItB,KAAJ,CAAU,yBAAV,CAAN;AACD;;AAGD,QAAMF,MAAM,GAAG,IAAImC,KAAJ,CAAUX,KAAV,EAAiBY,IAAjB,CAAsB,CAAtB,CAAf;;AACA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,QAAQ,GAAGV,KAA/B,EAAsCa,CAAC,EAAvC,EAA2C;AACzC,QAAId,MAAM,CAACM,MAAP,CAAcN,MAAM,CAACE,MAAP,GAAgBa,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAG,CAAf,CAA9B,IAAoD,KAAKA,CAAC,GAAG,CAAjE,EAAqE;AACnErC,MAAAA,MAAM,CAACsC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAGH,QAAf,CAAD,CAAN,IAAoC,KAAKG,CAAC,GAAGH,QAA7C;AACD;AACF;;AAEDX,EAAAA,MAAM,CAACE,MAAP,IAAiBS,QAAQ,IAAIV,KAAK,GAAG,CAAZ,CAAzB;AACA,SAAOxB,MAAP;AACD;;AAED,SAASgC,iBAAT,CACET,MADF,EAEEC,KAFF,EAGEvB,IAHF,EAIY;AAEV,QAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;AAEA,MAAIM,KAAK,GAAG,CAAZ;;AACA,OAAK,IAAI7B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG2B,IAAI,CAACG,IAAL,CAAUP,QAAQ,GAAG,CAArB,CAApB,EAA6CvB,CAAC,EAA9C,EAAkD;AAEhD6B,IAAAA,KAAK,IAAI,CAAT;AACAA,IAAAA,KAAK,IAAIjB,MAAM,CAACM,MAAP,CAAcN,MAAM,CAACE,MAArB,CAAT;AACAF,IAAAA,MAAM,CAACE,MAAP,IAAiB,CAAjB;AACD;;AAGD,SAAO,IAAIU,KAAJ,CAAUX,KAAV,EAAiBY,IAAjB,CAAsBI,KAAtB,CAAP;AACD;;AAED,SAAS1B,kBAAT,CAA4Bd,MAA5B,EAA8CC,IAA9C,EAAiF;AAE/E,QAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;;AAEA,OAAK,IAAIvB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGX,MAAM,CAACY,MAAP,GAAgB,CAApC,EAAuCD,CAAC,EAAxC,EAA4C;AAC1CX,IAAAA,MAAM,CAACgB,IAAP,CAAY,CAAZ;AACD;;AAED,QAAMV,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa8B,IAAI,CAACG,IAAL,CAAUP,QAAQ,IAAIlC,MAAM,CAACY,MAAP,GAAgB,CAApB,CAAlB,CAAb,CAAZ;;AACA,OAAK,IAAIyB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,QAAQ,GAAGlC,MAAM,CAACY,MAAtC,EAA8CyB,CAAC,EAA/C,EAAmD;AACjD,QAAI,CAACrC,MAAM,CAACsC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAGH,QAAf,CAAD,CAAN,GAAoC,KAAKG,CAAC,GAAGH,QAA9C,IAA2D,CAA/D,EAAkE;AAChE5B,MAAAA,GAAG,CAACgC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAG,CAAf,CAAD,CAAH,IAA0B,KAAKA,CAAC,GAAG,CAAnC;AACD;AACF;;AAED,SAAO9B,MAAM,CAACM,MAAP,CAAc,CAACN,MAAM,CAACmC,IAAP,CAAYf,gBAAOgB,MAAP,CAAgB3C,MAAM,CAACY,MAAP,GAAgB,CAAjB,IAAuB,CAAxB,GAA6B,CAA3C,CAAZ,CAAD,EAA6DN,GAA7D,CAAd,CAAP;AACD;;AAED,SAASS,iBAAT,CAA2ByB,KAA3B,EAA0ChB,KAA1C,EAAyDvB,IAAzD,EAA4F;AAE1F,QAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;AAEA,QAAM5B,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa8B,IAAI,CAACG,IAAL,CAAUP,QAAQ,GAAG,CAArB,CAAb,CAAZ;;AAEA,OAAK,IAAIvB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGL,GAAG,CAACM,MAAxB,EAAgCD,CAAC,EAAjC,EAAqC;AACnCL,IAAAA,GAAG,CAACsC,UAAJ,CAAeJ,KAAK,GAAG,IAAvB,EAA6B7B,CAA7B;AAEA6B,IAAAA,KAAK,IAAI,CAAT;AACD;;AAED,SAAOjC,MAAM,CAACM,MAAP,CAAc,CAACN,MAAM,CAACmC,IAAP,CAAYf,gBAAOgB,MAAP,CAAcnB,KAAK,IAAI,CAAvB,CAAZ,CAAD,EAAyClB,GAAzC,CAAd,CAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport type {PrimitiveType} from '../schema/declare';\nimport type {CursorBuffer, ParquetCodecOptions} from './declare';\nimport varint from 'varint';\n\n// eslint-disable-next-line max-statements, complexity\nexport function encodeValues(\n type: PrimitiveType,\n values: any[],\n opts: ParquetCodecOptions\n): Buffer {\n if (!('bitWidth' in opts)) {\n throw new Error('bitWidth is required');\n }\n\n switch (type) {\n case 'BOOLEAN':\n case 'INT32':\n case 'INT64':\n // tslint:disable-next-line:no-parameter-reassignment\n values = values.map((x) => parseInt(x, 10));\n break;\n\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n\n let buf = Buffer.alloc(0);\n let run: any[] = [];\n let repeats = 0;\n\n for (let i = 0; i < values.length; i++) {\n // If we are at the beginning of a run and the next value is same we start\n // collecting repeated values\n if (repeats === 0 && run.length % 8 === 0 && values[i] === values[i + 1]) {\n // If we have any data in runs we need to encode them\n if (run.length) {\n buf = Buffer.concat([buf, encodeRunBitpacked(run, opts)]);\n run = [];\n }\n repeats = 1;\n } else if (repeats > 0 && values[i] === values[i - 1]) {\n repeats += 1;\n } else {\n // If values changes we need to post any previous repeated values\n if (repeats) {\n buf = Buffer.concat([buf, encodeRunRepeated(values[i - 1], repeats, opts)]);\n repeats = 0;\n }\n run.push(values[i]);\n }\n }\n\n if (repeats) {\n buf = Buffer.concat([buf, encodeRunRepeated(values[values.length - 1], repeats, opts)]);\n } else if (run.length) {\n buf = Buffer.concat([buf, encodeRunBitpacked(run, opts)]);\n }\n\n if (opts.disableEnvelope) {\n return buf;\n }\n\n const envelope = Buffer.alloc(buf.length + 4);\n envelope.writeUInt32LE(buf.length, undefined);\n buf.copy(envelope, 4);\n\n return envelope;\n}\n\nexport function decodeValues(\n type: PrimitiveType,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n if (!('bitWidth' in opts)) {\n throw new Error('bitWidth is required');\n }\n\n if (!opts.disableEnvelope) {\n cursor.offset += 4;\n }\n\n let values: number[] = [];\n while (values.length < count) {\n const header = varint.decode(cursor.buffer, cursor.offset);\n cursor.offset += varint.encodingLength(header);\n if (header & 1) {\n const count = (header >> 1) * 8;\n values.push(...decodeRunBitpacked(cursor, count, opts));\n } else {\n const count = header >> 1;\n values.push(...decodeRunRepeated(cursor, count, opts));\n }\n }\n values = values.slice(0, count);\n\n if (values.length !== count) {\n throw new Error('invalid RLE encoding');\n }\n\n return values;\n}\n\nfunction decodeRunBitpacked(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n if (count % 8 !== 0) {\n throw new Error('must be a multiple of 8');\n }\n\n // tslint:disable-next-line:prefer-array-literal\n const values = new Array(count).fill(0);\n for (let b = 0; b < bitWidth * count; b++) {\n if (cursor.buffer[cursor.offset + Math.floor(b / 8)] & (1 << b % 8)) {\n values[Math.floor(b / bitWidth)] |= 1 << b % bitWidth;\n }\n }\n\n cursor.offset += bitWidth * (count / 8);\n return values;\n}\n\nfunction decodeRunRepeated(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n let value = 0;\n for (let i = 0; i < Math.ceil(bitWidth / 8); i++) {\n // eslint-disable-next-line\n value << 8; // TODO - this looks wrong\n value += cursor.buffer[cursor.offset];\n cursor.offset += 1;\n }\n\n // tslint:disable-next-line:prefer-array-literal\n return new Array(count).fill(value);\n}\n\nfunction encodeRunBitpacked(values: number[], opts: ParquetCodecOptions): Buffer {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n for (let i = 0; i < values.length % 8; i++) {\n values.push(0);\n }\n\n const buf = Buffer.alloc(Math.ceil(bitWidth * (values.length / 8)));\n for (let b = 0; b < bitWidth * values.length; b++) {\n if ((values[Math.floor(b / bitWidth)] & (1 << b % bitWidth)) > 0) {\n buf[Math.floor(b / 8)] |= 1 << b % 8;\n }\n }\n\n return Buffer.concat([Buffer.from(varint.encode(((values.length / 8) << 1) | 1)), buf]);\n}\n\nfunction encodeRunRepeated(value: number, count: number, opts: ParquetCodecOptions): Buffer {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n const buf = Buffer.alloc(Math.ceil(bitWidth / 8));\n\n for (let i = 0; i < buf.length; i++) {\n buf.writeUInt8(value & 0xff, i);\n // eslint-disable-next-line\n value >> 8; // TODO - this looks wrong\n }\n\n return Buffer.concat([Buffer.from(varint.encode(count << 1)), buf]);\n}\n"],"file":"rle.js"}
|
|
1
|
+
{"version":3,"sources":["../../../../src/parquetjs/codecs/rle.ts"],"names":["encodeValues","type","values","opts","Error","map","x","parseInt","buf","Buffer","alloc","run","repeats","i","length","concat","encodeRunBitpacked","encodeRunRepeated","push","disableEnvelope","envelope","writeUInt32LE","undefined","copy","decodeValues","cursor","count","offset","header","varint","decode","buffer","encodingLength","decodeRunBitpacked","decodeRunRepeated","slice","bitWidth","Array","fill","b","Math","floor","value","ceil","from","encode","writeUInt8"],"mappings":";;;;;;;;;;;;AAIA;;AAGO,SAASA,YAAT,CACLC,IADK,EAELC,MAFK,EAGLC,IAHK,EAIG;AACR,MAAI,EAAE,cAAcA,IAAhB,CAAJ,EAA2B;AACzB,UAAM,IAAIC,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,UAAQH,IAAR;AACE,SAAK,SAAL;AACA,SAAK,OAAL;AACA,SAAK,OAAL;AAEEC,MAAAA,MAAM,GAAGA,MAAM,CAACG,GAAP,CAAW,UAACC,CAAD;AAAA,eAAOC,QAAQ,CAACD,CAAD,EAAI,EAAJ,CAAf;AAAA,OAAX,CAAT;AACA;;AAEF;AACE,YAAM,IAAIF,KAAJ,6BAA+BH,IAA/B,EAAN;AATJ;;AAYA,MAAIO,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa,CAAb,CAAV;AACA,MAAIC,GAAU,GAAG,EAAjB;AACA,MAAIC,OAAO,GAAG,CAAd;;AAEA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGX,MAAM,CAACY,MAA3B,EAAmCD,CAAC,EAApC,EAAwC;AAGtC,QAAID,OAAO,KAAK,CAAZ,IAAiBD,GAAG,CAACG,MAAJ,GAAa,CAAb,KAAmB,CAApC,IAAyCZ,MAAM,CAACW,CAAD,CAAN,KAAcX,MAAM,CAACW,CAAC,GAAG,CAAL,CAAjE,EAA0E;AAExE,UAAIF,GAAG,CAACG,MAAR,EAAgB;AACdN,QAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMQ,kBAAkB,CAACL,GAAD,EAAMR,IAAN,CAAxB,CAAd,CAAN;AACAQ,QAAAA,GAAG,GAAG,EAAN;AACD;;AACDC,MAAAA,OAAO,GAAG,CAAV;AACD,KAPD,MAOO,IAAIA,OAAO,GAAG,CAAV,IAAeV,MAAM,CAACW,CAAD,CAAN,KAAcX,MAAM,CAACW,CAAC,GAAG,CAAL,CAAvC,EAAgD;AACrDD,MAAAA,OAAO,IAAI,CAAX;AACD,KAFM,MAEA;AAEL,UAAIA,OAAJ,EAAa;AACXJ,QAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMS,iBAAiB,CAACf,MAAM,CAACW,CAAC,GAAG,CAAL,CAAP,EAAgBD,OAAhB,EAAyBT,IAAzB,CAAvB,CAAd,CAAN;AACAS,QAAAA,OAAO,GAAG,CAAV;AACD;;AACDD,MAAAA,GAAG,CAACO,IAAJ,CAAShB,MAAM,CAACW,CAAD,CAAf;AACD;AACF;;AAED,MAAID,OAAJ,EAAa;AACXJ,IAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMS,iBAAiB,CAACf,MAAM,CAACA,MAAM,CAACY,MAAP,GAAgB,CAAjB,CAAP,EAA4BF,OAA5B,EAAqCT,IAArC,CAAvB,CAAd,CAAN;AACD,GAFD,MAEO,IAAIQ,GAAG,CAACG,MAAR,EAAgB;AACrBN,IAAAA,GAAG,GAAGC,MAAM,CAACM,MAAP,CAAc,CAACP,GAAD,EAAMQ,kBAAkB,CAACL,GAAD,EAAMR,IAAN,CAAxB,CAAd,CAAN;AACD;;AAED,MAAIA,IAAI,CAACgB,eAAT,EAA0B;AACxB,WAAOX,GAAP;AACD;;AAED,MAAMY,QAAQ,GAAGX,MAAM,CAACC,KAAP,CAAaF,GAAG,CAACM,MAAJ,GAAa,CAA1B,CAAjB;AACAM,EAAAA,QAAQ,CAACC,aAAT,CAAuBb,GAAG,CAACM,MAA3B,EAAmCQ,SAAnC;AACAd,EAAAA,GAAG,CAACe,IAAJ,CAASH,QAAT,EAAmB,CAAnB;AAEA,SAAOA,QAAP;AACD;;AAEM,SAASI,YAAT,CACLvB,IADK,EAELwB,MAFK,EAGLC,KAHK,EAILvB,IAJK,EAKK;AACV,MAAI,EAAE,cAAcA,IAAhB,CAAJ,EAA2B;AACzB,UAAM,IAAIC,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,MAAI,CAACD,IAAI,CAACgB,eAAV,EAA2B;AACzBM,IAAAA,MAAM,CAACE,MAAP,IAAiB,CAAjB;AACD;;AAED,MAAIzB,MAAgB,GAAG,EAAvB;;AACA,SAAOA,MAAM,CAACY,MAAP,GAAgBY,KAAvB,EAA8B;AAC5B,QAAME,MAAM,GAAGC,gBAAOC,MAAP,CAAcL,MAAM,CAACM,MAArB,EAA6BN,MAAM,CAACE,MAApC,CAAf;;AACAF,IAAAA,MAAM,CAACE,MAAP,IAAiBE,gBAAOG,cAAP,CAAsBJ,MAAtB,CAAjB;;AACA,QAAIA,MAAM,GAAG,CAAb,EAAgB;AAAA;;AACd,UAAMF,MAAK,GAAG,CAACE,MAAM,IAAI,CAAX,IAAgB,CAA9B;;AACA,iBAAA1B,MAAM,EAACgB,IAAP,iDAAee,kBAAkB,CAACR,MAAD,EAASC,MAAT,EAAgBvB,IAAhB,CAAjC;AACD,KAHD,MAGO;AAAA;;AACL,UAAMuB,OAAK,GAAGE,MAAM,IAAI,CAAxB;;AACA,kBAAA1B,MAAM,EAACgB,IAAP,kDAAegB,iBAAiB,CAACT,MAAD,EAASC,OAAT,EAAgBvB,IAAhB,CAAhC;AACD;AACF;;AACDD,EAAAA,MAAM,GAAGA,MAAM,CAACiC,KAAP,CAAa,CAAb,EAAgBT,KAAhB,CAAT;;AAEA,MAAIxB,MAAM,CAACY,MAAP,KAAkBY,KAAtB,EAA6B;AAC3B,UAAM,IAAItB,KAAJ,CAAU,sBAAV,CAAN;AACD;;AAED,SAAOF,MAAP;AACD;;AAED,SAAS+B,kBAAT,CACER,MADF,EAEEC,KAFF,EAGEvB,IAHF,EAIY;AAEV,MAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;;AAEA,MAAIV,KAAK,GAAG,CAAR,KAAc,CAAlB,EAAqB;AACnB,UAAM,IAAItB,KAAJ,CAAU,yBAAV,CAAN;AACD;;AAGD,MAAMF,MAAM,GAAG,IAAImC,KAAJ,CAAUX,KAAV,EAAiBY,IAAjB,CAAsB,CAAtB,CAAf;;AACA,OAAK,IAAIC,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,QAAQ,GAAGV,KAA/B,EAAsCa,CAAC,EAAvC,EAA2C;AACzC,QAAId,MAAM,CAACM,MAAP,CAAcN,MAAM,CAACE,MAAP,GAAgBa,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAG,CAAf,CAA9B,IAAoD,KAAKA,CAAC,GAAG,CAAjE,EAAqE;AACnErC,MAAAA,MAAM,CAACsC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAGH,QAAf,CAAD,CAAN,IAAoC,KAAKG,CAAC,GAAGH,QAA7C;AACD;AACF;;AAEDX,EAAAA,MAAM,CAACE,MAAP,IAAiBS,QAAQ,IAAIV,KAAK,GAAG,CAAZ,CAAzB;AACA,SAAOxB,MAAP;AACD;;AAED,SAASgC,iBAAT,CACET,MADF,EAEEC,KAFF,EAGEvB,IAHF,EAIY;AAEV,MAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;AAEA,MAAIM,KAAK,GAAG,CAAZ;;AACA,OAAK,IAAI7B,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAG2B,IAAI,CAACG,IAAL,CAAUP,QAAQ,GAAG,CAArB,CAApB,EAA6CvB,CAAC,EAA9C,EAAkD;AAEhD6B,IAAAA,KAAK,IAAI,CAAT;AACAA,IAAAA,KAAK,IAAIjB,MAAM,CAACM,MAAP,CAAcN,MAAM,CAACE,MAArB,CAAT;AACAF,IAAAA,MAAM,CAACE,MAAP,IAAiB,CAAjB;AACD;;AAGD,SAAO,IAAIU,KAAJ,CAAUX,KAAV,EAAiBY,IAAjB,CAAsBI,KAAtB,CAAP;AACD;;AAED,SAAS1B,kBAAT,CAA4Bd,MAA5B,EAA8CC,IAA9C,EAAiF;AAE/E,MAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;;AAEA,OAAK,IAAIvB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGX,MAAM,CAACY,MAAP,GAAgB,CAApC,EAAuCD,CAAC,EAAxC,EAA4C;AAC1CX,IAAAA,MAAM,CAACgB,IAAP,CAAY,CAAZ;AACD;;AAED,MAAMV,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa8B,IAAI,CAACG,IAAL,CAAUP,QAAQ,IAAIlC,MAAM,CAACY,MAAP,GAAgB,CAApB,CAAlB,CAAb,CAAZ;;AACA,OAAK,IAAIyB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGH,QAAQ,GAAGlC,MAAM,CAACY,MAAtC,EAA8CyB,CAAC,EAA/C,EAAmD;AACjD,QAAI,CAACrC,MAAM,CAACsC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAGH,QAAf,CAAD,CAAN,GAAoC,KAAKG,CAAC,GAAGH,QAA9C,IAA2D,CAA/D,EAAkE;AAChE5B,MAAAA,GAAG,CAACgC,IAAI,CAACC,KAAL,CAAWF,CAAC,GAAG,CAAf,CAAD,CAAH,IAA0B,KAAKA,CAAC,GAAG,CAAnC;AACD;AACF;;AAED,SAAO9B,MAAM,CAACM,MAAP,CAAc,CAACN,MAAM,CAACmC,IAAP,CAAYf,gBAAOgB,MAAP,CAAgB3C,MAAM,CAACY,MAAP,GAAgB,CAAjB,IAAuB,CAAxB,GAA6B,CAA3C,CAAZ,CAAD,EAA6DN,GAA7D,CAAd,CAAP;AACD;;AAED,SAASS,iBAAT,CAA2ByB,KAA3B,EAA0ChB,KAA1C,EAAyDvB,IAAzD,EAA4F;AAE1F,MAAMiC,QAAgB,GAAGjC,IAAI,CAACiC,QAA9B;AAEA,MAAM5B,GAAG,GAAGC,MAAM,CAACC,KAAP,CAAa8B,IAAI,CAACG,IAAL,CAAUP,QAAQ,GAAG,CAArB,CAAb,CAAZ;;AAEA,OAAK,IAAIvB,CAAC,GAAG,CAAb,EAAgBA,CAAC,GAAGL,GAAG,CAACM,MAAxB,EAAgCD,CAAC,EAAjC,EAAqC;AACnCL,IAAAA,GAAG,CAACsC,UAAJ,CAAeJ,KAAK,GAAG,IAAvB,EAA6B7B,CAA7B;AAEA6B,IAAAA,KAAK,IAAI,CAAT;AACD;;AAED,SAAOjC,MAAM,CAACM,MAAP,CAAc,CAACN,MAAM,CAACmC,IAAP,CAAYf,gBAAOgB,MAAP,CAAcnB,KAAK,IAAI,CAAvB,CAAZ,CAAD,EAAyClB,GAAzC,CAAd,CAAP;AACD","sourcesContent":["// Forked from https://github.com/kbajalc/parquets under MIT license (Copyright (c) 2017 ironSource Ltd.)\n\nimport type {PrimitiveType} from '../schema/declare';\nimport type {CursorBuffer, ParquetCodecOptions} from './declare';\nimport varint from 'varint';\n\n// eslint-disable-next-line max-statements, complexity\nexport function encodeValues(\n type: PrimitiveType,\n values: any[],\n opts: ParquetCodecOptions\n): Buffer {\n if (!('bitWidth' in opts)) {\n throw new Error('bitWidth is required');\n }\n\n switch (type) {\n case 'BOOLEAN':\n case 'INT32':\n case 'INT64':\n // tslint:disable-next-line:no-parameter-reassignment\n values = values.map((x) => parseInt(x, 10));\n break;\n\n default:\n throw new Error(`unsupported type: ${type}`);\n }\n\n let buf = Buffer.alloc(0);\n let run: any[] = [];\n let repeats = 0;\n\n for (let i = 0; i < values.length; i++) {\n // If we are at the beginning of a run and the next value is same we start\n // collecting repeated values\n if (repeats === 0 && run.length % 8 === 0 && values[i] === values[i + 1]) {\n // If we have any data in runs we need to encode them\n if (run.length) {\n buf = Buffer.concat([buf, encodeRunBitpacked(run, opts)]);\n run = [];\n }\n repeats = 1;\n } else if (repeats > 0 && values[i] === values[i - 1]) {\n repeats += 1;\n } else {\n // If values changes we need to post any previous repeated values\n if (repeats) {\n buf = Buffer.concat([buf, encodeRunRepeated(values[i - 1], repeats, opts)]);\n repeats = 0;\n }\n run.push(values[i]);\n }\n }\n\n if (repeats) {\n buf = Buffer.concat([buf, encodeRunRepeated(values[values.length - 1], repeats, opts)]);\n } else if (run.length) {\n buf = Buffer.concat([buf, encodeRunBitpacked(run, opts)]);\n }\n\n if (opts.disableEnvelope) {\n return buf;\n }\n\n const envelope = Buffer.alloc(buf.length + 4);\n envelope.writeUInt32LE(buf.length, undefined);\n buf.copy(envelope, 4);\n\n return envelope;\n}\n\nexport function decodeValues(\n type: PrimitiveType,\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n if (!('bitWidth' in opts)) {\n throw new Error('bitWidth is required');\n }\n\n if (!opts.disableEnvelope) {\n cursor.offset += 4;\n }\n\n let values: number[] = [];\n while (values.length < count) {\n const header = varint.decode(cursor.buffer, cursor.offset);\n cursor.offset += varint.encodingLength(header);\n if (header & 1) {\n const count = (header >> 1) * 8;\n values.push(...decodeRunBitpacked(cursor, count, opts));\n } else {\n const count = header >> 1;\n values.push(...decodeRunRepeated(cursor, count, opts));\n }\n }\n values = values.slice(0, count);\n\n if (values.length !== count) {\n throw new Error('invalid RLE encoding');\n }\n\n return values;\n}\n\nfunction decodeRunBitpacked(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n if (count % 8 !== 0) {\n throw new Error('must be a multiple of 8');\n }\n\n // tslint:disable-next-line:prefer-array-literal\n const values = new Array(count).fill(0);\n for (let b = 0; b < bitWidth * count; b++) {\n if (cursor.buffer[cursor.offset + Math.floor(b / 8)] & (1 << b % 8)) {\n values[Math.floor(b / bitWidth)] |= 1 << b % bitWidth;\n }\n }\n\n cursor.offset += bitWidth * (count / 8);\n return values;\n}\n\nfunction decodeRunRepeated(\n cursor: CursorBuffer,\n count: number,\n opts: ParquetCodecOptions\n): number[] {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n let value = 0;\n for (let i = 0; i < Math.ceil(bitWidth / 8); i++) {\n // eslint-disable-next-line\n value << 8; // TODO - this looks wrong\n value += cursor.buffer[cursor.offset];\n cursor.offset += 1;\n }\n\n // tslint:disable-next-line:prefer-array-literal\n return new Array(count).fill(value);\n}\n\nfunction encodeRunBitpacked(values: number[], opts: ParquetCodecOptions): Buffer {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n for (let i = 0; i < values.length % 8; i++) {\n values.push(0);\n }\n\n const buf = Buffer.alloc(Math.ceil(bitWidth * (values.length / 8)));\n for (let b = 0; b < bitWidth * values.length; b++) {\n if ((values[Math.floor(b / bitWidth)] & (1 << b % bitWidth)) > 0) {\n buf[Math.floor(b / 8)] |= 1 << b % 8;\n }\n }\n\n return Buffer.concat([Buffer.from(varint.encode(((values.length / 8) << 1) | 1)), buf]);\n}\n\nfunction encodeRunRepeated(value: number, count: number, opts: ParquetCodecOptions): Buffer {\n // @ts-ignore\n const bitWidth: number = opts.bitWidth;\n\n const buf = Buffer.alloc(Math.ceil(bitWidth / 8));\n\n for (let i = 0; i < buf.length; i++) {\n buf.writeUInt8(value & 0xff, i);\n // eslint-disable-next-line\n value >> 8; // TODO - this looks wrong\n }\n\n return Buffer.concat([Buffer.from(varint.encode(count << 1)), buf]);\n}\n"],"file":"rle.js"}
|