Add SVG

[mrlarssen]

Is it possible to add SVG to the document? For instance an svg-string such as

< svg xmlns="http://www.w3.org/2000/svg" width="200" height="100" version="1.1">
< rect width="200" height="100" stroke="black" stroke-width="6" fill="green" />
</ svg>

would be perfect. Or just a svg file?

[scanny]

Can you do that from the Word application UI?

[mrlarssen]

I haven't tried it myself, but it seems like it:

https://support.office.com/en-us/article/Insert-SVG-images-in-Microsoft-Office-69f29d39-194a-4072-8c35-dbe5e7ea528c

[scanny]

I suspect it translates them into Windows Metafile Format (WMF) and stores them that way. You'd need to try it and then inspect the .docx file (it's a zip archive) and see what the image was that got included. I expect the file extension of the image file name would reliably indicate the underlying file type its stored as.

[mrlarssen]

I guess the documentation is outdated, as this approach does not let me insert an svg into the word document. So I assume the limitation lays with word itself?

Then I need to look into how I can convert svg to png in Python.

Thanks for your response.

[scanny]

This page has some insights on why it might not be working for you:
https://www.quora.com/How-would-you-insert-a-svg-file-in-PowerPoint

It could be your PowerPoint version is not the very latest; I know mine isn't :)

It's still not clear whether the new PowerPoint 2016 supports SVG files natively or whether they are just converted to WMF/EMF and stored that way, like it does with PDF/EPS images.

[ftomassetti]

A PR related to this is pending #798

[kuri65536]

Thank you to python-docx, this is very useful software.

I had done to embed svg images with python-docx.

this is a part of my
mkdocs-theme-topdf .
code is here

BTW, I found the PR and this is just a sample workaround for
the current version.

(this preview from Word android)

from docx.image import image

def run(doc: Document):
    monkey()
    para = doc.add_paragraph()
    pic = para.add_run().add_picture("sample.svg")
    compose_asvg(pic)

def compose_asvg(pic: Tag) -> None:
    embed = '{%s}embed' % (
        "http://schemas.openxmlformats.org/officeDocument/2006/relationships")
    url = "http://schemas.microsoft.com/office/drawing/2016/SVG/main"
    url14 = "http://schemas.microsoft.com/office/drawing/2010/main"
    etree.register_namespace("asvg", url)
    etree.register_namespace("a14", url14)

    ablip = pic._inline[-1][-1][-1][1][0]
    elst = OxmlElement('a:extLst')
    ablip.append(elst)
    embed_id = ablip.attrib.get(embed, "")
    del(ablip.attrib[embed])

    aext = OxmlElement('a:ext')
    elst.append(aext)
    aext.set("uri", "{28A0092B-C50C-407E-A947-70E740481C1C}")
    ldpi = etree.Element('{%s}useLocalDpi' % url14)
    aext.append(ldpi)
    ldpi.set("val", "0")

    aext = OxmlElement('a:ext')
    elst.append(aext)
    aext.set("uri", "{96DAC541-7B7A-43D3-8B79-37D633B846F1}")

    asvg = etree.Element('{%s}svgBlip' % url)
    aext.append(asvg)
    asvg.set(embed, embed_id)

def monkey() -> None:
    global default_factory
    if default_factory is not None:
        return
    default_factory = image._ImageHeaderFactory
    image._ImageHeaderFactory = _ImageHeaderFactory

def _ImageHeaderFactory(stream: IO[Text]) -> image.BaseImageHeader:
    stream.seek(0)
    buf = stream.read(32)
    if "<svg".encode("latin") in buf:
        stream.seek(0)
        return Svg.from_stream(stream)
    assert default_factory is not None
    return default_factory(stream)

class Svg(image.BaseImageHeader):
    @classmethod
    def from_stream(cls, stream: IO[Text]) -> '_SvgParser':
        """Return a |Svg| instance having header properties parsed from image
             in *stream*."""
        # this is dummy.
        return cls(100, 100, 96, 96)

    @property
    def content_type(self) -> Text:
        return "image/svg"

[scanny]

Nice, @kuri65536, thanks for posting this :)

[tikuma-lsuhsc]

@kuri65536 - Thanks for the nice workaround!

For anyone else still waiting for the PR, @kuri65536 's code caused a small hiccup in my code. Specifically, _ImageHeaderFactory() is not accounting for valid preceding XML tags. Here is how I fixed it:

  import re

  def _ImageHeaderFactory(stream):
      stream.seek(0)
      buf = stream.read(32).decode("utf-8")
      while re.match(r"\s*\<\?|\s*\<\!", buf):
          m = re.match(r"(?:\s*<[\s\S]*?>\s*)+", buf)
          if m:
              buf = buf[m.end(0) :]
          if not m or len(buf) < 4:
              buf += stream.read(32).decode("utf-8")
      if buf.startswith("<svg"):
          stream.seek(0)
          return Svg.from_stream(stream)
      assert default_factory is not None
      return default_factory(stream)

[cristianocca]

It's been some time, any news or actual guide how to add an svg image to the docx template file?

[jfthuong]

If it might help,

Here are 2 Word with a similar image in PNG and SVG that have been added in the same place.

I have also extracted the files and pretty-formatted the XML files for easier comparison

png_vs_svg.zip

[takis]

Hi,

I had a first try at implementing this, but have only tested using simple SVGs. The files open without issues in Word on my Mac Mini.

Any feedback welcome. I haven't written any tests yet.