ignore whitespace in malformed floats #960

src/ezdxf/recover.py

@@ -1,6 +1,8 @@
 #  Copyright (c) 2020-2023, Manfred Moitzi
 #  License: MIT License
 from __future__ import annotations
+
+import string
 import typing
 from typing import (
     TYPE_CHECKING,
@@ -637,6 +639,11 @@ def synced_bytes_loader(stream: BinaryIO) -> Iterator[DXFTag]:
 ACADVER = b"$ACADVER"
 
 
+def _strip_whitespace(s: str) -> str:
+    ws = set(string.whitespace)
+    return "".join([c for c in s if c not in ws])
+
+
 def detect_encoding(tags: Iterable[DXFTag]) -> str:
     """Detect text encoding from header variables $DWGCODEPAGE and $ACADVER
     out of a stream of DXFTag objects.
@@ -721,6 +728,7 @@ def recover_int(s: Union[str, bytes]) -> int:
     def recover_float(s: Union[str, bytes]) -> float:
         if isinstance(s, bytes):
             s = s.decode(encoding="utf8", errors="ignore")
+        s = _strip_whitespace(s)
         value = _search_float(s)
         msg = f'recovered invalid floating point value "{s}" near line {line} as "{value}"'
         messages.append((AuditError.INVALID_FLOATING_POINT_VALUE, msg))

tests/test_00_dxf_low_level_structs/test_055_recover_loader.py

@@ -137,11 +137,17 @@ def test_underlines_in_numbers_are_handled_by_python(self):
         assert tags[1] == (70, 456000)
         assert len(msg) == 0
 
-    def test_spaces_in_floats_returns_invalid_value(self):
-        loader = bytes_loader(BytesIO(b"50\n1.000 e20\n"))
+    @pytest.mark.parametrize("b", [
+        b"50\n1.000 e20\n",
+        b"50\n1.000 E20\n",
+        b"50\n1.000\te20\n",
+        b"50\n1.000\tE20\n",
+    ])
+    def test_ignore_whitespace_in_floats(self, b):
+        loader = bytes_loader(BytesIO(b))
         msg = []
         tags = list(byte_tag_compiler(loader, messages=msg))
-        assert tags[0] == (50, 1), "invalid interpretation"
+        assert tags[0] == (50, 1e20)
         assert msg[0][0] == AuditError.INVALID_FLOATING_POINT_VALUE