Skip to content

Commit 770eb54

Browse files
Paul-CornellPaul-Cornell
authored
Streamlit in Snowflake app example (#613)
1 parent 6da8377 commit 770eb54

File tree

2 files changed

+205
-1
lines changed

2 files changed

+205
-1
lines changed

‎docs.json

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -265,7 +265,8 @@
265265
"examplecode/tools/langflow",
266266
"examplecode/tools/vectorshift",
267267
"examplecode/tools/mcp",
268-
"examplecode/tools/mcp-partition"
268+
"examplecode/tools/mcp-partition",
269+
"examplecode/tools/snowflake-streamlit"
269270
]
270271
}
271272
]

‎examplecode/tools/snowflake-streamlit.mdx

Lines changed: 203 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,203 @@
1+
---
2+
title: Streamlit in Snowflake
3+
---
4+
5+
[Streamlit](https://streamlit.io/) is an open-source Python framework for data scientists and AI/ML engineers to
6+
deliver dynamic data apps with only a few lines of code.
7+
8+
[Streamlit in Snowflake](https://www.snowflake.com/en/product/features/streamlit-in-snowflake/) enables data scientists
9+
and Python developers to combine Streamlit's component-rich, open-source Python library with the scale, performance and
10+
security of the Snowflake platform. Streamlit Python scripts can define user interface (UI) components such as
11+
filters, graphs, sliders, and more to interact with your data.
12+
13+
In this example, you use Snowsight in your Snowflake account to create a simple Streamlit app that uses
14+
[Snowflake Cortex Search for RAG](https://docs.snowflake.com/user-guide/snowflake-cortex/cortex-search/cortex-search-overview)
15+
to ask natural-language questions about an existing table in your Snowflake account. This table
16+
contains data that was generated by Unstructured. Answers are returned in natural-language,
17+
chatbot-style format.
18+
19+
## Prerequisites
20+
21+
- A table in Snowflake that contains data that was generated by Unstructured. The
22+
target Snowflake table must have a column named `EMBEDDINGS` that will contains vector embeddings for the text in the table's `TEXT` column.
23+
The following Streamlit example app assumes that the `EMBEDDINGS` column contains 1,024 vector embeddings and has a data type of `VECTOR(FLOAT, 1024)`.
24+
25+
To create this table, you can [create a custom Unstructured workflow](/ui/workflows#create-a-custom-workflow)
26+
that uses any supported [source connector](/ui/sources/overview) along with the
27+
[Snowflake destination connector](/ui/destinations/snowflake). Then
28+
[run](/ui/workflows#edit%2C-delete%2C-or-run-a-workflow) the workflow to
29+
generate the data and then insert that generated data into the target Snowflake table.
30+
31+
After the data is inserted into the target Snowflake table, you can run the following Snowflake SQL statement to
32+
generate the 1,024 vector embeddings for the text in the table's `TEXT` column and then insert those generated vector
33+
embeddings into the table's `EMBEDDINGS` column. The model specified here for generating the vector embeddings is the
34+
same one that is used by the Streamlit example app:
35+
36+
```sql
37+
UPDATE ELEMENTS
38+
SET EMBEDDINGS = SNOWFLAKE.CORTEX.EMBED_TEXT_1024(
39+
'snowflake-arctic-embed-l-v2.0',
40+
TEXT
41+
);
42+
```
43+
44+
To learn how to run Snowflake SQL statements, see for example
45+
[Querying data using worksheets](https://docs.snowflake.com/user-guide/ui-snowsight-query).
46+
47+
- You must have the appropriate privileges to create and use a Streamlit app in your Snowflake account. These
48+
privileges include ones for the target table's parent database and schema as well as the Snowflake warehouse that
49+
runs the Streamlit app. For details, see
50+
[Getting started with Streamlit in Snowflake](https://docs.snowflake.com/developer-guide/streamlit/getting-started).
51+
52+
## Create and run the example app
53+
54+
<Steps>
55+
<Step title="Create the Streamlit app">
56+
1. In Snowsight for your Snowflake account, on the sidebar, click **Projects > Streamlit**.
57+
2. Click **+ Streamlit App**.
58+
3. For **App title**, enter a name for your app, such as `Unstructured Demo Streamlit App`.
59+
4. For **App location**, chose the target database and schema to store the app in.
60+
5. For **App warehouse**, choose the warehouse that you want to use to run your app and execute its queries.
61+
6. Click **Create**.
62+
</Step>
63+
<Step title="Add code to the Streamlit app">
64+
In this step, you add Python code to the Streamlit app that you created in the previous step.
65+
66+
This step explains each part of the code as you add it. If you want to skip past these explanations, add the
67+
code in the [complete code example](#complete-code-example) all at once, and then skip ahead to
68+
the next step, "Run the Streamlit app."
69+
70+
1. Import Python dependencies that get the current connection to the Snowflake database and schema and get Streamlit functions and features.
71+
72+
```python
73+
from snowflake.snowpark.context import get_active_session
74+
import streamlit as st
75+
```
76+
77+
2. Get the current connection to the Snowflake database and schema.
78+
79+
```python
80+
session = get_active_session()
81+
```
82+
83+
3. Display the title of the app in the Streamlit UI, and get the user's search query from the Streamlit UI.
84+
85+
```python
86+
st.title("Snowflake Cortex Search for RAG with Data from Unstructured")
87+
88+
query = st.text_input("Enter your search query:")
89+
```
90+
91+
4. Get the user's search query and display a progress indicator in the UI.
92+
93+
```python
94+
if query:
95+
with st.spinner("Embedding and retrieving..."):
96+
```
97+
5. Use the user's search query to get the top result from the `ELEMENTS` table.
98+
The `ELEMENTS` table contains the data that was generated by Unstructured. The code uses the
99+
`SNOWFLAKE.CORTEX.EMBED_TEXT_1024` function to generate vector embeddings for the user's search query and the `VECTOR_COSINE_SIMILARITY`
100+
function to get the similarity between the vector embeddings for the user's search query and the vector embeddings for the `TEXT` column
101+
for each rown in the `ELEMENTS` table. The code then orders the results by similarity and limits the results to the row with the greatest similarity
102+
between the search query and the target text.
103+
104+
```python
105+
top_result_df = session.sql(f"""
106+
WITH query_embedding AS (
107+
SELECT SNOWFLAKE.CORTEX.EMBED_TEXT_1024(
108+
'snowflake-arctic-embed-l-v2.0', '{query}'
109+
) AS EMBED
110+
)
111+
SELECT
112+
e.TEXT,
113+
VECTOR_COSINE_SIMILARITY(e.EMBEDDINGS, q.EMBED) AS similarity
114+
FROM ELEMENTS e, query_embedding q
115+
ORDER BY similarity DESC
116+
LIMIT 1
117+
""").to_pandas()
118+
```
119+
6. Get the `TEXT` column from the top result and use it as context for the user's search query.
120+
121+
```python
122+
context = top_result_df["TEXT"][0]
123+
```
124+
125+
7. Use the user's search query and the context from the top result to get a response from Snowflake Cortex Search for RAG.
126+
The code uses the `SNOWFLAKE.CORTEX.COMPLETE` function to generate a response to the user's search query based on the context from the top result.
127+
128+
```python
129+
completion_df = session.sql(f"""
130+
SELECT SNOWFLAKE.CORTEX.COMPLETE(
131+
'snowflake-arctic',
132+
CONCAT('Context: ', $$ {context} $$, ' \\n\\nQuestion: {query}\\nAnswer:')
133+
) AS RESPONSE
134+
""").to_pandas()
135+
```
136+
137+
8. Display the generated response in the Streamlit UI.
138+
139+
```python
140+
st.write("Answer:")
141+
st.write(completion_df["RESPONSE"][0])
142+
```
143+
</Step>
144+
<Step title="Run the Streamlit app">
145+
1. In the upper right corner, click **Run**.
146+
2. For **Enter your search query**, enter some natural-language question about the `TEXT` column in the table.
147+
3. Press **Enter**.
148+
149+
Snowflake Cortex Search for RAG returns its answer to your question in natural-language, chatbot-style format.
150+
</Step>
151+
</Steps>
152+
153+
## Complete code example
154+
155+
The full code example for the Streamlit app is as follows:
156+
157+
```python
158+
from snowflake.snowpark.context import get_active_session
159+
import streamlit as st
160+
161+
session = get_active_session()
162+
163+
st.title("Snowflake Cortex Search for RAG with Data from Unstructured")
164+
165+
query = st.text_input("Enter your search query:")
166+
167+
if query:
168+
with st.spinner("Embedding and retrieving..."):
169+
170+
top_result_df = session.sql(f"""
171+
WITH query_embedding AS (
172+
SELECT SNOWFLAKE.CORTEX.EMBED_TEXT_1024(
173+
'snowflake-arctic-embed-l-v2.0', '{query}'
174+
) AS EMBED
175+
)
176+
SELECT
177+
e.TEXT,
178+
VECTOR_COSINE_SIMILARITY(e.EMBEDDINGS, q.EMBED) AS similarity
179+
FROM ELEMENTS e, query_embedding q
180+
ORDER BY similarity DESC
181+
LIMIT 1
182+
""").to_pandas()
183+
184+
context = top_result_df["TEXT"][0]
185+
186+
completion_df = session.sql(f"""
187+
SELECT SNOWFLAKE.CORTEX.COMPLETE(
188+
'snowflake-arctic',
189+
CONCAT('Context: ', $$ {context} $$, ' \\n\\nQuestion: {query}\\nAnswer:')
190+
) AS RESPONSE
191+
""").to_pandas()
192+
193+
st.write("Answer:")
194+
st.write(completion_df["RESPONSE"][0])
195+
```
196+
197+
## Additional resources
198+
199+
- [Streamlit in Snowflake documentation](https://docs.snowflake.com/developer-guide/streamlit/about-streamlit)
200+
- [Create and deploy Streamlit apps using Snowsight](https://docs.snowflake.com/developer-guide/streamlit/create-streamlit-ui)
201+
- [Snowflake Solutions Developer Center for Streamlit](https://www.snowflake.com/en/developers/solutions-center/?tags=technology%2Fstreamlit)
202+
- [Streamlit documentation](https://docs.streamlit.io/)
203+

0 commit comments

Comments
 (0)