page.get_image_info() returns outdated cached results after replacing image

[stijnvermeeren-swisstopo]

Description of the bug

When calling page.get_image_info(xrefs=True), then applying page.replace_image() to one of the images on the page, and then calling page.get_image_info(xrefs=True) again on the same page object, the details of the old image (that has been replaced) are still returned, and the value of xref if 0.

The reason seems to be that the result of the page.get_image_info() call is cached on the Page object (

PyMuPDF/src/utils.py

Line 837 in 0a9c2e8

page._image_info = imginfo

), and this cache is not automatically cleared when calling page.replace_image().

I would expect the second call to page.get_image_info(xrefs=True) to return the details of the new image (including the correct xref), or at least for this caching behavior to be clearly described in the documentation.

How to reproduce the bug

import pymupdf

doc = pymupdf.Document()
page = doc.new_page()
image1x1 = pymupdf.Pixmap(pymupdf.Colorspace(pymupdf.CS_RGB), pymupdf.IRect(0, 0, 1, 1))
image2x2 = pymupdf.Pixmap(pymupdf.Colorspace(pymupdf.CS_RGB), pymupdf.IRect(0, 0, 2, 2))
xref = page.insert_image(page.rect, pixmap=image1x1)

print("Before replace_image")
for image_info in page.get_image_info(xrefs=True):
    print(image_info)
    # {..., 'width': 1, 'height': 1, ..., 'size': 59, 'digest': b'&9-[t\ro\x84y\xba\x80\xfeA?\xb3\xe9', ..., 'xref': 5}

page.replace_image(xref, pixmap=image2x2)

print("After replace_image, same page object")
for image_info in page.get_image_info(xrefs=True):
    print(image_info)
    # {..., 'width': 1, 'height': 1, ..., 'size': 59, 'digest': b'&9-[t\ro\x84y\xba\x80\xfeA?\xb3\xe9', ..., 'xref': 0}

print("After replace_image, new page object")
for image_info in doc[page.number].get_image_info(xrefs=True):
    print(image_info)
    # {..., 'width': 2, 'height': 2, ..., 'size': 68, 'digest': b"'\x87\x15P,\xd9\x01ze\x08\x01\xa4\x1e\xda\xd9\xce", ..., 'xref': 7}

PyMuPDF version

1.25.3

Operating system

Linux

Python version

3.10

[JorjMcKie]

Thanks for the report. You can circumvent this for the time being by reloading the page: page = doc.reload_page(page).

[julian-smith-artifex-com]

Fixed in PyMuPDF-1.25.4.