PyMuPDF-1.21.1 has been released.

Wheels for Windows, Linux and MacOS, and the sdist, are available on pypi.org and can be installed in the usual way, for example:

python -m pip install --upgrade pymupdf

Changes in Version 1.21.1 (2022-12-13)

• This release uses MuPDF-1.21.1.

• Bug fixes:

  • Fixed #2110: Fully embedded font is extracted only partially if it occupies more than one object
  • Fixed #2094: Rectangle Detection Logic
  • Fixed #2088: Destination point not set for named links in toc
  • Fixed #2087: Image with Filter "[/FlateDecode/JPXDecode]" not extracted
  • Fixed #2086: Document.save() owner_pw & user_pw has buffer overflow bug
  • Fixed #2076: Segfault in fitz.py
  • Fixed #2051: Missing DPI Parameter
  • Fixed #2048: Invalid size of TextPage and bbox with newest version 1.21.0
  • Fixed #2045: SystemError: returned a result with an error set
  • Fixed #2039: 1.21.0 fails to build against system libmupdf
  • Fixed #2036: Archive::Archive defined twice

• Other

  • Swallow "&zoom=nan" in link uri strings.
  • Add new Page utility methods Page.replace_image() and Page.delete_image().

• Documentation:

  • #2040: Added note about test failure with non-default build of MuPDF, to tests/README.md.
  • #2037: In docs/installation.rst, mention incompatibility with chocolatey.org on Windows.
  • #2061: Fixed description of Annot.file_info.
  • #2065: Show how to insert internal PDF link.
  • Improved description of building from source without an sdist.
  • Added information about running tests.
  • #2084: Fixed broken link to PyMuPDF-Utilities.

This is largely unchanged from 1.21.0rc1, except that it builds with the official MuPDF-1.21.0 release.

Install with: python -m pip install pymupdf==1.21.0rc2

• Uses mupdf-1.21.0.
• New Story support.
• Added wheels for Python-3.11.
• Docs: https://pymupdf.readthedocs.io/en/1.21.0rc2/
  • Changelog: https://pymupdf.readthedocs.io/en/1.21.0rc2/changes.html

Install with: python -m pip install pymupdf==1.21.0rc1

• Uses mupdf-1.21.0-rc1.
• New Story support.
• Added wheels for Python-3.11.
• Docs: https://pymupdf.readthedocs.io/en/1.21.0rc1/
  • Changelog: https://pymupdf.readthedocs.io/en/1.21.0rc1/changes.html

• Built with MuPDF-1.20.3.
• Fix #1787.
• Fix #1824.
• Improvements to documentation:
  • Moved old docs/faq.rst into separate docs/recipes-* files.
  • Improved information about building from source in docs/installation.rst.
  • Clarified memory allocation setting JM_MEMORY in docs/tools.rst.
  • Fixed link to PDF Reference manual in docs/app3.rst.
  • Fixed building of html documentation on OpenBSD.

Wheels for Windows, Linux and MacOS, and the sdist, are available on pypi.org and can be installed in the usual way, for example:

pip install --upgrade pymupdf

• Fix https://github.com/pymupdf/PyMuPDF/pull/1724.
• Fix https://github.com/pymupdf/PyMuPDF/issues/1771.
• Fix https://github.com/pymupdf/PyMuPDF/issues/1751.
• Fix https://github.com/pymupdf/PyMuPDF/issues/1645.
• Improvements to sphinx-generated documentation.

This release integrates the recently-released MuPDF-1.20.0, and has fixes for #1733 and #1738. The latter also contains an additional fix for occasional SEGVs when freeing documents.

Building from source works slightly differently from before:

• We now automatically download the required MuPDF source and build it into PyMuPDF.
• Python sdists (source distributions) already contain the required MuPDF source and build without downloading.
• One can override the default build behaviour by setting environmental variables, for example to build with a system-installed mupdf. See the doc-comment at the start of setup.py for details.

Fixes: #1620, #1601

Enhancements:

• new method Page.load_widget() to load a widget from its xref
• new dictionary pdfcolor which contains 500 predefined PDF colors
• Quad class supports operator algebra
• text search and extraction default flags now accessible as predefined constants
• iterators Page.annots() and Page.widgets() now prohibit reloading the page within their scope
• removed multiple utility functions from the Tools class and redefined them as standalone
• Parameter new in Document.update_stream() is now obsolete.

Fixes: #1583, #1552, #1550, #1521, #1518, #1513, #1510, #1417, #1550. Also fixed some undocumented errors that caused the span["origin"] to be incorrectly set in corner cases.

Added new items "orientation" and associated transformtion matrix to the output of fitz.image_properties(), which contains EXIF data of supporting image files.

A new method Document.xref_copy() allows making xref objects duplicates of each other.

Fixes: #1505, #1484, #1479, #1474.

Changes:

• Full support of PDF page rectangles like /ArtBox etc.
• New global variable TESSDATA_PREFIX for comfortably checking presence of OCR support
• Changed Document.xref_set_key() such that dictionary keys will physically be removed if set to value "null".
• Changed Document.extract_font() to optionally return a dictionary (instead of a tuple).

Fixes: #1351, #1417, #1418, #1430, #1433

• New or changed Pixmap methods color_topusage(), color_count(), warp(). Some of them solve #1397.
• New Annot method and property irt_xref, set_irt_xref(). Implements #1450.
• New Rect / IRect method torect() which creates a matrix to transform between given rectangles.
• Page.get_texttrace() now also supports non-horizontal text.

Improvements:

• Page.get_drawings() now includes area orientation for rectangles
• Page pixmap creation has a new parameter "dpi"
• New check for monochrome / unicolor pixmaps and number of colors

Fixes: #1388, #1375, #1364, #1342, #1355, #1397, #1408.

OCR of a document page has been improved a lot compared to v1.19.0. Text extractions now also come with an integrated sort. Fixes: #1328

Introduces major new features like PDF journalling and OCR support by directly invoking Tesseract-OCR. In addition, it is possible to detect whether object are covered (hidden) by other objects.

As part of the new version, the following issues have resolved: #1313, #1311, #1290, #1286, #1287, #1284.

Fixes #1266

This version fixes #1257, #1252, #1244, #1241, #1234, #1236, #1227.

The fitz module now supports text extraction via a new subcommand "gettext". Among a couple of modes, preserving the original layout can be chosen.

Also fixed #1187, #1184, #1154, #1152 and #1146.

Apart from some minor fixes, this release introduces support for small caps in TextWriter based text output.

In addition, method Document.subset_fonts() now prefixes subsetted font names with the 6 upper case letter prefix as prescribed by the PDF standard.

List of fixed issues: #1088, #1081, #1078, #1085.

The following habe been fixed:

• #1043
• #1053
• undocumented occasional error calculating envelopping rectangle for paths in Page.get_drawings()
• undocumented occasional loop in TextWriter.fill_textbox()
• added method Font.char_lengths() which returns a tuple of all character widths for a given string. An improved version of Font.text_length()
• greatly improved performance of Font.text_length()
• added various ways to delete multiple PDF pages, among them are slices and the Python del statement
• changed method Document.del_toc_item(): the item's title text will no longer be removed - instead the item is shown grayed-out to indicate its deletion.

Method Page.insert_image has been rewritten for improved performance in standard cases. Also introduced option to re-use pre-existing images in the file directly to provide another performance boost. Other changes:

• implemented or fixed #1042, #1041, #1037
• minor improvements in PDF EmbeddedFiles handling for better support of building PDF collections apps.

Meta information for images embedded in document pages has been enriched by the so-called transformation matrix. It can be used to find out, what "happened" to the image rectangle to make it fit in its bbox on the page, like scaling and rotation.

Other changes are mostly minor bug fixes: #990 #972

A new Page method get_image_info() is also available, which extracts image meta information from the page's TextPage - much like the corresponding Page.get_text("dict"), but without extracting any text or the image binary data themselves.

Fixed: #941 #929 #927

• included PDF trailer access in Document.xref_get_key()
• added a number of functions for recovering text quads in "dict" / "rawdict" text extractions

Fixed #888, #895, #896, #885, #922 Implemented #897 (text output right-to-left).

• Font subsetting now works without rewriting the respective text.
• Added a utility function to compute the quad of a text span for "dict" and "rawdict" text extraction.

Fixes:

• #881
• #878

Fixes:

• #844, #838, #823, #818, #814

Implemented enhancement requests:

• #855, which allows font subsetting using package fontTools

• #870, which allows convert_to_pdf method also for PDF documents.

• #843, Document.tobytes() (formerly Document.write()) now also support linearized output. Plus several extensions / improvements around supporting Python fileobjects.

• Added new methods to quickly determine whether a PDF has annotations or links.

• Extended the Document.scrub() method with a new parameter, which allows to also remove page thumbnails.

• Added methods to directly inquire and set values in PDF objects - without the need to manipulating PDF object sources in an unwieldy way - see methods Document.xref_set_key() / Document.xref_get_key().

Continued the process of changing the naming convention for class methods and attributes to "snake_case". As announced before, this is a tedious, error-prone process, and requires special care to maintain a high backlevel support for existing scripts. In future versions - probably synchronously to MuPDF v1.19.0 - we will remove definitions of old names, but a method for re-activating old aliases will remain available.

The recent introduction of "Discussions" by Github has been very motivating for our users. Based on their feedback, several enhancement have been implemented. Here is a selection:

• Most Python functions now have typing / annotation support .
• For PDF table-of-contents items, colors are now supported (reading and writing)
• PDF page label support for reading and writing
• Support personalized tagging of new annotations, fields and links for easier selection of relevant objects.

There also is a number of fixes - please consult the documentation.

Font metrics handling has been improved: text box writing now observes the relevant font properties when determining line heights. In this course a new option has been introduced, which allows getting text bboxes (glyphs, spans, text search quads, etc.) that more exactly wrap the text only - as opposed to always returning line height bboxes.

Fixes:

• #771
• #768
• #750
• #739
• #728
• #727

• Improved PDF Optional Content support

• Started overhaul of method and attribute naming

• Introduced support of Popup annotations

• Implemented the following fixes:

  • #727
  • #726
  • #724

As a major new feature, the PDF Optional Content concept is now widely supported.

The following fixes have been implemented:

• #714
• #711
• #707
• #713
• #709

This resolves

• #575
• #697
• #691

and removes the hit_max parameter from text searching. In addition, hyphenated words around line breaks are still found.

The use of the clip parameter in text searches and text extractions now only includes characters whose bboxes are fully contained in the clip rctangle.