pdf_compare.py

# SPDX-FileCopyrightText: 2025 James R. Barlow
# SPDX-License-Identifier: MIT

"""Compare two PDFs."""

from __future__ importannotations

importos
fromioimportBytesIO
frompathlibimportPath
fromtempfileimportTemporaryDirectory

importpikepdf
importpymupdf
importstreamlitasst
fromlxmlimportetree
fromstreamlit_pdf_viewerimportpdf_viewer


defdo_metadata(pdf):
withpikepdf.open(pdf) aspdf:
withpdf.open_metadata() asmeta:
xml_txt=str(meta)
parser=etree.XMLParser(remove_blank_text=True)
tree=etree.fromstring(xml_txt, parser=parser)
st.code(
etree.tostring(tree, pretty_print=True).decode("utf-8"),
language="xml",
 )
st.write(pdf.docinfo)
st.write("Number of pages:", len(pdf.pages))


defmain():
st.set_page_config(layout="wide")

st.title("PDF Compare")
st.write("Compare two PDFs.")

col1, col2=st.columns(2)
withcol1:
uploaded_pdf1=st.file_uploader("Upload a PDF", type=["pdf"], key='pdf1')
withcol2:
uploaded_pdf2=st.file_uploader("Upload a PDF", type=["pdf"], key='pdf2')
ifuploaded_pdf1isNoneoruploaded_pdf2isNone:
return

pdf_bytes1=uploaded_pdf1.getvalue()
pdf_bytes2=uploaded_pdf2.getvalue()

withst.expander("PDF Metadata"):
col1, col2=st.columns(2)
withcol1:
do_metadata(BytesIO(pdf_bytes1))
withcol2:
do_metadata(BytesIO(pdf_bytes2))

withTemporaryDirectory() asd:
Path(d, "1.pdf").write_bytes(pdf_bytes1)
Path(d, "2.pdf").write_bytes(pdf_bytes2)

withst.expander("Text"):
doc1=pymupdf.open(os.path.join(d, "1.pdf"))
doc2=pymupdf.open(os.path.join(d, "2.pdf"))
fori, page1_2inenumerate(zip(doc1, doc2)):
st.write(f"Page {i+1}")
page1, page2=page1_2
col1, col2=st.columns(2)
withcol1, st.container(border=True):
st.write(page1.get_text())
withcol2, st.container(border=True):
st.write(page2.get_text())

withst.expander("PDF Viewer"):
col1, col2=st.columns(2)
withcol1:
pdf_viewer(Path(d, "1.pdf"), key='pdf_viewer1', render_text=True)
withcol2:
pdf_viewer(Path(d, "2.pdf"), key='pdf_viewer2', render_text=True)


if__name__=="__main__":
main()