uniqdiff 1.1.0: Streaming Field Diff Engine

uniqdiff 1.1.0

uniqdiff 1.1.0 is an additive engine release focused on structured diff, streaming output, schema awareness, and better production usage in ETL / CI pipelines.

This release keeps the stable 1.0 comparison engine contract and extends it with new engine-level primitives for field-level diff, schema diff, sorted streaming diff, JSONL event output, and downstream UniqTools integrations.

Highlights

• Added field-level diff by key.
• Added schema-aware diff for columns, inferred value types, and nullability.
• Added sorted streaming diff APIs for already sorted inputs.
• Added production-grade uniqdiff.jsonl event stream output.
• Added lazy JSONL/event readers and summary helpers.
• Added explicit uniqdiff.engine facade for stable engine imports.
• Improved mode="auto" metadata and disk strategy selection.
• Improved JSONL/file output performance.
• Optimized composite key token extraction and duplicate collection.
• Added profiling and realistic cross-tool benchmark tooling.

New Engine APIs

New documented APIs include:

• compare_sorted_iter()
• write_sorted_diff_file()
• compare_fields()
• compare_fields_sorted()
• compare_fields_files()
• compare_fields_files_sorted()
• compare_file_fields()
• compare_file_fields_sorted()
• iter_field_diff_rows()
• iter_field_diff_events()
• iter_sorted_field_diff_events()
• infer_schema()
• compare_schema()
• compare_file_schema()
• iter_compare_events()
• iter_event_rows()
• summarize_events()
• summarize_event_file()
• uniqdiff.engine

CLI Improvements

The CLI now supports:

• --format jsonl
• --field-diff
• --sorted-input
• --exclude-columns
• --max-rows
• --max-bytes
• --max-output-rows
• --max-output-bytes
• --schema-diff
• --schema-sample-size
• --empty-string-not-null
• --loose-numeric-types

Example:

uniqdiff diff old.csv new.csv \
  --key id \
  --field-diff \
  --format jsonl \
  --output diff.jsonl