Product feed tools with an API and regex rules

What 'API' means in product feed management

Source API: the tool pulls your catalog from an endpoint instead of a file. Textfinity reads a CSV or TSV, or XML in Google Shopping's RSS or Atom shape, from an HTTPS or SFTP URL, so a platform's CSV export endpoint, a PIM's Google feed or a custom script's CSV output all work as a shop source, alongside the native Shopify connection and Google Sheets. It does not read JSON: a JSON API response has to be flattened to a CSV (one row per variant) first, and a JSON URL gets a prompt to export CSV instead. A merged source joins a second file or sheet onto the catalog by key to add columns the main export lacks.

Channel API: the tool delivers through the channel's own interface rather than a file. Textfinity does not push through channel APIs: Google Merchant Center and Meta Commerce Manager fetch the published feed URL on a schedule, and we add that URL to the merchant's account with them. TikTok, Pinterest, Microsoft, Snapchat and custom feeds for channels such as Idealo fetch the feed URL the same way; Amazon and eBay files are downloaded and uploaded by the merchant.

Programmatic control: something other than a person drives the tool, creating feeds, changing rules, reading output. This is what developers usually mean by 'feed management with API', and it is the one to be explicit about: Textfinity has no REST API. What it has is an MCP server, available to clients: an AI assistant that speaks MCP, such as Claude, connects with an API token or OAuth, and the assistant can list organizations, shops and feeds, search products, create a feed, add, update or delete mapping rules, add filter rules, run feed hygiene and read a feed's output. Around that, the input side is any URL or sheet your systems write to and the output side is a stable feed URL any system can fetch. If a deployment pipeline must call HTTP endpoints directly, Textfinity is not that tool today.

What regex rules in a feed tool should do

Four capabilities cover almost every cleanup. Remove: delete everything a pattern matches, such as HTML tags or trailing SKU codes. Replace: rewrite matches, with capture groups, such as normalizing units or reordering words. Match as a condition: include or exclude products whose value fits a pattern, or apply a mapping rule only when it does. Match against a list of patterns: one rule, many brands or category codes. Textfinity implements these as the remove from regexp and replace from regexp transforms, the matches regexp and does not match regexp filter operators, is in list (regexp) and is not in list (regexp), and a matches regexp condition on any mapping rule. The mapping condition is anchored at the start of the value (begin the pattern with .* to match mid-value); the filter operator is not.

What regex cannot do well is logic across columns: 'if stock is zero write out of stock, else in stock' or 'brand plus title plus size unless size is empty'. That is the job of the advanced DSL, which references attributes in brackets, concatenates with +, and supports nested if/then/else with string tests (equals, not equals, includes, is empty), numeric comparisons on calc{} values, AND, OR and parentheses, plus calc{} arithmetic, round(), and now(), date(), time() and isoformat() for date fields.

Regex product feed examples

Each line is a remove from regexp or replace from regexp transform on one column. Patterns are standard; test on a sample before applying to 5,000 rows.

strip HTML from description      remove:  <[^>]+>
collapse whitespace              replace: \s{2,}            -> " "
drop trailing SKU in title       remove:  \s*\(?[A-Z]{2,}-\d{3,}\)?$
normalize 'cm' spacing           replace: (\d+)\s*cm         -> "\1 cm"
extract size from title          replace: ^.*\b(XS|S|M|L|XL)\b.*$ -> "\1"
fix double currency              replace: ^(\d+\.\d{2}) (EUR|USD) \2$ -> "\1 \2"
filter: valid GTIN only          include if [gtin] matches regexp ^\d{8}(\d{4,6})?$
filter: brand list by pattern    include if [brand] is in list (regexp) ^(Nike|Adidas|Puma)$

DSL examples where regex runs out

The advanced transform is where cross-column logic lives. These are expressions for the advanced transform on a column.

availability:
  if calc{[stock]} > 0 then "in_stock" else "out_of_stock"

title with optional size:
  if [size] is empty then [brand] + " " + [title]
  else [brand] + " " + [title] + ", " + [size]

custom_label_0 by sale flag and margin band:
  if [price] is empty then ""
  else if [is_sale] equals "true" then "sale"
  else if [margin_band] equals "high" then "hero" else "standard"

sale_price_effective_date (end column carries its time):
  isoformat([sale_start]) + "/" + isoformat([sale_end])

Choosing a feed tool when the API question is real

Be honest about which API you need. If the answer is 'my platform has an export endpoint and the channel needs a file on a URL', almost any feed tool qualifies and the regex and rule depth is what to compare; Textfinity covers that with EU hosting, set up for you and quoted per catalog. If the answer is 'my deployment pipeline must create and rebuild feeds programmatically', ask vendors for a REST API reference before the demo; Textfinity has none. If the answer is 'I want to set up and change feeds by asking an assistant', Textfinity's MCP server, available to clients, covers feed creation, mapping and filter rules, hygiene runs and reading output.

Then test the rules on your ugliest data. A tool that offers regex but no capture groups, or conditions but no else branch, will push you back to spreadsheets within a month. Run the examples above on a 100-product sample before you commit; on Textfinity we run them on your own catalog during setup, and feed hygiene tells you whether the cleanups actually cleaned.