sep and tables pattern must be provided together

Author: svittoz

edsnlp/pipes/misc/tables/patterns.py

@@ -1,4 +1,4 @@
-sep = r"¦"
+sep = r"'¦|\|'"
 regex = dict(
     tables=rf"(\b.*{sep}.*\n)+",
 )

edsnlp/pipes/misc/tables/tables.py

@@ -127,14 +127,16 @@ def __init__(
         attr: Union[Dict[str, str], str] = "TEXT",
         ignore_excluded: bool = True,
     ):
-        if tables_pattern is None:
+        if tables_pattern is None and sep_pattern is None:
             self.tables_pattern = patterns.regex
-        else:
-            self.tables_pattern = tables_pattern
-
-        if sep_pattern is None:
             self.sep = patterns.sep
+        elif tables_pattern is None or sep_pattern is None:
+            raise ValueError(
+                "Both tables_pattern and sep_pattern must be provided "
+                "for custom eds.table pipeline."
+            )
         else:
+            self.tables_pattern = tables_pattern
             self.sep = sep_pattern
 
         super().__init__(