srcset.py

_SPACE = "\u0020\u0009\u000a\u000c\u000d"
_POS_OUTSIDE = 0
_POS_URL = 1
_POS_DESCRIPTOR = 2


def urls(srcset):
	# URLs may contain commas, so we can't just .split(',')
	pos = _POS_OUTSIDE
	url = ""
	for c in srcset:
		if pos == _POS_OUTSIDE and c not in _SPACE:
			if c == ",":  # stray comma
				continue
			url += c
			pos = _POS_URL
		elif pos == _POS_URL:
			if c in _SPACE:
				if len(url) >= 1 and url[-1] == ",":  # no descriptor
					pos = _POS_OUTSIDE
				else:
					pos = _POS_DESCRIPTOR
				yield url.rstrip(",")
				url = ""
			else:
				url += c
		elif pos == _POS_DESCRIPTOR:
			if c == ",":
				pos = _POS_OUTSIDE

	if url:
		yield url.rstrip(",")


# https://raw.githubusercontent.com/web-platform-tests/wpt/704d739c5a54658e6ea09efaa76824b134b36504/html/semantics/embedded-content/the-img-element/srcset/parse-a-srcset-attribute.html
_splitting_loop = r"""
<img srcset='' data-expect=''>
<img srcset=',' data-expect=''>
<img srcset=',,,' data-expect=''>
<img srcset='  data:,a  1x  ' data-expect='data:,a'>
<img srcset='&#x9;&#x9;data:,a&#x9;&#x9;1x&#x9;&#x9;' data-expect='data:,a'>
<img srcset='&#xA;&#xA;data:,a&#xA;&#xA;1x&#xA;&#xA;' data-expect='data:,a'>
<img srcset='&#xB;&#xB;data:,a&#xB;&#xB;1x&#xB;&#xB;' data-expect='&#xB;&#xB;data:,a&#xB;&#xB;1x&#xB;&#xB;' data-resolve>
<!-- <img srcset='&#xC;&#xC;data:,a&#xC;&#xC;1x&#xC;&#xC;' data-expect='data:,a'> <!-- this one fails on lxml; srcset in the parsed tag is empty -->
<img srcset='&#xD;&#xD;data:,a&#xD;&#xD;1x&#xD;&#xD;' data-expect='data:,a'>
<img srcset='&#xE;&#xE;data:,a&#xE;&#xE;1x&#xE;&#xE;' data-expect='&#xE;&#xE;data:,a&#xE;&#xE;1x&#xE;&#xE;' data-resolve>
<img srcset='&#xF;&#xF;data:,a&#xF;&#xF;1x&#xF;&#xF;' data-expect='&#xF;&#xF;data:,a&#xF;&#xF;1x&#xF;&#xF;' data-resolve>
<img srcset='&#x10;&#x10;data:,a&#x10;&#x10;1x&#x10;&#x10;' data-expect='&#x10;&#x10;data:,a&#x10;&#x10;1x&#x10;&#x10;' data-resolve>
<img srcset='data:,a' data-expect='data:,a'>
<img srcset='data:,a ' data-expect='data:,a'>
<img srcset='data:,a ,' data-expect='data:,a'>
<img srcset='data:,a,' data-expect='data:,a'>
<img srcset='data:,a, ' data-expect='data:,a'>
<img srcset='data:,a,,,' data-expect='data:,a'>
<img srcset='data:,a,, , ' data-expect='data:,a'>
<img srcset=' data:,a' data-expect='data:,a'>
<img srcset=',,,data:,a' data-expect='data:,a'>
<img srcset=' , ,,data:,a' data-expect='data:,a'>
<img srcset='&nbsp;data:,a' data-expect='&nbsp;data:,a' data-resolve>
<img srcset='data:,a&nbsp;' data-expect='data:,a&nbsp;' data-resolve>
"""

_descriptor_tokenizer = r"""
<img srcset='data:,a 1x' data-expect='data:,a'>
<img srcset='data:,a 1x ' data-expect='data:,a'>
<img srcset='data:,a 1x,' data-expect='data:,a'>
<img srcset='data:,a ( , data:,b 1x, ), data:,c' data-expect='data:,c'>
<img srcset='data:,a ((( , data:,b 1x, ), data:,c' data-expect='data:,c'>
<img srcset='data:,a [ , data:,b 1x, ], data:,c' data-expect='data:,b'>
<img srcset='data:,a { , data:,b 1x, }, data:,c' data-expect='data:,b'>
<img srcset='data:,a " , data:,b 1x, ", data:,c' data-expect='data:,b'>
<img srcset='data:,a \,data:;\,b, data:,c' data-expect='data:;\,b'>
<img srcset='data:,a, data:,b (' data-expect='data:,a'>
<img srcset='data:,a, data:,b (  ' data-expect='data:,a'>
<img srcset='data:,a, data:,b (,' data-expect='data:,a'>
<img srcset='data:,a, data:,b (x' data-expect='data:,a'>
<img srcset='data:,a, data:,b ()' data-expect='data:,a'>
<img srcset='data:,a (, data:,b' data-expect=''>
<img srcset='data:,a /*, data:,b, data:,c */' data-expect='data:,b'>
<img srcset='data:,a //, data:,b' data-expect='data:,b'>
"""


def test():
	import bs4

	for _html in (_splitting_loop, _descriptor_tokenizer):
		soup = bs4.BeautifulSoup(_html)
		total = 0
		success = 0
		for img in soup.find_all("img"):
			total += 1
			parsed = set(urls(img['srcset']))
			expect = {img['data-expect']} if img['data-expect'] else set()
			if expect == parsed:
				success += 1
			elif expect.issubset(parsed):
				# we don't mind getting some strictly invalid descriptors/URLs
				# as long as the valid ones are extracted correctly
				print("%2d" % total, "(w) extracted too much:", img['srcset'].__repr__(), "→", parsed)
				success += 1
			else:
				print(
					"%2d" % total, "(F)",
					#img,
					img['srcset'].__repr__(),
					"→",
					parsed,
					"(expected", [img['data-expect']], ")"
				)
		print(f"{success}/{total}")