Beautiful Soup

#!/usr/bin/env python3

# Above, the shebang. Need to add execute permission also

# BS4, Beautiful Soup's new abbreviated name, is
# pre-installed on Ubuntus. As is the libxml2
# wrap, 'lxml'. So you shouldn't need to apt-get
# anything. But see the note below.
from bs4 import BeautifulSoup


# At first, my Ubuntu did not recognise lxml was
# installed. No idea why. I tried an import,
# then it started to work regardless. Go figure.
# As for 're', always useful here.
#import lxml
import re


# Instructions usually say use 'lxml' as a
# parameter to summon the tolerant-but-accurate
# libxml2 parser.
# However, Beautiful Soup works by default as an
# HTML parser. One effect is it will wrap output
# in compliant HTML tags like BODY and HTML.
# This is not wanted for SVG, so state 'xml' as
# a style, not a parser. Beautiful Soup will
# look for an XML parser. If 'lxml' is present
# Beautiful Soup will use that.
with open("funkyweb.svg") as fp:
    soup = BeautifulSoup(fp,  "xml")
    #soup = BeautifulSoup(fp,  "lxml")

# Not necessary.
# A sign if all is right or wrong.
count = 0

# Let's jamb a lowlight filter in
# Now, Beautiful Soup works as an
# object-orientated API. This is is slow and
# clumsy for extensive building. Also, the
# example that follows has an attribute name
#'type', which clashes with a Python keyword.
# That means the object-orientated API can't
# express it. But another feature saves this
# script. Two Beautiful Soup object-trees can
# be linked. So XML can be parsed in.
soupFilter = BeautifulSoup(
'''<filter id="lo-light" x="0" y="0" width="100%" height="100%">
     <feComponentTransfer>
       <feFuncR type="gamma" amplitude="0.5" exponent="1" offset="0"></feFuncR>
       <feFuncG type="gamma" amplitude="0.5" exponent="1" offset="0"></feFuncG>
       <feFuncB type="gamma" amplitude="0.5" exponent="1" offset="0"></feFuncB>
     </feComponentTransfer>
   </filter>''',
"xml"
)

# However, can use the object-orientated API to
# find the 'def' section of the SVG, replace it
# with the filter (or append the filter)
defsTag = soup.find("defs")
defsTag.replace_with(soupFilter)

# Right, let's add links round several of the
# paths in the SVG. This is the kind of job that
# is a waste of time manually, inefficient and
# with erratic, non-replicable results.

# Need to add the web-link namespace.
# From the attribute API, there is no way to
# generate namespaced attributes
# ('svgTag.xlms:xlink' ?) . However, the
# consistent and clunky constructor works as
# presented.
svgTag['xmlns:xlink'] = "http://www.w3.org/1999/xlink"

# Let's wrap paths in links. Assuming the paths
# have a meaningful 'id'.
for tag in soup.findAll("path"):
        urlId = id.lower()
        urlId = urlId.replace("_", "-")

        # With the Beautiful Soup API, you need
        # to make an element from parts, not
        # plaster in text (but see the XML
        # solution above)
        newTag = soup.new_tag("a", title=f"{id}")

        # Mamespaced attribute issue again. See
        # above.
        newTag['xlink:href'] = f"/area/{urlId}"
        tag.wrap(newTag)
        count = count + 1

# Right, lets talk about XML editors. Like
# Inkscape. Inkscape generates much of it's own
# data to fill SVGs. For example, it knows how
# and what size it must open an SVG (all SVG
# editors will do this). It records this
# information in the XML in an SVG file. However,
# this information is surplus to data needed to
# render the SVG. This script is cleaning, so
# lets get rid of that data.
# NB: Before Inkscape was forked, it was called
# Sodipodi (and Sodipodi continues to this day)
# This code needs to remove those references too.

# There is a tag called Sodipodi, unused for
# final rendering. Remove.
editorTag = soup.find("sodipodi:namedview")
editorTag.decompose()

# At base, this section of code compresses, and
# tidies, SVG code. Not going far with this. Why?
# Because there are programs designed specially.
# And those programs tackle ideas that can make a
# big difference. For example, cutting down on
# decimal point length in SVG paths. That can
# change the look of the SVG, so you'll need to
# try parameters. However, reduction to one
# decimal point can reduce SVG sizes by half.

# Specifically, remove Inkscape/Sodipodi editor
# attributes
# Can reduce size by 1/10. But makes final SVG
# much cleaner.

# I know XML users are self-important, but this
# is the only way I could think of to iterate
# the Soup structure
# Only need the key (of the attribute), not
# the value
for tag in soup.findAll(recursive=True):
    # Ugly code, but offhand I don't know how
    # Python can remove multiple dict entries
    # in a loop
    removeAttrs = [attrName for attrName in tag.attrs if (attrName.startswith("sodipodi:") or attrName.startswith("inkscape:"))]
    [tag.attrs.pop(attr, None) for attr in removeAttrs]

# One more compression. Editors heap stack
# attributes on text elements. These
# attributes are CSS-style lengthy e.g.
# "font-style:normal;font-variant:normal;font-weight:normal;...' etc.
# You can see, all of this, and maybe more,
# can in some circumstances be reduced to a
# simple CSS-like style="" on the main element,
# to be inherited. So strip all attributes from
# text tags. Is there anything to preserve? No,
# not even x and y attributes, because text
# elements wrap TSPAN elements that position.
# That said, check afterwards.
# Could save 1/10  file size.. More to the point,
# is better code.

# Put in the new overall style attributes
svgTag = soup.find('svg')

# display:inline is a basic. May be there already
svgTag['style'] = "display:inline; font-size:12px;"

# Strip all attributes from TEXT elements by
# replacement
for tag in soup.findAll('text', recursive=True):
    tag.attrs = {}


# Viewbox fix. Shift 'x' and 'y' positioning
# values to a viewbox. Will stop an SVG falling
# out of view, or exploding across an HTML page
width = soup.svg['width']
height = soup.svg['height']
soup.svg['viewBox'] = f"0 0 {width} {height}"
del soup.svg['width']
del soup.svg['height']

# Right, that's it, enough. Write the result out
with open("funkyweb_clean.svg", "w") as fp:
    # BS turns all text into UTF8. This will
    # destroy HTML entities. So those must be
    # re-encoded on save(), Hence, 'minimal'
    # format paramater.
    # Also, Beautiful Soup usually outputs as
    # pretty-print. But the output is tidy
    # anyway, and compact results are easy to
    # cut and paste. Use a simple 'decode'
    #fp.write(soup.prettify(formatter="minimal"))
    fp.write(soup.decode(formatter="minimal"))

# Checksum
print(f"count: {count}")