Field processing pipeline nodes

A field processing pipeline builds one field’s value, node by node, before it is written: into the product in an import, into the feed’s column in a channel. It has 26 nodes and 2 sources, and all of them are offered in imports and in channel feeds alike. This page lists every one with its settings, and says how to test a node before you run an import or a feed.

How a pipeline works

A pipeline starts with a source: a Source field, which reads a field of the record, or a Static value. Each node after it takes the values connected into it and gives one value on. The value of the last node is what is written. When a node has nothing to give, it gives No output, and the field is left as it is — see No output and Empty value.

To open a field’s pipeline, use the arrow button beside the field. Pointing at it shows Edit pipeline; the number on it counts the field’s nodes. The button is in three places:

Where	The field’s arrow
A channel’s feed	On each row of the feed, between the source and the output column
A data source	The dataset’s Mapping tab, at the left of a mapped field. It is disabled until the row has a field to map into
An import from the product list	The Map columns step, at the left of each row

An empty pipeline offers Add source. In a pipeline, point at a node: a + after it adds the next node, a + before it adds a node in front, and Add source beside it adds another input. Each + opens Add node, with a search box and the categories down the side: Text, Numbers, Boolean, Date & time, Lists, Assets and Code. A node can be in more than one category. Add source opens the same panel with Sources only. Select a node to open its settings, with its Preview below them.

A node that isn’t complete is refused when you save a channel’s feed or a data source’s dataset: it is marked This node isn’t complete., with the settings it still needs marked in red, and the field’s row says A node in this pipeline isn’t complete. Select Edit pipeline and fix the settings marked in red.

No output and Empty value

These are two different results, and in an import they do opposite things:

A node gives	In an import	In a feed
No output	The field’s stored value is left unchanged	The record has no value for the field
Empty value	The field’s stored value is cleared	The column is empty; a JSON feed writes null

Two nodes let you choose what they give when they have nothing to give: Pick item from list, under Output if no item at this index, and Replace text, under Output if no match. Both show the hint In an import, No output leaves the field’s stored value unchanged, and Empty value clears it. A Source field gives No output for a field the record doesn’t have, unless Return empty value if field is missing is ticked.

To see which of the two a node gives, before you import anything, test it in the Preview.

Sources

Added with Add source only — they start a pipeline, or give a node another input.

Source	What it does	Settings: choices (default)
Source field	Reads the value of a field of the record	Source field: the record’s fields. Return empty value if field is missing (off), with the hint In an import, this clears the field’s stored value instead of leaving it unchanged.
Static value	Gives a fixed value	Value (empty)

Nodes

In alphabetical order, as Add node sorts them. A setting in bold is the label on screen; the default is in brackets.

Node	Categories	What it does	Settings: choices (default)
Asset to public URL	Assets	Gives the public URL of an asset’s file	Image variant: Original (key default), Original (key original), then your image variants by name (Original (key default)). Both Original options give the same file under a different URL.
Build price	Numbers	Makes one price from columns or fixed values, and finds or creates its price list	Nine fields, each a value source: Amount (Constant value 0) · Sale price · Currency code · Valid from · Valid until · Price list name · Price list external ID · Price list external ID system (Constant value erp) · Price list type. Amount, Currency code and Price list name are required, and without all three it makes no price. An amount of 0 is a price. It needs at least one input connected, even when every field is a constant value.
Calculate	Numbers	Adds, subtracts, multiplies or divides the inputs, or applies a formula	Operation: Add · Subtract · Multiply · Divide · Use formula (Add). Value (0): one more number the operation applies, after the inputs — the node sums itself up as a + b + … + 0. With Multiply or Divide, set Value to 1 to calculate with the inputs alone. With Use formula, Formula, in which the inputs are a, b, c and so on, in input order. Subtract and Divide start from the first input.
Change text case	Text	Changes the text’s case	Case (Uppercase): Uppercase · Lowercase · Title Case · Sentence case · camel · PascalCase · snake · UP · kebab-case · Train-Case · SCREAMING-KEBAB-CASE · Capitalize Words · Capitalize first letter · Invert case · Altern
Clean up HTML	Text	Tidies HTML markup	Remove element attributes, such as class or style (off). It keeps src and alt, so an image still shows.
Combine yes/no values	Boolean	Gives one yes/no value from several	Operation: All inputs must be true · One or more inputs must be true · Exactly one input must be true (All inputs must be true). Negate (off)
Convert unit	Numbers	Converts a number from one unit to another, such as cm to mm	See Measurement nodes
Extract part of text	Text	Takes part of the text by position and length	First character (1) · Length (1). Characters count from 1; a negative First character counts from the end, so -1 is the last character. A Length of 0 takes the rest of the text, and a negative one takes the characters before First character.
JavaScript	Code	Runs your own function on the value	Script: a new node starts with function convert(input, record) { return input; }. See JavaScript node
Join lists	Lists	Joins several lists into one list	None. It skips inputs that give No output or Empty value. When every input gives No output, it gives No output; when an input gives Empty value and none gives an item, it gives an empty list, which an import stores as cleared.
Join texts	Text	Writes the inputs one after another, with a separator, or fills a pattern with them	Separator (,) · Maximum length (empty): an input is added only while the text stays shorter · Default value (empty). Under Pattern: Pattern (empty), in which {0} is the first input, {1} the second, as in Hello {0}! · Apply when the inputs are empty (off)
Limit to range	Numbers	Keeps the number between a minimum and a maximum	Minimum value · Maximum value (both empty). The minimum must be below the maximum.
Measurements to text	Text, Numbers	Writes measurements and their unit as text, such as 20 x 10 x 5 cm	See Measurement nodes
Number to text	Text, Numbers	Writes a number as text	Number of decimals (2) · Decimal separator (.) · Thousands separator (empty) · Insert before number · Insert after number (empty). The panel shows an example output.
Pick input as text	Text, Lists	Takes one of its inputs, by position, as text	Index (0): the first input is 0 · Default value (empty), given when there is no input at that index. It counts the node’s inputs, not list items — unlike Pick item from list: one list connected to it is one input.
Pick item from list	Lists	Joins all inputs into one list and takes one item, such as the first image	Index (0): the first item is 0 · Output if no item at this index: No output · Empty value · Specific value (No output); Output value for Specific value. It skips inputs that give No output or Empty value, so they take no position; empty text is an item.
Regular expression	Text, Code	Matches the text with a regular expression and replaces the match	Pattern (required), without the enclosing slashes, such as [0-9a-z] or \s · Replacement (empty), where $1, $2 and so on insert the match’s groups, counted from 1 · Case-insensitive matching (off) · Return an empty value if the pattern doesn’t match (off)
Replace text	Text	Replaces one or more texts	One row per replacement: Search, Case sensitive, Match word, Pattern is regular expression, Replacement; Add replacement adds a row, and the replacements can be pasted from and copied to the clipboard. Output if no match: Unchanged input value · No output · Empty value · Specific value (Unchanged input value); Output value for Specific value. Here Empty value gives empty text, which an import also stores as cleared.
Text template	Text, Code	Builds text from a template	Template: sources[0], sources[1] and so on are the inputs, and record. followed by an attribute code reads any field · Per language (on): applies the template for each language. The template language is Scriban; inside the node most export.* functions return nothing, and record._categories isn’t available.
Text to date	Text, Date & time	Reads a date, or a date and time, from text	The format the text is in: ISO standard (the default), such as 2021-10-24 or 2021-10-24T15:30 · RFC standard, such as Sun, 21 Oct 2018 12:16:24 GMT · Language-specific, with Language · Custom pattern, with Pattern. Under Output conversions, Output time zone. Under Empty value handling, Treat as empty before: a date before it gives an empty value. Both are value sources.
Text to hash	Text	Hashes the text	Hash algorithm: MD5 · SHA-1 · SHA-256 · SHA-512 (MD5)
Text to measurements	Text, Numbers	Reads numbers and their unit from text, such as 20x10x5 cm	See Measurement nodes
Text to number	Numbers	Reads a number from text	Decimal separator (.). When the text holds more than the number, it reads the first run of digits, spaces, points, commas and minus signs, so 12 kg gives 12. When that run isn’t a number on its own, as in ca. 12 kg or 10 - 20 cm, it gives No output.
Text to yes/no	Text, Boolean	Tests the text against a condition and outputs one value if it holds and another if it doesn’t	Operator: = (equals) · != (not equal) · contains · does not contain · starts with · does not start with · ends with · does not end with · is empty · is not empty · matches regex (= (equals)) · Value (empty) · Case sensitive (off). Under Output: True - condition matches and False - no match
Trim whitespace	Text	Removes whitespace at the start or end, or replaces duplicate whitespace with one space	Remove whitespace at the start (on) · Remove whitespace at the end (on) · Replace duplicate whitespace with one space (off)
Yes/no to text	Text, Boolean	Writes a yes/no value as text	Output if yes (1) · Output if no (0) · Output if empty (0). An empty input gives the text of Output if no; Output if empty doesn’t change that.

Value sources

Each field of Build price, and Output time zone and Treat as empty before of Text to date, says where its value comes from:

Choice	The value
No value	None
Constant value	What you type or pick beside it
Input value	One of the node’s inputs, by Input number. Here the first input is 1 — unlike Pick item from list, Pick input as text and Join texts, which count from 0

Test a node in the Preview

Preview, under a node’s settings, shows what the node gives for test inputs you type. Nothing you enter there is saved, and nothing is imported or exported. Every node has one except the two sources.

Each input connected to the node has a row, named after the node it comes from; a node with nothing connected has one row, Input. For each row, choose what it gives the node:

Test input	What to enter
Text (the default)	The text. Leave it empty for empty text
List of texts	One item per line. A blank line is no item; leave it all empty for an empty list
No output	Nothing — the input gives no output
Empty value	Nothing — the input gives an empty value
Number, Measurements	On the first row, when its Source field has a unit — see Measurement nodes

A row fed by a Source field also says what that field gives when the record doesn’t have it: If “‹field›” is missing, Source field gives No output. — or gives Empty value. when Return empty value if field is missing is ticked. For a node that reads other fields of the record, such as Text template or JavaScript, add them under Record (optional), one per line: color: red.

Output names what the node gives:

Output	What it says
No output	In an import, this leaves the field’s stored value unchanged.
Empty value	In an import, this clears the field’s stored value.
Empty text	In an import, this clears the field’s stored value, and a JSON export writes it as “”.
Empty list	In an import, this clears the field’s stored value, and a JSON export writes it as [].
List of 3 items	The items, one per line. An empty item reads Empty text or Empty value
The value itself	—

Changing the node’s settings keeps your test inputs, so you can switch Output if no match from No output to Empty value and watch the output change.

The Preview reads numbers and dates with fixed settings: a decimal point, ISO dates, and no account language. It does not use your import’s Data Format or a channel’s settings. So a number with a decimal comma, or a date written the way your language writes it, can come out differently in the Preview than in your import. Type test numbers with a decimal point, and test dates in ISO format, such as 2026-10-24.

When a node can’t build a value

A node that can’t build a value leaves that field out of the record. The rest of the record is written, and the run goes on. In an import the value stored before stays; in a feed the record has no value for the field. That happens when:

• a JavaScript node’s script fails — wherever it is in the pipeline, the nodes after it don’t run. A script whose result can’t be used is the exception: the node after it gets that error as its input, as below;
• a node gets another node’s error as its input, and can’t do anything with it;
• a node gets a value of a type it can’t take. A feed’s log entry begins Couldn’t use a value because a node can’t take it. Check the source data and the node’s settings.; an import’s says This value couldn’t be converted for this field. Check the value in the file and the field’s mapping, and import again.

The run’s step counts them. An import’s step ends, for example, products.csv: imported — 4,812 records created; 12 values that the field processing pipeline couldn’t build — see the log file. A feed’s step reads Created feed “‹feed›” with warnings. 1,200 records written, 12 values that the field processing pipeline couldn’t build — see the log file for details.

The step’s log file has an entry per value, naming the field, and the record: by its row in an import, by its identifier in a feed. Its message says why — see reading a log file. The measurement nodes’ own messages are listed on Measurement nodes.

Related articles

• JavaScript node — the script, its limits, and every message about a failed script
• Measurement nodes — read, convert and write measurements and their units
• Price lists — how Build price finds or creates a price list
• Import or delete products from a file — the import’s Map columns step
• Data sources — a dataset’s Mapping tab
• Custom product feeds — a feed’s fields