Merge pull request #621 from nextstrain/improve-validate

Improve validation of Auspice JSONs

augur/data/schema-export-v2.json

@@ -445,9 +445,9 @@
                 },
                 "children": {
                     "description": "Child nodes. Recursive structure. Terminal nodes do not have this property.",
-                    "$comment": "Polytomies (more than 2 items) allowed.",
+                    "$comment": "Polytomies (more than 2 items) allowed, as are nodes with a single child.",
                     "type": "array",
-                    "minItems": 2,
+                    "minItems": 1,
                     "items": {"$ref": "#/properties/tree"}
                 }
             }

augur/validate_export.py

@@ -7,6 +7,20 @@
 import sys
 from collections import defaultdict
 
+def ensure_no_duplicate_names(root, ValidateError):
+    """
+    Check that all node names are identical, which is required for auspice (v2) JSONs.
+    """
+    names = set()
+    def recurse(node):
+        if node["name"] in names:
+            raise ValidateError(f"Node {node['name']} appears multiple times in the tree.")
+        names.add(node["name"])
+        if "children" in node:
+            [recurse(child) for child in node["children"]]
+    recurse(root)
+
+
 def collectTreeAttrsV2(root, warn):
     """
     Collect all keys specified on `node["node_attrs"]` throughout the tree
@@ -82,6 +96,8 @@ def warn(msg):
 
     print("Validating that the JSON is internally consistent...")
 
+    ensure_no_duplicate_names(data["tree"], ValidateError)
+
     if "entropy" in data["meta"]["panels"] and "genome_annotations" not in data["meta"]:
         warn("The entropy panel has been specified but annotations don't exist.")
 

tests/test_validate_export.py

@@ -0,0 +1,30 @@
+import Bio.Phylo
+from io import StringIO
+from pathlib import Path
+import pytest
+import sys
+
+# we assume (and assert) that this script is running from the tests/ directory
+sys.path.append(str(Path(__file__).parent.parent.parent))
+
+from augur.export_v2 import convert_tree_to_json_structure
+from augur.validate import ValidateError
+from augur.validate_export import ensure_no_duplicate_names
+
+
+class TestValidateExport():
+    def test_export_without_duplicate_names(self):
+        # Create a tree with unique tip names.
+        tree = Bio.Phylo.read(StringIO("root(A, internal(B, C))"), "newick")
+        metadata = {"A": {}, "B": {}, "C": {}, "root": {}, "internal": {}}
+        root = convert_tree_to_json_structure(tree.root, metadata)
+        ensure_no_duplicate_names(root, ValidateError)
+
+    def test_export_with_duplicate_names(self):
+        # Create a tree with duplicate tip names.
+        tree = Bio.Phylo.read(StringIO("root(A, internal(B, B))"), "newick")
+        metadata = {"A": {}, "B": {}, "root": {}, "internal": {}}
+        root = convert_tree_to_json_structure(tree.root, metadata)
+
+        with pytest.raises(ValidateError):
+            ensure_no_duplicate_names(root, ValidateError)