Pypdf

Performance Improvements (PI)
- optimize _decode_png_prediction (2068)

Bug Fixes (BUG)
- Fix incorrect tm_matrix in call to visitor_text (2060)
- Writing German characters into form fields (2047)
- Prevent stall when accessing image in corrupted pdf (2081)
- append() fails when articles do not have /T (2080)

Robustness (ROB)
- Cope with xref not followed by separator (2083)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.15.0...3.15.1)

New Features (ENH)
- Add `level` parameter to compress_content_streams (2044)
- Process /uniHHHH for text_extract (2043)

Bug Fixes (BUG)
- Fix AnnotationBuilder.link (2066)
- JPX image without ColorSpace (2062)
- Added check for field /Info when cloning reader document (2055)
- Fix indexed/CMYK images (2039)

Maintenance (MAINT)
- Cryptography as primary dependency (2053)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.14.0...3.15.0)

New Features (ENH)
- Accelerate image list keys generation (2014)
- Use `cryptography` for encryption/decryption as a fallback for PyCryptodome (2000)
- Extract LaTeX characters (2016)
- ASCIIHexDecode.decode now returns bytes instead of str (1994)

Bug Fixes (BUG)
- Add RunLengthDecode filter (2012)
- Process /Separation ColorSpace (2007)
- Handle single element ColorSpace list (2026)
- Process lookup decoded as TextStringObjects (2008)

Robustness (ROB)
- Cope with garbage collector during cloning (1841)

Maintenance (MAINT)
- Cleanup of annotations (1745)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.13.0...3.14.0)

New Features (ENH)
- Add is_open in outlines in PdfReader and PdfWriter (1960)

Bug Fixes (BUG)
- Search /DA in hierarchy fields (2002)
- Cope with different ISO date length (1999)
- Decode Black only/CMYK deviceN images (1984)
- Process CMYK in deflate images (1977)

Developer Experience (DEV)
- Add mypy to pre-commit (2001)
- Release automation (1991, 1985)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.12.2...3.13.0)

Bug Fixes (BUG)
- Accept calRGB and calGray color_spaces (1968)
- Process 2bits and 4bits images (1967)
- Check for AcroForm and ensure it is not None (1965)

Developer Experience (DEV)
- Automate the release process (1970)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.12.1...3.12.2)

Bug Fixes (BUG)
- Prevent updating page contents after merging page (stamping/watermarking) (1952)
- % to be hex encoded in names (1958)
- Inverse color in CMYK images (1947)
- Dates conversion not working with Z00\'00\' (1946)
- Support UTF-16-LE Strings (1884)

[Full Changelog](https://github.com/py-pdf/pypdf/compare/3.12.0...3.12.1)