"object (___) was not found in its object stream" error

[brandxntu]

I am experiencing an issue with a handful of PDF files.

After opening these specific files with fitz.open(file) , I am greeted with a string of the following errors:

mupdf: object (2065 0 R) was not found in its object stream
mupdf: object (2068 0 R) was not found in its object stream
mupdf: object (2071 0 R) was not found in its object stream
mupdf: object (2075 0 R) was not found in its object stream

The program does not crash however.
Information can still be extracted from the file at this point including metadata, page.mediabox, and page.get_text().

However, when attempting to save the file using pdf.save() the program will crash.

mupdf: object (1924 0 R) was not found in its object stream
Traceback (most recent call last):

File "", line 1, in
pdf.save("file.pdf")

RuntimeError: object (1924 0 R) was not found in its object stream

Using metadata, I found that the problem files all had a metadata['creator'] value of "PDF-XChange Editor 7.0.326.1"

Any idea what is going on or how this issue could be fixed would be much appreciated.

[JorjMcKie]

These PDFs are severely damaged. Impossible to tell why from this end.
The message may sound cryptic, but that PDF viewer compresses PDF objects into so-called objects streams (object type "Objstm" to save space. These are objects that contain other object definitions. Here, MuPDF cannot find objects in some of these compressed data.

Try to clean the file via mutool clean -gggg damaged.pdf .... May not work if the problems are too severe. If it fails you may be able to at least rescue a few pages via insert_pdf using a new file.

[JorjMcKie]

@shmyyy -

fitz.open works on document - so means the document is not damaged.

No, wrong conclusion! It only means that some of the most severe errors either did not occur or were successfully repaired. Nothing else.

You PDF may still have all sorts of other errors: damaged images, fonts, page tree, objects not found, ... you name it.
PDF is not a file format that can ensure consistency.

PyMuPDF does not "mess up" anything.

[JorjMcKie]

So means it is not damaged ...

No, you cannot conclude this!
Displaying a PDF (paging through it with some PDF reader) does not necessarily access everything in the file. So you may not touch any messy part within the file.
Whereas saving the file accesses yet other parts - among them damaged pieces.

Please accept: your file is damaged!
You may be able to rescue (parts of) it in one or the other way, but there definitely are damages.

[renzo-montesa]

I have a similar case
the error code is : object (2135 0 R) was not found in its object stream

[JorjMcKie]

Reiterating: Your PDF is damaged. If you have the latest PyMuPDF version you should be able to save it.

[renzo-montesa]

can i share you the code that im using PyMuPDF on.
I am trying to batch process multiple pdf files. and obtaining annotations based on
annotation_subject or subject in annotation_title or subject in annotation_name . It checks all annotations and only keeps the annotation that the user inputted then saves it in a new pdf in a new directory.

Most of the files works properly and it saves the pdf with no error. but some of the pdfs that It process receives the error object (2135 0 R) . It saves the file but when I open it its corrupted.

The original files before the process can be opened and checked in pdf viewer. Only after its processed that it becomes corrupted.

Im curious to know why its only corrupted when it is processed by the program and not corrupted when not processed. but this happens only to some files not all of it

import os
import fitz
import pandas as pd
import webbrowser

def process_pdfs(dir_path, comment_subjects_file):
# Check if a directory is selected
if not dir_path:
print("No directory selected.")
return

# Check if the specified directory exists
if not os.path.isdir(dir_path):
    print("Directory does not exist.")
    return

# Check if the comment_subjects_file is provided
if not comment_subjects_file:
    print("No comment subjects file provided."),
    return

# Check if the comment_subjects_file is a valid CSV file
if not comment_subjects_file.lower().endswith(".xlsx"):
    print("Invalid comment subjects file. Please provide a CSV file.")
    return

# Load the comment subjects from the CSV file
try:
    df = pd.read_excel(comment_subjects_file)
    print(df)
    df["Sub System Description"] = df["Sub System Description"].str.rstrip()
    df["Sub System Description"] = df["Sub System Description"].str.replace("/", " ").str.replace("\\", " ")
    comment_subjects = df["Sub System"].tolist()
    sub_system_description=df["Sub System Description"].tolist()
    # Remove trailing spaces in the "Sub System Description" column
    
except Exception as e:
    print(f"Error loading comment subjects from the file: {e}")
    return
# Check if comment_subjects is empty
if not comment_subjects:
    print("No comment subjects found in the file.")
    return
if not sub_system_description:
    print("No Sub System Description found in the file.")
    return

print(dir_path)

pdf_files = [file_name for file_name in os.listdir(dir_path) if file_name.lower().endswith(".pdf")]
if len(pdf_files) == 0:
    print("No PDF files found in the directory ")
    return

loop_counter=0
for subject in comment_subjects:
    dir_path_subject=dir_path+"/"+subject
    print(subject)
    # Check if there are no PDF files in the directory
    # Create a folder for the modified PDFs
    description=sub_system_description[loop_counter]
    output_folder = os.path.join(dir_path, subject+"_"+description)
    os.makedirs(output_folder, exist_ok=True)
    loop_counter+=1

    # List to store PDFs that don't have any of the comment_subjects
    pdfs_without_comment = []
    file_code_error=['-----------SYNTAX ERRORS AHEAD----------']

    
    # Loop through all PDF files in the directory
    for file_name in pdf_files:
        pdfs_with_comment=[]
        file_name_error=[]
        file_path = os.path.join(dir_path, file_name)
        # Open the PDF file
        pdf_document = fitz.open(file_path)

        # Flag to indicate if the PDF has any comment with any of the comment_subjects
        has_comment = False

        for page in pdf_document:
            print(page)
            try:  
                # Get all annotations (comments) on the page
                annotations = page.annots()
                comment_name=[]
                # Loop through all comments and delete them
                annot_counter=0
                for annotation in annotations:
                
                    annot_counter+=1
                    annotation_info = annotation.info
                    annotation_subject = str(annotation_info.get('subject', ''))
                    annotation_title=  str(annotation_info.get('title',''))
                    print(annotation_title)
                    annotation_name= str(annotation_info.get('name',''))
                   
                    has_matching_subject = False
    
                    if subject in annotation_subject or subject in annotation_title or subject in annotation_name :
                        has_matching_subject = True
                        if subject not in comment_name:
                            comment_name.append(subject)
                        
                    if has_matching_subject:
                        has_comment = True
                        
                    else:
                        page.delete_annot(annotation)
            except RuntimeError as e:
                error_message = str(e)
                if error_message.startswith("xref") and error_message.endswith("is not an annot of this page"):
                    xref_value = int(error_message.split()[1])
                    print(f"Error: xref {xref_value} is not an annot of this page")
                    error=f"Error: xref {xref_value} is not an annot of this page"
                    file_code_error.append(error)
                    file_name_error.append(file_name)
                    if file_name_error:
                        print("i am here")
                        notepad_files = os.path.join(output_folder, "files_skipped_because_error.txt")
                        with open(notepad_files, "w") as file:
                            for file_name in file_name_error:

                                file.write(str(file_name) + "\n")
                            
                            for file_name in file_code_error:
                                file.write(str(file_name) + "\n") 
                    continue  # Skip this xref and move to the next one

                else:
                    # Handle other runtime errors if needed
                    print("An error occurred:", str(e))        
       

        # Create new filename with comment_subjects included
        new_file_name = file_name
        new_file_path = os.path.join(output_folder, new_file_name)

        # # Check if the PDF has any comment with any of the comment_subjects
        # if has_comment:
        #     # Save changes to new file
        #     pdf_document.save(new_file_path)
        # else:
        #     # Add the file name to the list of PDFs without comments
        #     pdfs_without_comment.append(file_name)


                    # Check if the PDF has any comment with any of the comment_subjects
        if has_comment:
            # Open the PDF file for manual inspection
            webbrowser.open(new_file_path)
        else:
            # Add the file name to the list of PDFs without comments
            pdfs_without_comment.append(file_name)


        # Close the PDF file
        pdf_document.close()

    # Print the list of PDF files without comments
    if pdfs_without_comment:
        notepad_file = os.path.join(output_folder, "files_without_comment.txt")
        with open(notepad_file, "w") as file:
            for file_name in pdfs_without_comment:

                file.write(str(file_name) + "\n")

    print("Processing complete.")

Get user input for directory path

directory_path = input("Enter the directory path: ")

Get user input for file path

comment_subjects_file = input("Enter the file path for comment_subjects.csv: ")

process_pdfs(directory_path, comment_subjects_file)

[JorjMcKie]

This is a huge amount of code. I don't see how you save the changed documents: a lot depends on the save options - if you use too "soft" ones, a potential damage of the input may be propagated.
I believe you are using insufficient save options, because (Py-) MuPDF never creates those PDF objects which exclusively can cause that error message.

Message RuntimeError: object (nnnn 0 R) was not found in its object stream only occurs if there is reference to xref nnnn that is contained in a compressed "super"-object (object stream), but cannot be located in there. When using strong save options (garbage=3, deflate=True), object streams are dissolved and become normal PDF xrefs. Opening a file that has been saved like this never can generate this message.

What is your pymupdf version? You must use the most current one, because MuPDF has been too picky before that and raised an exception when it encountered that situation.

[JorjMcKie]

Another comment: A loop structured like this

for annot in page.annots():
    ...
    page.delete_annot(annot)

is dangerous and will not work in older pymupdf versions, because you are altering the iterator page.annots() from within the loop that depends on it.