Skip to content

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

Fix test failures: update hash test for MD5, handle unicode surrogates

chopratejascfd44b3f6ab1
diff --git a/headroom/utils.py b/headroom/utils.py
index 8840b4046..b9d94f604 100644
--- a/headroom/utils.py
+++ b/headroom/utils.py
@@ -23,7 +23,7 @@ def generate_request_id() -> str:
def compute_hash(data: str | bytes) -> str:
"""Compute SHA256 hash, returning hex string."""
if isinstance(data, str):
- data = data.encode("utf-8")
+ data = data.encode("utf-8", errors="surrogatepass")
return hashlib.sha256(data).hexdigest()
diff --git a/tests/test_compression_store.py b/tests/test_compression_store.py
index 6e77f650a..2e2f4b5a2 100644
--- a/tests/test_compression_store.py
+++ b/tests/test_compression_store.py
@@ -1249,10 +1249,10 @@ class TestHashCollisionDetection:
# Should not have collision warning
assert "Hash collision detected" not in caplog.text
- def test_hash_uses_sha256_truncated(self, store: CompressionStore):
- """Hash is SHA256 truncated to 24 characters."""
+ def test_hash_uses_md5_truncated(self, store: CompressionStore):
+ """Hash is MD5 truncated to 24 characters (fast, non-crypto)."""
content = "test content"
- expected_hash = hashlib.sha256(content.encode()).hexdigest()[:24]
+ expected_hash = hashlib.md5(content.encode()).hexdigest()[:24]
hash_key = store.store(original=content, compressed="[]")