I tried to use update_object() to fix the cross-reference, but an error occurred.

[xiaolibuzai-ovo]

code:

        for xref in range(1, self.doc.xref_length()):
            try:
                _ = self.doc.xref_object(xref)
            except:
                try:
                    self.doc.update_object(xref, "<<>>")
                except Exception as e:
                    logger.error(f"save update_object error: {str(e)}, {traceback.format_exc()}")

error:

PageProcess error: RAISEPY() takes 2 positional arguments but 3 were given, Traceback (most recent call last):
  File "/usr/local/trpc/bin/process/pymupdf_process.py", line 152, in save
    _ = self.doc.xref_object(xref)
  File "/usr/local/trpc/bin/lib/pymupdf/__init__.py", line 6032, in xref_object
    ret = extra.xref_object( self.this, xref, compressed, ascii)
  File "/usr/local/trpc/bin/lib/pymupdf/extra.py", line 120, in xref_object
    return _extra.xref_object(*args)
RuntimeError: bad xref

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/local/trpc/bin/pdf_atom_process_py.py", line 126, in PageProcess
    url = await pymupdf_cli.save(file_id, request.scene, request.namespace_id)
  File "/usr/local/trpc/bin/process/pymupdf_process.py", line 154, in save
    self.doc.update_object(xref, "<<>>")
  File "/usr/local/trpc/bin/lib/pymupdf/__init__.py", line 5829, in update_object
    RAISEPY("bad xref", MSG_BAD_XREF, PyExc_ValueError)
TypeError: RAISEPY() takes 2 positional arguments but 3 were given

version: pymupdf==1.25.5
platform: linux

[julian-smith-artifex-com]

pymupdf-1.25.5 is very old, please upgrade to the latest version.

In particular, the incorrect call of RAISEPY() with "bad xref" was fixed a while ago.

[xiaolibuzai-ovo]

I have already upgraded to the latest version, but the issue still persists.

v1.26.4

[xiaolibuzai-ovo]

PageProcess error: bad xref, Traceback (most recent call last):
File "/usr/local/trpc/bin/process/pymupdf_process.py", line 152, in save
_ = self.doc.xref_object(xref)
File "/usr/local/trpc/bin/lib/pymupdf/init.py", line 6095, in xref_object
ret = extra.xref_object( self.this, xref, compressed, ascii)
File "/usr/local/trpc/bin/lib/pymupdf/extra.py", line 120, in xref_object
return _extra.xref_object(*args)
RuntimeError: bad xref

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/usr/local/trpc/bin/pdf_atom_process_py.py", line 125, in PageProcess
url = await pymupdf_cli.save(file_id, request.scene, request.namespace_id)
File "/usr/local/trpc/bin/process/pymupdf_process.py", line 154, in save
self.doc.update_object(xref, "<<>>")
File "/usr/local/trpc/bin/lib/pymupdf/init.py", line 5892, in update_object
RAISEPY("bad xref", MSG_BAD_XREF)
File "/usr/local/trpc/bin/lib/pymupdf/init.py", line 17449, in RAISEPY
raise Exception( msg)
Exception: bad xref

[xiaolibuzai-ovo]

file:
01995b6ca7837b52abaa24e38e8c076d.pdf

[julian-smith-artifex-com]

I have created a test with your code, and with the current pymupdf-1.26.4 it runs without raising an exception, though there is a warning from mupdf repairing PDF document.

def test_4702():
    path = util.download(
            'https://github.com/user-attachments/files/22403483/01995b6ca7837b52abaa24e38e8c076d.pdf',
            'test_4702.pdf',
            )
    with pymupdf.open(path) as document:
        for xref in range(1, document.xref_length()):
            print(f'{xref=}')
            _ = document.xref_object(xref)
    wt = pymupdf.TOOLS.mupdf_warnings()
    assert wt == 'repairing PDF document'

Is your code modifying the document before it does the xref loop?

[xiaolibuzai-ovo]

Yes, I am working on PDF layout restoration. After translating the English text, I cover it with a white rectangle and then insert the translated text using insert_html. However, an error occurs when saving the file.

[xiaolibuzai-ovo]

    Something like this.
    
    page = self.doc[page_num]
    if del_links:
        for link in page.get_links():
            page.delete_link(link)

    for i in range(len(blocks)):
        block = blocks[i]
        white = pymupdf.pdfcolor["white"]
        page.draw_rect(tuple(block.bbox), color=None, fill=white)
        
        
     for i in range(len(blocks)):
        block = blocks[i]
        try:
           
            page.insert_htmlbox(tuple(block.bbox), block.text, css=block.css, rotate=block.rotate, archive=ARCHIVE)
        except OverflowError as e:
            logger.error_context(self.ctx, f"restore err: {str(e)}")
            continue

[julian-smith-artifex-com]

Please post a full reproducer. For example it needs to specify page_num.