jupytext

Fixed

• Include the lab extension that was missing in the conda package (#703).

Changed

• The Jupytext extension for JupyterLab is compatible with Jupyter Lab 3.0, thanks to Martin Renou's awesome contribution #683.

Changed

• Jupytext 1.8.2 depends on python>=3.6. The last version of Jupytext explicitly tested with Python 2.7 and 3.5 was Jupytext 1.7.1, cf. #697.

Changed

• The dependency on markdown-it-py is conditional on python>=3.6 (#697)

Changed

• Removed support for Python 2.7 and 3.5, a preliminary step towards a JupyterLab 3.0-compatible extension (#683)
• The MyST Markdown format uses markdown-it-py~=0.6.0 (#692)

1.7.1 (2020-11-16)

Fixed

• Text notebooks have the same format and mimetype as ipynb notebooks. This fixes the File Load Error - content.indexOf is not a function error on text notebooks (#659)

Changed

• Jupytext's contents manager uses the parent CM's get and save methods to read and save text files, and explicitly calls jupytext.reads and jupytext.writes to do the conversion. We don't use mock nor internal parent methods any more. Thanks to Max Klein for helping making this work! (#634, #635)
• Thanks to the above, Jupytext can work on top of contents manager that don't derive from FileContentsManager, and in particular it works with jupyterfs (#618)
• The documentation was reorganized. README.md was simplified and now includes many links to the documentation.
• The documentation now uses myst_parser rather than recommonmark. And we use conda on RTD (#650, #652)
• The readf and writef functions were dropped (they had been deprecated in favor of read and write in June 2019, v1.2.0)
• The description & dependencies of the JupyterLab extension were updated (#654)
• The --set-kernel - command, on a Python notebook, gives an explicit error when no kernel is not found that matches the current Python executable.
• All the GitHub workflow files were concatenated into a unique file, and we have added an pypi-publish step to automatically publish the package on PyPi when new releases are created.
• The CHANGELOG.md file was moved under docs to better expose the history of changes.

Added

• Configuration errors are reported in the console and/or in Jupyter (#613)
• Jupytext's Contents Manager internal errors are logged on the console, and trigger an HTTP Error 500 (#638)
• The GitHub actions run on both push events and pull requests, and duplicate jobs are skipped (#605)
• Jupytext has a tox.ini file, thanks to Chris Sewell (#605)
• Jupytext is tested against Python 3.9
• The execution cell metadata is now filtered by default (#656)

Fixed

• Optional dependency on sphinx-gallery frozen to version ~=0.7.0 (#614)
• Codecov/patch reports should be OK now (#639)
• Jupytext tests work on non-English locales (#636)
• Cell metadata that are already present in text notebook can be filtered out using a config file (#656)
• Optional cell attributes like attachments are preserved (#671)

Changed

• Jupytext's contents manager uses the parent CM's get and save methods to read and save text files, and explicitly calls jupytext.reads and jupytext.writes to do the conversion. We don't use mock nor internal parent methods any more. Thanks to Max Klein for helping making this work! (#634, #635)
• Thanks to the above, Jupytext can work on top of contents manager that don't derive from FileContentsManager, and in particular it works with jupyterfs (#618)
• The documentation was reorganized. README.md was simplified and now includes many links to the documentation.
• The documentation now uses myst_parser rather than recommonmark. And we use conda on RTD (#650, #652)
• The readf and writef functions were dropped (they had been deprecated in favor of read and write in June 2019, v1.2.0)
• The description & dependencies of the JupyterLab extension were updated (#654)
• The --set-kernel - command, on a Python notebook, gives an explicit error when no kernel is not found that matches the current Python executable.
• All the GitHub workflow files were concatenated into a unique file, and we have added an pypi-publish step to automatically publish the package on PyPi when new releases are created.
• The CHANGELOG.md file was moved under docs to better expose the history of changes.

Added

• Configuration errors are reported in the console and/or in Jupyter (#613)
• Jupytext's Contents Manager internal errors are logged on the console, and trigger an HTTP Error 500 (#638)
• The GitHub actions run on both push events and pull requests, and duplicate jobs are skipped (#605)
• Jupytext has a tox.ini file, thanks to Chris Sewell (#605)
• Jupytext is tested against Python 3.9
• The execution cell metadata is now filtered by default (#656)

Fixed

• Optional dependency on sphinx-gallery frozen to version ~=0.7.0 (#614)
• Codecov/patch reports should be OK now (#639)
• Jupytext tests work on non-English locales (#636)
• Cell metadata that are already present in text notebook can be filtered out using a config file (#656)
• Optional cell attributes like attachments are preserved (#671)

Changed

• Jupytext's contents manager uses the parent CM's get and save methods to read and save text files, and explicitly calls jupytext.reads and jupytext.writes to do the conversion. We don't use mock nor internal parent methods any more. Thanks to Max Klein for helping making this work! (#634, #635)
• Thanks to the above, Jupytext can work on top of contents manager that don't derive from FileContentsManager, and in particular with jupyterfs (#618)
• The documentation was reorganized. README.md was simplified and now includes many links to the documentation.

Added

• Configuration errors are reported in the console and/or in Jupyter (#613)
• Jupytext's Contents Manager internal errors are logged on the console, and trigger an HTTP Error 500 (#638)
• The GitHub actions run on both push events and pull requests, and duplicate jobs are skipped (#605)
• Jupytext has a tox.ini file, thanks to Chris Sewell (#605)

Fixed

• Optional dependency on sphinx-gallery frozen to version ~=0.7.0 (#614)
• Codecov/patch reports should be OK now (#639)
• Jupytext tests work on non-English locales (#636)

1.6.0 (2020-09-01)

Added

• New option hide_notebook_metadata to encapsulate the notebook metadata in an HTML comment (#527)
• New option root_level_metadata_as_raw_cell. Set it to False if you don't want to see root level metadata of R Markdown notebooks as a raw cell in Jupyter (#415)
• New option doxygen_equation_markers to translate Markdown equations into Doxygen equations (#517)
• New option custom_cell_magics to comment out cells starting with user-specific cell magics (#513)
• Documented how to use isort on notebooks (#553)
• jupytext notebook.ipynb --to filename.py will warn that --to is used in place of --output.
• jupytext --set-formats filename.py will suggest to use --sync instead of --set-formats (#544)
• Warn if 'Include Metadata' is off when saving text files in Jupyter (#561)
• Test that notebooks paired through a configuration file are left unmodified (#598)
• Test that metadata filters in the configuration files are taken into account when using jupytext --to (#543)
• New argument --run-path to execute the notebooks at the desired location (#595)
• Activated GitHub code scanning alerts

Changed

• Jupytext now depends on markdown-it-py (Python 3.6 and above) and always features the MyST-Markdown format, thanks to Chris Sewell (#591)
• The md:myst and md:pandoc are always included in the Jupytext formats, and an informative runtime error will occur if the required dependencies, resp. markdown-it-py and pandoc, are not installed. (#556)
• The # %% cell marker has the same indentation as the first line in the cell (#562)
• Jupytext is now installed from source on MyBinder to avoid cache issues (#567)
• The tests that execute a notebook are now skipped on Windows to avoid timeout issues (#489)

Fixed

• Only scripts can have an encoding comment, not Markdown or R Markdown files (#576)
• Spaces in --pipe commands are supported (#562)
• Bash commands starting with special characters are now correctly detected, thanks to Aaron Gokaslan (#587)
• MyST Markdown files are recognized as such even if MyST-Markdown is not available (#556)
• Build JupyterLab with dev-build=False and minimize=False on mybinder to avoid build errors
• Configured coverage targets in codecov.yml

Added

• Activated GitHub code scanning alerts
• New option hide_notebook_metadata to encapsulate the notebook metadata in an HTML comment (#527)
• New option root_level_metadata_as_raw_cell. Set it to False if you don't want to see root level metadata of R Markdown notebooks as a raw cell in Jupyter (#415)
• New option doxygen_equation_markers to translate Markdown equations into Doxygen equations (#517)
• New option custom_cell_magics to comment out cells starting with user-specific cell magics (#513)
• Tested isort on notebooks (#553)
• jupytext notebook.ipynb --to filename.py will warn that --to is used in place of --output.
• jupytext --set-formats filename.py will suggest to use --sync instead of --set-formats (#544)
• Warn if 'Include Metadata' is off when saving text files in Jupyter (#561)
• Test that notebooks paired through a configuration file are left unmodified (#598)
• Test that metadata filters in the configuration files are taken into account when using jupytext --to (#543)
• New argument --run-path to execute the notebooks at the desired location (#595)

Changed

• Install Jupytext from source on MyBinder to avoid cache issues (#567)
• Skip the tests that execute a notebook on Windows to avoid timeout issues (#489)
• The # %% cell marker has the same indentation as the first line in the cell (#562)
• The md:myst and md:pandoc are always included in the Jupytext formats, and an informative runtime error will occur if the required dependencies, resp. markdown-it-py and pandoc, are not installed. (#556)
• Jupytext now depends on markdown-it-py and always features the MyST-Markdown format (Python 3.6 and above, #591)

Fixed

• Configured coverage targets in codecov.yml
• Only scripts can have an encoding comment, not Markdown or R Markdown files (#576)
• Spaces in --pipe commands are supported (#562)
• Bash commands starting with special characters are now correctly detected, thanks to Aaron Gokaslan (#587)
• MyST Markdown files are recognized as such even if myst is missing (#556)
• Build JupyterLab with dev-build=False and minimize=False on mybinder to avoid build errors

1.5.2 (2020-07-21)

Changed

• The documentation uses the Alabaster theme

Fixed

• Preserve indentation when commenting out magic commands (#437)
• Fixed MyBinder - jupytext.py is not a configuration file (#559, #567)

1.5.1 (2020-07-05)

Fixed

• Added toml as a dependency (#552).
• Filtered out __pycache__ and .pyc files from the pip package.
• Fixed coverage upload by adding coverage as a dependency to the conda CI workflow.
• Fixed the conda CI / Python 2.7 job by explicitly installing backports.functools_lru_cache (#554).

1.5.0 (2020-06-07)

Added

• Jupytext can use a local or global configuration file (#508)
• Jupytext can pair notebooks in trees. Use e.g. notebooks///ipynb,scripts///py:percent if you want to replicate the tree of notebooks under notebooks in a folder named scripts (#424)
• The extension for Jupyter Notebook has a New Text Notebook menu that creates text-only notebooks (#443)
• Groovy and Java are now supported, thanks to Przemek Wesołek (#500)
• The Coconut language is also supported, thanks to Thurston Sexton (#532)
• Resource files with .resource extension from the Robot Framework are supported, thanks to Hiski Valli (#535)
• Jupytext is tested in pip and conda environments, on Linux, Mac OS and Windows, using Github actions (#487)
• Jupytext uses pre-commit checks and automatic reformatting with pre-commit, black and flake8 (#483)
• Documentation improvements:
  • Mention that the YAML header can be created with either --set-kernel, --set-formats, or both (#485)
  • Mention that one should use double quotes, not single quotes, around jupytext --check commands like "pytest {}" on Windows (#475)
  • Improved error message when a file is in a version that can't be read by Jupytext (#531)

Fixed

• Skip the jupytext --execute tests when the warning Timeout waiting for IOPub output occurs, which is the case intermittently on Windows (#489)
• Fixed wrong paired paths when syncing with the --pre-commit flag (#506)

Added

• Jupytext can use a local or global configuration file (#508)
• Jupytext can pair notebooks in trees. Use e.g. notebooks///ipynb,scripts///py:percent if you want to replicate the arborescence of notebooks under notebooks in a folder named scripts (#424)
• The extension for Jupyter Notebook adds a New Text Notebook menu that creates text-only notebooks (#443)
• Jupytext is tested in pip and conda environments, on Linux, Mac OS and Windows, using Github actions (#487)
• Pre-commit checks and automatic reformatting of Jupytext's code with pre-commit, black and flake8 (#483)
• Groovy and Java are now supported, thanks to Przemek Wesołek's contribution (#500)
• Mention that the YAML header can be created with either --set-kernel, --set-formats, or both (#485)
• Mention that one should use double quotes, not single quotes, around jupytext --check commands like "pytest {}" on Windows (#475)

Fixed

• Skip the jupytext --execute tests when the warning Timeout waiting for IOPub output occurs, which is the case intermittently on Windows (#489)
• Fixed wrong paired paths when syncing with the --pre-commit flag (#506)

Added

• Jupytext can use a local or global configuration file (#508)
• Jupytext can pair notebooks in trees. Use e.g. notebooks///ipynb,scripts///py:percent if you want to replicate the arborescence of notebooks under notebooks in a folder named scripts (#424)
• The extension for Jupyter Notebook adds a New Text Notebook menu that creates text-only notebooks (#443)
• Jupytext is tested in pip and conda environments, on Linux, Mac OS and Windows, using Github actions (#487)
• Pre-commit checks and automatic reformatting of Jupytext's code with pre-commit, black and flake8 (#483)
• Groovy and Java are now supported, thanks to Przemek Wesołek's contribution (#500)
• Mention that the YAML header can be created with either --set-kernel, --set-formats, or both (#485)
• Mention that one should use double quotes, not single quotes, around jupytext --check commands like "pytest {}" on Windows (#475)

Fixed

• Skip the jupytext --execute tests when the warning Timeout waiting for IOPub output occurs, which is the case intermittently on Windows (#489)
• Fixed wrong paired paths when syncing with the --pre-commit flag (#506)

Added

• Jupytext can use a local or global configuration file (#508)
• Jupytext can pair notebooks in trees. Use e.g. notebooks///ipynb,scripts///py:percent if you want to replicate the arborescence of notebooks under notebooks in a folder named scripts (#424)
• Jupytext is tested in pip and conda environments, on Linux, Mac OS and Windows, using Github actions (#487)
• Pre-commit checks and automatic reformatting of Jupytext's code with pre-commit, black and flake8 (#483)
• Groovy and Java are now supported, thanks to Przemek Wesołek's contribution (#500)
• Mention that the YAML header can be created with either --set-kernel, --set-formats, or both (#485)
• Mention that one should use double quotes, not single quotes, around jupytext --check commands like "pytest {}" on Windows (#475)

Fixed

• Skip the jupytext --execute tests when the warning Timeout waiting for IOPub output occurs, which is the case intermittently on Windows (#489)
• Fixed wrong paired paths when syncing with the --pre-commit flag (#506)

Added

• Added an example with custom notebook metadata (#469)
• Added an example with custom cell tags (#478)

Changed

• The outputs from the .ipynb file are matched with the input cells from the text file with less strict rules. In this version, a search and replace on the text file will not remove the outputs any more (#464).
• Update parsing of myst notebooks to new (markdown-it based) parser (please upgrade to myst-parser to version ~0.8) (#473)
• Use os.path.samefile when searching for the kernel that corresponds to the current environment (--set-kernel -)

Fixed

• Fixed the CLI example for not commenting out magic commands: --opt comment_magics=false. In addition, most of the jupytext commands in using-cli.md are now tested! (#465)
• jupytext.read and jupytext.write now give more meaningful errors when the format information is incorrect (#462)
• Multiline comments starting or ending with quadruple quotes should not cause issues anymore (#460)
• Fixed active cells in the py:percent format (#477)

1.4.1 (2020-03-19)

Added

• Script of script (SoS) notebooks are now supported. Thanks to Thomas Pernet-coudrier for contributing the sample notebook (#453).
• New MyST Markdown format (md:myst), developed by the ExecutableBookProject team. Read more about the MyST Markdown format in the documentation. And many thanks to Chris Sewell for contributing the actual implementation! (#447 #456 #458)

Fixed

• When using jupytext --pipe cmd, the output of cmd should not appear in the terminal (#432)

Changed

• The new jupyterlab extension (in version 1.2.0) is compatible with JupyterLab 2.0. Many thanks to Jean Helie! (#449)
• It is not compatible with JupyterLab 1.x anymore. If you wish, you can install manually the previous version of the extension with jupyter labextension install jupyterlab-jupytext@1.1.1.

Fixed

• Display the output/errors of command executed with jupytext --pipe or jupytext --check (#432)

Fixed

• Removed leading slash in notebook paths (#444)
• Fixed jupytext --set-formats when using formats with prefix and/or suffix (#450)

Added

• C# and F# Jupyter notebooks are now supported (#427, #429)

Fixed

• jupytext --to script *.ipynb now computes the script extension for each notebook (#428)
• Fix shebang handling for languages with non-# comments, by Jonas Bushart (#434)
• Indented bash commands are now commented out (#437)
• The main formats are documented in jupytext --help (#426, #433)

1.3.3 (2020-01-27)

Added

• Jupytext has a logo! Many thanks to Kyle Kelley for contributing the logo (#423), and to Chris Holdgraf for suggesting this (#260).
• Nested metadata filtering is now supported! You can use this to rid of jupytext_version if you wish (#416).

Fixed

• Code cells in the Markdown format can contain triple backticks inside multiline strings (#419).
• Changes in the YAML header when running jupytext --test on text files are ignored (#414).

Fixed

• The --pre-commit mode now ignores non-notebook files in the index (#338).
• nbformat_minor is taken from the notebook with outputs when inputs and outputs are merged.

Added

• New nomarker format in the Jupyter Notebook and JupyterLab extensions (#397)

Changed

• The normarker format replaces the one previously named bare.

Fixed

• Multiline strings are now accepted in the YAML header (#404). Fix contributed by ZHUO Qiang (#405).
• Fixed the instructions on how to use multiline comments for all Markdown cells in the py:percent format, by ZHUO Qiang (#403).
• Added cd to the list of magic commands.
• Do not read paired files when --set-formats is being used (fixes #399).

See also What's new in Jupytext 1.3?

Added

• Pairing a notebook to both .md and .py is now supported. Input cells are loaded from the most recent text representation (#290)
• Both the Jupyter Notebook and the Jupyter Lab extension for Jupytext were updated (#290)
• Raw cells are now encoded using HTML comments (<!-- #raw --> and <!-- #endraw -->) in Markdown files (#321)
• Markdown cells can be delimited with any of <!-- #region -->, <!-- #markdown --> or <!-- #md --> (#344)
• Code blocks from Markdown files, when they don't have an explicit language, appear in Markdown cells in Jupyter (#321)
• Code blocks with an explicit language and a .noeval attribute are inactive in Jupyter (#347)
• Markdown and raw cells can be quoted with triple quotes in the py:percent format. And Markdown cells can start with just # %% [md] (#305)
• The light format uses # + [markdown] rather than the previous cell_type metadata to identify markdown cells with metadata (#356)
• Explicit Markdown cells in the light format # + [markdown] can use triple quotes (#356)
• IPython magic help commands like float? are classified as magics, and thus commented in Python scripts (#297)
• Cell metadata can be encoded as either key=value (the new default) or in JSON. An automatic option cell_metadata_json should help minimize the impact on existing files (#344)
• R Markdown hidden inputs, outputs, or cells are now mapped to the corresponding Jupyter Book tags by default (#337)
• The notebook metadata filter is automatically updated when extra keys are added to the YAML header (#376)
• The Jupyter Notebook extension for Jupytext is compatible with Jupyter Notebook 6.0 (#346)
• jupytext notebook.py --to ipynb updates the timestamp of notebook.py so that the paired notebook still works in Jupyter (#335, #254)
• jupytext --check pytest notebook.ipynb can be used to run test functions in a notebook (#286)
• jupytext --check and jupytext --pipe can run commands that only operate on files: when {} is found in the text of the command, jupytext saves the text representation of the notebook in a temp file, and replaces {} with the name of that file before executing the command. (#286)
• Documented how to sync notebooks in a pre-commit hook (#338)
• Added support for Rust/Evxcr, by Jonas Bushart (#351)
• Added support for Robot Framework, by Asko Soukka (#378)
• Added support for PowerShell (#349)

Changed

• Use CHANGELOG.md rather than HISTORY.rst

Fixed

• Commands like cat = x are not magic commands, so they are not commented any more (#339)
• Fixed an inconsistent round trip (code cell with "cat" being converted to a markdown cell) in the py:light format (#339)
• jupytext --test textfile.ext now really compares the text file to its round trip (rather than the corresponding notebook) (#339)
• Markdown cells that contain code are now preserved in a round trip through the Markdown and R Markdown formats (#361)
• Code cells with a %%python3 cell magic are now preserved in a round trip through the Markdown format (#365)
• jupytext --execute runs the notebook in its folder (#382)
• Strings in the metadata of code cells are quoted in the Rmd representation. And we escape R code in chunk options with #R_CODE# (#383)

Read a more user friendly version here!

Added

• Pairing a notebook to both .md and .py is now supported. Input cells are loaded from the most recent text representation (#290)
• Both the Jupyter Notebook and the Jupyter Lab extension for Jupytext were updated (#290)
• Raw cells are now encoded using HTML comments (<!-- #raw --> and <!-- #endraw -->) in Markdown files (#321)
• Markdown cells can be delimited with any of <!-- #region -->, <!-- #markdown --> or <!-- #md --> (#344)
• Code blocks from Markdown files, when they don't have an explicit language, appear in Markdown cells in Jupyter (#321)
• Code blocks with an explicit language and a .noeval attribute are inactive in Jupyter (#347)
• Markdown and raw cells can be quoted with triple quotes in the py:percent format. And Markdown cells can start with just # %% [md] (#305)
• The light format uses # + [markdown] rather than the previous cell_type metadata to identify markdown cells with metadata (#356)
• Explicit Markdown cells in the light format # + [markdown] can use triple quotes (#356)
• IPython magic help commands like float? are classified as magics, and thus commented in Python scripts (#297)
• Cell metadata can be encoded as either key=value (the new default) or in JSON. An automatic option cell_metadata_json should help minimize the impact on existing files (#344)
• R Markdown hidden inputs, outputs, or cells are now mapped to the corresponding Jupyter Book tags by default (#337)
• The notebook metadata filter is automatically updated when extra keys are added to the YAML header (#376)
• The Jupyter Notebook extension for Jupytext is compatible with Jupyter Notebook 6.0 (#346)
• jupytext notebook.py --to ipynb updates the timestamp of notebook.py so that the paired notebook still works in Jupyter (#335, #254)
• jupytext --check pytest notebook.ipynb can be used to run test functions in a notebook (#286)
• jupytext --check and jupytext --pipe can run commands that only operate on files: when {} is found in the text of the command, jupytext saves the text representation of the notebook in a temp file, and replaces {} with the name of that file before executing the command. (#286)
• Documented how to sync notebooks in a pre-commit hook (#338)
• Added support for Rust/Evxcr, by Jonas Bushart (#351)
• Added support for Robot Framework, by Asko Soukka (#378)
• Added support for PowerShell (#349)

Changed

• Use CHANGELOG.md rather than HISTORY.rst

Fixed

• Commands like cat = x are not magic commands, so they are not commented any more (#339)
• Fixed an inconsistent round trip (code cell with "cat" being converted to a markdown cell) in the py:light format (#339)
• jupytext --test textfile.ext now really compares the text file to its round trip (rather than the corresponding notebook) (#339)
• Markdown cells that contain code are now preserved in a round trip through the Markdown and R Markdown formats (#361)
• Code cells with a %%python3 cell magic are now preserved in a round trip through the Markdown format (#365)
• jupytext --execute runs the notebook in its folder (#382)
• Strings in the metadata of code cells are quoted in the Rmd representation. And we escape R code in chunk options with #R_CODE# (#383)

Read a more user friendly version here!

Improvements

• Pairing a notebook to both .md and .py is now supported. Input cells are loaded from the most recent text representation (#290)
• Both the Jupyter Notebook and the Jupyter Lab extension for Jupytext were updated (#290)
• Raw cells are now encoded using HTML comments (<!-- #raw --> and <!-- #endraw -->) in Markdown files (#321)
• Markdown cells can be delimited with any of <!-- #region -->, <!-- #markdown --> or <!-- #md --> (#344)
• Code blocks from Markdown files, when they don't have an explicit language, appear in Markdown cells in Jupyter (#321)
• Code blocks with an explicit language and a .noeval attribute are inactive in Jupyter (#347)
• Markdown and raw cells can be quoted with triple quotes in the py:percent format. And Markdown cells can start with just # %% [md] (#305)
• The light format uses # + [markdown] rather than the previous cell_type metadata to identify markdown cells with metadata (#356)
• Explicit Markdown cells in the light format # + [markdown] can use triple quotes (#356)
• IPython magic help commands like float? are classified as magics, and thus commented in Python scripts (#297)
• Cell metadata can be encoded as either key=value (the new default) or in JSON. An automatic option cell_metadata_json should help minimize the impact on existing files (#344)
• R Markdown hidden inputs, outputs, or cells are now mapped to the corresponding Jupyter Book tags by default (#337)
• The Jupyter Notebook extension for Jupytext is compatible with Jupyter Notebook 6.0 (#346)
• jupytext notebook.py --to ipynb updates the timestamp of notebook.py so that the paired notebook still works in Jupyter (#335, #254)
• jupytext --check pytest notebook.ipynb can be used to run test functions in a notebook (#286)
• jupytext --check and jupytext --pipe can run commands that only operate on files: when {} is found in the text of the command, jupytext saves the text representation of the notebook in a temp file, and replaces {} with the name of that file before executing the command. (#286)
• Added support for Rust/Evxcr, by Jonas Bushart (#351)
• Added support for PowerShell (#349)

BugFixes

• Commands like cat = x are not magic commands, so they are not commented any more (#339)
• Fixed an inconsistent round trip (code cell with "cat" being converted to a markdown cell) in the py:light format (#339)
• jupytext --test textfile.ext now really compares the text file to its round trip (rather than the corresponding notebook) (#339)
• Markdown cells that contain code are now preserved in a round trip through the Markdown and R Markdown formats (#361)
• Code cells with a %%python3 cell magic are now preserved in a round trip through the Markdown format (#365)

Improvements

• Raw cells are now encoded using HTML comments (<!-- #raw --> and <!-- #endraw -->) in Markdown files. And code blocks from Markdown files, when they don't have an explicit language, are displayed as Markdown cells in Jupyter (#321)
• Markdown and raw cells can use multiline comments in the py:percent format (#305)
• jupytext notebook.py --to ipynb updates the timestamp of notebook.py so that the paired notebook still works in Jupyter (#335, #254)

BugFixes

• Fixed an inconsistent round trip (code cell with "cat" being converted to a markdown cell) in the py:light format (#339)
• Commands like cat = x are not magic commands {#339)
• jupytext --test now really compares the text (rather than the corresponding notebook) when run on text files (#339)

Improvements

• The documentation includes a mention on how to set metadata filters at the command line (#330)
• Jupytext will not catch any error when the flag --warn-only is not set (#327)

BugFixes

• Now the flag --warn-only catches every possible error (#263)
• .md and .markdown files are treated identically (#325)
• Fixed --set-kernel when using pipes (#326)
• Fixed utf-8 encoding on stdout on Python 2 (#331)

BugFixes

• Dependency on setuptools in pandoc.py made optional to fix the build of the conda package (#310, #323)

Improvements

• Documentation includes a section on how to use Jupytext as a Python library (#317)
• Mention of the server extension in the documentation (#304)
• Text notebooks can be tested with jupytext --execute notebook.md (#303)
• The default value of as_version in jupytext.read is nbformat.NO_CONVERT, as for nbformat.read
• Jupytext tests are now included in sdist (#310)

BugFixes

• Fixed the usability of the fmt argument in jupytext.read (#312)
• Fixed the download notebook error when c.notebook_extensions has a custom value (#318)
• String delimiters in commented text are now ignored (#307)

Improvements

• Added documentation on how to use Jupytext with JupyterLab 0.35 (#299)
• and on using Jupytext with the pre-commit package manager (#292)
• The read and write functions are easier to use (#292)

BugFixes

• Jupytext now ships the jupyterlab-jupytext extension in version 1.0.2. The version 1.0.1 erroneously introduces a target_formats metadata in the jupytext section, instead of formats, and works only after two clicks.

Improvements

• New --execute option in Jupytext CLI (#231)
• The --set-formats option in Jupytext CLI also triggers --sync, allowing shorter commands.
• jupytext's read and write functions can be used as drop-in replacements for nbformat's ones (#262).
• jupytext --sync will now skip unpaired notebooks (#281).
• The JupyterLab extension was updated. It now works on on text files (#213) and has a new option to include (or not) the metadata in the text representation of the notebook.
• Jupytext's contents manager class is derived dynamically from the default CM class for compatibility with jupyter_server (#270)
• Removed dependency on mock and testfixtures, thanks to Jean-Sebastien Dieu (#279)
• Added support for .markdown extension (#288)

BugFixes

• The jupyterlab-jupytext extension shipped with the python package is in version 1.0.1, and is compatible only with JupyterLab >= 1.0. If you use an earlier version of JupyterLab, please install the version 0.19 of the extension with jupyter labextension install jupyterlab-jupytext@0.19 (#276, #278)
• Text files can be unpaired (#289)

Improvements

• New --execute option in Jupytext CLI (#231)
• The --set-formats option in Jupytext CLI also triggers --sync, allowing shorter commands.
• jupytext's read and write functions can be used as drop-in replacements for nbformat's ones (#262).
• jupytext --sync will now skip unpaired notebooks (#281).
• The JupyterLab extension was updated. It now works on on text files (#213) and has a new option to include (or not) the metadata in the text representation of the notebook.
• Jupytext's contents manager class is derived dynamically from the default CM class for compatibility with jupyter_server (#270)
• Removed dependency on mock and testfixtures, thanks to Jean-Sebastien Dieu (#279)

BugFixes

• The jupyterlab-jupytext extension shipped with the python package is in version 1.0, and is compatible only with JupyterLab >= 1.0. If you use an earlier version of JupyterLab, please install the version 0.19 of the extension with jupyter labextension install jupyterlab-jupytext@0.19 (#276, #278)

Improvements

• Added support for Scala notebook, by Tobias Frischholz (#253)
• Added a getting started notebook for jupytext (and Binder), by Chris Holdgraf (#257)
• The Markdown and R Markdown representations are now tested for all the languages
• The Jupytext notebook extension also works when the notebook is a text file (#213)

BugFixes

• The Jupytext Menu in Jupyter Notebook is now compatible with jupyter_nbextensions_configurator (#178)
• Entries in the Jupytext menu are updated when the menu is hovered on (#248)
• Fixed link to .md files in the documentation (#255)

Improvements

• Added support for Scala notebook, by Tobias Frischholz (#253)
• Markdown and R Markdown representations are now tested on all languages
• Jupytext extensions now also work when the notebook is a text file (#213)

BugFixes

• The Jupytext Menu in Jupyter Notebook is now compatible with jupyter_nbextensions_configurator (#178)
• Entries in the Jupytext menu are updated when the menu is hovered on (#248)
• Fixed link to .md files in the documentation (#255)

Improvements

• Jupytext now supports Javascript and Typescript, thanks to Hatem Hosny (#250)
• Jupytext works with Python 3.8 as well

BugFixes

• Fixed global auto pairing (#249)

Fixed

• Fixed implicit dependency on jupyter_client (#245)

Improvements

• New argument --set-kernel in Jupytext command line (#230)
• Jupytext now accepts --to script or --to auto (#240)
• Jupytext now has a real Sphinx documentation on readthedocs, thanks to Chris Holdgraf (#237)

BugFixes

• Invalid notebooks may cause a warning, but not a fatal error (#234)
• Jupyter server extension leaves the contents manager unchanged if it is a sub-class of Jupytext's CM (#236)
• Fixed format inference when metadata is present but not format information (#239)
• Preserve executable and encoding information in scripts with metadata (#241)

Improvements

• New argument --set-kernel in Jupytext command line (#230)

BugFixes

• Invalid notebooks may cause a warning, but not a fatal error (#234)
• Jupyter server extension leaves the contents manager unchanged if it is a sub-class of Jupytext's CM (#236)

Improvements

• New argument --kernel in Jupytext command line (#230)

BugFixes

• Invalid notebooks may cause a warning, but not a fatal error (#234)
• Jupyter server extension leaves the contents manager unchanged if it is a sub-class of Jupytext's CM (#236)

Improvements

• Support for IDL notebooks and .pro scripts (#232)

Improvements

• Jupytext's content manager has a new notebook_extensions option (#224, #183)
• Cells can be made inactive in scripts with the active-ipynb cell tag (#226)

BugFixes

• Directories ending in .jl (or .ipynb) are not notebooks (#228)
• Empty notebooks have no language (#227)

Improvements

• Jupytext server extension leaves the contents manager unchanged when it is already a subclass of TextFileContentsManager (#218)
• The base class for TextFileContentsManager defaults to FileContentsManager when LargeFileManager is not available (#217)

Improvements

• Markdown and R Markdown formats now support metadata (#66, #111, #188)
• The light format for Scripts can use custom cell markers, e.g. Vim or VScode/PyCharm folding markers (#199)
• Pandoc's Markdown format for Jupyter notebooks is available in Jupytext as md:pandoc (#208)

BugFixes

• Jupytext's contents manager is now based on LargeFileManager to allow large file uploads (#210)
• YAML header parsed with yaml.safe_load rather than yaml.load (#215)
• IPython line magic can be split across lines (#209)
• jupytext --to py rather than --to python in the README (#216)

Improvements

• Markdown and R Markdown formats now support metadata (#66, #111, #188)
• The light format for Scripts can use custom cell markers, e.g. Vim or VScode/PyCharm folding markers (#199)
• Pandoc's Markdown format for Jupyter notebooks is available in Jupytext as md:pandoc (#208)

BugFixes

• Jupytext's contents manager is now based on LargeFileManager to allow large file uploads (#210)
• YAML header parsed with yaml.safe_load rather than yaml.load
• IPython line magic can be split across lines (#209)

Improvements

• Markdown and R Markdown formats now support metadata (#66, #111, #188)
• The light format for Scripts can use custom cell markers, e.g. Vim or VScode/PyCharm folding markers (#199)
• Pandoc's Markdown format for Jupyter notebooks is available in Jupytext as md:pandoc (#208)

BugFixes

• Jupytext's contents manager is now based on LargeFileManager to allow large file uploads (#210)
• YAML header parsed with yaml.safe_load rather than yaml.load
• IPython line magic can be split across lines (#209)

Improvements

• Markdown and R Markdown formats now support metadata (#66, #111, #188)
• The light format for Scripts can use custom cell markers, e.g. Vim or VScode/PyCharm folding markers (#199)

BugFixes

• Jupytext's contents manager is now based on LargeFileManager to allow large file uploads (#210)
• YAML header parsed with yaml.safe_load rather than yaml.load
• IPython line magic can be split across lines (#209)

BugFixes

• Fix the error 'notebook file has changed on disk' when saving large notebooks (#207)

Improvements

• Wildcard are now supported on Windows (#202)
• Trusted notebooks remain trusted when inputs cells are modified (#203)

BugFixes

• Pre-commit mode adds the result of conversion to the commit (#200)

Improvements

• Matlab and Octave notebooks and scripts are now supported (#197)

BugFixes

• notebook_metadata_filter = "all" now works (#196)
• Default pairing in subfolders fixed in Jupyter Lab (#180)

Improvements

• Rename notebooks in pairs in the tree view (#190)
• Associate .scm file extension with Scheme scripts (#192)
• Added support for Clojure, by bzinberg (#193)

BugFixes

• Allow spaces between ? or ! and python or bash command (#189)