Pin evidence JSON files to LF

.gitattributes

@@ -3,3 +3,4 @@ tests/data/gguf_gemma4_tokenizer_oracle.json text eol=lf
 tests/data/gguf_gemma4_qualification_inputs.tar.xz binary
 tests/data/gguf_tokenizer_artifact_evidence.json text eol=lf
 tests/data/gguf_tokenizer_artifact_inputs.tar.xz binary
+testdata/evidence/**/*.json text eol=lf

src/mobius/integrations/gguf/_draft_runtime_evidence_test.py

@@ -5,28 +5,125 @@
 
 import hashlib
 import json
+import os
 import re
+import shutil
+import subprocess
 from pathlib import Path
 
+_REPO_ROOT = Path(__file__).parents[4]
+_EVIDENCE_DIR = _REPO_ROOT / "testdata" / "evidence"
+
 
 def _evidence() -> dict:
-    path = (
-        Path(__file__).parents[4]
-        / "testdata"
-        / "evidence"
-        / "gguf_draft_runtime_evidence.json"
-    )
+    path = _EVIDENCE_DIR / "gguf_draft_runtime_evidence.json"
     return json.loads(path.read_text(encoding="utf-8"))
 
 
 def _independent_trace(record: dict) -> dict:
     metadata = record["independent_direct_ort_trace"]
-    path = Path(__file__).parents[4] / "testdata" / "evidence" / metadata["filename"]
+    path = _EVIDENCE_DIR / metadata["filename"]
     payload = path.read_bytes()
     assert hashlib.sha256(payload).hexdigest() == metadata["sha256"]
     return json.loads(payload)
 
 
+def _effective_git_attributes(
+    repository: Path, paths: list[str], *, global_attributes: Path
+) -> dict[str, dict[str, str]]:
+    environment = os.environ.copy()
+    environment["GIT_ATTR_NOSYSTEM"] = "1"
+    result = subprocess.run(
+        [
+            "git",
+            "-c",
+            f"core.attributesFile={global_attributes}",
+            "check-attr",
+            "-z",
+            "text",
+            "eol",
+            "--",
+            *paths,
+        ],
+        cwd=repository,
+        check=True,
+        capture_output=True,
+        env=environment,
+    )
+    fields = result.stdout.decode("utf-8").split("\0")
+    assert fields.pop() == ""
+    assert len(fields) % 3 == 0
+    attributes: dict[str, dict[str, str]] = {}
+    for path, attribute, value in zip(fields[0::3], fields[1::3], fields[2::3], strict=True):
+        attributes.setdefault(path, {})[attribute] = value
+    return attributes
+
+
+def test_evidence_json_files_are_lf_normalized_for_raw_hashes(tmp_path: Path) -> None:
+    isolated_repo = tmp_path / "repository"
+    isolated_repo.mkdir()
+    subprocess.run(["git", "init", "--quiet"], cwd=isolated_repo, check=True)
+    shutil.copyfile(_REPO_ROOT / ".gitattributes", isolated_repo / ".gitattributes")
+
+    global_attributes = tmp_path / "global-attributes"
+    global_attributes.touch()
+    evidence_json = sorted(
+        path.relative_to(_REPO_ROOT).as_posix() for path in _EVIDENCE_DIR.rglob("*.json")
+    )
+    assert any(path.startswith("testdata/evidence/causal-lm/") for path in evidence_json)
+    future_binary = "testdata/evidence/future-evidence.bin"
+    for relative_path in [*evidence_json, future_binary]:
+        path = isolated_repo / relative_path
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.touch()
+    attributes = _effective_git_attributes(
+        isolated_repo,
+        [*evidence_json, future_binary],
+        global_attributes=global_attributes,
+    )
+
+    for relative_path in evidence_json:
+        assert attributes[relative_path] == {"text": "set", "eol": "lf"}
+    assert attributes[future_binary] == {
+        "text": "unspecified",
+        "eol": "unspecified",
+    }
+
+    expected_trace_hashes = {
+        "gguf_dflash_independent_ort_trace.json": {
+            "crlf": "e0e70e909f33ed44aa87da39bab721ebbee0c3e730e2f89dbc509c6683701e42",
+            "semantic": "ea1f1414d5bc28c2b7776fa66f2a3b4bc72eeda2b38ecaa8d1ca3f8f8bfe8fcf",
+        },
+        "gguf_eagle3_independent_ort_trace.json": {
+            "crlf": "78aa94f24d5f1b9ae020de1ffa5a70c6a30f30ff45eb5310a540cdd2b3d8fab2",
+            "semantic": "4dbd07a42b89f2067deb1eb249ce8f06ccb973cd700724592f3892deaf78a5db",
+        },
+    }
+    traces = {
+        record["independent_direct_ort_trace"]["filename"]: record[
+            "independent_direct_ort_trace"
+        ]
+        for record in _evidence()["routes"]
+    }
+    assert traces.keys() == expected_trace_hashes.keys()
+    for filename, metadata in traces.items():
+        payload = (_EVIDENCE_DIR / filename).read_bytes()
+        assert b"\n" in payload
+        assert b"\r\n" not in payload
+        assert hashlib.sha256(payload).hexdigest() == metadata["sha256"]
+        semantic_payload = json.dumps(
+            json.loads(payload), sort_keys=True, separators=(",", ":")
+        ).encode()
+        assert (
+            hashlib.sha256(semantic_payload).hexdigest()
+            == expected_trace_hashes[filename]["semantic"]
+        )
+        crlf_payload = payload.replace(b"\n", b"\r\n")
+        crlf_sha256 = hashlib.sha256(crlf_payload).hexdigest()
+        assert crlf_sha256 == expected_trace_hashes[filename]["crlf"]
+        assert crlf_sha256 != metadata["sha256"]
+
+
 def test_draft_runtime_evidence_is_bounded_immutable_and_complete() -> None:
     evidence = _evidence()
     policy = evidence["policy"]

testdata/evidence/gguf_dflash_independent_ort_trace.json

@@ -1 +1,2 @@
+
 {"counters":{"rounds":5,"proposed":75,"accepted":18,"multi_token_rounds":3,"rejections":5},"final_draft_cache":{"length":28,"sha256":"f605ea72eeaad1ef074a36551390d3aa14b93159e03b15483e4358b719cc6456"},"final_target_cache":{"length":47,"sha256":"0107c259dfd716ad347b6d6f5ee6a5a590111bf627172aa96a91ac082edb4467"},"generated_tokens_sha256":"78c55f7b34e9a14f4c02850a631ed9f85e0a155f25901b5d99db36f1755b7a35","reorder":{"supported":false,"reason":"reference coordinator is batch-size 1"},"rounds":[{"proposal_ids":[9665,397,1067,366,3215,1339,1339,1004,366,3215,3215,3215,3215,3215,262],"proposal_tokens":[9665,397,1067,366,3215,1339,1339,1004,366,3215,3215,3215,3215,3215,262],"proposal_logits_sha256":"7d4394456101a75eac8208a09bd4d9394021c43b622c5bbac5775ff91198de09","posterior_tokens":[9665,397,970,366,3215,397,970,970,3974,271,4159,4159,29,3215,3215,737],"accepted_prefix":2,"correction_token":970,"target":{"before_length":16,"before_sha256":"5e7df359df34adde23fa09f2eb47d892025c83721563aebd080b4ad9a6b6aa90","tentative_length":32,"tentative_sha256":"ef2539592e67a1bca8d82e695c2e507dcded79b443bb1ae1c09e0914597a7997","committed_length":19,"committed_sha256":"569fa3bd3ee7f69974bc68dd7b8e80b848872933ed391813f4909a8652809c68","replay_length":19,"replay_sha256":"844f3a07602a104a1e2af6700fdd5c833378861326aa94dab3f48daf920f7404","replay_max_abs":0.38140740990638733,"replay_logits_sha256":"8c37b3902528b3c48c9ff007b97c6671d2e546eb84d48cf70d3054e8ab96947f","replay_tokens":[9665,397,970],"replay_tokens_match":true},"draft":{"before_length":0,"before_sha256":"2256da259b2ba5263517c15404741b28f4ad539e981103e75a09120017c4a8aa","tentative_length":32,"tentative_sha256":"89e21cd47e81c23f6b0b0e6773c54ed531520482d6914ad16a5ee3f8e8fe5fec","committed_length":16,"committed_sha256":"1ccaa4079c3ff0d1784e21bb1ed701a90addb4cb4123b1a7838108656c1ee144","replay_length":16,"replay_sha256":"1ccaa4079c3ff0d1784e21bb1ed701a90addb4cb4123b1a7838108656c1ee144","replay_max_abs":0.0}},{"proposal_ids":[4473,19688,1339,1004,4473,1548,1548,2890,11,526,3347,10303,12995,2890,2890],"proposal_tokens":[4473,19688,1339,1004,4473,1548,1548,2890,11,526,3347,10303,12995,2890,2890],"proposal_logits_sha256":"100711fefb9b6633de2581544dda1044d5930c2eefb5878e5ae85a92b7000818","posterior_tokens":[4473,1460,401,1004,3974,1460,2890,9,11,526,3347,11,11,526,11,11],"accepted_prefix":1,"correction_token":1460,"target":{"before_length":19,"before_sha256":"569fa3bd3ee7f69974bc68dd7b8e80b848872933ed391813f4909a8652809c68","tentative_length":35,"tentative_sha256":"49ce90b4b89c117215575f24c4e3e74e46f8927155d085b0fabdbdd438a56137","committed_length":21,"committed_sha256":"3a90d2e4e5bede1223c479d8c37bbcef20a712ca59abb3d9de44def548998e21","replay_length":21,"replay_sha256":"28e403a97d6ea2c5f6c7adb38ce65e6cd1df29edcb0dcab0e9b86345dbb8a045","replay_max_abs":0.42594051361083984,"replay_logits_sha256":"882877c702243681fb9b1ba65af4a2fd42d9415fd43ff968c847a39abfd1e3fb","replay_tokens":[4473,1460],"replay_tokens_match":true},"draft":{"before_length":16,"before_sha256":"1ccaa4079c3ff0d1784e21bb1ed701a90addb4cb4123b1a7838108656c1ee144","tentative_length":35,"tentative_sha256":"6a231365b75a07eef8a18f5d57b2bfce80b518d5312e1ef3d185a0c9da9eafc2","committed_length":19,"committed_sha256":"4547e247919a66d0f26b6a8b0c2ca4373f4fcfd3b9b496dccc8b1f521be3a353","replay_length":19,"replay_sha256":"4547e247919a66d0f26b6a8b0c2ca4373f4fcfd3b9b496dccc8b1f521be3a353","replay_max_abs":0.0}},{"proposal_ids":[1460,401,1004,14291,1548,2890,11,262,2890,11,11,11,11,11,11],"proposal_tokens":[1460,401,1004,14291,1548,2890,11,262,2890,11,11,11,11,11,11],"proposal_logits_sha256":"6ed1f7fa13163fd3dc6685910765ae8f93fbc92a5c2c1f1539dd68cd35551cf7","posterior_tokens":[401,401,1004,3974,1548,353,12995,526,526,11,526,526,526,526,526,1154],"accepted_prefix":0,"correction_token":401,"target":{"before_length":21,"before_sha256":"3a90d2e4e5bede1223c479d8c37bbcef20a712ca59abb3d9de44def548998e21","tentative_length":37,"tentative_sha256":"9a4f7926144625ddf6bd08b34aff64dc6dd96ef5be7ccdaa051615af4b2c6de9","committed_length":22,"committed_sha256":"8d0972cf012af1f15fe7abaa0924ce364426b068042242cb922b1c916fbbcd09","replay_length":22,"replay_sha256":"d79f0b1431b2983185338a2ccea149fe255c860872d7b8de5620408925fe4ba6","replay_max_abs":0.42594051361083984,"replay_logits_sha256":"bb1b1bedb240fa5ba5f0f5cc03d54af17754412dca571302b7a52e7c43d60265","replay_tokens":[401],"replay_tokens_match":true},"draft":{"before_length":19,"before_sha256":"4547e247919a66d0f26b6a8b0c2ca4373f4fcfd3b9b496dccc8b1f521be3a353","tentative_length":37,"tentative_sha256":"acccd907957d5f5099191761c075a7c1411153f34c1c1143126ad347c7401726","committed_length":21,"committed_sha256":"491310117ccb8b31bb0ec19813b5cbc3553d3a3754a23866e01de1002f1932a6","replay_length":21,"replay_sha256":"491310117ccb8b31bb0ec19813b5cbc3553d3a3754a23866e01de1002f1932a6","replay_max_abs":0.0}},{"proposal_ids":[1004,3974,10231,1548,2890,3347,526,11,11,11,2890,11,11,11,11],"proposal_tokens":[1004,3974,10231,1548,2890,3347,526,11,11,11,2890,11,11,11,11],"proposal_logits_sha256":"77f7266c190ca90cb49f5b505a5ecd5a055781c8cc8b25bca096a1b7e13a0149","posterior_tokens":[1004,3974,10231,1548,2890,12995,11,1550,526,526,526,11,526,526,526,526],"accepted_prefix":5,"correction_token":12995,"target":{"before_length":22,"before_sha256":"8d0972cf012af1f15fe7abaa0924ce364426b068042242cb922b1c916fbbcd09","tentative_length":38,"tentative_sha256":"55c77188f51c200ff87de3cda8996b4c34153ee2ffb0eb99e4977dbb1dcc6651","committed_length":28,"committed_sha256":"9308aee1b626c47d6205c13db448e40b2c6114630c99587b1fad5abede557f8c","replay_length":28,"replay_sha256":"dfed2bc06bb7c0c1df97d5ca6788230b749b1db18e3754a9300879db8698088a","replay_max_abs":0.42594051361083984,"replay_logits_sha256":"8390ad34c283728352a3bad080e0e41ec5b73a21d2918a5f8fa43937a6d6ae03","replay_tokens":[1004,3974,10231,1548,2890,12995],"replay_tokens_match":true},"draft":{"before_length":21,"before_sha256":"491310117ccb8b31bb0ec19813b5cbc3553d3a3754a23866e01de1002f1932a6","tentative_length":38,"tentative_sha256":"c7c82c1f8ddc5d25ef7d2551d69342d1ecb274204406f240e6975f2914ee1f5c","committed_length":22,"committed_sha256":"bee2317b6bb1ee062cadf209a54e46bc359ca868ce2335a046fd22ff363af48b","replay_length":22,"replay_sha256":"bee2317b6bb1ee062cadf209a54e46bc359ca868ce2335a046fd22ff363af48b","replay_max_abs":0.0}},{"proposal_ids":[526,3347,11,526,1550,8,341,262,421,320,366,1550,8,341,3347],"proposal_tokens":[526,3347,11,526,1550,8,341,262,421,320,366,1550,8,341,3347],"proposal_logits_sha256":"34779cc62958835e854c331118a0669f9cd63f4f0caf775e03082995406e4a36","posterior_tokens":[526,3347,11,526,1550,8,341,262,421,320,10303,220,8,341,286,284],"accepted_prefix":10,"correction_token":10303,"target":{"before_length":28,"before_sha256":"9308aee1b626c47d6205c13db448e40b2c6114630c99587b1fad5abede557f8c","tentative_length":44,"tentative_sha256":"6dd6d7d8a2363b689ac193c5a1d929fdb0a4188c3515652ef752da99b553b40a","committed_length":39,"committed_sha256":"15ea231d4104dfec54a7b4801c9975214d502cb41fe5e020339ce342244f3b5b","replay_length":39,"replay_sha256":"fe857aa5c9169817aaef93b4ad84b796f7330ea191d519ef1a89aff5d59ef295","replay_max_abs":1.9676249027252197,"replay_logits_sha256":"a5dde6b14cbcf60f319b9d60bdf514bdd18f6dee27380db0b1088c7da0d74b6b","replay_tokens":[526,3347,11,526,1550,8,341,262,421,320,10303],"replay_tokens_match":true},"draft":{"before_length":22,"before_sha256":"bee2317b6bb1ee062cadf209a54e46bc359ca868ce2335a046fd22ff363af48b","tentative_length":44,"tentative_sha256":"ea4068ba98caef1960a4fee2010087e3cd181f40f20183f51d58cdebf6618a29","committed_length":28,"committed_sha256":"f605ea72eeaad1ef074a36551390d3aa14b93159e03b15483e4358b719cc6456","replay_length":28,"replay_sha256":"f605ea72eeaad1ef074a36551390d3aa14b93159e03b15483e4358b719cc6456","replay_max_abs":0.0}}],"tokens_equal":true}

testdata/evidence/gguf_draft_runtime_evidence.json

@@ -142,7 +142,7 @@
       },
       "independent_direct_ort_trace": {
         "filename": "gguf_dflash_independent_ort_trace.json",
-        "sha256": "d81f0d1ec7294289679a04d637c56a0c46cba0153e6e0aa6a2e079d3fb2a9519",
+        "sha256": "1062e5ed665aeafbcf0d0fb250f9b48c520091ebc4b2ad6b19dbd341518c481f",
         "mutation_discriminators": {
           "draft_mapping": "semantic.draft_mapping",
           "proposal_order": "semantic.proposal_order",
@@ -251,7 +251,7 @@
       },
       "independent_direct_ort_trace": {
         "filename": "gguf_eagle3_independent_ort_trace.json",
-        "sha256": "2f45ddddd60bbc480914b0a95c58291c209b39524517baa3f872d32df52e56c6",
+        "sha256": "9d6c6fdb630bc0a040eacc2a2357a2991609ea0d904ff16574a9b55f7d010d55",
         "mutation_discriminators": {
           "draft_mapping": "semantic.draft_mapping",
           "proposal_order": "semantic.proposal_order",