Author line: drop bare e-mail domains, attach combining marks to the following letter

pdf_pipeline.py

@@ -96,8 +96,8 @@ def prepare(pdf, job, instruction, canvas, accent, columns, custom_spec=''):
             if re.search(r'published as a conference|under review|preprint|anonymous author|proceedings of|copyright|workshop',txt,re.I): continue
             author_lines.append(txt)
         authors=' '.join(author_lines)
-        authors=re.sub(r'\{[^}]*\}?|\S+@\S+','',authors)            # e-mail lists
-        authors=re.sub(r'\s+([\u0300-\u036f])',r'\1',authors)          # 'Balunovi ́c' -> 'Balunović'
+        authors=re.sub(r'\{[^}]*\}?|\S*@\S+','',authors)            # e-mail lists
+        authors=re.sub(r'\s+([\u0300-\u036f])(\w)',r'\2\1',authors)    # 'Balunovi ́c' -> 'Balunović' (mark belongs to the next letter)
         authors=clean_text(unicodedata.normalize('NFC',authors))[:500] or clean_text(doc.metadata.get('author',''))[:500]
         asset_lines=[];assets=[]
         for a in chosen: