Svglib

A pure-Python library for reading and converting SVG

About

Svglib is a Python library for reading SVG files and converting them (to a reasonable degree) to other formats using the ReportLab Open Source toolkit.

Used as a package you can read existing SVG files and convert them into ReportLab Drawing objects that can be used in a variety of contexts, e.g. as ReportLab Platypus Flowable objects or in RML. As a command-line tool it converts SVG files into PDF ones (but adding other output formats like bitmap or EPS is really easy and will be better supported, soon).

Tests include a huge W3C SVG test suite plus ca. 200 flags from Wikipedia and some selected symbols from Wikipedia (with increasingly less pointing to missing features).

Features

• convert SVG files into ReportLab Graphics Drawing objects
• handle plain or compressed SVG files (.svg and .svgz)
• allow patterns for output files on command-line
• install a Python package named svglib
• install a Python command-line script named svg2pdf
• support SVG linear and radial gradients
• support SVG <switch> elements with conditional rendering
• provide a PyTest test suite with over 90% code coverage
• test entire W3C SVG test suite after pulling from the internet
• test all SVG flags from Wikipedia after pulling from the internet
• test selected SVG symbols from Wikipedia after pulling from the net
• support Python 3.9+ and PyPy3

Known limitations

• @import rules in stylesheets are ignored. CSS is supported, but the range of supported attributes is still limited
• clip paths support path, rect, circle, ellipse and polygon shapes, but only one clip-rule per clipPath.
• SVG mask elements are not supported
• SVG ForeignObject elements are not supported.

Examples

You can use svglib as a Python package e.g. like in the following interactive Python session:

>>> from svglib.svglib import svg2rlg
>>> from reportlab.graphics import renderPDF
>>>
>>> drawing = svg2rlg("file.svg")
>>> renderPDF.drawToFile(drawing, "file.pdf")

Note that the second parameter of drawToFile can be any Python file object, like a BytesIO buffer if you don't want the result to be written on disk.

In addition a script named svg2pdf can be used more easily from the system command-line. Here is the output from svg2pdf -h:

usage: svg2pdf [-h] [-v] [-o PATH_PAT] [PATH [PATH ...]]

svg2pdf v. x.x.x
A converter from SVG to PDF (via ReportLab Graphics)

positional arguments:
  PATH                  Input SVG file path with extension .svg or .svgz.

optional arguments:
  -h, --help            show this help message and exit
  -v, --version         Print version number and exit.
  -o PATH_PAT, --output PATH_PAT
                        Set output path (incl. the placeholders: dirname,
                        basename,base, ext, now) in both, %(name)s and {name}
                        notations.

examples:
  # convert path/file.svg to path/file.pdf
  svg2pdf path/file.svg

  # convert file1.svg to file1.pdf and file2.svgz to file2.pdf
  svg2pdf file1.svg file2.svgz

  # convert file.svg to out.pdf
  svg2pdf -o out.pdf file.svg

  # convert all SVG files in path/ to PDF files with names like:
  # path/file1.svg -> file1.pdf
  svg2pdf -o "%(base)s.pdf" path/file*.svg

  # like before but with timestamp in the PDF files:
  # path/file1.svg -> path/out-12-58-36-file1.pdf
  svg2pdf -o {dirname}/out-{now.hour}-{now.minute}-{now.second}-%(base)s.pdf path/file*.svg

issues/pull requests:
    https://github.com/deeplook/svglib

Units and output sizes

SVG is anchored to a 96 dpi coordinate system: one user unit (bare number or px) equals one CSS pixel, which is 1/96 of an inch. ReportLab works in points (1 pt = 1/72 inch). The conversion factor is 1 px = 0.75 pt.

The Drawing object returned by svg2rlg always has its width and height expressed in ReportLab points. The table below shows what you get for common unit choices on the root <svg> element:

The physical size is the same in every row — the unit only affects which number appears in the file.

Font sizes follow the same rule. font-size="16" (bare or px) produces a 12 pt font in the PDF (16 × 0.75). Use font-size="12pt" when you need to specify a precise typographic size directly.

Choosing units

• Web-first SVGs: use bare numbers throughout and a viewBox. The browser renders at 96 dpi; svglib scales to 72 dpi points. Physical dimensions are preserved in the PDF.
• Print at an exact physical size: declare width/height in mm, cm, or in on the <svg> element and use a matching viewBox. Both screen and print will show the declared physical size.
• Control exact PDF point dimensions: use pt units on width/height; the numeric value is preserved in the Drawing.

Adjusting the output size at render time

If you need to rescale the returned Drawing (e.g. to fit a fixed page size or restore the pre-2.0 behaviour), scale the Drawing object directly:

drawing = svg2rlg("file.svg")
factor = 4 / 3   # restore 1.x size (1 user unit → 1 pt, non-spec)
drawing.width  *= factor
drawing.height *= factor
drawing.scale(factor, factor)

Dependencies

Svglib depends mainly on the reportlab package, which provides the abstractions for building complex Drawings which it can render into different fileformats, including PDF, EPS, SVG and various bitmap ones. Other dependencies are lxml which is used in the context of SVG CSS stylesheets.

PDF output does not require Cairo. SVG images embedded in input files are included in generated PDFs through ReportLab's PDF renderer.

Bitmap output

Rendering ReportLab drawings to bitmap formats such as PNG uses reportlab.graphics.renderPM and requires a renderPM backend.

The default ReportLab 4.x backend is rlPyCairo:

$ pip install "svglib[bitmaps]"

Depending on the platform, pycairo may also require the system Cairo library to be installed. For installation instructions, see the official website: https://www.cairographics.org/download/

Alternatively, users can install ReportLab's legacy _renderPM backend:

$ pip install svglib rl-renderPM

To choose a backend explicitly, set ReportLab's renderPM backend before calling renderPM:

from reportlab import rl_config
from reportlab.graphics import renderPM

rl_config.renderPMBackend = "rlPyCairo"  # default in ReportLab 4.x
# or:
rl_config.renderPMBackend = "_renderPM"

renderPM.drawToFile(drawing, "file.png", fmt="PNG")

Installation

There are three ways to install svglib.

1. Using pip

With the pip command on your system and a working internet connection you can install the newest version of svglib with only one command in a terminal:

$ pip install svglib

You can also use pip to install the very latest version of the repository from GitHub, but then you won't be able to conveniently run the test suite:

$ pip install git+https://github.com/deeplook/svglib

2. Using conda

If you use Anaconda or Miniconda you are surely using its respective package manager, Conda, as well. In that case you should be able to install svglib using these simple commands:

$ conda config --add channels conda-forge
$ conda install svglib

Svglib was kindly packaged for conda by nicoddemus. See here more about svglib with conda.

3. Manual installation

Alternatively, you can install a tarball like svglib-<version>.tar.gz after downloading it from the svglib page on PyPI or the svglib releases page on GitHub and installing it via pip:

$ pip install svglib-<version>.tar.gz

This will install a Python package named svglib in the site-packages subfolder of your Python installation and a script tool named svg2pdf in your bin directory, e.g. in /usr/local/bin.

Development and Testing

To develop svglib, install uv and run uv sync to install the requirements and development dependencies. To run the test suite with pytest, run: uv run pytest.

Distribution Testing

The svglib tarball distribution contains a PyTest test suite in the tests directory. There, in tests/README.rst, you can also read more about testing. You can run the testsuite e.g. like shown in the following lines on the command-line:

$ make test
========================= test session starts =========================
platform darwin -- Python 3.9.6, pytest-8.4.2, pluggy-1.6.0
rootdir: /Users/dinu/dev/svglib
configfile: pyproject.toml
plugins: cov-7.0.0
collected 115 items

tests/test_basic.py ............................................ [ 43%]
..................                                               [ 53%]
tests/test_fonts.py ............s.....................s.....     [ 88%]
tests/test_samples.py .s.s.s.s.....                              [100%]

============== 109 passed, 6 skipped, 1 warning in 33.32s =============

Bug reports

Please report bugs on the svglib issue tracker on GitHub (pull requests are also appreciated)! If necessary, please include information about the operating system, as well as the versions of svglib, ReportLab and Python being used!