Tweak: improve date matching regex for dates after numbers (#8964)

diff --git a/src/documents/parsers.py b/src/documents/parsers.py
index 2d73dc63ffc4618933da41ac0300361b5b491113..d840817e4f5ba8091a9695205246ac1c0c74de9e 100644 (file)
--- a/src/documents/parsers.py
+++ b/src/documents/parsers.py
@@ -41,7 +41,7 @@ DATE_REGEX = re.compile(
     r"(\b|(?!=([_-])))(\d{1,2}[\. ]+[a-zéûäëčžúřěáíóńźçŞğü]{3,9} \d{4}|[a-zéûäëčžúřěáíóńźçŞğü]{3,9} \d{1,2}, \d{4})(\b|(?=([_-])))|"
     r"(\b|(?!=([_-])))([^\W\d_]{3,9} \d{1,2}, (\d{4}))(\b|(?=([_-])))|"
     r"(\b|(?!=([_-])))([^\W\d_]{3,9} \d{4})(\b|(?=([_-])))|"
-    r"(\b|(?!=([_-])))(\d{1,2}[^ ]{2}[\. ]+[^ ]{3,9}[ \.\/-]\d{4})(\b|(?=([_-])))|"
+    r"(\b|(?!=([_-])))(\d{1,2}[^ 0-9]{2}[\. ]+[^ ]{3,9}[ \.\/-]\d{4})(\b|(?=([_-])))|"
     r"(\b|(?!=([_-])))(\b\d{1,2}[ \.\/-][a-zéûäëčžúřěáíóńźçŞğü]{3}[ \.\/-]\d{4})(\b|(?=([_-])))",
     re.IGNORECASE,
 )