Update frontend UI app (#174)

Co-authored-by: dorbaker <[email protected]>

Author: jgbradley1

.devcontainer/devcontainer.json

@@ -12,15 +12,16 @@
 	],
 	"remoteUser": "vscode",
 	"remoteEnv": {
-        // We add the .venv to the beginning of the path env in the Dockerfile
-        // so that we use the proper python, however vscode rewrites/overwrites
-        // the PATH in the image and puts /usr/local/bin in front of our .venv
-        // path. This fixes that issue.
-        "PATH": "${containerEnv:PATH}",
-        // Add src folder to PYTHONPATH so that we can import modules that
-        // are in the source dir
-        "PYTHONPATH": "/graphrag-accelerator/backend/:$PATH"
-    },
+		// We add the .venv to the beginning of the path env in the Dockerfile
+		// so that we use the proper python, however vscode rewrites/overwrites
+		// the PATH in the image and puts /usr/local/bin in front of our .venv
+		// path. This fixes that issue.
+		"PATH": "${containerEnv:PATH}",
+		// Add src folder to PYTHONPATH so that we can import modules that
+		// are in the source dir
+		"PYTHONPATH": "/graphrag-accelerator/backend/:$PATH",
+		"AZURE_CLI_DISABLE_CONNECTION_VERIFICATION": "1"
+	},
 	"mounts": [
 		// NOTE: we reference both HOME and USERPROFILE environment variables to simultaneously support both Windows and Unix environments
 		//       in most default situations, only one variable will exist (Windows has USERPROFILE and unix has HOME) and a reference to the other variable will result in an empty string

backend/manage-indexing-jobs.py

@@ -16,6 +16,7 @@
     client,
     config,
 )
+
 from src.api.azure_clients import AzureStorageClientManager
 from src.api.common import sanitize_name
 from src.models import PipelineJob

poetry.lock renamed to backend/poetry.lock

@@ -9,7 +9,6 @@ authors = [
     "Gabriel Nieves <[email protected]>",
     "Douglas Orbaker <[email protected]>",
     "Chris Sanchez <[email protected]>",
-    "Katy Smith <[email protected]>",
     "Shane Solomon <[email protected]>",
 ]
 readme = "README.md"
@@ -33,7 +32,7 @@ pytest = ">=8.2.1"
 wikipedia = ">=1.4.0"
 
 [tool.poetry.group.backend.dependencies]
-adlfs = ">=2023.10.0"
+adlfs = ">=2024.7.0"
 applicationinsights = ">=0.11.10"
 attrs = ">=23.2.0"
 azure-core = ">=1.30.1"
@@ -66,12 +65,6 @@ tiktoken = ">=0.6.0"
 uvicorn = ">=0.23.2"
 urllib3 = ">=2.2.2"
 
-[tool.poetry.group.frontend.dependencies]
-python-dotenv = ">=0.19.1"
-requests = "*"
-streamlit = ">=0.88.0"
-streamlit-nested-layout = "==0.1.3"
-
 [tool.ruff]
 target-version = "py310"
 line-length = 88

pyproject.toml renamed to backend/pyproject.toml

@@ -127,7 +127,9 @@ async def setup_indexing_pipeline(
             )
         # if indexing job is in a failed state, delete the associated K8s job and pod to allow for a new job to be scheduled
         if PipelineJobState(existing_job.status) == PipelineJobState.FAILED:
-            _delete_k8s_job(f"indexing-job-{sanitized_index_name}", os.environ["AKS_NAMESPACE"])
+            _delete_k8s_job(
+                f"indexing-job-{sanitized_index_name}", os.environ["AKS_NAMESPACE"]
+            )
         # reset the pipeline job details
         existing_job._status = PipelineJobState.SCHEDULED
         existing_job._percent_complete = 0

backend/src/api/index.py

@@ -68,11 +68,15 @@ async def lifespan(app: FastAPI):
         ] = pod.spec.service_account_name
         # retrieve list of existing cronjobs
         batch_v1 = client.BatchV1Api()
-        namespace_cronjobs = batch_v1.list_namespaced_cron_job(namespace=os.environ["AKS_NAMESPACE"])
+        namespace_cronjobs = batch_v1.list_namespaced_cron_job(
+            namespace=os.environ["AKS_NAMESPACE"]
+        )
         cronjob_names = [cronjob.metadata.name for cronjob in namespace_cronjobs.items]
         # create cronjob if it does not exist
         if manifest["metadata"]["name"] not in cronjob_names:
-            batch_v1.create_namespaced_cron_job(namespace=os.environ["AKS_NAMESPACE"], body=manifest)
+            batch_v1.create_namespaced_cron_job(
+                namespace=os.environ["AKS_NAMESPACE"], body=manifest
+            )
     except Exception as e:
         print("Failed to create graphrag cronjob.")
         reporter = ReporterSingleton().get_instance()

backend/src/main.py

@@ -10,6 +10,7 @@
 )
 from azure.storage.blob import BlobServiceClient
 from azure.storage.blob.aio import BlobServiceClient as BlobServiceClientAsync
+
 from src.api.azure_clients import AzureStorageClientManager
 
 

backend/src/tests/api/test_azure_clients.py

@@ -11,11 +11,11 @@ ENV PIP_DISABLE_PIP_VERSION_CHECK=1
 ENV SETUPTOOLS_USE_DISTUTILS=stdlib
 ENV PYTHONPATH=/backend
 
-COPY poetry.lock pyproject.toml /
 COPY backend /backend
-RUN pip install poetry \
+RUN cd backend \
+    && pip install poetry \
     && poetry config virtualenvs.create false \
-    && poetry install --without frontend
+    && poetry install
 
 # download all nltk data that graphrag requires
 RUN python -m nltk.downloader punkt averaged_perceptron_tagger maxent_ne_chunker words wordnet

docker/Dockerfile-backend

@@ -7,11 +7,11 @@ ENV PIP_ROOT_USER_ACTION=ignore
 ENV PIP_DISABLE_PIP_VERSION_CHECK=1
 ENV SETUPTOOLS_USE_DISTUTILS=stdlib
 
-COPY poetry.lock pyproject.toml /
 COPY frontend /frontend
-RUN pip install poetry \
+RUN cd frontend \
+    && pip install poetry \
     && poetry config virtualenvs.create false \
-    && poetry install --without backend
+    && poetry install
 
 WORKDIR /frontend
 EXPOSE 8080

docker/Dockerfile-frontend

@@ -4,6 +4,7 @@
 import os
 
 import streamlit as st
+
 from src.components import tabs
 from src.components.index_pipeline import IndexPipeline
 from src.enums import EnvVars
@@ -16,29 +17,28 @@
 
 
 def graphrag_app(initialized: bool):
-    # main entry point for app interface
     st.title("Microsoft GraphRAG Copilot")
-    main_tab, prompt_gen_tab, prompt_edit_tab, index_tab, query_tab = st.tabs(
-        [
-            "**Intro**",
-            "**1. Prompt Generation**",
-            "**2. Prompt Configuration**",
-            "**3. Index**",
-            "**4. Query**",
-        ]
-    )
+    main_tab, prompt_gen_tab, prompt_edit_tab, index_tab, query_tab = st.tabs([
+        "**Intro**",
+        "**1. Prompt Generation**",
+        "**2. Prompt Configuration**",
+        "**3. Index**",
+        "**4. Query**",
+    ])
+    # display only the main tab if a connection to an existing APIM has not been initialized
     with main_tab:
         tabs.get_main_tab(initialized)
-
-    # if not initialized, only main tab is displayed
     if initialized:
-        # assign API request information
+        # setup API request information
         COLUMN_WIDTHS = [0.275, 0.45, 0.275]
-        api_url = st.session_state[EnvVars.DEPLOYMENT_URL.value]
+        apim_url = st.session_state[EnvVars.DEPLOYMENT_URL.value]
         apim_key = st.session_state[EnvVars.APIM_SUBSCRIPTION_KEY.value]
-        client = GraphragAPI(api_url, apim_key)
+        # perform health check to verify connectivity
+        client = GraphragAPI(apim_url, apim_key)
+        if not client.health_check_passed():
+            st.error("APIM Connection Error")
+            st.stop()
         indexPipe = IndexPipeline(client, COLUMN_WIDTHS)
-
         # display tabs
         with prompt_gen_tab:
             tabs.get_prompt_generation_tab(client, COLUMN_WIDTHS)
@@ -48,9 +48,7 @@ def graphrag_app(initialized: bool):
             tabs.get_index_tab(indexPipe)
         with query_tab:
             tabs.get_query_tab(client)
-
     deployer_email = os.getenv("DEPLOYER_EMAIL", "[email protected]")
-
     footer = f"""
         <div class="footer">
             <p> Responses may be inaccurate; please review all responses for accuracy. Learn more about Azure OpenAI code of conduct <a href="https://learn.microsoft.com/en-us/legal/cognitive-services/openai/code-of-conduct"> here</a>. </br> For feedback, email us at <a href="mailto:{deployer_email}">{deployer_email}</a>.</p>