Contact

LibreOffice Calc · Data preparation · Intermediate

Preview Data Provider transformations in LibreOffice Calc

Prepare a local CSV Data Provider import, check transformation order and distinguish a documented refresh command from verified refresh behavior.

Documentation-qualified; native import, refresh and persistence remain untested · Editorial standards

Calc’s Data Provider is a separate import-and-transform workflow, not another name for a registered Base source. Use this guide to prepare a small local-file test, inspect its preview and decide whether the result is safe to load. It does not establish a production refresh pipeline.

Scope and stopping conditions

The menu workflow follows the free desktop Calc Guide 26.2. Supplemental Help was labelled 26.8 when checked on 23 October 2026; those details are identified below rather than assumed to prove identical behavior in every 26.2 installation. English Windows/Linux menu names are used. No native Data Provider import, refresh, save/reopen test or screenshot was performed for this article.

Work in a disposable ODS, not a production workbook. If Data > Data Provider is absent, stop this exercise. Do not install components, change expert settings or enable experimental features to reveal it. The cited provider pages do not establish that every build exposes the command, or that an experimental setting is universally required. Availability is a separate check from correctness.

Choose the correct import owner

For a one-time import with delimiter choices, use CSV import. For a registered table or query, use the Base-source guide. For the separate external-table link workflow, use linked external data.

Data Provider’s distinctive job here is transforming source data before it reaches a predefined database range. Guide 26.2 documents CSV, HTML and XML input, a local-file path or source URL, and a transformation list whose order cannot be rearranged directly. This example uses only a local CSV: no account, website scraping, authentication or remote API is involved.

Prepare a small, auditable fixture

Save a working copy of the existing sales-by-region CSV as provider-sales.csv. It has three columns, Region, Channel and Amount, plus eight records. Do not edit the supplied original.

The eight amounts are 120, 80, 90, 140, 60, 30, 40 and 20. Their independent arithmetic total is 580. North totals 300 and South totals 280. These are fixture checks, not observed Calc execution results.

In a new scratch workbook, create an empty sheet named ProviderTest. Reserve A1:C20 exclusively for this exercise and define it as database range ProviderScratch through Data > Define Range. See named and database ranges if the distinction is unfamiliar. Keep notes and checking formulas outside the target; do not assume refresh preserves anything written inside it.

Open Data > Data Provider, choose ProviderScratch, select CSV and supply the full path to your local working copy. The Identifier field is documented for XPath selection of HTML/XML, not as a CSV delimiter control. Do not paste an XPath into this CSV exercise.

Make the preview earn your trust

First inspect the untransformed data: three separate fields, eight records, recognizable headers and numeric amounts. If the file becomes one long column, or numbers appear as unsuitable text, cancel. Resolve the input interpretation before adding transformations; a plausible total cannot repair misaligned records.

For a narrowly scoped trial, select a text transformation that converts the Region values to uppercase. The cited Help 26.8 describes Add for appending a transformation and Apply for previewing it, with OK loading the sheet. Use those controls only if they match your installed dialog.

Do not guess whether a column-index field begins at zero or one: the cited user documentation does not settle that contract. Verify against the preview that Region changes to NORTH/SOUTH while Channel and Amount stay unchanged. If the mapping cannot be established safely, cancel and leave this action untested. The header may also change case; check it explicitly rather than silently assuming header exclusion.

Expected invariants are eight records and the same eight amounts, still totaling 580. A text-only transformation must not change the monetary values. If a proposed second operation deletes or splits columns, reconsider the indexes used by later operations. Because sequence matters, rebuild an incorrectly ordered list instead of treating transformations as interchangeable.

Load once, then evaluate refresh separately

Only confirm OK after the preview passes. Inspect the destination’s record count, row alignment and amounts again. Save a separate checkpoint before trying any update.

Guide 26.2 names a Refresh Data Provider tool. It does not supply enough detail here to promise automatic timing, cross-session retention, row-shrink cleanup or a universal menu location. Do not substitute Refresh Range from the registered Base workflow and call that a provider refresh test.

For a controlled future test, change just the working CSV’s first Amount from 120 to 125. Expected total: 585; North: 305; South: 280. Use the installed provider refresh command only when positively identified. Recheck every record, then save, close and reopen a copy to test retention separately. A workbook that opens showing old values has not demonstrated a working connection. Native refresh and persistence remain unverified in this article.

Diagnose without broadening access

  • No database range in the list: confirm that the range was defined in this workbook, not merely given a cell-name label.
  • Wrong column changes: remove the transformation and check the index-to-preview mapping. Do not load uncertain results.
  • A refresh appears unchanged: verify the exact local path and that the edited file was saved. Recalculation alone is not evidence that the file was reread.
  • Results differ after a structural change: restore the scratch checkpoint and retest with the original three-column schema. Do not repair production data by repeatedly guessing transformations.

Sources and evidence boundary

Calc Guide 26.2, Chapter 15 supplies the provider workflow and separate refresh-tool reference. Data Provider Help, labelled 26.8 supplies the Apply/OK distinction and text-transformation description. Advanced options Help, labelled 26.8 explains that experimental-feature availability varies by version; this guide makes no settings changes. Independent CSV arithmetic does not establish native compatibility.