feat: add optional validation of input doc

Signed-off-by: Daniel Bluhm <[email protected]>

did_peer_4/__init__.py

@@ -5,6 +5,7 @@
 from hashlib import sha256
 
 from .doc_visitor import DocVisitor
+from .valid import validate_input_document
 
 # Regex patterns
 BASE58_ALPHABET = "123456789ABCDEFGHJKLMNPQRSTUVWXYZabcdefghijkmnopqrstuvwxyz"
@@ -56,8 +57,11 @@ def _hash_encoded_doc(encoded_doc: str) -> str:
 
 def encode(
     document: Dict[str, Any],
+    validate: bool = True,
 ) -> str:
     """Encode an input document into a did:peer:4."""
+    if validate:
+        document = dict(validate_input_document(document))
     encoded_doc = _encode_doc(document)
     hashed = _hash_encoded_doc(encoded_doc)
     return f"did:peer:4{hashed}:{encoded_doc}"
@@ -156,3 +160,14 @@ def resolve_short_from_doc(
             raise ValueError("Document does not match DID")
 
     return resolve_short(long)
+
+
+__all__ = [
+    "encode",
+    "encode_short",
+    "decode",
+    "resolve",
+    "resolve_short",
+    "resolve_short_from_doc",
+    "validate_input_document",
+]

did_peer_4/valid.py

@@ -0,0 +1,78 @@
+"""Validate input documents."""
+from typing import Any, Mapping
+
+
+def resources(document: Mapping[str, Any]):
+    """Yield all resources in a document, skipping references."""
+    keys = (
+        "verificationMethod",
+        "authentication",
+        "assertionMethod",
+        "keyAgreement",
+        "capabilityDelegation",
+        "capabilityInvocation",
+        "service",
+    )
+    for key in keys:
+        if key in document:
+            if not isinstance(document[key], list):
+                raise ValueError(f"{key} must be a list")
+
+            for index, resource in enumerate(document[key]):
+                if isinstance(resource, dict):
+                    yield key, index, resource
+
+
+def validate_input_document(document: Mapping[str, Any]) -> Mapping[str, Any]:
+    """Validate did:peer:4 input document.
+
+    This validation is deliberately superficial. It is intended to catch mistakes
+    in the input document that would cause the peer DID to be invalid. It is not
+    intended to validate the contents of the document, which is left to the caller
+    after resolution.
+
+    The following checks are performed:
+
+    - The document must be a Mapping.
+    - The document must not be empty.
+    - The document must not contain an id.
+    - If present, alsoKnownAs must be a list.
+    - verificationMethod, authentication, assertionMethod, keyAgreement,
+      capabilityDelegation, capabilityInvocation, and service must be lists, if
+      present.
+    - All resources (verification methods, embedded verification methods,
+      services) must have an id.
+    - All resource ids must be strings.
+    - All resource ids must be relative.
+    - All resources must have a type.
+    """
+    if not isinstance(document, Mapping):
+        raise ValueError("document must be a Mapping")
+
+    if not document:
+        raise ValueError("document must not be empty")
+
+    if "id" in document:
+        raise ValueError("id must not be present in input document")
+
+    if "alsoKnownAs" in document:
+        if not isinstance(document["alsoKnownAs"], list):
+            raise ValueError("alsoKnownAs must be a list")
+
+    for key, index, resource in resources(document):
+        if "id" not in resource:
+            raise ValueError(f"{key}[{index}]: resource must have an id")
+
+        ident = resource["id"]
+        if not isinstance(ident, str):
+            raise ValueError("r{key}[{index}]: esource id must be a string")
+
+        if not ident.startswith("#"):
+            raise ValueError(
+                "A{key}[{index}]: ll document resource ids must be relative"
+            )
+
+        if "type" not in resource:
+            raise ValueError("r{key}[{index}]: esource must have a type")
+
+    return document

pyproject.toml

@@ -22,4 +22,14 @@ dev = [
     "black>=23.7.0",
     "ruff>=0.0.285",
     "pre-commit>=3.3.3",
+    "pytest-cov>=4.1.0",
 ]
+
+[tool.coverage.report]
+exclude_lines = [
+    "pragma: no cover",
+    "@abstract"
+]
+precision = 2
+skip_covered = true
+show_missing = true

tests/__init__.py

@@ -0,0 +1,11 @@
+"""Tests for did:peer:4."""
+from pathlib import Path
+
+
+def examples():
+    """Load json from examples directory and return generator over examples."""
+    examples_dir = Path(__file__).parent / "examples"
+    yield from examples_dir.glob("*.json")
+
+
+EXAMPLES = list(examples())

tests/test_readme_examples.py

@@ -1,17 +1,9 @@
 import json
 import pytest
-from pathlib import Path
 
 from did_peer_4 import encode, long_to_short, resolve, resolve_short
 
-
-def examples():
-    """Load json from examples directory and return generator over examples."""
-    examples_dir = Path(__file__).parent / "examples"
-    yield from examples_dir.glob("*.json")
-
-
-EXAMPLES = list(examples())
+from . import EXAMPLES
 
 
 def print_example(