improve form regex

vplan-fr · Mar 10, 2024 · a2fab17 · a2fab17
1 parent 6e28848
commit a2fab17
Showing 1 changed file with 10 additions and 6 deletions.
diff --git a/backend/vplan_utils.py b/backend/vplan_utils.py
@@ -10,17 +10,21 @@
 _parse_form_pattern = re.compile(
     r"(?<!\S)(?:"
     r"(?P<major>\d{1,2}(?!\d)|[A-Za-zÄÖÜäöüß]+(?![A-Za-zÄÖÜäöüß]))"
-    r"(?P<sep>[/.] |[^A-Za-zÄÖÜäöüß0-9() \n]?)"
-    r"(?P<minor>(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+?)(?:,(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+?))*)"
-    r"|(?P<alpha>\d{1,2})"
+    r"(?P<sep> |[/._]|[^A-Za-zÄÖÜäöüß0-9() \n]?)(?:(?<! ) )?"
+    r"(?P<minor>"
+    r"(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+\d{0,2})"
+    r"(?:,(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+\d{0,2}))*"
+    r")|(?P<alpha>\d{1,2})"
     r")(?![^\s,:])"
 )
 _loose_parse_form_pattern = re.compile(
     r"(?<!\S)(?:"
     r"(?P<major>\d{1,2}(?!\d)|[A-Za-zÄÖÜäöüß]+(?![A-Za-zÄÖÜäöüß]))"
-    r"(?P<sep>[/.]? |[^A-Za-zÄÖÜäöüß0-9() \n]?)"
-    r"(?P<minor>(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+?)(?:,(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+?))*)"
-    r"|(?P<alpha>\d{1,2})"
+    r"(?P<sep> |[/._]|[^A-Za-zÄÖÜäöüß0-9() \n]?)(?:(?<! ) )?"
+    r"(?P<minor>"
+    r"(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+\d{0,2})"
+    r"(?:,(?:\d{1,2}[A-Za-zÄÖÜäöüß]?|[A-Za-zÄÖÜäöüß]+\d{0,2}))*"
+    r")|(?P<alpha>\d{1,2})"
     r")(?![^\s,:()])"
 )