Merge pull request #100 from galaxyproject/fix_#98

Improve document root detection method

server/galaxyls/server.py

@@ -33,6 +33,8 @@
 )
 from pygls.workspace import Document
 
+from galaxyls.services.validation import DocumentValidator
+
 from .config import CompletionMode, GalaxyToolsConfiguration
 from .features import AUTO_CLOSE_TAGS, CMD_GENERATE_COMMAND, CMD_GENERATE_TEST
 from .services.language import GalaxyToolLanguageService
@@ -182,4 +184,4 @@ def _get_xml_document(document: Document) -> XmlDocument:
 
 def _is_document_supported(document: Document) -> bool:
     """Returns True if the given document is supported by the server."""
-    return XmlDocument.has_valid_root(document)
+    return DocumentValidator().has_valid_root(document)

server/galaxyls/services/validation.py

@@ -0,0 +1,33 @@
+import re
+
+from typing import Optional
+from pygls.workspace import Document
+
+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]*)"
+
+
+class DocumentValidator:
+    """Provides some utilities to quickly check documents without completely parse them beforehand."""
+
+    def has_valid_root(self, document: Document) -> bool:
+        """Checks if the document's root element matches one of the supported types."""
+        root = self._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 in supported
+        return False
+
+    def _get_document_root_tag(self, 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)
+        if match:
+            group = match.group(TAG_GROUP_NAME)
+            return group
+        return None

server/galaxyls/services/xml/document.py

@@ -1,7 +1,6 @@
 from typing import Dict, Optional
 
 from anytree.search import findall
-from lxml import etree
 from pygls.types import Position, Range
 from pygls.workspace import Document
 
@@ -118,17 +117,3 @@ def get_position_after(self, element: XmlElement) -> Position:
         if element.is_self_closed:
             return convert_document_offset_to_position(self.document, element.end)
         return convert_document_offset_to_position(self.document, element.end_offset)
-
-    @staticmethod
-    def has_valid_root(document: Document) -> bool:
-        """Checks if the document's root element matches one of the supported types."""
-        try:
-            xml = etree.parse(str(document.path))
-            root = xml.getroot()
-            if root is not None and root.tag:
-                root_tag = root.tag.upper()
-                supported = [e.name for e in DocumentType if e != DocumentType.UNKNOWN]
-                return root_tag in supported
-            return False
-        except BaseException:
-            return False

server/galaxyls/tests/unit/test_validation.py

@@ -1,6 +1,8 @@
 import pytest
 from lxml import etree
 
+from galaxyls.services.validation import DocumentValidator
+
 from ...services.xsd.constants import TOOL_XSD_FILE
 from ...services.xsd.validation import GalaxyToolValidationService
 from .sample_data import (
@@ -62,3 +64,34 @@ def test_validate_macro_file_returns_diagnostics_when_syntax_error(self, xsd_sch
         actual = service.validate_document(xml_document)
 
         assert len(actual) == 1
+
+
+class TestDocumentValidatorClass:
+    @pytest.mark.parametrize(
+        "source, expected",
+        [
+            ("<tool>", True),
+            ("  <tool>", True),
+            ("\n<tool>", True),
+            ("\n  <tool>", True),
+            ("  \n  <tool>", True),
+            ("unexpected  <tool>", True),
+            ("  unexpected  <tool>", True),
+            ("\nunexpected\n  <tool>", True),
+            ('<?xml version="1.0" encoding="UTF-8"?><tool>', True),
+            ('<?xml version="1.0" encoding="UTF-8"?>\n<tool>', True),
+            ("<macros>", True),
+            ('<?xml version="1.0" encoding="UTF-8"?>\n<macros>', True),
+            ("", False),
+            ("   ", False),
+            ("test", False),
+            ("<test>", False),
+        ],
+    )
+    def test_has_valid_root_returns_expected(self, source: str, expected: bool) -> None:
+        document = TestUtils.to_document(source)
+        validator = DocumentValidator()
+
+        actual = validator.has_valid_root(document)
+
+        assert actual == expected