@reviewpipe/exporter-csv

Flattens a reviewpipe AnalysisResult's themes and trend into a single tidy CSV table. This is a lossy, spreadsheet-friendly view — for the complete, lossless result use @reviewpipe/exporter-json.

import { Pipeline } from '@reviewpipe/core';
import { CsvExporter } from '@reviewpipe/exporter-csv';

const csv = await new Pipeline()
  .source(adapter)
  .provider(provider)
  .export(new CsvExporter())
  .run(input);

Schema

One CSV table with a header row. Every row carries a section column that says what kind of record it is; columns that don't apply to a row are left blank.

| Column | theme rows | trend rows | | ------------------ | -------------------------------------- | ---------------------------------------- | | section | "theme" | "trend" | | label | theme label | — | | keywords | keywords joined by ; | — | | mentions | number of reviews in the theme | — | | sentiment | mean sentiment of the theme, [-1, 1] | — | | exampleReviewIds | example review ids joined by ; | — | | period | — | time bucket, e.g. "2026-01" | | averageSentiment | — | mean sentiment for the bucket, [-1, 1] | | reviewCount | — | number of reviews in the bucket |

Example:

section,label,keywords,mentions,sentiment,exampleReviewIds,period,averageSentiment,reviewCount
theme,shipping,shipping;delivery,3,-0.1,a;b,,,
trend,,,,,,2026-01,0.2,4

Notes:

• List-valued fields (keywords, exampleReviewIds) use ; as the delimiter, exported as LIST_DELIMITER.
• Numbers are written unrounded, so the table round-trips exactly.
• An empty result (no themes, no trend) produces just the header row.