The Digital Notary

1# File: c2pa_validator.py
2# Dependency: c2pa-python (official Coalition for Content Provenance
3#             and Authenticity bindings, contentauth/c2pa-python on GitHub)
4# Install: pip install c2pa-python
5 
6import hashlib
7import json
8import urllib.request
9import c2pa
10 
11 
12# Public trust list maintained by the Content Authenticity Initiative.
13# Anchors include Adobe, Microsoft, the BBC, Leica, Sony, Nikon, etc.
14TRUST_ANCHORS_URL = "https://contentcredentials.org/trust/anchors.pem"
15 
16 
17class RealityValidator:
18    """
19    The Digital Notary: a logic gate for verifying Content Credentials.
20    Simulates the 'reality handshake' performed by a browser, social
21    platform, or newsroom verification pipeline.
22    """
23 
24    def __init__(self):
25        # Load the 'Iron Alliance' trust anchors and bind them to a
26        # c2pa.Settings object that the Reader will consult during validation.
27        with urllib.request.urlopen(TRUST_ANCHORS_URL) as response:
28            anchors_pem = response.read().decode("utf-8")
29 
30        self.settings = c2pa.Settings.from_dict({
31            "verify": {"verify_cert_anchors": True},
32            "trust": {"trust_anchors": anchors_pem},
33        })
34 
35    def validate_asset(self, file_path: str) -> dict:
36        """
37        Input:  Path to an image or video file.
38        Output: Verification status (VERIFIED, WARNING, TAMPERED, UNVERIFIED).
39        """
40        # Step 1: Extract the 'Digital Nutrition Label' (Manifest).
41        try:
42            with c2pa.Context(self.settings) as context:
43                with c2pa.Reader(file_path, context=context) as reader:
44                    manifest_json = json.loads(reader.detailed_json())
45        except Exception:
46            return {"status": "UNVERIFIED",
47                    "reason": "No credentials found (Gray Zone)"}
48 
49        active_label = manifest_json.get("active_manifest")
50        manifests = manifest_json.get("manifests", {})
51        manifest = manifests.get(active_label, {})
52 
53        # Step 2: Verify the identity (who signed this?).
54        # The Reader has already checked the signing certificate against the
55        # trust anchors loaded above. A trust failure surfaces here as a
56        # 'validation_status' entry with a code starting with 'signingCredential'.
57        for status in manifest.get("validation_status", []):
58            if status.get("code", "").startswith("signingCredential"):
59                return {"status": "WARNING",
60                        "reason": "Identity unknown or self-signed"}
61 
62        # Step 3: Verify the integrity (has it been tampered with?).
63        # The c2pa library validates hard-binding hashes during Reader
64        # construction. Any mismatch is reported as 'assertion.dataHash.mismatch'.
65        for status in manifest.get("validation_status", []):
66            if status.get("code") == "assertion.dataHash.mismatch":
67                return {"status": "TAMPERED",
68                        "reason": "Hash mismatch (deepfake injection)"}
69 
70        # Step 4: The handshake is complete.
71        return {
72            "status": "VERIFIED",
73            "issuer": manifest.get("signature_info", {}).get("issuer"),
74            "claim_generator": manifest.get("claim_generator"),
75            "ingredients": manifest.get("ingredients", []),  # the chain of edits
76        }
77 
78    @staticmethod
79    def _calculate_sha256(file_path: str) -> str:
80        """Standard cryptographic hash of the binary file (illustrative;
81        the c2pa Reader performs hard-binding validation internally)."""
82        sha = hashlib.sha256()
83        with open(file_path, "rb") as f:
84            for block in iter(lambda: f.read(4096), b""):
85                sha.update(block)
86        return sha.hexdigest()