Why the font size of some text became different after rewriting to a new pdf?

[nissansz]

1-test_2.pdf
1-test.pdf

import pymupdf


def adjust_matrix(font, bbox, text, fontsize):
    """Compute matrix performing a horizontal scale.

    Args:
        font: Font object
        bbox: bbox of the text to fill
        text: text
        fontsize: fontsize to use
    Returns:
        Horizontal scaling matrix
    """
    tl = font.text_length(text, fontsize=fontsize)
    width = bbox[2] - bbox[0]
    scale = width / tl
    return pymupdf.Matrix(scale, scale)

path =  r"L:\doc88\doc88_extractor-V1.3-WINDOWS-X64\docs\2025年新青岛版数学一年级上册整册课件\pdf\1.pdf"
path =  r"L:\doc88\doc88_extractor-V1.3-WINDOWS-X64\docs\2025年新青岛版数学一年级上册整册课件.pdf"

path =  r"L:\doc88\doc88_extractor-V1.3-WINDOWS-X64\docs\（正式版）GB∕T 14848-2017 地下水质量标准\swf\1-test.pdf"

doc = pymupdf.open(path)
font = pymupdf.Font("cjk")
fontname = "newfont"
flags = pymupdf.TEXTFLAGS_TEXT | pymupdf.TEXT_ACCURATE_BBOXES
for page in doc:
    blocks = page.get_text("dict", flags=flags)["blocks"]
    spans = [s for b in blocks for l in b["lines"] for s in l["spans"]]
    page.add_redact_annot(page.rect)
    page.apply_redactions(
        images=pymupdf.PDF_REDACT_IMAGE_NONE,
        graphics=pymupdf.PDF_REDACT_LINE_ART_REMOVE_IF_TOUCHED,
        text=pymupdf.PDF_REDACT_TEXT_REMOVE,
    )
    page.insert_font(fontname=fontname, fontbuffer=font.buffer)
    for s in spans:
        color = pymupdf.sRGB_to_pdf(s["color"])
        bbox = pymupdf.Rect(s["bbox"])
        fsize = bbox.height
        matrix = adjust_matrix(font, bbox, s["text"], fsize)
        page.insert_text(
            bbox.bl
            + (0, -5),  # hard coded: seems to be needed to get correct vertical pos
            s["text"],
            fontsize=fsize,
            color=color,
            fill=color,
            border_width=0.1,
            fontname=fontname,
            morph=(bbox.bl, matrix),
        )
doc.subset_fonts(fallback=True)
# doc.ez_save("out.pdf")
doc.ez_save(path.replace('.pdf', '_2.pdf'))

[robinwatts]

So, if I'm reading that right, you're asking for the bounding box of the text to replace, and then using the height of that box as the fontsize.

Consider the bounding box of 'ocean' and the bounding box of 'Aphrodite'. Same font. Radically different heights. Hopefully that makes it clear how bad an idea it is to try to estimate the font size from the bbox.

Using s["size"] would be better, I think. But even then, attempting to match text in this way is going to be hit and miss unless you have exactly the same font.

[nissansz]

I tried below. No improvement, any other solution?

fsize = bbox.height

    fsize = s["size"]

[nissansz]

what is origin? How to use full info like origin, ascender, descender, origin, bbox to improve font size?

s={'size': 15.0865478515625, 'flags': 0, 'font': 'MYFONT7', 'color': 0, 'ascender': 0.86328125, 'descender': -0.1357421875, 'text': '中华人民共和国国家质量监督检验检疫总局', 'origin': (138.10000610351562, 762.4000244140625), 'bbox': (138.10000610351562, 749.3633422851562, 432.238037109375, 764.4498901367188)}
s={'size': 15.0865478515625, 'flags': 0, 'font': 'MYFONT7', 'color': 0, 'ascender': 0.86328125, 'descender': -0.1357421875, 'text': '中', 'origin': (138.56500244140625, 779.0882568359375), 'bbox': (138.56500244140625, 766.0515747070312, 152.32559204101562, 781.1381225585938)}
s={'size': 15.0865478515625, 'flags': 0, 'font': 'MYFONT7', 'color': 0, 'ascender': 0.86328125, 'descender': -0.1357421875, 'text': '国', 'origin': (165.68978881835938, 779.08837890625), 'bbox': (165.68978881835938, 766.0516967773438, 179.5682373046875, 781.1382446289062)}
s={'size': 15.0865478515625, 'flags': 0, 'font': 'MYFONT7', 'color': 0, 'ascender': 0.86328125, 'descender': -0.1357421875, 'text': '国家标准化管理委员会', 'origin': (192.34957885742188, 779.08837890625), 'bbox': (192.34957885742188, 766.0516967773438, 434.98968505859375, 781.1388549804688)}
s={'size': 12.140579223632812, 'flags': 0, 'font': 'MYFONT39', 'color': 0, 'ascender': 0.8837890625, 'descender': -0.115234375, 'text': '发布', 'origin': (450.70001220703125, 769.3499755859375), 'bbox': (450.70001220703125, 758.6098022460938, 483.830322265625, 770.7503662109375)}

[nissansz]

I improved it.
1-test_4.pdf