`substrings_to_isolate` no longer works on multi-line strings

[alexlembcke]

Description of bug / unexpected behavior

Previously, if you wanted to put a long sentence on the screen you could break it up into multiple lines using triple quotation marks and substrings_to_isolate would still pick up all of the pieces you wanted to highlight. Since PR #4515 that is no longer the case. Now it only picks up on substrings in the first line of the string.

Expected behavior

substrings_to_isolate should work on multi-line strings.

How to reproduce the issue

Code for reproducing the problem

from manim import *

class SubstringsTest(Scene):
    def construct(self):
        multiline_string = Tex(
            """This is a very long string, 
            which will test how well the new implementation of Tex handles such long strings.""",
            substrings_to_isolate=[
                "This",
                "implementation"
                ],
            tex_template=TexFontTemplates.libertine
            )
        print(multiline_string.get_part_by_tex("This"))
        print(multiline_string.get_part_by_tex("implementation"))

        self.play(Write(multiline_string))

Note how in the log output below it doesn't separate out the second substring on the second line, only the first substring on the first line. Then when you do get_part_by_tex it returns None for the second substring.

Logs

Terminal output

Manim Community v0.20.1

[02/28/26 08:17:16] INFO     Writing \special{dvisvgm:raw <g                     tex_file_writing.py:110
                             id='unique000'>}\special{dvisvgm:raw <g                                    
                             id='unique000substring'>}This\special{dvisvgm:raw                          
                             </g>} is a very long string, \special{dvisvgm:raw                          
                             </g>} to media/Tex/71f2c05dd57efc72.tex
VGroup of 4 submobjects
None

[henrikmidtiby]

Thanks for raising this issue.
I think that it is related to the installed version of dvisvgm.

On my computer (Ubuntu 22.04), the default version of dvisvgm is 2.13.1 when installed through the apt package manager.
From TeXLive I can install version 3.6 of dvisvgm.
The issue is that they generate slightly different output, which changes how the rendered file is imported into manim.

In the following I have changed line 471 - 472 in tex_mobject.py to the following.

        for substring in substrings_to_isolate:
            match = re.match(
                f"(.*?)({re.escape(substring)})(.*)",
                unprocessed_string,
                flags=re.DOTALL,
            )

Then the output from manim when using dvisvgm version 2.13.1 is as follows.

While the output when using dvisvgm version 3.6 is this

For reference the code snippet that I use is this

from manim import *

class SubstringsTest(Scene):
    def construct(self):
        multiline_string = Tex(
            """This is a very long string, 
            which will test how well the new implementation of Tex handles such long strings.""",
            substrings_to_isolate=[
                "This",
                "implementation"
                ],
            ).scale(0.75)
        print(multiline_string.get_part_by_tex("This"))
        print(multiline_string.get_part_by_tex("implementation"))
        self.add(multiline_string)