Fix document detection

server/galaxyls/services/validation.py

@@ -5,9 +5,10 @@
 
 from galaxyls.services.xml.types import DocumentType
 
-MAX_PEEK_CONTENT = 100
-TAG_GROUP_NAME = "tag"
-TAG_REGEX = rf"[\n\s]*?.*?[\n\s]*?<(?!\?)(?P<{TAG_GROUP_NAME}>[\w]*)"
+MAX_PEEK_CONTENT = 1000
+TAG_GROUP_NAME = "root_tag"
+TAG_REGEX = r"[\n\s]*?.*?[\n\s]*?<(?!\?)(?!\!)(?P<root_tag>[\w]*)"
+SUPPORTED_ROOT_TAGS = [e.name.lower() for e in DocumentType if e != DocumentType.UNKNOWN]
 
 
 class DocumentValidator:
@@ -19,17 +20,15 @@ def has_valid_root(cls, document: Document) -> bool:
         or is an empty document."""
         if DocumentValidator.is_empty_document(document):
             return True
-        root = DocumentValidator._get_document_root_tag(document)
-        if root is not None:
-            root_tag = root.upper()
-            supported = [e.name for e in DocumentType if e != DocumentType.UNKNOWN]
-            return root_tag == "" or root_tag in supported
+        root_tag = DocumentValidator.get_document_root_tag(document)
+        if root_tag is not None:
+            return root_tag == "" or root_tag in SUPPORTED_ROOT_TAGS
         return False
 
     @classmethod
     def is_tool_document(cls, document: Document) -> bool:
         """Checks if the document's root element is <tool>."""
-        root = DocumentValidator._get_document_root_tag(document)
+        root = DocumentValidator.get_document_root_tag(document)
         if root is not None:
             root_tag = root.upper()
             return root_tag == DocumentType.TOOL.name
@@ -41,11 +40,11 @@ def is_empty_document(cls, document: Document) -> bool:
         return not document.source or document.source.isspace()
 
     @classmethod
-    def _get_document_root_tag(cls, document: Document) -> Optional[str]:
+    def get_document_root_tag(cls, document: Document) -> Optional[str]:
         """Checks the first MAX_PEEK_CONTENT characters of the document for a root tag and
         returns the name of the tag if found."""
         content_peek = document.source[:MAX_PEEK_CONTENT]
-        match = re.match(TAG_REGEX, content_peek)
+        match = re.search(TAG_REGEX, content_peek)
         if match:
             group = match.group(TAG_GROUP_NAME)
             return group

server/galaxyls/tests/unit/test_validation.py

@@ -36,6 +36,19 @@ class TestDocumentValidatorClass:
             ('<?xml version="1.0" encoding="UTF-8"?>\n<macros>', True),
             ("test", False),
             ("<test>", False),
+            ("<visualization>", False),
+            (
+                '<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE visualization SYSTEM "../../visualization.dtd">\n<visualization>',
+                False,
+            ),
+            (
+                '<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE test SYSTEM "../../test.dtd">\n',
+                False,
+            ),
+            (
+                '<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE test SYSTEM "../../test.dtd">\n<tool',
+                True,
+            ),
         ],
     )
     def test_has_valid_root_returns_expected(self, source: str, expected: bool) -> None:
@@ -45,3 +58,43 @@ def test_has_valid_root_returns_expected(self, source: str, expected: bool) -> N
         actual = validator.has_valid_root(document)
 
         assert actual == expected
+
+    @pytest.mark.parametrize(
+        "source, expected",
+        [
+            ("", None),
+            ("   ", None),
+            ("  \n ", None),
+            ("test", None),
+            ("<", ""),
+            ("<tes", "tes"),
+            ("<test   ", "test"),
+            ("<tool>", "tool"),
+            ("  <tool>", "tool"),
+            ("\n<tool>", "tool"),
+            ("\n  <tool>", "tool"),
+            ("  \n  <tool>", "tool"),
+            ("unexpected  <tool>", "tool"),
+            ("  unexpected  <tool>", "tool"),
+            ("\nunexpected\n  <tool>", "tool"),
+            ('<?xml version="1.0" encoding="UTF-8"?><tool>', "tool"),
+            ('<?xml version="1.0" encoding="UTF-8"?>\n<tool>', "tool"),
+            ("<macros>", "macros"),
+            ('<?xml version="1.0" encoding="UTF-8"?>\n<macros>', "macros"),
+            ("<test>", "test"),
+            ('<test attribute="0">', "test"),
+            ("<visualization>", "visualization"),
+            ('<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE test SYSTEM "../../test.dtd">\n', None),
+            (
+                '<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE visualization SYSTEM "../../visualization.dtd">\n<visualization>',
+                "visualization",
+            ),
+        ],
+    )
+    def test_get_document_root_tag_returns_expected(self, source: str, expected: bool) -> None:
+        document = TestUtils.to_document(source)
+        validator = DocumentValidator()
+
+        actual = validator.get_document_root_tag(document)
+
+        assert actual == expected