glyphive
Glyphive archives a file tree to compact, OCR-friendly, printable pages and restores it from a transcript or scan. Its format is designed for the failure modes of printed text: every encoded line is independently checked, page metadata is protected, and Reed-Solomon parity can repair scattered OCR errors.
The pages stay readable and re-typeable. Restore never accepts a guess merely because decompression progressed; checksums, error correction, and the final document digest are the correctness oracles.
Alpha: the current format and CLI may change before 1.0.
Installation
pip install glyphive
| Extra | Adds | Needed for |
|---|---|---|
pdf |
fpdf2 |
PDF output |
docx |
python-docx |
Word output |
zstd |
zstandard |
zstd compression |
ocr |
Pillow, pytesseract |
Tesseract image bridge |
all |
all packages above | All lightweight integrations |
The OCR extra does not install the operating-system Tesseract program or its language data.
30-second tour
glyphive create -f backup.txt --compression gzip -C project .
glyphive list -f backup.txt
glyphive extract -f backup.txt -C restored
The create command archives one root directory. Root-level .gitignore and
.ignore files are honored by default, and .git/ is always omitted. The list
and extract commands read the protected machine metadata rather than trusting
the human-facing header text.
Learn more
- Create an archive — formats, compression, metadata, and ignore behavior.
- Restore an archive — inspect, verify, and safely extract.
- Wire format — archive records, line frames, parity, and page metadata.
- OCR — provider installation, selection, and measurement.
- Benchmarks — reproducible performance and density methodology.
- API reference — the public Python modules.
- Changelog — user-visible changes.