render math in job tab info

[RobBuchananCompPhys]

Description of work
Currently, the Job tab info does not render the embedded LaTex expression strings, which can look less informative and cluttered. This PR proposes an enhancement in which images (rendered in a simple matplotlib figure) of the expression are inserted into the html of the Job tab, allowing mathematical expressions to be displayed in the text.

Fixes
Before:

After:

To test
Please describe the tests that were run to verify the changes.

[MBartkowiakSTFC]

Looks like it's almost there. A few possible points that could still be worked on:

1. If an equation is not inline, it is not picked up:

2. The ignored expressions still affect the line breaks:

3. In principle, it may be worthwhile to try to inline the images. With some tweaking, I got to the stage where it looked like this:

But if the vertical alignment cannot be corrected easily and the whole thing looks to awkward, then it should be fine to stick to the current version with a line break per maths symbol.

[ChiCheng45]

Just a few things.

White background with dark mode theme, and the inline equations' font size is very large. For the white background, I guess we could make the text white and the background transparent. The font size issue might be trickier. I wonder if we should latex the whole thing?

[RobBuchananCompPhys]

Just a few things.

White background with dark mode theme, and the inline equations' font size is very large. For the white background, I guess we could make the text white and the background transparent. The font size issue might be trickier. I wonder if we should latex the whole thing?

Good question - we may be able to get away with transparency and white font when in dark mode.

[oerc0122]

I think the MathRenderer can be simplified by using REs to their full extent. Though you may have to be careful and make a copy of the text since replace may disrupt finditer.

[oerc0122]

Really, this would be an ideal use-case of 3.14's t-strings

[oerc0122]

This is a somewhat strange way to do this. The usual way would be to use Regular expression groups.

Suggested change

	def processInlineExpressions(text: str) -> list[tuple[str, bool]]:
	pattern = r"(:math:`.*?`)"
	substrings = re.split(pattern, text)
	scanned = []
	for s in substrings:
	if s.startswith(":math:"):
	expr = s[len(":math:`") : -1]
	scanned.append((expr, True))
	else:
	scanned.append((s, False))
	return scanned
	def processInlineExpressions(text: str) -> list[tuple[str, bool]]:
	pattern = r"(:math:`(.*?)`)" # <-- Should really be class-level constant
	last = 0
	for match in re.finditer(pattern, text):
	scanned.append((text[last:match.start], False))
	scanned.append((match.group(1), True))
	last = match.end
	scanned.append((text[last:], False))
	return scanned

Although I'm still not sure whether you really want to capture the non-converting bits or whether you just want to substitute the matches in-place for their image equivalents, which simplifies the logic massively.

Suggested change

	def processInlineExpressions(text: str) -> list[tuple[str, bool]]:
	pattern = r"(:math:`.*?`)"
	substrings = re.split(pattern, text)
	scanned = []
	for s in substrings:
	if s.startswith(":math:"):
	expr = s[len(":math:`") : -1]
	scanned.append((expr, True))
	else:
	scanned.append((s, False))
	return scanned
	def processInlineExpressions(self, text: str) -> None:
	pattern = r"(:math:`(.*?)`)" # <-- Should really be class-level constant
	for match in re.finditer(pattern, text):
	text.replace(match[0], self.render(match[1]))

[oerc0122]

Again, you can use pattern matching here.

Suggested change

	@staticmethod
	def processMultiLineBlockExpression(
	strings: list[str], n: int
	) -> tuple[list[tuple[str, bool]], int]:
	substrings = strings[n + 1 :]
	group = []
	for s in substrings:
	if (not s) or (s == r"<br /><br />"):
	break
	group.append(s)
	group = [s.replace(r"<br />", "") for s in group]
	total = "".join(group)
	if total.startswith(r"<br />") and total.endswith(r"<br />"):
	result = total.split(r"<br />")[1]
	else:
	result = total.split(r"<br /><br />")[0]
	return [(result, True)], n + len(group) + 1
	def processMultiLineBlockExpression(
	self, strings: str
	) -> None:
	pattern = r".. math::\s*\n(?:\s*\n)?(\s+)(.+)\n(\1.+\n)*" # <- Should be class level variable
	for match in re.finditer(pattern, text):
	if match[3] is not None:
	body = "".join(more_itertools.prepend(match[2], match[3].splitlines()))
	else:
	body = match[2]
	text.replace(match[0], self.render(body))

[oerc0122]

As above, we can use pattern matching:

Suggested change

	@staticmethod
	def processBlockExpression(text: str) -> list[tuple[str, bool]]:
	return [(text[len(".. math:: ") :], True)]
	def processBlockExpression(text: str) -> list[tuple[str, bool]]:
	pattern = ".. math::\s*(.+)"
	for match in re.finditer(pattern, text):
	text.replace(match[0], self.render(match[1]))

[oerc0122]

Bear in mind that latex tokens may not correspond to large images:

\underbrace{\mathbf{x}} = \mathbf{3}

$\underbrace{\mathbf{x}} = \mathbf{3}$

[ChiCheng45]

I don't think this needs to be fix here but something I discover in #1038 which also will happen for the equation images.

[ChiCheng45]

I don't think this needs to be fix here but something I discover in #1038 which also will happen for the equation images.

I've been able to fix some dark light mode system change GUI issues. See #1062 might be worth trying to add a similar fix here I think.

[ChiCheng45]

A few more things. Some text has been dropped. The "for a set of atoms" is missing. Also the equations are on their own line, I think it would be nice if you could inline them.

here is the text on protos.

@RobBuchananCompPhys Can you resolve the merge conflict, I can have a look to see if dark/light mode change can be fixed for the equation images.