24153816851

Committed 08 Apr 2026 07:17PM UTC coverage: 53.247% (-12.1%) from 65.319%

Build # 24153816851

Build Type

push

github

Committed by

web-flow

Commit Message

chore: Move from tape to vitest (#3351)

Coverage Stats

8651 of 17291 branches covered (50.03%)

Branch coverage included in aggregate %.

7 of 7 new or added lines in 1 file covered. (100.0%)

2031 existing lines in 296 files now uncovered.

17563 of 31940 relevant lines covered (54.99%)

5279.54 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

89.53

/modules/csv/src/csv-loader.ts

// loaders.gl
// SPDX-License-Identifier: MIT
// Copyright (c) vis.gl contributors

import type {LoaderWithParser, LoaderOptions} from '@loaders.gl/loader-utils';
import type {Schema, ArrayRowTable, ObjectRowTable, TableBatch} from '@loaders.gl/schema';

import {log, toArrayBufferIterator} from '@loaders.gl/loader-utils';
import {
  AsyncQueue,
  deduceTableSchema,
  TableBatchBuilder,
  convertToArrayRow,
  convertToObjectRow
} from '@loaders.gl/schema-utils';
import Papa from './papaparse/papaparse';
import AsyncIteratorStreamer from './papaparse/async-iterator-streamer';
import {CSVFormat} from './csv-format';

// __VERSION__ is injected by babel-plugin-version-inline
// @ts-ignore TS2304: Cannot find name '__VERSION__'.
const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';

const DEFAULT_CSV_SHAPE = 'object-row-table';

export type CSVLoaderOptions = LoaderOptions & {
  csv?: {
    // loaders.gl options
    shape?: 'array-row-table' | 'object-row-table';
    /** optimizes memory usage but increases parsing time. */
    optimizeMemoryUsage?: boolean;
    columnPrefix?: string;
    header?: 'auto';

    // CSV options (papaparse)
    // delimiter: auto
    // newline: auto
    quoteChar?: string;
    escapeChar?: string;
    // Convert numbers and boolean values in rows from strings
    dynamicTyping?: boolean;
    comments?: boolean;
    skipEmptyLines?: boolean | 'greedy';
    // transform: null?
    delimitersToGuess?: string[];
    // fastMode: auto
  };
};

export const CSVLoader = {
  ...CSVFormat,

  dataType: null as unknown as ObjectRowTable | ArrayRowTable,
  batchType: null as unknown as TableBatch,
  version: VERSION,
  parse: async (arrayBuffer: ArrayBuffer, options?: CSVLoaderOptions) =>
    parseCSV(new TextDecoder().decode(arrayBuffer), options),
  parseText: (text: string, options?: CSVLoaderOptions) => parseCSV(text, options),
  parseInBatches: parseCSVInBatches,
  // @ts-ignore
  // testText: null,
  options: {
    csv: {
      shape: DEFAULT_CSV_SHAPE, // 'object-row-table'
      optimizeMemoryUsage: false,
      // CSV options
      header: 'auto',
      columnPrefix: 'column',
      // delimiter: auto
      // newline: auto
      quoteChar: '"',
      escapeChar: '"',
      dynamicTyping: true,
      comments: false,
      skipEmptyLines: true,
      // transform: null?
      delimitersToGuess: [',', '\t', '|', ';']
      // fastMode: auto
    }
  }
} as const satisfies LoaderWithParser<ObjectRowTable | ArrayRowTable, TableBatch, CSVLoaderOptions>;

async function parseCSV(
  csvText: string,
  options?: CSVLoaderOptions
): Promise<ObjectRowTable | ArrayRowTable> {
  // Apps can call the parse method directly, so we apply default options here
  const csvOptions = {...CSVLoader.options.csv, ...options?.csv};

  const firstRow = readFirstRow(csvText);
  const header: boolean =
    csvOptions.header === 'auto' ? isHeaderRow(firstRow) : Boolean(csvOptions.header);

  const parseWithHeader = header;

  const papaparseConfig = {
    // dynamicTyping: true,
    ...csvOptions,
    header: parseWithHeader,
    download: false, // We handle loading, no need for papaparse to do it for us
    transformHeader: parseWithHeader ? duplicateColumnTransformer() : undefined,
    error: e => {
      throw new Error(e);
    }
  };

  const result = Papa.parse(csvText, papaparseConfig);
  const rows = result.data as any[];

  const headerRow = result.meta.fields || generateHeader(csvOptions.columnPrefix, firstRow.length);

  const shape = csvOptions.shape || DEFAULT_CSV_SHAPE;
  let table: ArrayRowTable | ObjectRowTable;
  switch (shape) {
    case 'object-row-table':
      table = {
        shape: 'object-row-table',
        data: rows.map(row => (Array.isArray(row) ? convertToObjectRow(row, headerRow) : row))
      };
      break;
    case 'array-row-table':
      table = {
        shape: 'array-row-table',
        data: rows.map(row => (Array.isArray(row) ? row : convertToArrayRow(row, headerRow)))
      };
      break;
    default:
      throw new Error(shape);
  }
  table.schema = deduceTableSchema(table!);
  return table;
}

// TODO - support batch size 0 = no batching/single batch?
function parseCSVInBatches(
  asyncIterator:
    | AsyncIterable<ArrayBufferLike | ArrayBufferView>
    | Iterable<ArrayBufferLike | ArrayBufferView>,
  options?: CSVLoaderOptions
): AsyncIterable<TableBatch> {
  // Papaparse does not support standard batch size handling
  // TODO - investigate papaparse chunks mode
  options = {...options};
  if (options?.core?.batchSize === 'auto') {
    options.core.batchSize = 4000;
  }

  // Apps can call the parse method directly, we so apply default options here
  const csvOptions = {...CSVLoader.options.csv, ...options?.csv};

  const asyncQueue = new AsyncQueue<TableBatch>();

  let isFirstRow: boolean = true;
  let headerRow: string[] | null = null;
  let tableBatchBuilder: TableBatchBuilder | null = null;
  let schema: Schema | null = null;

  const config = {
    // dynamicTyping: true, // Convert numbers and boolean values in rows from strings,
    ...csvOptions,
    header: false, // Unfortunately, header detection is not automatic and does not infer shapes
    download: false, // We handle loading, no need for papaparse to do it for us
    // chunkSize is set to 5MB explicitly (same as Papaparse default) due to a bug where the
    // streaming parser gets stuck if skipEmptyLines and a step callback are both supplied.
    // See https://github.com/mholt/PapaParse/issues/465
    chunkSize: 1024 * 1024 * 5,
    // skipEmptyLines is set to a boolean value if supplied. Greedy is set to true
    // skipEmptyLines is handled manually given two bugs where the streaming parser gets stuck if
    // both of the skipEmptyLines and step callback options are provided:
    // - true doesn't work unless chunkSize is set: https://github.com/mholt/PapaParse/issues/465
    // - greedy doesn't work: https://github.com/mholt/PapaParse/issues/825
    skipEmptyLines: false,

    // step is called on every row
    // eslint-disable-next-line complexity, max-statements
    step(results) {
      let row = results.data;

      if (csvOptions.skipEmptyLines === 'greedy') {
        // Manually reject lines that are empty
        const collapsedRow = row.flat().join('').trim();
        if (collapsedRow === '') {
          return;
        }
      } else if (csvOptions.skipEmptyLines === true) {
        row = normalizePapaStreamingRow(row);
        if (row.length === 1 && row[0] === null) {
          return;
        }
      }
      const bytesUsed = results.meta.cursor;

      // Check if we need to save a header row
      if (isFirstRow && !headerRow) {
        // Auto detects or can be forced with csvOptions.header
        const header = csvOptions.header === 'auto' ? isHeaderRow(row) : Boolean(csvOptions.header);
        if (header) {
          headerRow = row.map(duplicateColumnTransformer());
          return;
        }
      }

      // If first data row, we can deduce the schema
      if (isFirstRow) {
        isFirstRow = false;
        if (!headerRow) {
          headerRow = generateHeader(csvOptions.columnPrefix, row.length);
        }
        schema = deduceCSVSchema(row, headerRow);
      }

      if (csvOptions.optimizeMemoryUsage) {
        // A workaround to allocate new strings and don't retain pointers to original strings.
        // https://bugs.chromium.org/p/v8/issues/detail?id=2869
        row = JSON.parse(JSON.stringify(row));
      }

      const shape = (options as any)?.shape || csvOptions.shape || DEFAULT_CSV_SHAPE;
      if (shape === 'object-row-table' && headerRow && row.length > headerRow.length) {
        row = convertToPapaObjectRow(row, headerRow);
      }

      // Add the row
      tableBatchBuilder =
        tableBatchBuilder ||
        new TableBatchBuilder(
          // @ts-expect-error TODO this is not a proper schema
          schema,
          {
            shape,
            ...(options?.core || {})
          }
        );

      try {
        tableBatchBuilder.addRow(row);
        // If a batch has been completed, emit it
        const batch = tableBatchBuilder && tableBatchBuilder.getFullBatch({bytesUsed});
        if (batch) {
          asyncQueue.enqueue(batch);
        }
      } catch (error) {
        asyncQueue.enqueue(error as Error);
      }
    },

    // complete is called when all rows have been read
    complete(results) {
      try {
        const bytesUsed = results.meta.cursor;
        // Ensure any final (partial) batch gets emitted
        const batch = tableBatchBuilder && tableBatchBuilder.getFinalBatch({bytesUsed});
        if (batch) {
          asyncQueue.enqueue(batch);
        }
      } catch (error) {
        asyncQueue.enqueue(error as Error);
      }

      asyncQueue.close();
    }
  };

  Papa.parse(toArrayBufferIterator(asyncIterator), config, AsyncIteratorStreamer);

  // TODO - Does it matter if we return asyncIterable or asyncIterator
  // return asyncQueue[Symbol.asyncIterator]();
  return asyncQueue;
}

/**
 * Checks if a certain row is a header row
 * @param row the row to check
 * @returns true if the row looks like a header
 */
function isHeaderRow(row: string[]): boolean {
  return row && row.every(value => typeof value === 'string');
}

/**
 * Reads, parses, and returns the first row of a CSV text
 * @param csvText the csv text to parse
 * @returns the first row
 */
function readFirstRow(csvText: string): any[] {
  const result = Papa.parse(csvText, {
    dynamicTyping: true,
    preview: 1
  });
  return result.data[0];
}

/**
 * Creates a transformer that renames duplicate columns. This is needed as Papaparse doesn't handle
 * duplicate header columns and would use the latest occurrence by default.
 * See the header option in https://www.papaparse.com/docs#config
 * @returns a transform function that returns sanitized names for duplicate fields
 */
function duplicateColumnTransformer(): (column: string) => string {
  const observedColumns = new Set<string>();
  return col => {
    let colName = col;
    let counter = 1;
    while (observedColumns.has(colName)) {
      colName = `${col}.${counter}`;
      counter++;
    }
    observedColumns.add(colName);
    return colName;
  };
}

/**
 * Generates the header of a CSV given a prefix and a column count
 * @param columnPrefix the columnPrefix to use
 * @param count the count of column names to generate
 * @returns an array of column names
 */
function generateHeader(columnPrefix: string, count: number = 0): string[] {
  const headers: string[] = [];
  for (let i = 0; i < count; i++) {
    headers.push(`${columnPrefix}${i + 1}`);
  }
  return headers;
}

function normalizePapaStreamingRow(row: unknown[]): unknown[] {
  return row.map(value => (Array.isArray(value) && value.length === 0 ? null : value));
}

function convertToPapaObjectRow(
  row: unknown[],
  headerRow: string[]
): {[columnName: string]: unknown} {
  const objectRow = convertToObjectRow(row, headerRow);
  const parsedExtra = row.slice(headerRow.length);
  if (parsedExtra.length > 0) {
    objectRow.__parsed_extra = parsedExtra;
  }
  return objectRow;
}

function deduceCSVSchema(row, headerRow): Schema {
  const fields: Schema['fields'] = [];
  for (let i = 0; i < row.length; i++) {
    const columnName = (headerRow && headerRow[i]) || i;
    const value = row[i];
    switch (typeof value) {
      case 'number':
        fields.push({name: String(columnName), type: 'float64', nullable: true});
        break;
      case 'boolean':
        fields.push({name: String(columnName), type: 'bool', nullable: true});
        break;
      case 'string':
        fields.push({name: String(columnName), type: 'utf8', nullable: true});
        break;
      default:
        log.warn(`CSV: Unknown column type: ${typeof value}`)();
        fields.push({name: String(columnName), type: 'utf8', nullable: true});
    }
  }
  return {
    fields,
    metadata: {
      'loaders.gl#format': 'csv',
      'loaders.gl#loader': 'CSVLoader'
    }
  };
}

// TODO - remove
// type ObjectField = {name: string; index: number; type: any};
// type ObjectSchema = {[key: string]: ObjectField} | ObjectField[];

// function deduceObjectSchema(row, headerRow): ObjectSchema {
//   const schema: ObjectSchema = headerRow ? {} : [];
//   for (let i = 0; i < row.length; i++) {
//     const columnName = (headerRow && headerRow[i]) || i;
//     const value = row[i];
//     switch (typeof value) {
//       case 'number':
//       case 'boolean':
//         // TODO - booleans could be handled differently...
//         schema[columnName] = {name: String(columnName), index: i, type: Float32Array};
//         break;
//       case 'string':
//       default:
//         schema[columnName] = {name: String(columnName), index: i, type: Array};
//       // We currently only handle numeric rows
//       // TODO we could offer a function to map strings to numbers?
//     }
//   }
//   return schema;
// }

1	// loaders.gl
2	// SPDX-License-Identifier: MIT
3	// Copyright (c) vis.gl contributors
4
5	import type {LoaderWithParser, LoaderOptions} from '@loaders.gl/loader-utils';
6	import type {Schema, ArrayRowTable, ObjectRowTable, TableBatch} from '@loaders.gl/schema';
7
8	import {log, toArrayBufferIterator} from '@loaders.gl/loader-utils';
9	import {
10	AsyncQueue,
11	deduceTableSchema,
12	TableBatchBuilder,
13	convertToArrayRow,
14	convertToObjectRow
15	} from '@loaders.gl/schema-utils';
16	import Papa from './papaparse/papaparse';
17	import AsyncIteratorStreamer from './papaparse/async-iterator-streamer';
18	import {CSVFormat} from './csv-format';
19
20	// __VERSION__ is injected by babel-plugin-version-inline
21	// @ts-ignore TS2304: Cannot find name '__VERSION__'.
22	const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';	7!
23
24	const DEFAULT_CSV_SHAPE = 'object-row-table';	7✔
25
26	export type CSVLoaderOptions = LoaderOptions & {
27	csv?: {
28	// loaders.gl options
29	shape?: 'array-row-table' \| 'object-row-table';
30	/** optimizes memory usage but increases parsing time. */
31	optimizeMemoryUsage?: boolean;
32	columnPrefix?: string;
33	header?: 'auto';
34
35	// CSV options (papaparse)
36	// delimiter: auto
37	// newline: auto
38	quoteChar?: string;
39	escapeChar?: string;
40	// Convert numbers and boolean values in rows from strings
41	dynamicTyping?: boolean;
42	comments?: boolean;
43	skipEmptyLines?: boolean \| 'greedy';
44	// transform: null?
45	delimitersToGuess?: string[];
46	// fastMode: auto
47	};
48	};
49
50	export const CSVLoader = {	7✔
51	...CSVFormat,
52
53	dataType: null as unknown as ObjectRowTable \| ArrayRowTable,
54	batchType: null as unknown as TableBatch,
55	version: VERSION,
56	parse: async (arrayBuffer: ArrayBuffer, options?: CSVLoaderOptions) =>
57	parseCSV(new TextDecoder().decode(arrayBuffer), options),	1✔
58	parseText: (text: string, options?: CSVLoaderOptions) => parseCSV(text, options),	50✔
59	parseInBatches: parseCSVInBatches,
60	// @ts-ignore
61	// testText: null,
62	options: {
63	csv: {
64	shape: DEFAULT_CSV_SHAPE, // 'object-row-table'
65	optimizeMemoryUsage: false,
66	// CSV options
67	header: 'auto',
68	columnPrefix: 'column',
69	// delimiter: auto
70	// newline: auto
71	quoteChar: '"',
72	escapeChar: '"',
73	dynamicTyping: true,
74	comments: false,
75	skipEmptyLines: true,
76	// transform: null?
77	delimitersToGuess: [',', '\t', '\|', ';']
78	// fastMode: auto
79	}
80	}
81	} as const satisfies LoaderWithParser<ObjectRowTable \| ArrayRowTable, TableBatch, CSVLoaderOptions>;
82
83	async function parseCSV(
84	csvText: string,
85	options?: CSVLoaderOptions
86	): Promise<ObjectRowTable \| ArrayRowTable> {
87	// Apps can call the parse method directly, so we apply default options here
88	const csvOptions = {...CSVLoader.options.csv, ...options?.csv};	51✔
89
90	const firstRow = readFirstRow(csvText);	51✔
91	const header: boolean =
92	csvOptions.header === 'auto' ? isHeaderRow(firstRow) : Boolean(csvOptions.header);	51✔
93
94	const parseWithHeader = header;	51✔
95
96	const papaparseConfig = {	51✔
97	// dynamicTyping: true,
98	...csvOptions,
99	header: parseWithHeader,
100	download: false, // We handle loading, no need for papaparse to do it for us
101	transformHeader: parseWithHeader ? duplicateColumnTransformer() : undefined,	51✔
102	error: e => {
103	throw new Error(e);	×
104	}
105	};
106
107	const result = Papa.parse(csvText, papaparseConfig);	51✔
108	const rows = result.data as any[];	51✔
109
110	const headerRow = result.meta.fields \|\| generateHeader(csvOptions.columnPrefix, firstRow.length);	51✔
111
112	const shape = csvOptions.shape \|\| DEFAULT_CSV_SHAPE;	51!
113	let table: ArrayRowTable \| ObjectRowTable;
114	switch (shape) {	51!
115	case 'object-row-table':
116	table = {	46✔
117	shape: 'object-row-table',
118	data: rows.map(row => (Array.isArray(row) ? convertToObjectRow(row, headerRow) : row))	89,315✔
119	};
120	break;	46✔
121	case 'array-row-table':
122	table = {	4✔
123	shape: 'array-row-table',
124	data: rows.map(row => (Array.isArray(row) ? row : convertToArrayRow(row, headerRow)))	10!
125	};
126	break;	4✔
127	default:
UNCOV 128	throw new Error(shape);	×
129	}
130	table.schema = deduceTableSchema(table!);	50✔
131	return table;	50✔
132	}
133
134	// TODO - support batch size 0 = no batching/single batch?
135	function parseCSVInBatches(
136	asyncIterator:
137	\| AsyncIterable<ArrayBufferLike \| ArrayBufferView>
138	\| Iterable<ArrayBufferLike \| ArrayBufferView>,
139	options?: CSVLoaderOptions
140	): AsyncIterable<TableBatch> {
141	// Papaparse does not support standard batch size handling
142	// TODO - investigate papaparse chunks mode
143	options = {...options};	27✔
144	if (options?.core?.batchSize === 'auto') {	27✔
145	options.core.batchSize = 4000;	24✔
146	}
147
148	// Apps can call the parse method directly, we so apply default options here
149	const csvOptions = {...CSVLoader.options.csv, ...options?.csv};	27✔
150
151	const asyncQueue = new AsyncQueue<TableBatch>();	27✔
152
153	let isFirstRow: boolean = true;	27✔
154	let headerRow: string[] \| null = null;	27✔
155	let tableBatchBuilder: TableBatchBuilder \| null = null;	27✔
156	let schema: Schema \| null = null;	27✔
157
158	const config = {	27✔
159	// dynamicTyping: true, // Convert numbers and boolean values in rows from strings,
160	...csvOptions,
161	header: false, // Unfortunately, header detection is not automatic and does not infer shapes
162	download: false, // We handle loading, no need for papaparse to do it for us
163	// chunkSize is set to 5MB explicitly (same as Papaparse default) due to a bug where the
164	// streaming parser gets stuck if skipEmptyLines and a step callback are both supplied.
165	// See https://github.com/mholt/PapaParse/issues/465
166	chunkSize: 1024 * 1024 * 5,
167	// skipEmptyLines is set to a boolean value if supplied. Greedy is set to true
168	// skipEmptyLines is handled manually given two bugs where the streaming parser gets stuck if
169	// both of the skipEmptyLines and step callback options are provided:
170	// - true doesn't work unless chunkSize is set: https://github.com/mholt/PapaParse/issues/465
171	// - greedy doesn't work: https://github.com/mholt/PapaParse/issues/825
172	skipEmptyLines: false,
173
174	// step is called on every row
175	// eslint-disable-next-line complexity, max-statements
176	step(results) {
177	let row = results.data;	7,867✔
178
179	if (csvOptions.skipEmptyLines === 'greedy') {	7,867✔
180	// Manually reject lines that are empty
181	const collapsedRow = row.flat().join('').trim();	8✔
182	if (collapsedRow === '') {	8✔
183	return;	5✔
184	}
185	} else if (csvOptions.skipEmptyLines === true) {	7,859!
186	row = normalizePapaStreamingRow(row);	7,859✔
187	if (row.length === 1 && row[0] === null) {	7,859✔
188	return;	6✔
189	}
190	}
191	const bytesUsed = results.meta.cursor;	7,856✔
192
193	// Check if we need to save a header row
194	if (isFirstRow && !headerRow) {	7,856✔
195	// Auto detects or can be forced with csvOptions.header
196	const header = csvOptions.header === 'auto' ? isHeaderRow(row) : Boolean(csvOptions.header);	27✔
197	if (header) {	27✔
198	headerRow = row.map(duplicateColumnTransformer());	15✔
199	return;	15✔
200	}
201	}
202
203	// If first data row, we can deduce the schema
204	if (isFirstRow) {	7,841✔
205	isFirstRow = false;	27✔
206	if (!headerRow) {	27✔
207	headerRow = generateHeader(csvOptions.columnPrefix, row.length);	12✔
208	}
209	schema = deduceCSVSchema(row, headerRow);	27✔
210	}
211
212	if (csvOptions.optimizeMemoryUsage) {	7,841!
213	// A workaround to allocate new strings and don't retain pointers to original strings.
214	// https://bugs.chromium.org/p/v8/issues/detail?id=2869
UNCOV 215	row = JSON.parse(JSON.stringify(row));	×
216	}
217
218	const shape = (options as any)?.shape \|\| csvOptions.shape \|\| DEFAULT_CSV_SHAPE;	7,841!
219	if (shape === 'object-row-table' && headerRow && row.length > headerRow.length) {	7,867✔
220	row = convertToPapaObjectRow(row, headerRow);	4✔
221	}
222
223	// Add the row
224	tableBatchBuilder =	7,841✔
225	tableBatchBuilder \|\|	7,868✔
226	new TableBatchBuilder(
227	// @ts-expect-error TODO this is not a proper schema
228	schema,
229	{
230	shape,
231	...(options?.core \|\| {})	28✔
232	}
233	);
234
235	try {	7,867✔
236	tableBatchBuilder.addRow(row);	7,867✔
237	// If a batch has been completed, emit it
238	const batch = tableBatchBuilder && tableBatchBuilder.getFullBatch({bytesUsed});	7,867✔
239	if (batch) {	7,867✔
240	asyncQueue.enqueue(batch);	84✔
241	}
242	} catch (error) {
UNCOV 243	asyncQueue.enqueue(error as Error);	×
244	}
245	},
246
247	// complete is called when all rows have been read
248	complete(results) {
249	try {	27✔
250	const bytesUsed = results.meta.cursor;	27✔
251	// Ensure any final (partial) batch gets emitted
252	const batch = tableBatchBuilder && tableBatchBuilder.getFinalBatch({bytesUsed});	27✔
253	if (batch) {	27✔
254	asyncQueue.enqueue(batch);	25✔
255	}
256	} catch (error) {
257	asyncQueue.enqueue(error as Error);	×
258	}
259
260	asyncQueue.close();	27✔
261	}
262	};
263
264	Papa.parse(toArrayBufferIterator(asyncIterator), config, AsyncIteratorStreamer);	27✔
265
266	// TODO - Does it matter if we return asyncIterable or asyncIterator
267	// return asyncQueue[Symbol.asyncIterator]();
268	return asyncQueue;	27✔
269	}
270
271	/**
272	* Checks if a certain row is a header row
273	* @param row the row to check
274	* @returns true if the row looks like a header
275	*/
276	function isHeaderRow(row: string[]): boolean {
277	return row && row.every(value => typeof value === 'string');	254✔
278	}
279
280	/**
281	* Reads, parses, and returns the first row of a CSV text
282	* @param csvText the csv text to parse
283	* @returns the first row
284	*/
285	function readFirstRow(csvText: string): any[] {
286	const result = Papa.parse(csvText, {	51✔
287	dynamicTyping: true,
288	preview: 1
289	});
290	return result.data[0];	51✔
291	}
292
293	/**
294	* Creates a transformer that renames duplicate columns. This is needed as Papaparse doesn't handle
295	* duplicate header columns and would use the latest occurrence by default.
296	* See the header option in https://www.papaparse.com/docs#config
297	* @returns a transform function that returns sanitized names for duplicate fields
298	*/
299	function duplicateColumnTransformer(): (column: string) => string {
300	const observedColumns = new Set<string>();	46✔
301	return col => {	46✔
302	let colName = col;	277✔
303	let counter = 1;	277✔
304	while (observedColumns.has(colName)) {	277✔
305	colName = `${col}.${counter}`;	45✔
306	counter++;	45✔
307	}
308	observedColumns.add(colName);	277✔
309	return colName;	277✔
310	};
311	}
312
313	/**
314	* Generates the header of a CSV given a prefix and a column count
315	* @param columnPrefix the columnPrefix to use
316	* @param count the count of column names to generate
317	* @returns an array of column names
318	*/
319	function generateHeader(columnPrefix: string, count: number = 0): string[] {	31✔
320	const headers: string[] = [];	31✔
321	for (let i = 0; i < count; i++) {	31✔
322	headers.push(`${columnPrefix}${i + 1}`);	96✔
323	}
324	return headers;	31✔
325	}
326
327	function normalizePapaStreamingRow(row: unknown[]): unknown[] {
328	return row.map(value => (Array.isArray(value) && value.length === 0 ? null : value));	34,099✔
329	}
330
331	function convertToPapaObjectRow(
332	row: unknown[],
333	headerRow: string[]
334	): {[columnName: string]: unknown} {
335	const objectRow = convertToObjectRow(row, headerRow);	4✔
336	const parsedExtra = row.slice(headerRow.length);	4✔
337	if (parsedExtra.length > 0) {	4!
338	objectRow.__parsed_extra = parsedExtra;	4✔
339	}
340	return objectRow;	4✔
341	}
342
343	function deduceCSVSchema(row, headerRow): Schema {
344	const fields: Schema['fields'] = [];	27✔
345	for (let i = 0; i < row.length; i++) {	27✔
346	const columnName = (headerRow && headerRow[i]) \|\| i;	112!
347	const value = row[i];	112✔
348	switch (typeof value) {	112!
349	case 'number':
350	fields.push({name: String(columnName), type: 'float64', nullable: true});	38✔
351	break;	38✔
352	case 'boolean':
353	fields.push({name: String(columnName), type: 'bool', nullable: true});	×
354	break;	×
355	case 'string':
356	fields.push({name: String(columnName), type: 'utf8', nullable: true});	74✔
357	break;	74✔
358	default:
359	log.warn(`CSV: Unknown column type: ${typeof value}`)();	×
360	fields.push({name: String(columnName), type: 'utf8', nullable: true});	×
361	}
362	}
363	return {	27✔
364	fields,
365	metadata: {
366	'loaders.gl#format': 'csv',
367	'loaders.gl#loader': 'CSVLoader'
368	}
369	};
370	}
371
372	// TODO - remove
373	// type ObjectField = {name: string; index: number; type: any};
374	// type ObjectSchema = {[key: string]: ObjectField} \| ObjectField[];
375
376	// function deduceObjectSchema(row, headerRow): ObjectSchema {
377	// const schema: ObjectSchema = headerRow ? {} : [];
378	// for (let i = 0; i < row.length; i++) {
379	// const columnName = (headerRow && headerRow[i]) \|\| i;
380	// const value = row[i];
381	// switch (typeof value) {
382	// case 'number':
383	// case 'boolean':
384	// // TODO - booleans could be handled differently...
385	// schema[columnName] = {name: String(columnName), index: i, type: Float32Array};
386	// break;
387	// case 'string':
388	// default:
389	// schema[columnName] = {name: String(columnName), index: i, type: Array};
390	// // We currently only handle numeric rows
391	// // TODO we could offer a function to map strings to numbers?
392	// }
393	// }
394	// return schema;
395	// }

visgl / loaders.gl / 24153816851

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous