Question: is PyMuPDF rect compatible with reportlab rect?

Dear developers,

I was trying to use coordinates obtained from PyMuPDF in reportlab but it seems they are not compatible or I am missing something. I am using rect [72, 74, 103, 85] to place free text annotation and you can see that reportlab places it at the bottom meanwhile fitz on top. My system and environment info can be found in #347
Thank you so much for your help to demystify this.

word1.pdf
word1_fitz.pdf
word1_reportlab.pdf

The code that generates word1_fitz.pdf:

import fitz

doc = fitz.open("word1.pdf")
color1 = [0.941, 0.976, 0.91]
# text color
black = (0, 0, 0)
border = {"width": 1}

# work with first page only
page = doc[0]

# get list of text locations, restricting to the first occurrence
anno = page.addFreetextAnnot(fitz.Rect(72, 74, 103, 85), "DM.USBUJ", fontsize=7, fontname="ti")

anno.setFlags(0)
anno.setBorder(border)
anno.update(fill_color=color1)
doc.save("word1_fitz.pdf")

The code that generates word1_reportlab.pdf:

from reportlab.pdfgen import canvas

c = canvas.Canvas("word1_reportlab.pdf")
c.drawString(100, 750, "")
  
body_style = '(0 0 0 rg /Arial,Bold 9 Tf)'
rect = [72, 74, 103, 85]
color = '[0.941 0.976 0.91]'
anno = "DM.USBUJ"

c.freeTextAnnotation(anno, DA=body_style, Rect=rect, C=color)  
c.save()

I think I found an answer to my own question :)
fitz is using top-left corner as a start of coordinate system, meanwhile reportlab is using bottom-left corner. reportlab used to have bottomup argument in Canvas fo flip the coordinate system but apparently they deprecated it. The workaround is to convert fitz rect y values by the height of the page:

heigth <- fitz_page.MediaBoxSize.y
y0orig = 0
y0 = height - y0orig
y1orig = 100
y1 = height - y1orig

[JorjMcKie]

Currently without internet trying an answer while being “blind”. From a short look at RL documentation I concluded they are following the standard PDF coordinates: origin point on a page is bottom left. So rectangles (which normally are defined 2 diagonally opposite points) are defined by their bottom left and their top right point coordinates. MuPDF uses the top left page corner as (0, 0), and consequently our rectangles are given by top-left / bottom-right points. MuPDF provides a way to calculate between these two systems: the page transformation matrix, available via ctm = page._getTransformation(), translates PDF coordinates into MuPDF coordinates. For example, a rectangle r, given in PDF (or RL) coordinates, can be had in (Py-) MuPDF coordinates via “r * ctm”. Conversely, an object in PyMuPDF coordinates can be translated to PDF by multiplying it with ~ctm (the inverse matrix). Hope that helps Von meinem iPhone gesendet