Config selection

The problem

When importing a file you’ve seen before (or a similar one), you shouldn’t have to reconfigure columns from scratch.

Fingerprinting

When data is parsed, Crosswalker generates a fingerprint based on:

• Column names (ordered)
• Column count
• Sample data patterns (if pattern detection is enabled)

This fingerprint is stored with each saved config.

Matching algorithm

When a new file is imported:

1. Generate fingerprint for the new file
2. Score each saved config against the new fingerprint
3. Scoring considers:
   • Column name overlap — How many column names match
   • Column count similarity — Same number of columns
   • Pattern similarity — Similar data shapes (IDs, dates, URLs)
4. Configs scoring above the threshold (default: 50) are suggested

User flow

graph TD
    A[Parse file] --> B[Generate fingerprint]
    B --> C{Matching configs?}
    C -->|Yes, score >= threshold| D[Suggest config]
    C -->|No match| E[Manual configuration]
    D --> F{User accepts?}
    F -->|Yes| G[Apply config]
    F -->|No| E
    G --> H[Preview & generate]
    E --> H

Settings

Setting	Default	Description
configMatchThreshold	50	Minimum score to suggest
enablePatternDetection	true	Consider data patterns in matching
autoApplyExactMatch	false	Skip confirmation on 100% match