Be more intelligent about PDFs without OCR layer

scripts/import-generic.py

@@ -47,7 +47,9 @@ class OllamaImporter(Importer):
 
     def identify(self, filepath):
         mimetype, encoding = mimetypes.guess_type(filepath)
-        return mimetype == "application/pdf"
+        if mimetype != "application/pdf":
+            return False
+        return "".join(pdf_lines(filepath)) != ""
 
     def account(self, filepath):
         return self.file_account

scripts/ocr-pdf.sh

@@ -0,0 +1,6 @@
+#!/bin/bash -eux
+
+in=$1
+out=$(dirname "$in")/$(basename -- "$in" ".${in##*.}")-ocr.pdf
+
+ocrmypdf --language eng+nld --tesseract-pagesegmode 11 "$in" "$out"