Configure a CSV parser

Parsers define how an FTP file becomes instrument readings. The current parser type is CSV, including supported non-comma delimiters; the filename extension does not determine how the content is parsed.

Choose a layout

One data point per row fits a timestamp followed by measurement columns:

timestamp,pressure,temperature
2026-09-04T12:00:00Z,42.5,18.2
2026-09-04T12:05:00Z,42.7,18.3

Configure the timestamp column and map pressure and temperature to variables. Choose the appropriate value type for each mapped column.

One data point across multiple rows fits a file where variable names and values appear in separate rows. Select the variable-name and variable-value columns and the columns that identify one logical data point.

Configure file grammar

Select comma, semicolon, tab, or pipe as the delimiter and dot or comma as the decimal separator. Configure whether a header is present, its row number, and the first data row. All row and column positions shown to users are one-based.

Skipping a preamble can be useful for logger files that include project details or units before the measurements. Do not accidentally map a units row as data.

Parse timestamps

Select the timestamp column or an ordered set of columns. If date and time occupy separate columns, configure the join text between them. Choose RFC3339 or the supported format tokens shown in the editor.

Set the expected timestamp order to oldest-first or newest-first. Ordering mismatches produce diagnostics. The pipeline timezone supplies the context needed by timestamps without an explicit offset; verify the final timestamp in the pipeline preview.

Identify instruments and variables

Use the pipeline’s base instrument name or a data-row column as the instrument-name source. Check the final names in preview. An optional prefix setting combines the detected instrument name with mapped variable names.

For value conversion, choose Decimal, String, or Auto where supported. Auto depends on the sampled content and parser behavior; inspect effective types when mixed content is possible.

Missing values and validation

Enter source-specific missing-value markers one per line. Empty, absent, and whitespace-only cells are always treated as null. Avoid using an ordinary valid measurement as a missing-value marker.

Preview representative files, including a normal row, a missing value, a boundary timestamp, and any alternate instrument name. Review column mapping, converted samples, physical row references, and diagnostics. A preview is a sample, so continue checking processing attempts after enabling the pipeline.

Parsers can be reused by pipelines. An existing parser is read-only inside the pipeline wizard; use Duplicate and edit to make an independent copy for a different file format. Review which pipelines use a parser before changing a shared definition elsewhere.

The parser maps variable names and value types. Configure measurement units on the resulting variables. See FTP ingestion for pipeline setup and file reprocessing.