Detect when metadata is dropped during PDF/A conversion

src/ocrmypdf/_pipeline.py

@@ -878,19 +878,19 @@ def metadata_fixup(
     options = context.get_options()
 
     input_files = list(f for f in flatten_groups(input_files_groups))
-    metadata_file = next(
+    original_file = next(
         (ii for ii in input_files if ii.endswith('.repaired.pdf')), None
     )
     layers_file = next(
         (ii for ii in input_files if ii.endswith('layers.rendered.pdf')), None
     )
-    pdfa = next(
+    pdfa_file = next(
         (ii for ii in input_files if ii.endswith('pdfa.pdf')), None
     )
-    metadata = pikepdf.open(metadata_file)
-    docinfo = get_docinfo(metadata, options)
+    original = pikepdf.open(original_file)
+    docinfo = get_docinfo(original, options)
 
-    working_file = pdfa if pdfa else layers_file
+    working_file = pdfa_file if pdfa_file else layers_file
 
     pdf = pikepdf.open(working_file)
     with pdf.open_metadata() as meta:
@@ -899,6 +899,20 @@ def metadata_fixup(
         # match Ghostscript, for consistency
         if 'xmp:CreateDate' not in meta:
             meta['xmp:CreateDate'] = meta.get('xmp:ModifyDate', '')
+        if pdfa_file:
+            meta_original = original.open_metadata()
+            not_copied = set(meta_original.keys()) - set(meta.keys())
+            if not_copied:
+                log.warning(
+                    "Some input metadata could not be copied because it is not "
+                    "permitted in PDF/A. You may wish to examine the output "
+                    "PDF's XMP metadata."
+                )
+                log.debug(
+                    "The following metadata fields were not copied: %r",
+                    not_copied
+                )
+
     pdf.save(output_file, compress_streams=True,
              object_stream_mode=pikepdf.ObjectStreamMode.generate)
 

tests/test_metadata.py

@@ -21,8 +21,9 @@ import pytest
 from datetime import timezone
 from pathlib import Path
 from shutil import copyfile
-from unittest.mock import patch
+from unittest.mock import patch, MagicMock
 import datetime
+from shutil import copyfile
 
 import pikepdf
 from pikepdf.models.metadata import decode_pdf_date
@@ -278,3 +279,39 @@ def test_kodak_toc(resources, outpdf, spoof_tesseract_noop):
 
     if pikepdf.Name.First in p.root.Outlines:
         assert isinstance(p.root.Outlines.First, pikepdf.Dictionary)
+
+
+def test_metadata_fixup_warning(resources, outdir):
+    from ocrmypdf._pipeline import metadata_fixup
+
+    input_files = [
+        str(outdir / 'graph.repaired.pdf'),
+        str(outdir / 'layers.rendered.pdf'),
+        str(outdir / 'pdfa.pdf'),  # It is okay that this is not a PDF/A
+    ]
+    for f in input_files:
+        copyfile(resources / 'graph.pdf', f)
+
+    log = MagicMock()
+    context = MagicMock()
+    metadata_fixup(
+        input_files_groups=input_files,
+        output_file=outdir / 'out.pdf',
+        log=log,
+        context=context)
+    log.warning.assert_not_called()
+
+    # Now add some metadata that will not be copyable
+    graph = pikepdf.open(outdir / 'graph.repaired.pdf')
+    with graph.open_metadata() as meta:
+        meta['prism2:publicationName'] = 'OCRmyPDF Test'
+    graph.save(outdir / 'graph.repaired.pdf')
+
+    log = MagicMock()
+    context = MagicMock()
+    metadata_fixup(
+        input_files_groups=input_files,
+        output_file=outdir / 'out.pdf',
+        log=log,
+        context=context)
+    log.warning.assert_called_once()