Capture

Import that preserves the source

Import isn't file conversion. Every source runs through a staged pipeline that keeps the original record, the provenance, and the structure that gave it meaning.

AvailablePipeline

Import that preserves the source

Extract
Transform
De-duplicate
Diff
Build hierarchy
Parse
Persist
Clean up
Rides every stagesource IDsbyte rangeshashesoriginal bytes
466 notes106,558blocksAddedModifiedRemoved
two-phase commitbounded memoryfidelity graded
Import is a governed eight-stage pipeline, not a file conversion — origin IDs, byte ranges and hashes travel with every block, end to end.
01

A generic importer flattens every source into the same shape and loses what made it meaningful. The structure is the meaning — drop it and you're left with noise.

02
  • An eight-stage pipeline — extract, transform, de-duplicate, diff, build hierarchy, parse, persist, clean up — with bounded memory throughout.
  • Provenance travels with every block: source IDs, ordinals, byte ranges, and hashes, with the original bytes archived alongside.
  • Two-phase commit keeps the database and files consistent — an import either lands fully or not at all.
  • Import reports surface quality and warnings; some sources get a graded fidelity analysis.
03
  1. Select a source export or folder.
  2. The pipeline extracts, normalizes, and de-duplicates records.
  3. Review the import report for quality and any recoverable issues.
  4. Browse the result, every block traceable to its origin.
04
  • The original record stays available for reprocessing as parsing improves.
  • Source-specific metadata survives normalization.
  • Large imports run with bounded memory, so the app stays responsive.