Make the go/no-go decision pluggable

src/ocrmypdf/_pipeline.py

@@ -160,6 +160,12 @@ def validate_pdfinfo_options(context):
     pdfinfo = context.pdfinfo
     options = context.options
 
+    if options.plugin_validation:
+        validate = load_plugin(options.plugin_validation)
+        result = validate(context)
+        if result is not None:
+            return result
+
     if pdfinfo.needs_rendering:
         log.error(
             "This PDF contains dynamic XFA forms created by Adobe LiveCycle "

src/ocrmypdf/api.py

@@ -119,7 +119,9 @@ def create_options(*, input_file, output_file, **kwargs):
     for arg, val in kwargs.items():
         if val is None:
             continue
-        if arg.startswith('filter') and (callable(val) or isinstance(val, str)):
+        if (arg.startswith('plugin') or arg.startswith('filter')) and (
+            callable(val) or isinstance(val, str)
+        ):
             deferred.append((arg, val))
             continue
         elif arg == 'tesseract_env':
@@ -199,6 +201,7 @@ def ocrmypdf(  # pylint: disable=unused-argument
     keep_temporary_files=None,
     progress_bar=None,
     filter_ocr_image=None,
+    plugin_validation=None,
     tesseract_env=None,
 ):
     """Run OCRmyPDF on one PDF or image.

src/ocrmypdf/cli.py

@@ -468,10 +468,13 @@ advanced.add_argument(
     help="Specify the location of the Tesseract user patterns file.",
 )
 
-filters = parser.add_argument_group("Filters", argparse.SUPPRESS)
-filters.add_argument(
+plugins = parser.add_argument_group("Filters and Plugins", argparse.SUPPRESS)
+plugins.add_argument(
     '--filter-ocr-image', help=argparse.SUPPRESS, type=check_plugin_loadable
 )
+plugins.add_argument(
+    '--plugin-validation', help=argparse.SUPPRESS, type=check_plugin_loadable
+)
 
 debugging = parser.add_argument_group(
     "Debugging", "Arguments to help with troubleshooting and debugging"

src/ocrmypdf/pdfinfo/layout.py

@@ -185,7 +185,7 @@ class LTStateAwareChar(LTChar):
 
     def get_text(self):
         if isinstance(self._text, tuple):
-            return '�'
+            return '\ufffd'  # standard 'Unknown symbol'
         return self._text
 
     def __repr__(self):