Dataset Format Converter

Turn any Apify dataset, or a JSON list, into the exact file the next system needs, in one run: CSV in the right style for UK, US or European Excel, Salesforce, HubSpot or Airtable; SQL scripts for seven databases with column types worked out; a Microsoft Dataverse table definition and import file; BigQuery load files; maps; calendars; feeds; sitemaps; real .xlsx workbooks; contact cards; search, time-series and document-database load files; fixed-width text for legacy importers; Parquet for data lakes; shopping feeds for Google, Meta and Shopify; bank-statement imports for Xero, QuickBooks, Sage and FreeAgent; and Teams and Slack cards. 44 targets.

It is deterministic: field mapping is explicit (or everything is flattened), so the same input always gives the same files and you can check them. Nothing goes to an AI service. Every format is read back by the test suite: CSV is parsed, JSON parsed, XML checked for well-formedness, the SQLite script is executed, and calendars pass an RFC 5545 validator 4.

Targets

Target	File(s)	What's special	For
csv	output.csv	Plain RFC 4180 CSV (UTF-8, ISO dates or csvDateFormat)	Any tool, Google Sheets, Power BI
tsv	output.tsv	Tab-separated	Databases, command-line tools
excel-uk-csv	output-excel-uk.csv	BOM, dd/mm/yyyy dates, formula-injection guard	Excel (UK)
excel-us-csv	output-excel-us.csv	BOM, mm/dd/yyyy dates, formula guard	Excel (US)
excel-eu-csv	output-excel-eu.csv	BOM, ; separator, decimal commas, formula guard	Excel (most of Europe)
salesforce-csv	output-salesforce.csv	UTF-8, ISO dates, UTC date-times, formula guard	Salesforce Data Import Wizard / Data Loader
hubspot-csv	output-hubspot.csv	UTF-8, ISO dates, UTC date-times, formula guard	HubSpot imports
airtable-csv	output-airtable.csv	UTF-8, ISO dates, formula guard	Airtable CSV import
dataverse	dataverse-import.csv + dataverse-table.json	Import CSV with display-name headers, plus a table definition (column types, schema names with your publisher prefix, lengths)	Microsoft Dataverse / Power Apps / Power Automate
json	output.json	Array of flat objects	APIs, web apps
jsonl	output.jsonl	One object per line	Data pipelines, Elasticsearch, Splunk
bigquery	bigquery-data.jsonl + bigquery-schema.json	Column names made BigQuery-safe, types and modes; exact NUMERICs	Google BigQuery (bq load --source_format=NEWLINE_DELIMITED_JSON)
xml	output.xml	Well-formed XML with safe element names	Legacy systems, ETL tools
yaml	output.yaml	YAML list (JSON-compatible scalars)	Config-driven tools
html	output.html	Accessible HTML table	Email, intranets
markdown	output.md	Markdown table (pipes escaped)	Docs, GitHub, Notion
sql-postgres	output-postgres.sql	CREATE TABLE with inferred types + batched INSERTs	PostgreSQL, Supabase
sql-sqlserver	output-sqlserver.sql	NVARCHAR/BIT/DECIMAL/DATETIMEOFFSET, batches of 1000	SQL Server, Azure SQL
sql-mysql	output-mysql.sql	Backtick identifiers, backslash-safe strings	MySQL, MariaDB
sql-sqlite	output-sqlite.sql	Executed for real in the tests	SQLite
sql-snowflake	output-snowflake.sql	NUMBER/TIMESTAMP_TZ	Snowflake
sql-oracle	output-oracle.sql	VARCHAR2/CLOB, TO_TIMESTAMP_TZ, one INSERT per row	Oracle
geojson	output.geojson	FeatureCollection of points (geoMapping)	QGIS, Mapbox, Leaflet, ArcGIS
kml	output.kml	Placemarks with ExtendedData (geoMapping)	Google Earth, Google My Maps
ics	output.ics	Events (icsMapping), validated against RFC 5545 before saving	Outlook, Google Calendar, Apple Calendar
rss	feed.xml	RSS 2.0 feed (rssMapping)	Feed readers, Slack/Teams RSS apps
sitemap	sitemap.xml	XML sitemap of URLs (sitemapMapping)	Search engines
xlsx	output.xlsx	Real Excel workbook: numbers, true/false and dates stored as typed cells, bold frozen header row, filter on, column widths set; formula-like text stays text; byte-identical for the same input 11	Excel, SharePoint, Teams, Google Sheets, Power BI
vcard	contacts.vcf	vCard 4.0 contacts (vcardMapping), escaped and line-folded per RFC 6350 6	Outlook, Google Contacts, iPhone, CRMs
gpx	output.gpx	GPX 1.1 waypoints (geoMapping) 7	GPS units, Garmin, Strava, field-survey apps, QGIS
elasticsearch-bulk	elasticsearch-bulk.ndjson	_bulk body: an index action line before each document, index = tableName in lower case 8	Elasticsearch, OpenSearch, Kibana (POST _bulk)
influx-line	influx.lp	Line protocol (influxMapping): tags and keys escaped, integers marked i, nanosecond timestamps 9	InfluxDB, Telegraf, QuestDB and other time-series stores (meter and SCADA data)
mongodb-jsonl	mongodb.jsonl	Extended JSON (relaxed): date-times become {"$date": ...} so they load as real dates 10	MongoDB (mongoimport), Atlas, Cosmos DB for MongoDB
fixed-width	output-fixed-width.txt + fixed-width-layout.json	Space-padded columns (numbers right-aligned), CRLF, plus a layout file giving each field's start, width and type	Mainframe, COBOL, banking and utility billing importers
parquet	output.parquet	Apache Parquet: typed columns (string, int64, double, boolean, date, UTC timestamp), GZIP pages; read back with pyarrow and DuckDB 12	Microsoft Fabric, Databricks, Snowflake, BigQuery, Athena, DuckDB, pandas

Platform presets

Target	File	What it does	For
google-merchant	google-merchant.tsv	Product feed (productMapping): 15.00 GBP prices, in_stock/out_of_stock/preorder/backorder, sale price from a higher wasPrice, GTIN check digits verified, identifier_exists set when there's no GTIN or brand + MPN, Google's length limits applied, commas in URLs encoded 13	Google Merchant Center (Shopping ads, free listings)
meta-catalog	meta-catalog.csv	Catalogue feed (productMapping): in stock/out of stock, up to 20 extra images within 2,000 characters, Meta's length limits; warns when brand is missing 14	Meta Commerce Manager (Facebook and Instagram shops, Advantage+ catalogue ads)
shopify-products	shopify-products.csv	Product import (productMapping): unique handles from titles, HTML body, compare-at price from wasPrice, extra images as extra rows, out-of-stock items as drafts 15	Shopify admin > Products > Import
xero-bank-csv	xero-bank-statement.csv	Date, Amount (money out negative), Payee, Description, Reference; UK or US date order (bankDateStyle) 18	Xero bank statement import
quickbooks-bank-csv	quickbooks-bank-statement.csv	3-column Date, Description, Amount 19	QuickBooks Online bank upload
sage-bank-csv	sage-bank-statement.csv	Date, Description, Amount in that order, a description on every row, plus Reference and Payee name 16	Sage Accounting bank import
freeagent-bank-csv	freeagent-bank-statement.csv	No header; dd/mm/yyyy, amount to 2 decimal places, description with no commas, quotes or line breaks 17	FreeAgent bank statement upload
teams-card	teams-card.json	Adaptive Card 1.4 message (cardMapping): up to 20 items with links and facts; warns over Teams' 28 KB limit 21	Teams incoming webhooks and Workflows, Power Automate
slack-blocks	slack-blocks.json	Block Kit message (cardMapping): header, up to 20 linked items with fields, within Slack's block and text limits 20	Slack incoming webhooks, chat.postMessage

Platform presets load straight into the platform, so they don't add the spreadsheet formula guard (which would end up stored in the data). Item rows the platform would reject (no id, zero price, relative links, impossible dates) are skipped and counted in the warnings. Layouts were checked against each platform's help pages on 3 October 2026; platforms change their import screens, so check a small file first.

All spreadsheet targets except csv and tsv prefix cells that start with =, +, - or @ with an apostrophe, so a value can't run as a formula when opened (CSV injection) 5.

Input

{
  "items": [
    {
      "name": "Example site A",
      "site": {
        "region": "West Midlands",
        "lat": 52.4862,
        "lon": -1.8904
      },
      "start": "2026-11-02",
      "tags": [
        "solar",
        "battery"
      ],
      "capacityMw": 12.5,
      "live": true,
      "url": "https://example.com/a"
    },
    {
      "name": "Example site B, phase 2",
      "site": {
        "region": "East of England",
        "lat": 52.2053,
        "lon": 0.1218
      },
      "start": "2027-01-15",
      "tags": [
        "wind"
      ],
      "capacityMw": 40,
      "live": false,
      "url": "https://example.com/b"
    },
    {
      "name": "=Example site C",
      "site": {
        "region": "London",
        "lat": 51.5072,
        "lon": -0.1276
      },
      "start": "2026-12-01T09:30:00Z",
      "tags": [],
      "capacityMw": null,
      "live": true,
      "url": "https://example.com/c"
    }
  ],
  "targets": [
    "excel-uk-csv",
    "sql-postgres",
    "dataverse",
    "ics",
    "geojson",
    "xlsx"
  ],
  "tableName": "sites",
  "icsMapping": {
    "summary": "name",
    "start": "start",
    "location": "site.region"
  },
  "geoMapping": {
    "lat": "site.lat",
    "lon": "site.lon",
    "name": "name"
  }
}

• Source: items (a JSON list) or datasetId (any Apify dataset this run's token can read 1; up to 20,000 items).
• targets: one or more from the table.
• fields (optional): [{"from": "site.region", "to": "Region", "type": "string"}]. Paths use dots for nesting and numbers for list positions (tags.0). Without fields, nested objects are flattened to dotted columns, lists of plain values are joined with arrayJoiner (default "; "), and lists of objects become JSON text.
• Types are inferred per column (boolean, integer, number, date, date-time, string) and drive SQL, Dataverse and BigQuery column types. Override them with fields[].type.
• Mappings for the formats that need them (each value is a field path):
  • icsMapping: {summary, start, end?, description?, location?, uid?}
  • geoMapping (geojson, kml, gpx): {lat, lon, name?}
  • rssMapping: {title, link, channelLink, description?, date?, channelTitle?, channelDescription?}
  • sitemapMapping: {loc, lastmod?}
  • vcardMapping: {fn, org?, title?, email?, tel?, url?, address?, note?} (a list of emails or phone numbers gives one line each)
  • productMapping (shopping feeds): {id, title, price, link, description?, imageLink?, additionalImages?, availability?, availabilityDate?, condition?, brand?, gtin?, mpn?, sku?, productType?, wasPrice?}, plus currency (ISO 4217, default GBP). Prices may be numbers or text such as £1,299.00; availability may be true/false, InStock or a schema.org URL.
  • bankMapping (bank statements): {date, amount, description?, payee?, reference?}. Dates are ISO or dd/mm/yyyy; amounts are signed (money out negative) or in brackets. bankDateStyle: uk (default) or us.
  • cardMapping (Teams and Slack): {title, text?, link?, fields?}, where fields is a comma-separated list of up to 10 columns shown as facts.
  • influxMapping: {time, tags?}. time is an ISO timestamp with Z or an offset; tags is a comma-separated list of columns to index as tags (every other column becomes a field). The measurement is tableName.
• Energy add-on: "enrich": {"settlementPeriodFrom": "timestamp"} adds GB electricity settlementDate and settlementPeriod columns (46, 48 or 50 periods a day around clock changes), so half-hourly data lines up with Elexon and supplier settlement data.

Output

Each file is saved in the run's key-value store under the name shown. The SCHEMA record holds the inferred columns, and the dataset gets one row per file (where it is, rows, bytes, warnings). The first row for the input above:

{
  "target": "excel-uk-csv",
  "key": "output-excel-uk.csv",
  "contentType": "text/csv; charset=utf-8",
  "rows": 3,
  "columns": 9,
  "bytes": 369,
  "url": null,
  "warnings": []
}

Warnings say when rows were skipped (for example, missing coordinates for a map) and why.

Limits

• Up to 20,000 items and 1,000 columns per run.
• SQL scripts are for loading data; review the inferred types before production use.
• Dataverse: create the table from dataverse-table.json (Power Apps > Tables > New table), then import dataverse-import.csv (Import > Import data from Excel/CSV, or a dataflow) 2.
• Excel cells hold at most 32,767 characters; longer text is cut, with a warning. Date-times go into .xlsx in UTC.
• Oracle string literals over 4,000 characters need loading another way (the column becomes CLOB).

Pricing

Pay per event: $5 per 1,000 results, plus Apify's tiny per-run start fee ($0.00005). See the Pricing tab for the current price.

Local use

With Node.js 22 or later, run node cli.js examples/input.json (writes files to ./out), or npm test.

Sources

1. Apify: Get dataset items (API). Accessed 3 October 2026.
2. Microsoft Learn: Import data into Dataverse (Power Apps). Accessed 3 October 2026.
3. RFC 4180: Common Format and MIME Type for CSV Files. Accessed 3 October 2026.
4. RFC 5545: iCalendar. Accessed 3 October 2026.
5. OWASP: CSV Injection. Accessed 3 October 2026.
6. RFC 6350: vCard Format Specification. Accessed 3 October 2026.
7. GPX 1.1 Schema Documentation. Accessed 3 October 2026.
8. Elastic: Bulk API. Accessed 3 October 2026.
9. InfluxData: Line protocol (InfluxDB v2). Accessed 3 October 2026.
10. MongoDB: Extended JSON. Accessed 3 October 2026.
11. Ecma International: ECMA-376 Office Open XML File Formats. Accessed 3 October 2026.
12. Apache Parquet: File format. Accessed 3 October 2026.
13. Google Merchant Center: Product data specification. Accessed 3 October 2026.
14. Meta: Catalog fields reference. Accessed 3 October 2026.
15. Shopify Help Center: Product CSV columns. Accessed 3 October 2026.
16. Sage: Supported file formats for bank statement imports. Accessed 3 October 2026.
17. FreeAgent: Format a CSV file to upload a bank statement. Accessed 3 October 2026.
18. Xero Central: Import a bank statement. Accessed 3 October 2026.
19. QuickBooks: Format CSV files to get bank transactions into QuickBooks. Accessed 3 October 2026.
20. Slack: Block Kit blocks reference. Accessed 3 October 2026.
21. Microsoft Learn: Create an Incoming Webhook (Teams). Accessed 3 October 2026.