CatalogOps — Shopify CSV preflight

Prepare supplier CSVs for simple Shopify products and inspect price changes and blank fields before importing. This independent tool is not affiliated with Shopify. Version 0.2.2 adds pay-per-event billing. The preparation engine in version 0.2.1 passed six private Apify cloud runs with synthetic data, including 100 and 1,000 rows and rejection of empty or over-limit input; version 0.2.2 also passed prepared/rejected private cloud checks before pricing was enabled. PPE pricing is configured at $0.05 per prepared file. Three private cloud checks of version 0.2.2 verified one prepared-file event after successful preparation, no event on rejected input, and no event when maxChargeEvents=0 rejected preparation. These owner-account checks are not external sales. No external payment or Shopify import has been verified.

Input

Paste UTF-8 CSV into supplierCsv, declare target-to-source columns in mapping, and include baselineCsv from Shopify when preparing updates. The prefilled two-product example is synthetic. No store credentials are required.

The baseline must identify each simple product once and include default Option1 name/value. Every mapped field for a matching product must be present in the baseline; a missing column is unknown and rejects preparation. Repeated handles, extra image rows, variants, linked options and unknown default options are rejected. Without a baseline, use only for products you have confirmed are new; existing variants cannot be verified.

Mapping requires a handle, title and price. Both current and supported legacy Shopify target names work. Output uses current headings plus default option columns. Mapped blank values that would erase baseline data reject the entire file. For an optional column you intend to preserve, omit it from mapping. SKU/barcode strings retain leading zeros. Decimal separator is explicit; NaN, infinity, currency symbols, thousands separators and fractional precision above two decimals are rejected. Status, when mapped, must be active, draft or archived. Handles are restricted to ASCII letters, digits and dashes.

Results

Read OUTPUT first: status is prepared or rejected, and preparedKey is null on rejection. report.json includes validation issues and, when the input reaches product comparison, before/after field changes and input hashes. Structural input or limit failures produce a short rejection report. prepared.csv exists only on success. A run can finish successfully with a rejected business input; use the manifest status to decide whether to import.

Successful preparation also saves one summary row in the default dataset before requesting any prepared-file charge. It contains status=prepared, the run ID, record count, storage keys for the manifest/report/CSV, and input/result hashes; it does not contain the full CSV or assert payment. Read OUTPUT.billingStatus and the platform charge record for billing. A fresh rejected run has no dataset row. Resurrecting the same run reuses an identical summary instead of appending another row; a conflicting dataset fails for manual reconciliation. The dataset summary adds no charge event.

The output has no claim of guaranteed import acceptance. Review changes and your store settings before importing. CSV text may be interpreted as formulas by spreadsheet applications. Inspect JSON or import the CSV directly instead of opening untrusted product data in a spreadsheet.

Limits

Each CSV defaults to 1 MB and 1,000 rows; hard ceilings are 5 MB and 1,000 rows. The mapping is at most 20 KB. Total generated output is capped at 20 MB. Preparation defaults to 15 seconds with a 30-second ceiling. The adapter allows 90 seconds including storage; set the platform run timeout to 120 seconds. Memory is fixed to 512 MB in the Actor definition. Preparation retries are zero; SDK storage may retry within the overall timeout.

Actor event price: $0.05 per prepared file, up to 1,000 rows per input CSV. PPE pricing is configured and the successful-preparation and rejection paths have been checked in private cloud runs. The Store pricing displayed for your run is authoritative. One prepared-file event is requested only after your CSV and report are saved and accessible. Rejected inputs have no prepared-file charge. No startup or dataset events are part of this offer. The default maxChargeEvents=1 permits at most one event; setting it to 0 rejects preparation in PPE mode. Your platform spending limit must allow one event. Outside PPE mode, preparation does not request a charge; standard cloud resource usage may apply.

OUTPUT.billingStatus records charged, not_charged, not_ppe, pending, uncertain or limit_reached. chargeEvents is 0, 1, or null when the result is unresolved. On an interrupted run, consult the platform charge record; do not interpret a pending receipt as payment confirmation. Already delivered files remain accessible. Resurrecting the same run does not request a second charge; unresolved charges require reconciliation. There is no proxy, AI API, web crawling, external URL download or Shopify write.

Inventory, image availability, Markets pricing, metafields, custom fulfillment and additional Shopify import requirements are outside this version's checks. Do not send customer/order CSVs or credentials. Input and output are stored in the run's Apify storage; access and retention follow that account's settings.