Overview
The Explode transform takes array-valued columns and produces one output row per array element. Non-array columns are duplicated on each output row. This is the standard way to unnest arrays from JSON payloads, event batches, or document stores into the flat row shape required by joins, aggregations, and relational destinations.When to Use
- After JSON sources that deliver arrays of items (e.g.
line_items,tags,events) - When you need to aggregate, filter, or join on individual array elements
- Before writing to relational databases that expect one row per entity
How It Works
Example
Input:
Output (explode on
items):
Configuration
Pipeline Patterns
E-commerce Line Items
Event Tags Analysis
Combined with Flatten
When data has both nested objects and arrays:Handling Edge Cases
Tips
- Empty arrays produce zero output rows for that input row — if you need to preserve such rows, add a default array value upstream
- Nested arrays (arrays of arrays) require chaining multiple Explode nodes
- Performance: Explode on high-cardinality arrays (hundreds of elements per row) can significantly increase dataset size — monitor row counts in the pipeline run summary
- Combine with Unique: After Explode, duplicates may appear if the same element exists in multiple source rows — use Unique if needed
Related
Flatten
Flatten nested objects into columns
Row Transforms
Filter, sort, and deduplicate after exploding
Aggregation
Aggregate per-element after exploding arrays
JSON Parser
Parse raw JSON before exploding