Ignore collector diagnostics in metadata evidence

docs/module-internals.md

@@ -137,7 +137,9 @@ metadata extraction from verdict logic:
 - `extract_provenance_evidence` reads the supported metadata signals into
   `ProvenanceEvidence`.
 - `evidence_from_metadata_record` normalizes an externally collected nested
-  metadata record into the same evidence type without file access.
+  metadata record into the same evidence type without file access. Diagnostic
+  values under `error` and `kind` are excluded from evidence while nested raw
+  bytes remain available through encoded binary fields.
 - `identify_from_evidence` evaluates that evidence without reopening the source.
 - `identify` preserves the path-based API and adds the optional registered
   visible-mark and open invisible-watermark decoders after extraction.

docs/python-api.md

@@ -132,8 +132,10 @@ report = identify_from_evidence(evidence)
 
 The normalizer recursively preserves text and byte values. It also decodes
 strings prefixed with `hex:` and fields named `base64` or ending in
-`_base64`. Pass a C2PA manifest-store dictionary in `record["c2pa_store"]`, or
-through the explicit `c2pa_manifest_store` argument.
+`_base64`. Diagnostic values under `error` and `kind` are ignored because they
+describe the collector rather than the source file. Pass a C2PA manifest-store
+dictionary in `record["c2pa_store"]`, or through the explicit
+`c2pa_manifest_store` argument.
 
 `identify_from_evidence` does not reopen the source file. It evaluates metadata
 only; registered visible marks and pixel-backed invisible watermarks remain in

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "remove-ai-watermarks"
-version = "0.21.1"
+version = "0.21.2"
 description = "AI watermark remover: strip visible and invisible AI watermarks (Gemini / Nano Banana sparkle, SynthID) and provenance metadata (C2PA, EXIF) from images"
 readme = "README.md"
 requires-python = ">=3.10.1"

src/remove_ai_watermarks/__init__.py

@@ -25,7 +25,7 @@ _os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
 _warnings.filterwarnings("ignore", message=r".*ImageProcessorFast.*")
 
 
-__version__ = "0.21.1"
+__version__ = "0.21.2"
 
 __all__ = ["__version__", "remove_visible", "visible_provenance"]
 

src/remove_ai_watermarks/identify.py

@@ -169,6 +169,7 @@ def _external_metadata(value: Any) -> tuple[list[tuple[str, Any]], bytes]:
     """Index nested metadata and recover common encoded binary values in one pass."""
     pairs: list[tuple[str, Any]] = []
     parts: list[bytes] = []
+    diagnostic_keys = {"error", "kind"}
 
     def visit(item: Any) -> None:
         if isinstance(item, dict):
@@ -177,6 +178,8 @@ def _external_metadata(value: Any) -> tuple[list[tuple[str, Any]], bytes]:
                 key_text = str(key)
                 pairs.append((key_text, nested))
                 parts.append(key_text.encode("utf-8", "replace"))
+                if key_text.lower() in diagnostic_keys:
+                    continue
                 if isinstance(nested, str) and (key_text == "base64" or key_text.endswith("_base64")):
                     encoded = nested.split("...TRUNCATED", 1)[0]
                     with contextlib.suppress(ValueError, TypeError):

tests/test_identify.py

@@ -6,6 +6,7 @@ against the real committed C2PA / IPTC fixtures in data/fixtures/provenance/.
 
 from __future__ import annotations
 
+import base64
 import json
 import subprocess
 import sys
@@ -61,6 +62,48 @@ class TestProvenanceEvidence:
         assert report.is_ai_generated is True
         assert {signal.name for signal in report.signals} >= {"gen_params", "xai_signature"}
 
+    def test_external_scanner_diagnostics_do_not_create_c2pa_evidence(self, tmp_path: Path):
+        path = tmp_path / "plain.jpg"
+        record = {
+            "c2pa_store": {"error": "ManifestNotFound: no JUMBF data found"},
+            "jpeg": {
+                "segments": [
+                    {
+                        "marker": "APP11",
+                        "kind": "c2pa_or_jumbf",
+                        "base64": "AAA=",
+                    }
+                ]
+            },
+        }
+
+        report = identify_from_evidence(evidence_from_metadata_record(record, path=path))
+
+        assert report.is_ai_generated is None
+        assert report.signals == []
+        assert report.watermarks == []
+
+    def test_external_scanner_raw_bytes_still_create_c2pa_evidence(self, tmp_path: Path):
+        path = tmp_path / "signed.jpg"
+        manifest = b"jumb c2pa OpenAI trainedAlgorithmicMedia"
+        record = {
+            "jpeg": {
+                "segments": [
+                    {
+                        "marker": "APP11",
+                        "kind": "c2pa_or_jumbf",
+                        "base64": base64.b64encode(manifest).decode(),
+                    }
+                ]
+            }
+        }
+
+        report = identify_from_evidence(evidence_from_metadata_record(record, path=path))
+
+        assert report.is_ai_generated is True
+        assert report.platform == "OpenAI (ChatGPT / gpt-image / DALL-E / Sora)"
+        assert [signal.name for signal in report.signals] == ["c2pa"]
+
     @pytest.mark.parametrize(
         "filename",
         [

uv.lock

@@ -3187,7 +3187,7 @@ wheels = [
 
 [[package]]
 name = "remove-ai-watermarks"
-version = "0.21.1"
+version = "0.21.2"
 source = { editable = "." }
 dependencies = [
     { name = "c2pa-python" },