bug: sometimes impossible to save redactions, potential data loss #198

mahlzahn commented 2026年05月05日 16:14:36 +02:00

Description

Sometimes it is impossible to save redactions, which leads to potential data loss (at least the already performed redactions are lost)

Example files

• samples_openpreserve/jhove-errors/PDF-HUL-35/143424271191706722106594310729442158189.pdf
• samples_openpreserve/jhove-errors/PDF-HUL-35/A Defence of the New Tenseless Theory of Time.pdf
• samples_openpreserve/pdf-handbuilt-test-corpus/T02-02_004_page-tree-non-existing-object-as-kid.pdf
• samples_openpreserve/pdf-handbuilt-test-corpus/T02-03_006_page-object-missing.pdf

Logs

Traceback (most recent call last):
 File "/usr/share/censor/censor/pymupdf_helpers.py", line 23, in clean_page
 mupdf_page.clean_contents(sanitize=False)
 ~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^
 File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 10870, in clean_contents
 mupdf.pdf_filter_page_contents( page.doc(), page, filter_)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "/usr/lib/python3.14/site-packages/mupdf/__init__.py", line 56523, in pdf_filter_page_contents
 return _mupdf.pdf_filter_page_contents(doc, page, options)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^
mupdf.FzErrorArgument: code=4: not a dict (null)

and

WARNING: Saving Document Failed: code=4: not a dict (null)
Traceback (most recent call last):
 File "/usr/share/censor/censor/window.py", line 295, in on_save_response
 self.save_file(file)
 ~~~~~~~~~~~~~~^^^^^^
 File "/usr/share/censor/censor/window.py", line 302, in save_file
 self.document.save(file.get_path())
 ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
 File "/usr/share/censor/censor/document.py", line 156, in save
 mupdf_page.add_redact_annot(redaction_rotated, fill=False)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 10549, in add_redact_annot
 annot = self._add_redact_annot(quad, text=text, da_str=da_str,
 align=align, fill=fill)
 File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 9726, in _add_redact_annot
 annot = mupdf.pdf_create_annot(page, mupdf.PDF_ANNOT_REDACT)
 File "/usr/lib/python3.14/site-packages/mupdf/__init__.py", line 55925, in pdf_create_annot
 return _mupdf.pdf_create_annot(page, type)
 ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^
mupdf.FzErrorArgument: code=4: not a dict (null)

Implementation

1. Determine already when opening if a file cannot be saved (actually it is already saved once, but seems to be not enough, yet). Just run dummy mupdf_page.add_redact_annot and/or mupdf_page.clean_contents and stop opening the file
2. Alternatively, fix the issue (or let upstream fix it)

Upstream issue

• not reported, yet

## Description Sometimes it is impossible to save redactions, which leads to potential data loss (at least the already performed redactions are lost) ## Example files - samples_openpreserve/jhove-errors/PDF-HUL-35/143424271191706722106594310729442158189.pdf - samples_openpreserve/jhove-errors/PDF-HUL-35/A Defence of the New Tenseless Theory of Time.pdf - samples_openpreserve/pdf-handbuilt-test-corpus/T02-02_004_page-tree-non-existing-object-as-kid.pdf - samples_openpreserve/pdf-handbuilt-test-corpus/T02-03_006_page-object-missing.pdf ## Logs ```py Traceback (most recent call last): File "/usr/share/censor/censor/pymupdf_helpers.py", line 23, in clean_page mupdf_page.clean_contents(sanitize=False) ~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^ File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 10870, in clean_contents mupdf.pdf_filter_page_contents( page.doc(), page, filter_) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.14/site-packages/mupdf/__init__.py", line 56523, in pdf_filter_page_contents return _mupdf.pdf_filter_page_contents(doc, page, options) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^ mupdf.FzErrorArgument: code=4: not a dict (null) ``` and ```py WARNING: Saving Document Failed: code=4: not a dict (null) Traceback (most recent call last): File "/usr/share/censor/censor/window.py", line 295, in on_save_response self.save_file(file) ~~~~~~~~~~~~~~^^^^^^ File "/usr/share/censor/censor/window.py", line 302, in save_file self.document.save(file.get_path()) ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^ File "/usr/share/censor/censor/document.py", line 156, in save mupdf_page.add_redact_annot(redaction_rotated, fill=False) ~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 10549, in add_redact_annot annot = self._add_redact_annot(quad, text=text, da_str=da_str, align=align, fill=fill) File "/usr/lib/python3.14/site-packages/pymupdf/__init__.py", line 9726, in _add_redact_annot annot = mupdf.pdf_create_annot(page, mupdf.PDF_ANNOT_REDACT) File "/usr/lib/python3.14/site-packages/mupdf/__init__.py", line 55925, in pdf_create_annot return _mupdf.pdf_create_annot(page, type) ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^ mupdf.FzErrorArgument: code=4: not a dict (null) ``` ## Implementation 1. Determine already when opening if a file cannot be saved (actually it is already saved once, but seems to be not enough, yet). Just run dummy `mupdf_page.add_redact_annot` and/or `mupdf_page.clean_contents` and stop opening the file 2. Alternatively, fix the issue (or let upstream fix it) ## Upstream issue - not reported, yet