Digital cursor movement recorded in PDF creation visible in CorelDraw

Question

I am using CairoSVG to turn an SVG into a PDF. I have found a bug where each line of text has a 1 dimensional mark at the end when the PDF is loaded into CorelDraw.

I find this is true even with the simplest and barest of SVGs. I then decided to dig into the python source of CairoSVG to fix the issue at it's root. I have found text.py to be the place where the issue occurs.

https://github.com/Kozea/CairoSVG/blob/master/cairosvg/text.py

Specifically, the for [x, y, dx, dy, r], letter in letters_positions: for loop. I can actually manipulate the mark by doing this after the for loop.

surface.context.move_to( 100, 100 )
surface.context.save()
surface.context.text_path(' ')            
surface.context.restore()

Depending on the values I enter, it will move one of the points (in this case the lower right point of the mark). This suggests that when CairoSVG is done with a line of text, the surface will move the cursor to the start of the last letter it was drawing. For some reason, the last time it does this is actually recorded into the PDF, although for most PDF viewers it does not render. I have tried to fine tune this mark so the two points are the same, but because this location is different each time depending on letter and location and more, I can't fine tune this number to make the mark disappear, even when using available width and height variables

Note that surface.context is not a file or function and surface is pprinted as <cairosvg.surface.PDFSurface object at 0x7f647802ec90>, it's some sort of Cairo interface for interacting with the PDF file directly through some sort of PDF API, although I haven't found any documentation on it.

I have tried lots of cheap work arounds, like saving the PDF in another program then opening in Corel, however elements are lost or altered in ways that are not acceptable.

I have also of course tried commenting out this file 1 line at a time, then chunks of code at a time, but with no luck.

I have also looked into other files such as path.py, parser.py, and more, but text.py seems to be the most promising location.

How can I prevent this mark from appearing in my PDF files?

Answer 1

I have found I could get the desired results by going into the CairoSVG code and using a different function to create the text elements onto the PDF.

This collects the letters into a variable.

# Collect entire line instead of letter
text_line_letters = ''
for [x, y, dx, dy, r], letter in letters_positions:
    text_line_letters = text_line_letters + letter
    # Skip rest of for loop
    continue

This uses the new function to creates the actual text onto the PDF.

# Create Text with glyph_path    
surface.context.save()
font = surface.context.get_scaled_font()
glyphs, clusters, is_backwards = font.text_to_glyphs(
    0, -5, text_line_letters, with_clusters=True)
surface.context.glyph_path(glyphs)  
surface.context.restore()

The actual root cause of the issue is unknown, but likely deeper inside the Cairo library itself.

https://github.com/bonzini/cairo/blob/9099c7e7307a39bc630919faa65bba089fd15104/src/cairo.c#L3349