feat: Add finance advisor spanner demo

.github/actions/spelling/allow.txt

@@ -370,6 +370,7 @@ meme
 memes
 metadatas
 mgrs
+miranda
 mpn
 nbconvert
 nbfmt

.gitignore

@@ -186,4 +186,3 @@ testem.log
 .DS_Store
 Thumbs.db
 *.vscode*
-gemini/sample-apps/finance-advisor-spanner/graph_viz.html

gemini/sample-apps/finance-advisor-spanner/Dockerfile

@@ -7,4 +7,4 @@ RUN pip install --no-cache-dir -r requirements.txt
 
 COPY . .
 
-CMD ["streamlit", "run", "Home.py", "--server.enableCORS", "false", "--browser.serverAddress", "0.0.0.0", "--browser.gatherUsageStats", "false", "--server.port", "8080"]
+CMD ["streamlit", "run", "home.py", "--server.enableCORS", "false", "--browser.serverAddress", "0.0.0.0", "--browser.gatherUsageStats", "false", "--server.port", "8080"]

gemini/sample-apps/finance-advisor-spanner/README.md

@@ -16,7 +16,7 @@ Finally, before I recommend a fund, I also want to check the exposure to a parti
 
 With the power of Spanner's multimodel support, I can run complex workloads on a single database for relational, analytical, text and vector use cases with virtually unlimited scale, five nines of availability—including enterprise security and governance for mission critical workloads.
 
-This demo highlights [Spanner](https://cloud.google.com/spanner), integration with [Vertex AI LLMs](https://cloud.google.com/model-garden?hl=en) for both embeddings and text completion models. You will learn how Spanner can help with usecases where you run Full Text Search, Approximate Nearest neighbor search and vector similarity search.
+This demo highlights [Spanner](https://cloud.google.com/spanner), integration with [Vertex AI LLMs](https://cloud.google.com/model-garden?hl=en) for both embeddings and text completion models. You will learn how Spanner can help with use cases where you run Full Text Search, Approximate Nearest Neighbor search and vector similarity search.
 
 ## Tech Stack
 
@@ -71,7 +71,7 @@ The Finvest Spanner demo application was built using:
 
    Change the endpoint as per your project and the spanner instance location
 
-   ```bash
+   ```sql
    ALTER MODEL EmbeddingsModel SET OPTIONS (
       endpoint = '//aiplatform.googleapis.com/projects/'YOUR PROJECT ID HERE'/locations/'YOUR SPANNER INSTANCE LOCATION HERE'/publishers/google/models/text-embedding-003'
       )

gemini/sample-apps/finance-advisor-spanner/Schema-Operations.sql

@@ -3,7 +3,6 @@ ALTER MODEL EmbeddingsModel SET OPTIONS (
 endpoint = '//aiplatform.googleapis.com/projects/<project-name>/locations/<location>/publishers/google/models/text-embedding-003'
 )
 ;
-
 ALTER TABLE EU_MutualFunds ADD COLUMN  fund_name_Tokens TOKENLIST AS (TOKENIZE_FULLTEXT(fund_name)) HIDDEN;
 ALTER TABLE EU_MutualFunds ADD COLUMN  category_Tokens TOKENLIST AS (TOKENIZE_FULLTEXT(category)) HIDDEN;
 ALTER TABLE EU_MutualFunds ADD COLUMN  investment_strategy_Tokens TOKENLIST AS (TOKENIZE_FULLTEXT(investment_strategy)) HIDDEN;
@@ -13,15 +12,12 @@ ALTER TABLE EU_MutualFunds ADD COLUMN  morningstar_benchmark_Tokens TOKENLIST AS
 ALTER TABLE EU_MutualFunds ADD COLUMN  top5_regions_Tokens TOKENLIST AS (TOKENIZE_FULLTEXT(top5_regions)) HIDDEN;
 ALTER TABLE EU_MutualFunds ADD COLUMN  top5_holdings_Tokens TOKENLIST AS (TOKENIZE_FULLTEXT(top5_holdings)) HIDDEN;
 ALTER TABLE EU_MutualFunds ADD COLUMN  investment_managers_Substring_Tokens  TOKENLIST AS (TOKENIZE_SUBSTRING(investment_managers)) HIDDEN;
-
 ALTER TABLE
   EU_MutualFunds ADD COLUMN investment_managers_Substring_Tokens_NGRAM TOKENLIST AS ( TOKENIZE_SUBSTRING(investment_managers,
       ngram_size_min=>2,
       ngram_size_max=>3,
       relative_search_types=>["word_prefix",
       "word_suffix"])) HIDDEN;
-
-
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 1958 ;
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 2008;
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 2004;
@@ -66,8 +62,6 @@ UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_stra
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 2020;
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 2000;
 UPDATE EU_MutualFunds SET investment_strategy_Embedding_vector = investment_strategy_Embedding WHERE investment_strategy_Embedding is not NULL and  EXTRACT(YEAR from inception_date)  = 1992;
-
-
 CREATE SEARCH INDEX
   category_Tokens_IDX
 ON
@@ -109,8 +103,6 @@ CREATE SEARCH INDEX
 ON
   EU_MutualFunds(investment_managers_Substring_Tokens,
     investment_strategy_Tokens);
-
-
 CREATE SEARCH INDEX
   investment_managers_Substring_NgRAM_investment_Strategy_Tokens_Combo_IDX
 ON
@@ -124,15 +116,10 @@ WHERE
   investment_strategy_Embedding_vector IS NOT NULL OPTIONS ( tree_depth = 2,
     num_leaves = 40,
     distance_type = 'EUCLIDEAN' );
-
 CREATE SEARCH INDEX
   investment_managers_Substring_Tokens_with_vectors_NGRAM_IDX
 ON
   EU_MutualFunds(investment_managers_Substring_Tokens_NGRAM) STORING (investment_strategy_Embedding_vector);
-
-
-
-
 CREATE OR REPLACE PROPERTY GRAPH FundGraph NODE TABLES( Companies AS Company DEFAULT LABEL PROPERTIES ALL COLUMNS,
     EU_MutualFunds AS Fund DEFAULT LABEL PROPERTIES ALL COLUMNS EXCEPT (_Injected_SearchUid,
       _Injected_VectorIndex_InvestmentStrategyEmbeddingIndex_FP8,
@@ -147,6 +134,3 @@ CREATE OR REPLACE PROPERTY GRAPH FundGraph NODE TABLES( Companies AS Company DEF
     Company(CompanySeq) DESTINATION KEY(SectorSeq)
   REFERENCES
     Sector(SectorSeq) LABEL Belongs_To PROPERTIES ALL COLUMNS );
-
-
-

gemini/sample-apps/finance-advisor-spanner/database.py

@@ -1,6 +1,6 @@
 """This file is for database operations done by the application """
 
-# pylint: disable=line-too-long, invalid-name, use-dict-literal, use-list-literal
+# pylint: disable=line-too-long
 import os
 
 from dotenv import load_dotenv
@@ -23,42 +23,19 @@
 database = instance.database(database_id)
 
 
-def spanner_read_data(query: str) -> pd.DataFrame:
-    """This function helps read data from spanner"""
+def spanner_read_data(query: str, *vector_input: list) -> pd.DataFrame:
+    """This function helps read data from Spanner"""
     with database.snapshot() as snapshot:
-        results = snapshot.execute_sql(query)
-        rows = list()
-        for row in results:
-            rows.append(row)
+        if len(vector_input) != 0:
+            results = snapshot.execute_sql(
+                query,
+                params={"vector": vector_input[0]},
+            )
+        else:
+            results = snapshot.execute_sql(query)
+        rows = list(results)
         cols = [x.name for x in results.fields]
-        result_df = pd.DataFrame(rows, columns=cols)
-    return result_df
-
-
-def spanner_read_data_list(query: str) -> list:
-    """This function helps read data from spanner and returns data as a list"""
-    with database.snapshot() as snapshot:
-        results = snapshot.execute_sql(query)
-        rows = list()
-        for row in results:
-            rows.append(row)
-    return rows
-
-
-def spanner_read_data_withparameters(query: str, vector_input: list) -> pd.DataFrame:
-    """This function helps read data from spanner with input params"""
-    with database.snapshot() as snapshot:
-        results = snapshot.execute_sql(
-            query,
-            params={"vector": vector_input},
-        )
-        rows = list()
-        for row in results:
-            rows.append(row)
-        cols = [x.name for x in results.fields]
-        result_df = pd.DataFrame(rows, columns=cols)
-
-    return result_df
+        return pd.DataFrame(rows, columns=cols)
 
 
 def fts_query(query_params: list) -> dict:
@@ -120,7 +97,7 @@ def semantic_query_ann(query_params: list) -> dict:
         + query_params[0]
         + '" AS content) ) ;'
     )
-    vector_input = spanner_read_data_list(embedding_query)
+    vector_input = spanner_read_data(embedding_query).values.tolist()
 
     if query_params[1].strip() != "":
         ann_query = (
@@ -132,7 +109,10 @@ def semantic_query_ann(query_params: list) -> dict:
         )
     else:
         ann_query = "SELECT fund_name, investment_strategy, investment_managers, APPROX_EUCLIDEAN_DISTANCE(investment_strategy_Embedding_vector, @vector, options => JSON '{\"num_leaves_to_search\": 10}') AS distance FROM EU_MutualFunds @{force_index = InvestmentStrategyEmbeddingIndex} WHERE investment_strategy_Embedding_vector IS NOT NULL ORDER BY distance LIMIT 100;"
-    results_df = spanner_read_data_withparameters(ann_query, vector_input[0][0])
+    results_df = spanner_read_data(ann_query, vector_input[0][0])
+    results_df = spanner_read_data(ann_query, vector_input[0][0])
+    results_df = spanner_read_data(ann_query, vector_input[0][0])
+    results_df = spanner_read_data(ann_query, vector_input[0][0])
 
     return_vals = dict()
     return_vals["query"] = ann_query