More fixes towards swapped punctuation

hplt-project · Aug 8, 2023 · 61f4671 · 61f4671
1 parent dcaaff0
commit 61f4671
Showing 1 changed file with 6 additions and 0 deletions.
diff --git a/opuscleaner/filters/fix_sent_final_punct.py b/opuscleaner/filters/fix_sent_final_punct.py
@@ -17,6 +17,12 @@
     # Sometimes two punctuation marks are swapped...
     if len(src) >=2 and len(trg) >= 2 and src[-2] == trg[-1] and src[-1] == trg[-2]:
        trg = trg[:-2] + src[-2] + src[-1]
+    # Sometimes they are swapped with space around eg SPACE». -> .SPACE»
+    if len(src) >=3 and src[-1] in my_punct and src[-2] == '»' and src[-3] == ' ':
+       src = src[:-3] + src[-1] + ' ' + src[-2]
+    if len(trg) >=3 and trg[-1] in my_punct and trg[-2] == '»' and trg[-3] == ' ':
+       trg = trg[:-3] + trg[-1] + ' ' + trg[-2]
+
 
     # check for the french quotes special case
     if (src[-1] == '»' or src[-1] == '«') and trg[-1] not in my_punct: