Classify v2 fake, agent_data.create migration, downstream copier updates (#268)

.copier-answers.yml

@@ -1,3 +1,3 @@
 # Changes here will be overwritten by Copier; NEVER EDIT MANUALLY
-_commit: v0.7.2
+_commit: v0.7.3
 _src_path: https://github.com/run-llama/template-workflow-data-extraction

pyproject.toml

@@ -1,11 +1,11 @@
 [project]
-name = "extraction-review"
+name = "extract-reconcile-invoice"
 version = "0.1.0"
 description = "Extracts data"
 readme = "README.md"
 requires-python = ">=3.12"
 dependencies = [
-    "llama-cloud>=2.3.0,<3",
+    "llama-cloud>=2.4.1,<3",
     "json-schema-to-pydantic>=0.4.8",
     "llama-index-workflows>=2.16.0,<3.0.0",
     "python-dotenv>=1.1.0",
@@ -24,12 +24,17 @@ dev = [
     "pytest>=8.4.1",
     "hatch>=1.14.1",
     "pytest-asyncio>=1.3.0",
+    "pytest-timeout>=2.3.1",
+    "llama-cloud-fake>=0.1,<0.2",
 ]
 
 [build-system]
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
+[tool.hatch.build.targets.wheel]
+packages = ["src/extraction_review"]
+
 [tool.hatch.envs.default.scripts]
 "format" = "ruff format ."
 "format-check" = "ruff format --check ."
@@ -40,6 +45,10 @@ test = "pytest"
 "all-check" = ["format-check", "lint-check", "test"]
 "all-fix" = ["format", "lint", "test"]
 
+[tool.pytest.ini_options]
+timeout = 120
+timeout_method = "thread"
+
 [tool.llamadeploy]
 env_files = [".env"]
 llama_cloud = true

src/extraction_review/process_file.py

@@ -451,7 +451,7 @@ Provide your analysis in the specified format."""
                     f"Removed {delete_result.deleted_count} existing record(s) "
                     f"for file {extracted_data.file_name}"
                 )
-        item = await llama_cloud_client.beta.agent_data.agent_data(
+        item = await llama_cloud_client.beta.agent_data.create(
             data=data_dict,
             deployment_name=agent_name or "_public",
             collection=EXTRACTED_DATA_COLLECTION,

tests/conftest.py

@@ -0,0 +1,20 @@
+"""Pytest configuration: install the LlamaCloud fake server for all tests."""
+
+import logging
+import sys
+
+import pytest
+from llama_cloud_fake import FakeLlamaCloudServer
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
+    handlers=[logging.StreamHandler(sys.stdout)],
+)
+
+_fake = FakeLlamaCloudServer().install()
+
+
+@pytest.fixture
+def fake() -> FakeLlamaCloudServer:
+    return _fake

tests/test_placeholder.py

@@ -1,2 +0,0 @@
-def test_placeholder():
-    pass

tests/test_workflow.py

@@ -0,0 +1,25 @@
+import pytest
+from extraction_review.process_file import FileEvent
+from extraction_review.process_file import workflow as process_file_workflow
+from llama_cloud_fake import FakeLlamaCloudServer
+
+
+@pytest.mark.asyncio
+async def test_process_file_workflow(
+    monkeypatch: pytest.MonkeyPatch,
+    fake: FakeLlamaCloudServer,
+) -> None:
+    monkeypatch.setenv("LLAMA_CLOUD_API_KEY", "fake-api-key")
+    file_id = fake.files.preload(path="tests/files/test.pdf")
+    try:
+        result = await process_file_workflow.run(start_event=FileEvent(file_id=file_id))
+    except Exception:
+        result = None
+    assert result is not None
+    assert isinstance(result, str)
+    assert len(result) == 7
+
+
+# metadata_workflow here calls `pipelines.upsert` to ensure the contracts
+# index exists, which isn't covered by the fake in llama-cloud-fake<0.1.1.
+# Add a test for it once the floor is bumped.