Skip to content

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Effort-based Versioning. (Changes to features documented as “experimental” will not increment macro and meso version numbers.)

Unreleased

4.0.0 - 2026-09-02

Compatibility with StepUp Core 4, a refactored Zenodo synchronization and more reliable concurrent notebook conversion.

Note that all changes of the 4.0.0rc* release candidates are combined below.

Added

  • Support for os.PathLike objects in stepup.reprep.api functions.
  • The resources and duration arguments of stepup.core.api.run() are accepted by all stepup.reprep.api functions.
  • An optional inp argument in compile_typst() to specify additional input files on which the typst source may depend. If not given, these dependencies are detected automatically after the typst compilation. When some of these additional inputs are the outputs of other steps, specifying them may improve scheduling efficiency.
  • A shell argument in wrap_git(), to run the wrapped git command through a shell.
  • The configuration file of srr-sync-zenodo may be written in YAML, JSON or TOML. The parser is selected by the suffix of the file name: .yaml and .yml, .json or .toml. YAML remains the recommended format, because it is the only one that supports comments. A file named .zenodo.json is rejected, because that is the metadata Zenodo reads when it archives a GitHub release, which follows a different schema.
  • A --dry-run option in srr-sync-zenodo, which validates the configuration, resolves the description, prints the metadata that would be sent to Zenodo and exits. Unlike an unset REPREP_ZENODO_TOKEN, this does not depend on the environment.
  • A custom_fields section in the configuration file of srr-sync-zenodo, holding the fields with which Zenodo extends the InvenioRDM record. The former top level code_repository field moved into this section, and seven fields are new: development_status, programming_languages, rights_holder, journal, meeting, imprint and thesis. Zenodo deploys them as code:developmentStatus, code:programmingLanguage, dc:rightsHolder, journal:journal, meeting:meeting, imprint:imprint and thesis:thesis. Everything that can be checked without contacting Zenodo is checked locally: the ISSN of a journal, the ISBN of an imprint, the URL and the identifiers of a meeting, and a thesis date that looks like a plain calendar date. There is no creator custom field, because it would repeat metadata.creators as free text, without ORCIDs or affiliations.
  • A metadata.languages field in the configuration file of srr-sync-zenodo, which takes ISO 639-3 language codes, e.g. eng for English.

Changed

  • The StepUp RepRep source code has been relicensed under LGPL-3.0-or-later. This clarifies that users of StepUp can assign any license of their choice to the workflows they create with StepUp (e.g., plan.py and related files). This has always been the intention, but with this change, it becomes legally explicit.
  • Compatibility with StepUp Core 4. This removes the block argument from every stepup.reprep.api function, and replaces the pool argument of convert_jupyter() and execute_papermill() by the more general resources argument. Existing plan.py files must be updated accordingly.
  • Replace all stepup.actions and stepup.tools entry points by standard console scripts, all with the srr- prefix. For example, stepup make-inventory became srr-make-inventory.
  • The convert_jupyter() function passes parameters to the notebook like papermill does: the new parameters argument takes a dictionary whose items are assigned in a cell inserted after the cell tagged parameters. The nbargs argument and the REPREP_NBARGS environment variable are gone, and so are the jupyter argument and the REPREP_JUPYTER environment variable, because the conversion calls the nbconvert Python API instead of the jupyter executable.
  • Rename environment variables for consistency:
    • REPREP_KEEP_TECTONIC_DEPS -> REPREP_TECTONIC_KEEP_DEPS
    • REPREP_KEEP_TYPST_DEPS -> REPREP_TYPST_KEEP_DEPS
  • The Zenodo synchronization has been refactored, including some breaking changes:
    • The srr-sync-zenodo command no longer amends input files. Instead, all inputs are determined when sync_zenodo() is called. This has two consequences:
      • The files to upload to Zenodo are no longer listed in the paths field of the configuration file. Instead, they are passed as the second argument of sync_zenodo() and as positional arguments of srr-sync-zenodo, after the configuration file. Both entry points reject duplicate file names and more than 100 files, which Zenodo does not accept.
      • The description is no longer taken from the path_readme field of sync_zenodo.yaml. Instead, it is passed with the path_description argument of sync_zenodo() and with the --description option of srr-sync-zenodo. (Markdown is converted to HTML automatically.)
    • The token must be specified in the REPREP_ZENODO_TOKEN environment variable, instead of being specified in the configuration file. The old variable REPREP_PATH_ZENODO_TOKEN is no longer supported. Note that REPREP_ZENODO_TOKEN holds the token itself, whereas REPREP_PATH_ZENODO_TOKEN held the path to a file containing the token. When REPREP_ZENODO_TOKEN is unset, srr-sync-zenodo validates the configuration and exits without contacting Zenodo. The endpoint stays a field of the configuration file, endpoint, which is optional and defaults to https://sandbox.zenodo.org/api. Because StepUp tracks the configuration file, switching between the sandbox and the production instance invalidates the step.
    • Unsupported keys in sync_zenodo.yaml now raise an error instead of being ignored, and values that must be strings are no longer coerced silently. For example, an unquoted version: 1.0 is rejected instead of becoming "1.0".
    • The identifiers of an award in the metadata.funding section are written like those of a meeting, as entries with an identifier and an optional scheme. They used to be entries with the scheme as the only key and the identifier as its value, which described the same thing in two ways within one configuration file.
    • Every field that takes an identifier from a controlled vocabulary of Zenodo is validated against the identifiers Zenodo has deployed: languages, license, resource_type, relation_type, development_status and programming_languages. An identifier that Zenodo does not know makes a deposit fail, so it is rejected locally instead, with an error message naming the vocabulary and its URL. Two resource types that srr-sync-zenodo used to accept are gone, because Zenodo no longer offers them: audio, for which video is the replacement, and publication-thesis, for which publication-dissertation is. The identifiers live in zenodo_vocabularies.yaml, which is regenerated from the Zenodo API by tools/update_zenodo_vocabularies.py in the source repository.
    • The fields that Zenodo requires are also required in sync_zenodo.yaml, and the length bounds that Zenodo enforces are checked locally: metadata.creators holds at least one creator, every creator carries a family_name, and metadata.title counts at least three characters.
    • The metadata.publisher field is required. It used to be optional, but Zenodo refuses to publish a record without a publisher, because it needs one to register a DOI, so a missing publisher is now reported before the record is created.
    • The metadata.version field may follow any convention, as long as it is a non-empty string of at most 191 characters. Zenodo stores it as free text and does not order the versions of a dataset by it, so srr-sync-zenodo only tests it for equality with the versions published on Zenodo. It creates a new version on Zenodo when the local version was never published, and refuses a version that is already published, which is usually a stale checkout or a revert instead of a new release. For the same reason, it refuses to synchronize a record that is no longer the latest published version of the dataset.
    • Only the metadata keys that carry a value are sent to Zenodo. The keywords are deposited as subjects only, because the InvenioRDM metadata has no keyword field. As observed on the sandbox on 2026-09-01, Zenodo silently dropped the unknown keywords key and accepted the empty values, so this does not change the records deposited with earlier versions.
    • When a value in sync_zenodo.yaml is rejected, the error message now explains what is wrong with it, instead of only stating that the value is invalid.
    • Everything that the user has to correct, and every request that Zenodo refuses, is reported as a message on standard error, after which srr-sync-zenodo exits with code 1. It used to re-raise the exception after printing the message, so the diagnosis was followed by a traceback, which for a rejected config file was a nested exception group of about forty lines, most of them pointing into the structuring code that cattrs generates. Only the errors that this command raises on purpose are reported this way. An unexpected error still ends in a traceback, because that is a bug to be reported.
    • The documentation and the example now recommend the names sync_zenodo.yaml for the configuration file and zenodo_description.md for the description, instead of zenodo.yaml and zenodo.md, because the old names are easily confused with the legacy .zenodo.json file. These names are only a convention: srr-sync-zenodo and sync_zenodo() accept any path.

    • The documentation of the Zenodo synchronization is split into three pages:

      1. A guide, which also explains how the legacy .zenodo.json file and the two Zenodo APIs relate to each other.
      2. A reference of the configuration file, with one section per top level key.
      3. A page with the relevant Zenodo vocabularies, which is generated from the Zenodo API.

      The access section, which decides who can see the record and download its files, is documented for the first time.

Removed

  • The stepup.reprep.tile_pdf module, with its Figure and Panel classes, has been removed, as it is no longer needed. This is easily replaced with a simple typst input.
  • The compile_typst() function no longer scans the sysinp dictionary for Path objects to automatically mark them as input dependencies.

Fixed

  • Many issues have been fixed in srr-sync-zenodo:
    • In sync_zenodo.yaml, a single value written where a list of strings belongs is read as a one element list. For example, keywords: coffee is the same as a list holding one keyword. This works for every field that takes a list of strings: keywords, license, languages, programming_languages and rights_holder. It already worked for license, whereas keywords was previously split into a list of characters. A value that is neither a string nor a list of strings is rejected with a message naming the expected type.
    • A record that carries no version on Zenodo is read instead of causing a KeyError. Zenodo does not require a version, so a record deposited through its web interface before srr-sync-zenodo was adopted may well have none. Such a record never matches the local version, so a new version is created for it.
    • The --clean option removes the drafts that Zenodo serves beyond the first page of the records of a user. As observed on 2026-08-30, that page holds at most 25 records, so the drafts of a user with more records than that were silently left behind.
    • A run that is given no files to upload no longer fails. It declared an empty list of uploads, which Zenodo rejects, after it had already created the record, so every run left another draft behind without recording its id. Note that Zenodo refuses to publish a record without files, as observed on the sandbox instance on 2026-08-30.
    • The publication date of a record that is already published is kept. The date of the build was sent with every metadata update, which moved the publication date of a published version to the day of the last build. A record that is not published yet is still dated on the day it was last synchronized, and a new version still gets the date on which it is created.
    • Every request sent to Zenodo times out after 60 seconds, instead of waiting forever for a connection that stalls. A request that cannot reach Zenodo at all is reported as a message, like a request that Zenodo refuses, instead of as a traceback.
    • A record is read as published only when Zenodo answers that its draft does not exist. Any other response, such as a refused token or an error of a gateway, is reported instead of being mistaken for a record without a draft.
    • The file name in the address of an upload is escaped, so a name holding a character that delimits a URL, such as # or ?, no longer truncates that address.
    • When the local files differ from those of a published version, the message explains that the version in the configuration file has to be changed, instead of only reporting the difference.
    • The metadata of a draft is sent before the files are uploaded, because Zenodo only accepts an upload to a record whose metadata has enabled its files. A draft that was created without files could therefore not be given any. The metadata is sent again after the upload, so that Zenodo also applies the order of the files and the default preview.
    • A rejected value is reported with the message of the validator alone. A value outside a fixed set, such as an unknown access.record, used to be followed by the internal representation of the attribute, and an unknown related.scheme by the validators of every scheme Zenodo knows.
    • The run after a failed upload finishes the draft that was left behind. A file is declared to Zenodo before its content is sent, so an upload that broke off halfway left a file name without content, which the next run compared to a checksum that Zenodo does not have yet. It reported an unexpected checksum format instead, which left --clean as the only way forward.
    • The directory of the file named by path_record_id is created when it does not exist, and a failure to write that file is reported as a message. Because the file is not an output of the step, its directory is not prepared for it, so a path_record_id inside a directory that a build does not create ended in a traceback right after the record was created on Zenodo.
    • A string written where a list of sections belongs, such as creators: Jane Doe, is reported once instead of once per character, and a scalar written where a section belongs names the type of the value.
    • The files to upload are checked for existence before Zenodo is contacted.
    • A paginated search result is no longer collected beyond a hundred pages, so a Zenodo instance that keeps sending full pages no longer causes an endless loop.
    • A record, a file entry or a commit response in which Zenodo leaves out the id, the checksum or the size is reported as a message instead of a KeyError.
    • A trailing slash on the endpoint in sync_zenodo.yaml is removed, so that it does not end up in the middle of every address.
  • srr-compile-tectonic only scans the error: lines of Tectonic’s standard error stream for missing input files. As of Tectonic 0.17, a halted run also dumps the engine transcript to that stream, from which paths were picked up that are no files at all.
  • The Jupyter kernels started by convert_jupyter() and execute_papermill() now communicate over ZeroMQ IPC sockets in a private temporary directory, instead of TCP sockets on localhost. When many notebooks were converted in parallel on a single machine, some conversions could fail with “Address already in use”, because jupyter_client selects free TCP ports before the kernel binds them.

3.1.11 - 2026-06-16

Switch to typst 0.15 and use its improved dependency tracking in compile_typst.

Changed

  • The compile_typst() function has improved dependency tracking, based on the new JSON deps format of typst 0.15. Older versions of typst are no longer supported.

3.1.10 - 2026-04-28

This is a minor bugfix release.

Fixed

  • Allow for HTML output files in compile_typst.

3.1.9 - 2026-03-22

Refactor wrap_git to support more flexible specification of inputs and outputs.

Added

  • Support for extra inp, env, out, and vol arguments in the wrap_git API. For example, this allows one to specify that the git command uses a file like .gitattributes. Changes to this file will then trigger the git command to be re-executed in the next build.

Changed

  • The old out argument of wrap_git is renamed to stdout to clarify that it only refers to the stdout of the git command, and to allow for other output files to be specified with the new out argument. Existing plan.py files must be updated accordingly.

Fixed

3.1.8 - 2025-12-28

Bugfix release

Changed

  • Replace pybtex by simpler built-in BibTeX parser in bibsane written in Lark. This eliminates a external dependencies, in particular latexcodec, which caused some issues.
  • Only report unused citations in bibsane, instead of dropping them.
  • Downgrade cattrs dependency to facilitate installation.
  • Updated reference test outputs for StepUp Core 3.2.

Fixed

  • Gracefully handle missing pagination in bibsane when reformatting page ranges.

3.1.7 - 2025-12-20

Bugfix release

Added

  • Bibsane can automatically add braces around words in titles that contain uppercase letters, to avoid unwanted lowercasing by BibTeX styles. This rule applies to titles and journal titles.

Changed

  • Distribute smaller source package.

Fixed

  • Correctly report merged records in bibsane’s screen output.
  • Fix handling of LaTeX encoding in bibsane and abbreviation of journal names with non-ASCII characters.

3.1.6 - 2025-11-21

Preliminary Tectonic support and minor fixes.

Added

  • Experimental support for the Tectonic LaTeX engine via the compile_tectonic() function.

Fixes

  • Minor cleanups.

3.1.5 - 2025-11-11

Update for Typst 0.14, refactored bibsane using pybtex.

Changed

  • Refactor bibsane to use pybtex instead of bibtexparser for reading and writing BibTeX files. Several minor issues have been fixed in the process, which result in a small change in behavior. Most notably, the indentation style has changed because this is yet not configurable in pybtex.
  • Rename wrap_git action to wrap-git for consistency with other StepUp actions.
  • Update compile_typst to work with Typst 0.14.

3.1.4 - 2025-10-18

Improved sync_zenodo function.

Added

Fixed

  • Fix Zenodo draft publication REST API.

3.1.3 - 2025-09-28

Minor API improvements.

Changes

  • Add support for other objects than dict for the sysinp argument of compile_typst().
  • The dictionary passed to the sysinp argument of compile_typst() is now sanitized to only contain strings as keys and values. Integer, float and path values are converted to strings automatically.
  • Add sanity check on the positional argument of sanitize_bibtex() to ensure that it is a .bib file.

3.1.2 - 2025-08-25

This is a minor feature release.

Added

  • The function scan_latex_deps now has a amend=True optional argument. If it is used in a workflow, it will by default amend the current step with all scanned TeX sources as inputs.

3.1.1 - 2025-06-25

Minor improvements (Zenodo synchronization, LaTeX dependencies).

Added

  • Detect dependencies in LaTeX from includepdf commands.
  • Add copyright field to Zenodo metadata.

Fixed

  • Fix minor documentation inaccuracies.

3.1.0 - 2025-06-23

More powerful sync_zenodo command that uses the Invenio RDM API.

Changed

  • Switch from the official Zenodo API to the Invenio RDM API. This is a breaking change, with a new layout of the zenodo.yaml configuration file. See the Zenodo synchronization documentation for details.
  • Add support for more metadata fields in the zenodo.yaml configuration file. This includes support for keywords, multiple affiliations, RORs, related identifiers, funding information and code repository.
  • Add the --clean option to sync_zenodo to remove all draft uploads from Zenodo. (Mainly useful for testing purposes.)
  • Human-readable output when the zenodo.yaml configuration file contains schema errors.

3.0.5 - 2025-06-22

This is a bugfix release.

Fixed

  • Ignore links returned by the Zenodo API that are not strings in sync_zenodo.
  • Amend a make_inventory step with all files in an inventory as inputs.

3.0.4 - 2025-06-21

Support for ORCID field in sync_zenodo.

Changed

  • Added ORCID field to the author metadata in the sync_zenodo configuration file.

3.0.3 - 2025-06-14

Replace a few more markdown imports.

3.0.2 - 2025-06-14

Small improvements and a wrapper for git commands that depend on the commit id.

Added

  • wrap_git() to define shell commands (typically git ...) that need to be re-executed when the current git branch or commit changes.

Changed

  • Use the markdown-it-py package instead of markdown for more precise and faster markdown rendering.

Fixed

  • Make flatten_latex() work with other file extensions.

3.0.1 - 2025-05-31

Integration with papermill to execute notebooks, and a few bug fixes.

Added

Fixed

  • Fixed outdated information in the Zenodo synchronization documentation, and fixed corresponding outdated code.
  • Remove linear=True argument when saving a PDF with MuPDF because it is no longer supported as of MuPDF 1.26. (It was not terribly useful in the first place.) For more details, see https://artifex.com/blog/mupdf-removes-linearisation.

3.0.0 - 2025-05-11

Major release with breaking changes, compatible with StepUp Core 3.

Changed

  • Breaking:
    • Compatibility with StepUp Core 3.
    • Migrated render_jinja() to StepUp Core 3.

2.3.6 - 2025-04-24

Make sanitize_bibtex() work without LaTeX and add support for TOML files in render_jinja().

Added

  • Support for TOML files in render_jinja().

Changed

  • Make sanitize_bibtex() usable without LaTeX.
  • New output files in the LaTeX fls file are treated as volatile outputs.

2.3.5 - 2025-03-13

Bug fix in rr-cat-pdf and simplify journal abbreviation in rr-bibsane. This requires an update the bibsane.yaml configuration file.

Fixed

  • The --insert-blank opton of rr-cat-pdf was always active, even when not present on the command line, which is now fixed.

  • The abbreviate_journal feature of bibsane used to work with a cache file, which was a remnant from an older implementation. This no longer made much sense (because pyiso4 is fast enough) and it cache files may cause non-reproducible behavior. (They are both inputs and outputs.) For these reasons, the cache feature is replaced by two configuration fields in bibsane.yaml:

    • A boolean flag abbreviate_journals to enable abbreviations.
    • An optional mapping custom_abbreviations with abbreviation overrides for when pyiso4 does not give the desired result.

An external file with abbreviations is no longer needed. They are just included in the bibsane.yaml file. (This also means that old bibsane.yaml config files may need to be updated.) See Sanitizing BibTeX files for more details.

2.3.4 - 2025-03-09

Improved handling of LaTeX fls file and refactored make_inventory() function.

Changed

  • The arguments of make_inventory() are now variadic. Files to be included are passed as positional arguments and the last positional argument is the inventory file to be written. An optional argument path_def can be used to specify an inventory definition file.
  • The include-git and exclude-git commands in an inventory definition now accept arguments, which are passed to the git ls-files command.
  • LaTeX output files inferred from the .fls file are filtered in the same ways as input files. LaTeX sometimes writes output files to ~/.texlive2023/ which should be ignored by StepUp.

Fixed

  • rr-flatten-latex now correctly handles empty tex files.

2.3.3 - 2025-03-03

A few minor improvements related to LaTeX and BibTeX, and parameterized Jupyter notebooks.

Added

  • Improve detection of inputs and (volatile) outputs when compiling a LaTeX document:
    • The -recorder option of latex is used to identify inputs and outputs more precisely.
    • The manual override %REPREP input is replaced by %REPREP inp.
    • New manual overrides are supported for (volatile) outputs: %REPREP out and %REPREP vol.
  • Add nbargs option to convert_jupyter(), to call a notebook with arguments. If a list or a dict is given, the data is convert to JSON.

Changed

  • Add overwrite option to sanitize_bibtex(), to flag that path_out refers to an input file to be overwritten, instead of trying to track it as an output file.

2.3.2 - 2025-02-24

This release makes compile_typst compatible with Typst 0.13, and drops support for markdown-katex.

Added

  • Configuration for development with devenv

Changed

  • Drop support for the ailing markdown_katex integration. (Typst can be used to achieve similar results much more efficiently).
  • Update compile_typst for Typst 0.13

2.3.1 - 2025-02-12

This is a minor bugfix release.

Fixed

  • Remove some debug output.

2.3.0 - 2025-02-12

This release adds support for Jupyter notebooks with convert_jupyter() and introduces small breaking changes to the API. Other noteworthy changes include new options to the compile_typst() function, more ways to specify variables in render_jinja().

Added

  • Execute and convert Jupyter notebooks with convert_jupyter().
  • rr-bibsane is now part of StepUp RepRep, instead of using the (retired) bibsane package. The main difference, other than the improved integration with StepUp RepRep, is that journal abbreviations are now generated with pyiso4 instead of the abrevvIso Web API. It has a corresponding sanitize_bibtex() function in stepup.reprep.api.

Changed

  • Extend compile_typst() with additional options:
    • Specification of the output file.
    • Key-value pairs for the --input argument.
    • PNG and SVG output formats (multipage is not working yet).
    • Optional inventory output file.
    • Pass-through arguments for typst
  • Breaking changes to existing API:
    • convert_pdf() and related functions are renamed:
    • convert_svg() and related functions are renamed:
    • The inkscape_args of convert_inkscape() must now be a list instead of a string.
    • compile_latex() no longer creates an inventory file by default. To recover the old behavior, add inventory=True to the arguments or set the environment variable REPREP_LATEX_INVENTORY="1".
    • compile_latex() no longer calls bibsane when the LaTeX source has a BibTeX bibliography. If you want to sanitize the BibTeX file, call sanitize_bibtex() after compile_latex().
    • The paths_variables argument of render_jinja() has been replaced by a variadic positional parameter (i.e. *paths_variables).
  • Other changes
    • Change convert_weasyprint() to perform the conversion in a single step.
    • Improve handling of arguments and dependencies in convert_markdown()
    • render_jinja() now accepts JSON and YAML files with variables for Jinja2 templates. In addition, one may specify a dictionary with variables directly when calling the function.
    • Documentation improvements.

2.2.3 - 2025-02-05

This release uses the new STEPUP_EXTERNAL_SOURCES environment variable introduced in StepUp Core 2.0.6.

Changed

  • Switch from SemVer to EffVer.
  • Scripts that automatically detect dependencies (rr-compile-latex, rr-compile-typst and rr-convert-inkscape) now also use the new STEPUP_EXTERNAL_SOURCES environment variable introduced in StepUp Core 2.0.6.

2.2.2 - 2025-01-31

This is a bugfix release addressing minor issues in the Typst support.

Fixed

  • Only call sys.exit in rr-compile-typst if the returncode is non-zero. This allows for other scripts to call its main function without exiting in case of a succeeded typst build.

2.2.1 - 2025-01-31

This is a bugfix release addressing minor issues in the Typst support.

Fixed

  • Fix returncode of the rr-compile-typst command (now passes through returncode of typst build).
  • By default, depfiles created by typst build are now stored in a temporary directory, to avoid littering the working directory. You can keep depfiles by setting the environment variable REPREP_TYPST_KEEP_DEPS="1", or with the argument keep_deps=True in compile_typst(). (Either one will .)

2.2.0 - 2025-01-29

This release adds experimental support for typst. It also introduces a few breaking API changes for the sake of consistency. (More breaking changes should be expected in the near future.)

Added

  • Experimental support for Typst.

Changed

2.1.0 - 2025-01-27

This release contains a few minor breaking changes for the sake of internal consistency.

Changed

  • The conversion with Inkscape has become a single step. (It was first split up in a step scanning for dependencies followed by the actual conversion.)
  • Add more entrypoints for command-line utilities in StepUp RepRep. Existing ones were renamed from reprep-* to rr-*.
  • Rename module stepup.reprep.render to stepup.reprep.render_jinja.
  • Rename API function: render() -> render_jinja()

2.0.2 - 2025-01-22

This is the first release of StepUp RepRep that is compatible with StepUp Core 2.0.0. (Earlier 2.0 releases were yanked due to packaging issues.)

Added

  • Add smarty extension to markdown conversion.
  • Add option to insert blank page after odd-paged PDF when concatenating PDFs.

Changed

  • Compatibility with StepUp Core 2.0.0, which breaks compatibility with older StepUp Core versions.

Fixed

  • Use shlex for building shell commands in stepup.reprep.api to avoid shell injection.

1.4.1 - 2024-09-02

Added

  • An extra argument was added to convert_markdown to specify CSS files.
  • Support default arguments for convert_markdown defined as environment variables: ${REPREP_KATEX_MACROS} and ${REPREP_MARKDOWN_CSS}.

Fixed

  • Fix bug: put header output of markdown_katex plugin in the HTML header.
  • Fix bug: rewrite paths to CSS files in convert_markdown to be relative to the parent of the output HTML file.

1.4.0 - 2024-08-27

Added

  • The unplot script, a sanitized version of Depix. It converts paths from SVG files back into data, which can be used to reverse-engineer data sets from plots.

Changed

  • Move load_module_file from StepUp Core to stepup.reprep.render and improve it to facilitate local imports.

1.3.0 - 2024-06-28

Added

  • The script rr-sync-zenodo and corresponding StepUp API function sync_zenodo() synchronize your local data with a draft dataset on Zenodo.
  • Small documentation updates

Fixed

1.2.1 - 2024-05-27

Changed

  • Conversion from HTML to weasyprint is now a two-step process and includes detection of implicit input files used in the HTML to PDF conversion. (Images and external CCS)
  • Improved reusability of script modules: add_notes_pdf, check_hrefs, convert_inkscape, convert_markdown, convert_weasyprint, latex, latex_flat, make_inventory, normalized_pdf, nup_pdf, raster_pdf, render and zip_inventory.

Fixed

  • Fixed a few errors in the HTML generated by convert_markdown()

1.2.0 - 2024-05-20

Added

  • rr-zip-inventory command to manually create a reproducible ZIP file from an inventory.txt file.
  • More documentation on how to work with inventory files.
  • Tutorial for archiving StepUp publication Git repositories.

Changed

  • Renamed all MANIFEST and manifest occurrences to inventory and removed dependency of setuptools for processing such files.
  • The API of make_inventory is made simpler than that of make_manifest.
  • The commands supported in inventory.def files now differ from those in setuptools: include, exclude, include-git, exclude-git, include-workflow and exclude-workflow.
  • The css style has been made customizable in convert_markdown.
  • KaTeX is now optional in convert_markdown.

Fixed

  • An error message is raised when trying to a put a directory in an inventory file.
  • Symbolic links are no longer dereferenced when they are listed in an inventory file.
  • Symbolic links are archived in ZIP files without dereferencing.

1.1.2 - 2024-05-16

Fixed

1.1.1 - 2024-05-07

Fixed

Changed

  • Documentation improvements

1.1.0 - 2024-05-02

Changed

  • Documentation improvements
  • Unit tests are made compatible with StepUp Core 1.2.0.

1.0.0 - 2024-04-25

Initial release