name	archive-pubmed
description	Use when maintaining a local PubMed XML archive for offline literature search, indexing, or bulk processing.
disable-model-invocation	true
user-invocable	true

archive-pubmed

Maintain a local PubMed XML archive for offline literature search, text mining, or batch analysis.
Backfill missing baseline/update files after interrupted refreshes.
Verify downloaded gzip packages before rebuilding the search/posting layers.
Rebuild local PubMed postings after the archive content changes.

# 1) Download new PubMed files and populate the local archive
export EDIRECT_LOCAL_ARCHIVE=/data/edirect
archive-pubmed

# 2) Check which PubMed packages are missing locally
archive-pubmed -missing

# 3) Rebuild the local PubMed search/posting structures
archive-pubmed -index

Set EDIRECT_LOCAL_ARCHIVE and confirm the archive root is writable before starting.
Make sure the required helper commands for drive setup and versioned-record refresh are present.
Run archive-pubmed to download new files and populate the archive.
Use -verify or -missing if the refresh was interrupted or you suspect archive gaps.
Run -index only after the data files are stable, then inspect the timing summary before using the local archive with other EDirect tools.

--help and --version are not safe inspection switches; they still run archive setup and can emit real helper/env failures.
The script requires EDIRECT_LOCAL_ARCHIVE, pm-setup, pm-prepare, and pm-refresh.
-clean, -scrub, -scour, -erase, and -zap are destructive maintenance modes and must be run separately from -index.
Mixed-year archive states can force a cleanup path before reindexing; the script explicitly tells you to -zap or -scour in some stale-data scenarios.
-stem and -strict change how text is normalized and indexed, so record them in reproducible workflows.

来源信息