refactored filename extensions

core/TESTS/test_drawings.py

@@ -126,3 +126,22 @@ class DrawingsPathlibTests(TestCase):
 
         di = DataIssue.objects.filter(parser='Tunnel', message__contains="files named 'unique'")
         self.assertFalse(di.exists())
+
+    def test_extension_helpers_and_constants(self):
+        # Helpers should recognise supported/image suffixes (case-insensitive)
+        self.assertTrue(drawings._is_supported_suffix('.png'))
+        self.assertTrue(drawings._is_supported_suffix('.xml'))
+        self.assertTrue(drawings._is_supported_suffix('.TH'))
+        self.assertFalse(drawings._is_supported_suffix(''))
+        self.assertFalse(drawings._is_supported_suffix('.exe'))
+
+        self.assertTrue(drawings._is_image_suffix('.png'))
+        self.assertTrue(drawings._is_image_suffix('.JPEG'))
+        self.assertFalse(drawings._is_image_suffix('.xml'))
+        self.assertFalse(drawings._is_image_suffix(''))
+
+        # Constants should include expected values and be consistent
+        self.assertIn('.png', drawings.IMAGE_EXTS)
+        self.assertEqual(set(drawings.IMAGE_LIKE_EXTS), set(drawings.IMAGE_EXTS))
+        self.assertIn('.th', drawings.SUPPORTED_EXTENSIONS)
+        self.assertIn('.png', drawings.SUPPORTED_EXTENSIONS)

parsers/drawings.py

@@ -12,8 +12,7 @@ for tunnel and therion files
 """
 
 todo = """
-- Rename functions more consistently between tunnel and therion variants
+- Fix lost links to 1999 wallets now they have been renamed
-
 
 - implement: findimportinsert(therionfile, imp)
   Tries to link the scrap (Therion format) to the referenced therion scrap
@@ -30,7 +29,31 @@ rx_pcpath = re.compile(r'<pcarea area_signal="frame".*?sfsketch="([^"]*)" sfstyl
 rx_pctext = re.compile(r'pctext.*?\*ref&space;([^&]*)')
 
 
-IMAGE_EXTS = {".png", ".jpg", ".jpeg", ".pdf", ".gif", ".txt", ".svg"}
+# Supported suffixes are stored with a leading dot and lowercase (matches Path.suffix)
+# We factor out the image-like extensions which are common to both sets so they
+# are defined once and then reused to build the other sets.
+IMAGE_LIKE_EXTS = {".png", ".jpg", ".jpeg", ".pdf", ".gif", ".txt", ".svg"}
+
+# Extensions that we treat as "image-like" for the purposes of reference parsing
+IMAGE_EXTS = set(IMAGE_LIKE_EXTS)
+
+# Supported extensions include image-like ones plus drawing/text-specific ones
+SUPPORTED_EXTENSIONS = IMAGE_LIKE_EXTS.union({".xml", ".th", ".th2"})
+
+
+def _is_supported_suffix(suffix: str) -> bool:
+    """Return True if `suffix` (e.g. '.png') is a supported drawings extension."""
+    if not suffix:
+        return False
+    return suffix.lower() in SUPPORTED_EXTENSIONS
+
+
+def _is_image_suffix(suffix: str) -> bool:
+    """Return True if `suffix` looks like an image/scan-type suffix."""
+    if not suffix:
+        return False
+    return suffix.lower() in IMAGE_EXTS
+
 rx_wallet = re.compile(r"""
     # r"(\d\d\d\d#X?\d+\w?|1995-96kh|92-94Surveybookkh|1991surveybook|smkhs)/(.*?(?:png|jpg|pdf|jpeg|gif|txt))$", path
     # This regex is designed to extract a specific directory prefix and a filename
@@ -154,7 +177,7 @@ def _process_reference(dwgfile, path, parser_label="Tunnel"):
 
     # Not a wallet reference; check image extension and possibly drawing-to-drawing reference
     suffix = Path(path).suffix.lower()
-    if suffix in IMAGE_EXTS:
+    if _is_image_suffix(suffix):
         # It's an image/scanned file type; we don't treat it as a referenced drawing
         return None, None
 
@@ -325,8 +348,6 @@ def load_drawings_files():
     if os.path.isfile("therionrefs.log"):
         os.remove("therionrefs.log")
 
-    supported_extensions = {".txt", ".xml", ".th", ".th2", ".pdf", ".png", ".svg", ".jpg"}
-
     # Walk the tree with pathlib, skip hidden and backup files
     files_meta = []  # list of tuples (ext, rel_path, dwgname, pathobj)
     for p in drawdatadir.rglob('*'):
@@ -339,7 +360,7 @@ def load_drawings_files():
             continue
 
         suffix = p.suffix.lower()
-        if suffix in supported_extensions or suffix == '':
+        if _is_supported_suffix(suffix) or suffix == '':
             rel = p.relative_to(drawdatadir).as_posix()
             if suffix == '':
                 dwgname = p.name