@loaders.gl/parquet 3.3.2 → 3.4.0-alpha.2
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 +16 -16
- package/dist/dist.min.js.map +3 -3
- package/dist/es5/index.js +46 -5
- package/dist/es5/index.js.map +1 -1
- package/dist/es5/lib/arrow/convert-columns-to-row-group.js +2 -0
- package/dist/es5/lib/arrow/convert-columns-to-row-group.js.map +1 -0
- package/dist/es5/lib/arrow/convert-row-group-to-columns.js +20 -0
- package/dist/es5/lib/arrow/convert-row-group-to-columns.js.map +1 -0
- package/dist/es5/lib/arrow/convert-schema-from-parquet.js +98 -0
- package/dist/es5/lib/arrow/convert-schema-from-parquet.js.map +1 -0
- package/dist/es5/lib/{convert-schema.js → arrow/convert-schema-to-parquet.js} +5 -31
- package/dist/es5/lib/arrow/convert-schema-to-parquet.js.map +1 -0
- package/dist/es5/lib/geo/decode-geo-metadata.js +82 -0
- package/dist/es5/lib/geo/decode-geo-metadata.js.map +1 -0
- package/dist/es5/lib/geo/geoparquet-schema.js +83 -0
- package/dist/es5/lib/geo/geoparquet-schema.js.map +1 -0
- package/dist/es5/lib/parsers/parse-parquet-to-columns.js +177 -0
- package/dist/es5/lib/parsers/parse-parquet-to-columns.js.map +1 -0
- package/dist/es5/lib/{parse-parquet.js → parsers/parse-parquet-to-rows.js} +2 -2
- package/dist/es5/lib/parsers/parse-parquet-to-rows.js.map +1 -0
- package/dist/es5/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/dist/es5/parquet-loader.js +3 -2
- package/dist/es5/parquet-loader.js.map +1 -1
- package/dist/es5/parquet-wasm-loader.js +1 -1
- package/dist/es5/parquet-wasm-loader.js.map +1 -1
- package/dist/es5/parquet-wasm-writer.js +1 -1
- package/dist/es5/parquet-wasm-writer.js.map +1 -1
- package/dist/es5/parquet-writer.js +1 -1
- package/dist/es5/parquet-writer.js.map +1 -1
- package/dist/esm/index.js +12 -2
- package/dist/esm/index.js.map +1 -1
- package/dist/esm/lib/arrow/convert-columns-to-row-group.js +2 -0
- package/dist/esm/lib/arrow/convert-columns-to-row-group.js.map +1 -0
- package/dist/esm/lib/arrow/convert-row-group-to-columns.js +10 -0
- package/dist/esm/lib/arrow/convert-row-group-to-columns.js.map +1 -0
- package/dist/esm/lib/{convert-schema.js → arrow/convert-schema-from-parquet.js} +32 -16
- package/dist/esm/lib/arrow/convert-schema-from-parquet.js.map +1 -0
- package/dist/esm/lib/arrow/convert-schema-to-parquet.js +40 -0
- package/dist/esm/lib/arrow/convert-schema-to-parquet.js.map +1 -0
- package/dist/esm/lib/geo/decode-geo-metadata.js +64 -0
- package/dist/esm/lib/geo/decode-geo-metadata.js.map +1 -0
- package/dist/esm/lib/geo/geoparquet-schema.js +78 -0
- package/dist/esm/lib/geo/geoparquet-schema.js.map +1 -0
- package/dist/esm/lib/parsers/parse-parquet-to-columns.js +37 -0
- package/dist/esm/lib/parsers/parse-parquet-to-columns.js.map +1 -0
- package/dist/esm/lib/{parse-parquet.js → parsers/parse-parquet-to-rows.js} +2 -2
- package/dist/esm/lib/parsers/parse-parquet-to-rows.js.map +1 -0
- package/dist/esm/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/dist/esm/parquet-loader.js +3 -2
- package/dist/esm/parquet-loader.js.map +1 -1
- package/dist/esm/parquet-wasm-loader.js +1 -1
- package/dist/esm/parquet-wasm-loader.js.map +1 -1
- package/dist/esm/parquet-wasm-writer.js +1 -1
- package/dist/esm/parquet-wasm-writer.js.map +1 -1
- package/dist/esm/parquet-writer.js +1 -1
- package/dist/esm/parquet-writer.js.map +1 -1
- package/dist/index.d.ts +23 -3
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +24 -6
- package/dist/lib/arrow/convert-columns-to-row-group.d.ts +1 -0
- package/dist/lib/arrow/convert-columns-to-row-group.d.ts.map +1 -0
- package/dist/lib/arrow/convert-columns-to-row-group.js +1 -0
- package/dist/lib/arrow/convert-row-group-to-columns.d.ts +4 -0
- package/dist/lib/arrow/convert-row-group-to-columns.d.ts.map +1 -0
- package/dist/lib/arrow/convert-row-group-to-columns.js +12 -0
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts +9 -0
- package/dist/lib/arrow/convert-schema-from-parquet.d.ts.map +1 -0
- package/dist/lib/{convert-schema.js → arrow/convert-schema-from-parquet.js} +30 -18
- package/dist/lib/arrow/convert-schema-to-parquet.d.ts +7 -0
- package/dist/lib/arrow/convert-schema-to-parquet.d.ts.map +1 -0
- package/dist/lib/arrow/convert-schema-to-parquet.js +72 -0
- package/dist/lib/geo/decode-geo-metadata.d.ts +31 -0
- package/dist/lib/geo/decode-geo-metadata.d.ts.map +1 -0
- package/dist/lib/geo/decode-geo-metadata.js +73 -0
- package/dist/lib/geo/geoparquet-schema.d.ts +80 -0
- package/dist/lib/geo/geoparquet-schema.d.ts.map +1 -0
- package/dist/lib/geo/geoparquet-schema.js +69 -0
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts +5 -0
- package/dist/lib/parsers/parse-parquet-to-columns.d.ts.map +1 -0
- package/dist/lib/parsers/parse-parquet-to-columns.js +40 -0
- package/dist/lib/parsers/parse-parquet-to-rows.d.ts +4 -0
- package/dist/lib/parsers/parse-parquet-to-rows.d.ts.map +1 -0
- package/dist/lib/{parse-parquet.js → parsers/parse-parquet-to-rows.js} +1 -1
- package/dist/parquet-loader.d.ts +1 -0
- package/dist/parquet-loader.d.ts.map +1 -1
- package/dist/parquet-loader.js +2 -1
- package/dist/parquet-worker.js +19 -19
- package/dist/parquet-worker.js.map +3 -3
- package/package.json +5 -5
- package/src/index.ts +22 -2
- package/src/lib/arrow/convert-columns-to-row-group.ts +0 -0
- package/src/lib/arrow/convert-row-group-to-columns.ts +15 -0
- package/src/lib/{convert-schema.ts → arrow/convert-schema-from-parquet.ts} +41 -22
- package/src/lib/arrow/convert-schema-to-parquet.ts +102 -0
- package/src/lib/geo/decode-geo-metadata.ts +99 -0
- package/src/lib/geo/geoparquet-schema.ts +69 -0
- package/src/lib/parsers/parse-parquet-to-columns.ts +49 -0
- package/src/lib/{parse-parquet.ts → parsers/parse-parquet-to-rows.ts} +2 -2
- package/src/lib/wip/convert-schema-deep.rs.disabled +976 -0
- package/src/parquet-loader.ts +3 -1
- package/dist/es5/lib/convert-schema.js.map +0 -1
- package/dist/es5/lib/parse-parquet.js.map +0 -1
- package/dist/es5/lib/read-array-buffer.js +0 -43
- package/dist/es5/lib/read-array-buffer.js.map +0 -1
- package/dist/esm/lib/convert-schema.js.map +0 -1
- package/dist/esm/lib/parse-parquet.js.map +0 -1
- package/dist/esm/lib/read-array-buffer.js +0 -10
- package/dist/esm/lib/read-array-buffer.js.map +0 -1
- package/dist/lib/convert-schema.d.ts +0 -8
- package/dist/lib/convert-schema.d.ts.map +0 -1
- package/dist/lib/parse-parquet.d.ts +0 -4
- package/dist/lib/parse-parquet.d.ts.map +0 -1
- package/dist/lib/read-array-buffer.d.ts +0 -19
- package/dist/lib/read-array-buffer.d.ts.map +0 -1
- package/dist/lib/read-array-buffer.js +0 -29
- package/src/lib/read-array-buffer.ts +0 -31
- /package/dist/es5/lib/{convert-schema-deep.ts.disabled → wip/convert-schema-deep.java.disabled} +0 -0
- /package/dist/esm/lib/{convert-schema-deep.ts.disabled → wip/convert-schema-deep.java.disabled} +0 -0
- /package/src/lib/{convert-schema-deep.ts.disabled → wip/convert-schema-deep.java.disabled} +0 -0
|
@@ -0,0 +1,69 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
// loaders.gl, MIT license
|
|
3
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
4
|
+
/* eslint-disable camelcase */
|
|
5
|
+
/**
|
|
6
|
+
* Geoparquet JSON schema for geo metadata
|
|
7
|
+
* @see https://github.com/geoarrow/geoarrow/blob/main/metadata.md
|
|
8
|
+
* @see https://github.com/opengeospatial/geoparquet/blob/main/format-specs/geoparquet.md
|
|
9
|
+
*/
|
|
10
|
+
exports.default = {
|
|
11
|
+
$schema: 'http://json-schema.org/draft-07/schema#',
|
|
12
|
+
title: 'GeoParquet',
|
|
13
|
+
description: 'Parquet metadata included in the geo field.',
|
|
14
|
+
type: 'object',
|
|
15
|
+
required: ['version', 'primary_column', 'columns'],
|
|
16
|
+
properties: {
|
|
17
|
+
version: { type: 'string', const: '1.0.0-beta.1' },
|
|
18
|
+
primary_column: { type: 'string', minLength: 1 },
|
|
19
|
+
columns: {
|
|
20
|
+
type: 'object',
|
|
21
|
+
minProperties: 1,
|
|
22
|
+
patternProperties: {
|
|
23
|
+
'.+': {
|
|
24
|
+
type: 'object',
|
|
25
|
+
required: ['encoding', 'geometry_types'],
|
|
26
|
+
properties: {
|
|
27
|
+
encoding: { type: 'string', const: 'WKB' },
|
|
28
|
+
geometry_types: {
|
|
29
|
+
type: 'array',
|
|
30
|
+
uniqueItems: true,
|
|
31
|
+
items: {
|
|
32
|
+
type: 'string',
|
|
33
|
+
pattern: '^(GeometryCollection|(Multi)?(Point|LineString|Polygon))( Z)?$'
|
|
34
|
+
}
|
|
35
|
+
},
|
|
36
|
+
crs: {
|
|
37
|
+
oneOf: [
|
|
38
|
+
{
|
|
39
|
+
$ref: 'https://proj.org/schemas/v0.5/projjson.schema.json'
|
|
40
|
+
},
|
|
41
|
+
{ type: 'null' }
|
|
42
|
+
]
|
|
43
|
+
},
|
|
44
|
+
edges: { type: 'string', enum: ['planar', 'spherical'] },
|
|
45
|
+
orientation: { type: 'string', const: 'counterclockwise' },
|
|
46
|
+
bbox: {
|
|
47
|
+
type: 'array',
|
|
48
|
+
items: { type: 'number' },
|
|
49
|
+
oneOf: [
|
|
50
|
+
{
|
|
51
|
+
description: '2D bbox consisting of (xmin, ymin, xmax, ymax)',
|
|
52
|
+
minItems: 4,
|
|
53
|
+
maxItems: 4
|
|
54
|
+
},
|
|
55
|
+
{
|
|
56
|
+
description: '3D bbox consisting of (xmin, ymin, zmin, xmax, ymax, zmax)',
|
|
57
|
+
minItems: 6,
|
|
58
|
+
maxItems: 6
|
|
59
|
+
}
|
|
60
|
+
]
|
|
61
|
+
},
|
|
62
|
+
epoch: { type: 'number' }
|
|
63
|
+
}
|
|
64
|
+
}
|
|
65
|
+
},
|
|
66
|
+
additionalProperties: false
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
};
|
|
@@ -0,0 +1,5 @@
|
|
|
1
|
+
import { ColumnarTableBatch } from '@loaders.gl/schema';
|
|
2
|
+
import type { ParquetLoaderOptions } from '../../parquet-loader';
|
|
3
|
+
export declare function parseParquetInColumns(arrayBuffer: ArrayBuffer, options?: ParquetLoaderOptions): Promise<ColumnarTableBatch | null>;
|
|
4
|
+
export declare function parseParquetFileInColumnarBatches(blob: Blob, options?: ParquetLoaderOptions): AsyncIterable<ColumnarTableBatch>;
|
|
5
|
+
//# sourceMappingURL=parse-parquet-to-columns.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parse-parquet-to-columns.d.ts","sourceRoot":"","sources":["../../../src/lib/parsers/parse-parquet-to-columns.ts"],"names":[],"mappings":"AAGA,OAAO,EAAC,kBAAkB,EAAS,MAAM,oBAAoB,CAAC;AAE9D,OAAO,KAAK,EAAC,oBAAoB,EAAC,MAAM,sBAAsB,CAAC;AAO/D,wBAAsB,qBAAqB,CACzC,WAAW,EAAE,WAAW,EACxB,OAAO,CAAC,EAAE,oBAAoB,sCAO/B;AAED,wBAAuB,iCAAiC,CACtD,IAAI,EAAE,IAAI,EACV,OAAO,CAAC,EAAE,oBAAoB,GAC7B,aAAa,CAAC,kBAAkB,CAAC,CAWnC"}
|
|
@@ -0,0 +1,40 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
// loaders.gl, MIT license
|
|
3
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
4
|
+
exports.parseParquetFileInColumnarBatches = exports.parseParquetInColumns = void 0;
|
|
5
|
+
const loader_utils_1 = require("@loaders.gl/loader-utils");
|
|
6
|
+
const parquet_reader_1 = require("../../parquetjs/parser/parquet-reader");
|
|
7
|
+
const convert_schema_from_parquet_1 = require("../arrow/convert-schema-from-parquet");
|
|
8
|
+
const convert_row_group_to_columns_1 = require("../arrow/convert-row-group-to-columns");
|
|
9
|
+
const decode_geo_metadata_1 = require("../geo/decode-geo-metadata");
|
|
10
|
+
async function parseParquetInColumns(arrayBuffer, options) {
|
|
11
|
+
const blob = new Blob([arrayBuffer]);
|
|
12
|
+
for await (const batch of parseParquetFileInColumnarBatches(blob, options)) {
|
|
13
|
+
return batch;
|
|
14
|
+
}
|
|
15
|
+
return null;
|
|
16
|
+
}
|
|
17
|
+
exports.parseParquetInColumns = parseParquetInColumns;
|
|
18
|
+
async function* parseParquetFileInColumnarBatches(blob, options) {
|
|
19
|
+
const file = (0, loader_utils_1.makeReadableFile)(blob);
|
|
20
|
+
const reader = new parquet_reader_1.ParquetReader(file);
|
|
21
|
+
const parquetSchema = await reader.getSchema();
|
|
22
|
+
const parquetMetadata = await reader.getFileMetadata();
|
|
23
|
+
const schema = (0, convert_schema_from_parquet_1.convertSchemaFromParquet)(parquetSchema, parquetMetadata);
|
|
24
|
+
(0, decode_geo_metadata_1.unpackGeoMetadata)(schema);
|
|
25
|
+
const rowGroups = reader.rowGroupIterator(options?.parquet);
|
|
26
|
+
for await (const rowGroup of rowGroups) {
|
|
27
|
+
yield convertRowGroupToTableBatch(schema, rowGroup);
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
exports.parseParquetFileInColumnarBatches = parseParquetFileInColumnarBatches;
|
|
31
|
+
function convertRowGroupToTableBatch(schema, rowGroup) {
|
|
32
|
+
const data = (0, convert_row_group_to_columns_1.convertParquetRowGroupToColumns)(schema, rowGroup);
|
|
33
|
+
return {
|
|
34
|
+
shape: 'columnar-table',
|
|
35
|
+
batchType: 'data',
|
|
36
|
+
schema,
|
|
37
|
+
data,
|
|
38
|
+
length: rowGroup.rowCount
|
|
39
|
+
};
|
|
40
|
+
}
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
import type { ParquetLoaderOptions } from '../../parquet-loader';
|
|
2
|
+
export declare function parseParquet(arrayBuffer: ArrayBuffer, options?: ParquetLoaderOptions): Promise<import("../../parquetjs/schema/declare").ParquetRecord[] | null>;
|
|
3
|
+
export declare function parseParquetFileInBatches(blob: Blob, options?: ParquetLoaderOptions): AsyncGenerator<import("../../parquetjs/schema/declare").ParquetRecord[], void, unknown>;
|
|
4
|
+
//# sourceMappingURL=parse-parquet-to-rows.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parse-parquet-to-rows.d.ts","sourceRoot":"","sources":["../../../src/lib/parsers/parse-parquet-to-rows.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAC,oBAAoB,EAAC,MAAM,sBAAsB,CAAC;AAG/D,wBAAsB,YAAY,CAAC,WAAW,EAAE,WAAW,EAAE,OAAO,CAAC,EAAE,oBAAoB,4EAM1F;AAED,wBAAuB,yBAAyB,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,CAAC,EAAE,oBAAoB,2FAO1F"}
|
|
@@ -4,7 +4,7 @@ exports.parseParquetFileInBatches = exports.parseParquet = void 0;
|
|
|
4
4
|
// import type {LoaderWithParser, Loader, LoaderOptions} from '@loaders.gl/loader-utils';
|
|
5
5
|
// import {ColumnarTableBatch} from '@loaders.gl/schema';
|
|
6
6
|
const loader_utils_1 = require("@loaders.gl/loader-utils");
|
|
7
|
-
const parquet_reader_1 = require("
|
|
7
|
+
const parquet_reader_1 = require("../../parquetjs/parser/parquet-reader");
|
|
8
8
|
async function parseParquet(arrayBuffer, options) {
|
|
9
9
|
const blob = new Blob([arrayBuffer]);
|
|
10
10
|
for await (const batch of parseParquetFileInBatches(blob, options)) {
|
package/dist/parquet-loader.d.ts
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"parquet-loader.d.ts","sourceRoot":"","sources":["../src/parquet-loader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAC,MAAM,EAAE,aAAa,EAAC,MAAM,0BAA0B,CAAC;AAMpE,MAAM,MAAM,oBAAoB,GAAG,aAAa,GAAG;IACjD,OAAO,CAAC,EAAE;QACR,IAAI,CAAC,EAAE,kBAAkB,CAAC;QAC1B,GAAG,CAAC,EAAE,MAAM,CAAC;QACb,UAAU,CAAC,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,EAAE,CAAC;
|
|
1
|
+
{"version":3,"file":"parquet-loader.d.ts","sourceRoot":"","sources":["../src/parquet-loader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAC,MAAM,EAAE,aAAa,EAAC,MAAM,0BAA0B,CAAC;AAMpE,MAAM,MAAM,oBAAoB,GAAG,aAAa,GAAG;IACjD,OAAO,CAAC,EAAE;QACR,IAAI,CAAC,EAAE,kBAAkB,CAAC;QAC1B,GAAG,CAAC,EAAE,MAAM,CAAC;QACb,UAAU,CAAC,EAAE,MAAM,EAAE,GAAG,MAAM,EAAE,EAAE,CAAC;QACnC,UAAU,CAAC,EAAE,OAAO,CAAC;KACtB,CAAC;CACH,CAAC;AAWF,6BAA6B;AAC7B,eAAO,MAAM,aAAa;;;;;;;;;;;;CAYzB,CAAC;AAEF,eAAO,MAAM,uBAAuB,EAAE,MAAsB,CAAC"}
|