Fix

The fix subcommand is Step 3 of the workflow. It reads the original .bib file and results.json, applies InspireHEP canonical values to mismatched entries, and writes a new .bib file.

Usage

bib-checker fix FILE.bib results.json [OPTIONS]

Options

Option Default Description
--output FILE <name>_fixed.bib Output path for the fixed bib.
--fields FIELDS doi,eprint,year,title Comma-separated list of fields to update.
--dry-run off Show what would change without writing anything.

How it works

  1. Only entries with status == "mismatch" that have an inspire_record are patched. missing entries cannot be auto-fixed (there is no record to copy from).
  2. For each mismatch entry, the specified fields are updated to InspireHEP's canonical values.

  3. The output file is structured in two sections:

  4. Main section — all entries that are now clean (both originally-ok and freshly fixed).

  5. Separator — a comment block marking the boundary.
  6. Flagged section — entries that still need manual attention:
    • All missing entries.
    • Mismatch entries where fix found no fixable InspireHEP record.

The separator looks like:

% ##########################################################################
% These citations need validation/checking
% ##########################################################################

The source file is never modified

fix always writes to a new file (<name>_fixed.bib by default). The original .bib is read-only from fix's perspective.

Reviewing fixes before applying

Always use --dry-run first to inspect what will change:

bib-checker fix paper.bib results.json --dry-run

Pay special attention to entries where all of eprint, year, and title differ — this usually means InspireHEP's record for that texkey points to a different paper, and the fix would be wrong.

Restricting which fields are updated

Use --fields to update only a subset of fields:

# Fix only the year, leave title and DOI alone
bib-checker fix paper.bib results.json --fields year

After running fix

The entries at the end of <name>_fixed.bib (after the separator) still need manual attention. Common actions:

  • Wrong texkey — if InspireHEP's record for that key is a different paper, find the correct texkey using report.html (the suggest step usually identifies it) and rename the key in both the .bib and your .tex files.
  • Non-HEP entries — books, technical reports, and web references may never be on InspireHEP. Consider adding them to your .bibcheckerignore file.
  • Preprints without a texkey — entries with non-standard keys (e.g. chan2024first) cannot be looked up; rename them to the InspireHEP convention.