Fixed critical bug where validation script exited prematurely under
set -euo pipefail due to arithmetic operations returning non-zero exit
codes when evaluating to 0.
Changes:
- Replaced ((var++)) with var=$((var + 1))
- Replaced ((var += value)) with var=$((var + value))
- Restructured progress indicator to avoid modulo 0 result
Verification:
- Script now successfully validates all 1,260 corpus files
- Correctly reports 603 valid files with complete license info
- Detects 657 files with missing licenses (expected - PDFium test files)
- Exits with code 1 when validation fails (as designed)
Closes bf-632yt
Implement scripts/validate-corpus.sh to verify corpus integrity against
manifest.csv. Validates:
- File existence
- File size matching manifest
- SHA256 checksums
- License information presence
- Total counts (files, pages, size)
Exit 0 on success, non-zero on validation failure. Outputs summary with
total files, pages, size, and validation status.
Closes bf-632yt
