FPDF.bezier() to render quadratic and cubic Bézier curves, and many improvements

Note: a temporary 2.8.0 version was released but then erroneously removed on Pypi.
Versions 2.8.0 & 2.8.1 are exactly the same.

Added

• support for quadratic and cubic Bézier curves with FPDF.bezier() - thanks to @awmc000
• support for escape character for markers in markdown text issue #1215
• Wrapping words on spaces now considers all common space symbols in addition to regular spaces (' '), addressing issues with word-wrapping for languages like Thai, as per #1190 and #1191.
• Templates can now be also defined in JSON files.
• support to optionally set wrapmode in templates (default "WORD" can optionally be set to "CHAR" to support wrapping on characters for scripts like Chinese or Japanese) - cf. #1159 - thanks to @carlhiggs
• documentation on how to use fpdf2 with Rough.js: link to docs
• documentation on how to use fpdf2 with gunicorn: link to docs
• new translation of the tutorial in Türkçe, thanks to @Natgho: Türkçe
• feature to identify the Unicode script of the input text and break it into fragments when different scripts are used, improving text shaping results
• FPDF.image(): now handles keep_aspect_ratio in combination with an enum value provided to x
• FPDF.write_html(): now supports CSS page breaks properties : documentation
• FPDF.write_html(): new optional font_family parameter to set the default font family
• FPDF.write_html(): spacing before lists can now be adjusted via the tag_styles attribute - thanks to @lcgeneralprojects
• file names are mentioned in errors when fpdf2 fails to parse a SVG image

Fixed

• FPDF.local_context() used to leak styling during page breaks, when rendering footer() & header() - MR: #1207
• fpdf.drawing.DeviceCMYK objects can now be passed to FPDF.set_draw_color(), FPDF.set_fill_color() and FPDF.set_text_color() without raising a ValueError: documentation.
• FPDF.write_html(): fixing rendering of <hr> tags, that do not trigger a page break anymore
• FPDF.write_html(): fixed automatic page break when an image does not have enough vertical space to be rendered on a page
• individual /Resources directories are now properly created for each document page. This change ensures better compliance with the PDF specification but results in a slight increase in the size of PDF documents. You can still use the old behavior by setting FPDF().single_resources_object = True
• line size calculation for fragments when text shaping is used
• FPDF.write_html(): fixed incoherent indentation of long <ul> list entries - cf. issue #1073 - thanks to @lcgeneralprojects
• default values for top_margin and bottom_margin in HTML2FPDF._new_paragraph() calls are now correctly converted into chosen document units.
• In text_columns(), paragraph top/bottom margins didn't correctly trigger column breaks; issue #1214
• fpdf.drawing.color_from_hex_string did not test or mention accepting lowercase hex values.
• handling of bidirectional text on FPDF.get_string_width() [issue #1231]
• new translation of the tutorial in Indonesian - thanks to @odhyp
• RecursionError in some cases when calling FPDF.write_html() inside FPDF.footer() - issue #1222

Removed

• support for Python 3.7, that reached end-of-life in 2023
• an obscure and undocumented feature of FPDF.write_html(), which used to magically pass instance attributes as arguments.

Deprecated

• fpdf.TitleStyle has been renamed into fpdf.TextStyle
• FPDF.write_html(): tag_indents introduced in the last version - Now the indentation can be provided through the tag_styles parameter, using the .l_margin of TextStyle instances

Changed

• FPDF.local_context() used to treat font_size as a value in points. Now this is the role of font_size_pt, whereas font_size allows to set the font size into chosen document units (specified with FPDF(unit=)) - MR: #1207
• FPDF.circle() : the previous r parameter, that in fact defined the diameter, has been replaced by a new radius paremeter. The x & y parameters now define the circle center, instead of its top-left corner as it used to be - issue #1245
• FPDF.table() now raises an error when a single row is too high to be rendered on a single page
• FPDF.write_html(): indentation of HTML elements can now be non-integer (float), and is now independent of font size and bullet strings.
• improved performance of font glyph selection by using functools cache
• Clarified usage of the style attribute in FPDF.add_font()

Improvements on `table()` and `write_html()`, and several bugfixes.

Added

• new optional parameter repeat_headings for FPDF.table() that indicates whether to print table headings on every page
• support for overriding paragraph direction on bidirectional text
• new optional li_prefix_color parameter for FPDF.write_html()
• support for start & type attributes of <ol> tags, and type attribute of <ul> tags, when using FPDF.write_html()
• FPDF.write_html() now accepts a tag_styles parameter to control the font, color & size of HTML elements: <a>, <blockquote>, <li>...
• FPDF.write_html() now accepts a tag_indents parameter to control, for example, the indent of <blockquote> elements
• FPDF.write_html() now honors line-height attributes on <ol> & <li> elements, or the same CSS property in style attributes on those tags
• allow to define custom cell_fill_mode logic for tables: Set cells background - documentation section. Also added 2 new values: TableCellFillMode.EVEN_ROWS & TableCellFillMode.EVEN_COLUMNS: documentation

Fixed

• a bug when rendering vector images with dashed lines that caused a warning message in Adobe Acrobat Reader
• ordering RTL fragments on bidirectional texts
• fixed type hint of member level in class OutlineSection from str to int.
• SVG clipping paths being incorrectly painted - cf. issue #1147]
• new translation of the tutorial in Polski - thanks to @DarekRepos

Changed

• improved the performance of FPDF.start_section() - cf. issue #1092

Deprecated

• The dd_tag_indent & li_tag_indent parameters of FPDF.write_html() are replaced by the new tag_indents generic parameter.
• The heading_sizes & pre_code_font parameters of FPDF.write_html() are replaced by the new tag_styles generic parameter.

Table improvements, bidirectional text support, support for <path> in SVG and several bug fixes

Added

• support for <path> elements in SVG <clipPath> elements
• support for bidirectional text shaping - thanks to @andersonhc
• documentation on how to combine fpdf2 with mistletoe in order to generate PDF documents from Markdown (link)
• tutorial in Dutch: Handleiding - thanks to @Polderrider
• support for Table cells that span multiple rows via the rowspan attribute, which can be combined with colspan - thanks to @mjasperse
• TableSpan.COL and TableSpan.ROW enums that can be used as placeholder table entries to identify span extents - thanks to @mjasperse

Fixed

• when adding a link on a table cell, an extra link was added erroneously on the left. Moreover, now FPDF._disable_writing() properly disable link writing.
• FPDF.write_html() now handles linking directly to other pages - thanks to @mjasperse
• non-bold TitleStyle is now rendered as non-bold even when the current font is bold
• calling .table() inside the render_toc_function
• using .set_text_shaping(True) & .offset_rendering()
• fixed gutter handing when a pagebreak occurs within a table with header rows - thanks to @mjasperse
• fixed handling of border=0 in HTML table - thanks to @mjasperse
• FPDF.write_html() now properly honors align= attributes in <th> tags
• fixed problem using bold italic standard fonts in markdown - thanks to @Alan-Collins

Changed

• refactored FPDF.multi_cell() to generate fewer PDF component objects - thanks to @mjasperse
• outer table borders are now drawn continuously for nonzero gutter_width/gutter_height, with spacing applied inside the border similar to HTML tables - thanks to @mjasperse - cf. #1071
• removed the requirement that all rows in a Table have the same number of columns - thanks to @mjasperse

Deprecated

• font aliases (Arial → Helvetica, CourierNew → Courer, TimesNewRoman → Times). They will be removed in a later release.

Support for Images and Clipping Paths in SVG files - Python 3.12 Support - Several Bugfixes

Added

• Basic support for <image> elements in SVG vector graphics inserted
• SVG importing now supports clipping paths, and <defs> tags anywhere in the SVG file - thanks to @afriedman412 - cf. #968
• FPDF.fonts.FontFace: Now has a static combine method that allows overriding a default FontFace (e.g. for specific cells in a table). Unspecified properties of the override FontFace retain the values of the default - thanks to @TedBrookings - cf. #979
• TextColumns() can now have images inserted (both raster and vector) - thanks to @gmischler
• TextColumns() can now advance to the next column with the new new_column() method or a FORM_FEED character (\u000c) in the text - thanks to @gmischler
• Added support for Free Text annotations: documentation - thanks to @MarekT0v - cf. #1039
• Tutorial in Dutch: Handleiding - thanks to @Polderrider
• Python 3.12 is now officially supported

Fixed

• Links over text in tables were broken in release 2.7.6, this is now fixed
• FPDF.set_font_color() raised a TypeError when used in tables
• FPDF.image(x=Align.C) used to fail for SVG images - fixed thanks to @gmischler - cf. #1003
• Previously set dash patterns were not transferred correctly to new pages - fixed thanks to @gmischler - cf. #993
• Inserted Vector images used to ignore the keep_aspect_ratio argument.
• FPDF.write_html() now properly honor the current text font color when styling table cells
• FPDF.write_html() delays unescaping data so as not to confuse entity names as nested tags
• FPDF.multi_cell() has improved handling of new_x and new_y when padding is non-zero.
• FPDF.multi_cell(fill=True) now avoids overlapping multiline strings when padding is non-zero.

Changed

• the public .images, .icc_profiles & .image_filter attributes of FPDF instances have been moved inside a nested FPDF.image_cache attribute. Similarly, the FPDF.preload_image() is now a function in the fpdf.image_parsing module: documentation
• the fpdf.svg module now produces WARNING log messages for unsupported SVG tags & attributes.
  If those logs annoy you, you can suppress them: logging.getLogger("fpdf.svg").propagate = False
• FPDF.table(): If cell styles are provided for cells in heading rows, combine the cell style as an override with the overall heading style.

Improvements to FPDF.table() - Text regions layout management - several bugfixes

This release is the first performed from the @py-pdf GitHub org, where fpdf2 migrated.
This release also marks the arrival of two new maintainers: Georg Mischler (@gmischler) and Anderson Herzogenrath da Costa (@andersonhc).

Added

• The new experimental method text_columns() allows to render text within a single or multiple columns, including height balancing: documentation
• FPDF.write_html() now supports heading colors defined as attributes (e.g. <h2 color="#00ff00">...)
• FPDF.table(): Now supports padding in cells : documentation
• FPDF.table(): Now supports vertical alignment in cells : documentation
• FPDF.table(): Now supports outer border width for rendering the outer border of the table with a different line-width
• FPDF.table(): Now supports multiple heading rows : documentation
• documentation on how to use livereload to enable a "watch" mode with PDF generation: Combine with livereload

Changed

• The formatting output by write_html() has changed in some aspects. Vertical spacing around headings and paragraphs may be slightly different, and elements at the top of the page don't have any extra spacing above anymore.
• FPDF.table(): If the height of a row is governed by an image, then the default vertical alignment of the other cells is "center". This was "top".
• variable-width non-breaking space (NBSP) support issue #834
  This change was made for consistency between row-height governed by text or images. The old behaviour can be enforced using the new vertical alignment parameter.

Fixed

• In multi_cells and table cells with horizontal padding, the text was not given quite enough space.
• write_html() can now handle formatting tags within paragraphs without adding extra line breaks (except in table cells for now).
• the font size in HTML <pre> and <code> tags is not fixed to 11 pica anymore, but adapts to the preceding text.
• FPDF.ln(), when called before any text has been written, will now use the current font height instead of doing nothing - cf. issue #937
• FPDF.image(), when provided a BytesIO instance, does not close it anymore - cf. issue #881
• Invalid characters were being generated when a string contains parentheses - cf. issue #884
• Frozen Glyph dataclass was causing problems for FPDFRecorder with TTF fonts - cf. issue #890
• Edge case when parsing a Markdown link followed by a newline - cf. issue #916, and when bold/italics/underline markers are repeated
• Zoom not set correctly when a numeric value was set in set_display_mode() - cf. issue #926
• FPDF.table(): images no longer overlap with cell borders - cf. issue #892
• Encryption of strings containing non-latin characters - cf. issue #933
• Handling of fragments with zero-length - cf. issue #902

Deprecated

• to improve naming consistency, the txt parameters of FPDF.cell(), FPDF.multi_cell(), FPDF.text() & FPDF.write() have been renamed to text

Text shaping using Harfbuzz - New mirror() method - New AES-256 encryption - Several bugfixes

Added

• FPDF.set_text_shaping(): new method to perform text shaping using Harfbuzz - documentation - thanks to @andersonhc in PR #820
• FPDF.mirror() - New method: documentation page - Contributed by @sebastiantia
• FPDF.table(): new optional parameters gutter_height, gutter_width and wrapmode. Links can also be added to cells by passing a link parameter to Row.cell()
• FPDF.multi_cell(): has a new optional center parameter to position the cell horizontally at the center of the page
• New AES-256 encryption: documentation - thanks to @andersonhc in PR #872
• Added tutorial in Khmer language: ភាសខ្មែរ - thanks to @kuth-chi
• Added tutorial in 日本語 - thanks to @alcnaka
• Better documentation & errors when facing HTML rendering limitations for <table> tags: https://pyfpdf.github.io/fpdf2/HTML.html

Fixed

• FPDF.table(): the colspan setting has been fixed - documentation
• FPDF.image(): allowing images path starting with data to be passed as input
• text overflow is better handled by FPDF.write() & FPDF.write_html() - cf. issue #847
• the initial text color is preserved when using FPDF.write_html() - cf. issue #846
• PDF metadata not encrypted - cf. issue #865
• handle superscript and subscript correctly when rendering TextLine- thanks to @Tolker-KU - cf. Pull Request #862
• make sure warnings always point to the users code - cf. Pull request #869

Deprecated

• the center optional parameter of FPDF.cell() is no more deprecated, as it allows for horizontal positioning, which is different from text alignment control with align="C"

Images can now be embedded as CMYK - Docs for using Pygal & FastAPI - Various bugfixes

Added

• FPDF.image(): CMYK images can now be inserted directly by passing them into the image method. Contributed by @devdev29
• documentation on how to embed graphs and charts generated using Pygal lib: documentation section - thanks to @ssavi-ict
• documentation on how to use fpdf2 with FastAPI: https://pyfpdf.github.io/fpdf2/UsageInWebAPI.html#FastAPI - thanks to @KamarulAdha
• FPDF.write_html(): <table> elements can now be aligned left or right on the page using align=
• FPDF.write_html(): a custom font can now be specified for <code> & <pre> elements, using the new optional parameter pre_code_font

Fixed

• FPDF.table(): images no more overflow cells
• FPDF.table(): text overflow in the last cell of the header row is now properly handled
• FPDF.table(): when align="RIGHT" is provided, the page right margin is now properly taken in consideration

Changed

• FPDF.write_html() does not render the top row as a header, in bold with a line below, when no <th> are used, in order to be more backward-compatible with earlier versions of fpdf2 - cf. #740

Deprecated

• the split_only optional parameter of FPDF.multi_cell(), which is replaced by two new distincts optional parameters: dry_run & output

Removing debug print() statement

Fixed

• removed a debug print() statement left in output.py:OutputProducer._add_fonts() 🤦‍♂️ - A rule was also added to .pre-commit-config.yaml to avoid this to happen again.

SVG & table() bugfixes

Fixed

• custom fonts can be used with FPDF.table() without triggering a TypeError: cannot pickle 'dict_keys' object - thanks @aeris07 for the bug report
• the SVG parser now accepts <rect> with width / height defined as percents

New methods table() and set_fallback_fonts() - keep_aspect_ratio - ICC Profiles

Added

• new method FPDF.table(): documentation
• FPDF.image() has a new keep_aspect_ratio optional boolean parameter, to fit it inside a given rectangle: documentation
• FPDF.multi_cell() and FPDF.write() now accept a wrapmode argument for word or character based line wrapping ("WORD"/"CHAR").
• new methods: FPDF.preload_image() & FPDF.use_font_style()
• new method FPDF.set_fallback_fonts() allow alternative fonts to be provided if a character on the text is not available on the currently set font - thanks to @andersonhc
• for inserted images that have an ICC Profile, this profile is now extracted and embedded; they should now be honored by PDF viewers - thanks to @eroux
• new translation of the tutorial in 简体中文 - thanks to @Bubbu0129
• documentation on how to embed static Plotly charts: link to docs
• additional linter / static code analyser in GitHub Actions CI pipeline: semgrep

Fixed

• outlines & hyperlinks were not working on encrypted files - thanks to @andersonhc
• a bug was introduced in the previous release (2.6.1): FPDF.set_link() could not update links generated with add_link()
• unicode (non limited to ASCII) text can now be provided as metadata #685
• all TitleStyle constructor parameters are now effectively optional
• memory usage was reduced by 10 MiB in some cases, thanks to a small optimization in using fonttools

Changed

• FPDF.write_html() now uses the new FPDF.table() method to render <table> tags. As a consequence, vertical space before <table> tags has sometimes been reduced.

• vector images parsing is now more robust: fpdf2 can now embed SVG files without viewPort or no height / width
• bitonal images are now encoded using CCITTFaxDecode, reducing their size in the PDF document - thanks to @eroux
• when possible, JPG and group4 encoded TIFFs are now embedded directly without recompression - thanks to @eroux

Removed

• FPDF.write_html() now uses the new FPDF.table() method to render <table> tags. As a consequence, it does not support the height attribute defined on <td> / <th> tags anymore, nor height / width attributes defined on <img> tags inside cells, nor width attributes defined on <thead> / <tfoot> tags.