gedih3.doctor.diagnoses.metadata#

metadata diagnosis — partition meta JSON + dataset manifest health.

Detects:

  • Partitions whose merged meta JSON file is missing.

  • A stale or absent _manifest.txt R2 sentinel at the database root.

Remedies (safe):

  • Regenerate missing partition meta via h3_merge_metadata.

  • Regenerate manifest via generate_manifest.

Performance pillar (v0.8.x lessons backport):

  • Both the missing-meta check and the newest-parquet-mtime walk are pushed to workers via gedih3.doctor.parallel.parallel_map(). The driver receives a single bool / float per partition and aggregates — no driver-side recursive glob over h3_*///.parquet on continental DBs.

Functions#

metadata_check(→ gedih3.doctor.report.Report)

metadata_fix(→ gedih3.doctor.report.Report)

Module Contents#

gedih3.doctor.diagnoses.metadata.metadata_check(ctx: gedih3.doctor.report.DoctorContext) → gedih3.doctor.report.Report[source]#
gedih3.doctor.diagnoses.metadata.metadata_fix(ctx: gedih3.doctor.report.DoctorContext, report: gedih3.doctor.report.Report) → gedih3.doctor.report.Report[source]#