fpdf2 latest news

I wrote my last post on fpdf2 18 months ago.
We released 7 more versions of fpdf2 since then!

This article will present some of the major features introduced since v2.7.3 to v2.8.1 of fpdf2:
click on the buttons below to reveal the various changes brought to the library.

fpdf2 joined the @py-pdf GitHub organization, that also hosts pypdf, PyPDF-Builder, pdfly and pypdf_table_extraction.
The reasons for this move were detailed in discussion #752.

fpdf2 also welcomes two new maintainers: Anderson Herzogenrath da Costa (@andersonhc) and Georg Mischler (@gmischler).
More details about them in discussion #896 & discussion #898.

A new FPDF.set_text_shaping() method has been introduced to perform text shaping using Harfbuzz, thanks to @andersonhc: documentation page on text shaping.

Previously only supported when rendering SVG images,
quadratic & cubic Bezier curves can now be directly rendered using the new FPDF.bezier() method, thanks to @awmc000: documentation on Bezier curves.

Support for Free Text annotations was added by @MarekT0v in PR #1039 : documentation on Free Text annotations.

We added more guides detailing how to combine fpdf2 with other libraries:

Several improvements were made regarding tables :

• links, padding & vertical alignment in cells with v_align
• new parameters gutter_height, gutter_width and wrapmode
• control over outer border width
• allowing for multiple heading rows, and control over headings repetition over pages
• custom cell_fill_mode logic can now be provided
• cells that span multiple rows via the rowspan attribute, which can be combined with colspan

Various improvements regarding our basic HTML renderer:

• support for CSS page breaks properties
• support for start & type attributes of <ol> tags, and type attribute of <ul> tags
• tag_styles to control the font, color, size & indent of most HTML elements
• li_prefix_color to control the color of list prefixes (bullets & numbers)

• FPDF.write_html() now uses the new FPDF.table() method to render <table> tags.
• font aliases are deprecated (Arial → Helvetica, CourierNew → Courier, TimesNewRoman → Times), they will be removed in a later release.
• to improve naming consistency, the txt parameters of FPDF.cell(), FPDF.multi_cell(), FPDF.text() & FPDF.write() have been renamed to text.
• the split_only optional parameter of FPDF.multi_cell() has been replaced by two new distincts optional parameters: dry_run & output.
• fpdf.TitleStyle has been renamed into fpdf.TextStyle.
• we removed an obscure and undocumented feature of FPDF.write_html(), which used to magically pass local variables as arguments.

In order to do “our part” to contribute to a more secure Python ecosystem,
fpdf2 releases are now performed using Pypi Trusted Publishers : Pypi blog announcement.

New tutorials :

Organizational changes
Text shaping with Harfbuzz
Bezier curves
Free Text annotations
Other additions
Combining with other libs
Tables improvements
HTML
Performance improvements
Uniformisation
Pypi Trusted Publisher
New tutorials

I would also like to celebrate that, as spotted by @gmischler in discussion #1232, fpdf2 has exceeded 1000 stars on GitHub!
Yes, it’s a bit superficial, but I’d like to thank all the programmers who have expressed their support for fpdf2: it makes me happy to know that this project is useful to you 🙂

Because such retrospective time is also an opportunity to reflect on things that did not go well,
I would like to mention a few failures:

• we’ve had a couple of Pull Requests left unanswered for almost a year… 😞 I will try to be more vigilant about this, as I know not having feedback can really kill developers’ motivation.

• I spotted a bug while using the excellent pdfly CLI tool to extract pages from a PDF document built with fpdf2: the resulting file was excessively big. After adding a unit test to pdfly to investigate (in PR #45), I realized that this was due to a slightly invalid structure in the PDF documents produced by fpdf2. All the PDF readers I tested did not care, but the pypdf library (used by pdfly) was using this information to properly extract pages from a file. This is now all fixed!

• I initially performed a release for fpdf2 version 2.8.0, but then erroneously removed it on Pypi, erroneously fearing a regression 😅 Versions 2.8.0 & 2.8.1 are exactly the same.

Finally, Hacktoberfest is currently running, and we have many issues up-for-grabs at fpdf2, so come help and contribute some code! 🙂