Find metadata of images embedded in PDFs

.gitignore

venv/
__pycache__/
.mypy_cache/

Makefile

export PIP_DISABLE_PIP_VERSION_CHECK=1

venv: requirements.txt requirements_dev.txt
	@python3 -m venv $@
	@source $@/bin/activate && pip install -r $< -r requirements_dev.txt
	@echo "enter virtual environment: source $@/bin/activate"

pdf-exif.py

#!/usr/bin/env python3

import sys
import hashlib
from io import BytesIO
from pathlib import Path

import pymupdf  # type: ignore
from PIL import Image, UnidentifiedImageError

def main() -> None:
    """Entry point"""

    document_path = Path(sys.argv[1]).expanduser()
    with document_path.open("r") as fd:
        document = pymupdf.open(fd)

    print(document.get_xml_metadata())

    hashes: set[bytes] = set()
    for xref in range(1, document.xref_length()):
        if stream := document.xref_stream(xref):
            if document.xref_get_key(xref, "Subtype") == ("name", "/Image"):
                checksum = hashlib.md5(stream)
                if checksum.digest() not in hashes:
                    hashes.add(checksum.digest())
                    print(xref, "  \t", checksum.hexdigest())
                    try:
                        image = Image.open(BytesIO(stream))
                        if exif := image.getexif():
                            print(exif)
                    except UnidentifiedImageError:
                        print("Unidentified image type")

if __name__ == "__main__":
    main()

requirements.txt

pymupdf
pillow

requirements_dev.txt

pylint
mypy

Run

pip install -r requirements.txt

to install dependencies. Then change the path and execute with ./pdf-exif.py path/to/file.pdf.

To see metadata, pdfinfo -meta path/to/file.pdf or exiftool path/to/file.pdf.

To strip metadata (if embedded objects do not contain metadata!),

exiftool -all:all="" path/to/file.pdf              # to remove metadata pointers
mutool clean -gggglcs path/to/file.pdf             # to garbage collect, possibly unnecessary
qpdf --linearize path/to/file.pdf path/to/out.pdf  # to actually remove the metadata

and test:

• smoke test with pdfinfo -meta path/to/out.pdf,
• open the binary file (with e.g. vim -b or xxd) and search for metadata tags and ensure no matches.

You can then add metadata such as Title, Subject and so on using exiftool.

---

Tools (recommended to install with Homebrew):

• exiftool (www, brew: exiftool),
• mutool (www, brew: mupdf-tools),
• qpdf (www, brew: qpdf).