Update app.py

app.py

@@ -24,12 +24,12 @@ import urllib.parse
 import re
 from langchain_google_genai import ChatGoogleGenerativeAI
 from langchain_community.document_loaders import PyPDFLoader, UnstructuredCSVLoader, UnstructuredExcelLoader, Docx2txtLoader, UnstructuredPowerPointLoader
-from langchain.chains import StuffDocumentsChain
 from langchain.chains.llm import LLMChain
 from langchain.prompts import PromptTemplate
 from langchain.vectorstores import FAISS
 from langchain_google_genai import GoogleGenerativeAIEmbeddings
 from langchain.text_splitter import CharacterTextSplitter
+import PIL.Image
 
 app = FastAPI()
 app.mount("/static", StaticFiles(directory="static"), name="static")
@@ -177,11 +177,12 @@ def generate_plot(df, plot_path, plot_type):
     plt.close(fig)
     return plot_path
 
+
 @app.get("/", response_class=HTMLResponse)
 async def read_form(request: Request):
     return templates.TemplateResponse("upload.html", {"request": request})
 
-@app.post("/process", response_class=HTMLResponse)
+@app.post("/process/", response_class=HTMLResponse)
 async def process_file(request: Request, file: UploadFile = File(...)):
     global df, uploaded_file, document_analyzed, file_path, file_extension
     uploaded_file = file
@@ -213,6 +214,7 @@ async def process_file(request: Request, file: UploadFile = File(...)):
 @app.post("/result")
 async def result(request: Request,  
                  target: str = Form(...),
+                 question: str = Form(...),
                  algorithm: str = Form(...)):
     global df, api
     global plot1_path, plot2_path, plot3_path, plot4_path, plot5_path, plot6_path, plot7_path, plot8_path, plot9_path, plot10_path, plot11_path
@@ -221,6 +223,14 @@ async def result(request: Request,
 
     api = "AIzaSyD3drCF1KnfAfTNYGNIBJNS_nAry7kzlxg"
     excluded_words = ["name", "postal", "date", "phone", "address", "id"]
+
+    def generate_gemini_response(plot_path):
+        genai.configure(api_key=api)
+        model = genai.GenerativeModel('gemini-1.5-flash-latest')
+        img = Image.open(plot_path)
+        response = model.generate_content([question + " As a marketing consultant, I want to understand consumer insights based on the chart and the market context so I can use the key findings to formulate actionable insights.", img])
+        response.resolve()
+        return response.text
     
     if df[target].dtype in ['float64', 'int64']:
         unique_values = df[target].nunique()
@@ -296,21 +306,7 @@ async def result(request: Request,
         plt.savefig(plot3_path)
         plt.close(fig)
 
-        #response 3
-        def to_markdown(text):
-            text = text.replace('•', '  *')
-            return Markdown(textwrap.indent(text, '> ', predicate=lambda _: True))
-
-        genai.configure(api_key=api)
-
-        import PIL.Image
-
-        img = PIL.Image.open("static/multiclass_barplot.png")
-        model = genai.GenerativeModel('gemini-1.5-flash-latest')
-        #response = model.generate_content(img)
-        response = model.generate_content(["As a marketing consulant, I want to understand consumer insighst based on the chart and the market context so I can use the key findings to formulate actionable insights", img])
-        response.resolve()
-        response3 = format_text(response.text)
+        response3 = format_text(generate_gemini_response(plot3_path))
 
 
     if method == "Classification":
@@ -320,22 +316,7 @@ async def result(request: Request,
         pairplot_fig.savefig(plot6_path)  # Save the pairplot as a PNG file
         
 
-        # Google Gemini Integration
-        genai.configure(api_key=api)
-        img = PIL.Image.open(plot6_path)
-        model = genai.GenerativeModel('gemini-1.5-flash-latest')
-
-        # Generate response based on the pairplot
-        response = model.generate_content([
-            "You are a professional Data Analyst, write the complete conclusion and actionable insight based on the image. Explain it by points.",
-            img
-        ])
-        response.resolve()
-
-        # Assign the response to response6
-        response6 = format_text(response.text)
-
-        # Include response6 and plot6_path in the data dictionary to be passed to the template
+        response6 = format_text(generate_gemini_response(plot6_path))
         
 
     if method == "Classification":
@@ -357,7 +338,7 @@ async def result(request: Request,
         for i, var in enumerate(int_vars):
             top_categories = df[var].value_counts().nlargest(10).index
             filtered_df = df[df[var].notnull() & df[var].isin(top_categories)]
-            sns.histplot(data=df, x=var, hue=target, kde=False, ax=axs[i], stat="percent")
+            sns.histplot(data=df, x=var, hue=target, kde=True, ax=axs[i], stat="percent")
             axs[i].set_title(var)
 
             # Annotate the subplot with sample size
@@ -376,58 +357,20 @@ async def result(request: Request,
         plt.savefig(plot4_path)
         plt.close(fig)
 
-        #response 4
-        def to_markdown(text):
-            text = text.replace('•', '  *')
-            return Markdown(textwrap.indent(text, '> ', predicate=lambda _: True))
-
-        genai.configure(api_key=api)
-
-        import PIL.Image
-
-        img = PIL.Image.open("static/multiclass_histplot.png")
-        model = genai.GenerativeModel('gemini-1.5-flash-latest')
-        response4 = model.generate_content(img)
-        response4 = model.generate_content(["As a marketing consulant, I want to understand consumer insighst based on the chart and the market context so I can use the key findings to formulate actionable insights", img])
-        response4.resolve()
-        response4 = format_text(response4.text)
-
+        response4 = format_text(generate_gemini_response(plot4_path))
 
 
 
 
+    import PIL.Image
     # Generate Pairplot
     pairplot_fig = sns.pairplot(df)
     plot5_path = "static/pair2.png"  
     pairplot_fig.savefig(plot5_path)  # Save the pairplot as a PNG file
 
-    # Google Gemini Integration
-    genai.configure(api_key=api)
-    img = Image.open(plot5_path)
-    model = genai.GenerativeModel('gemini-1.5-flash-latest')
-
-    # Generate response based on the pairplot
-    response = model.generate_content([
-        "You are a professional Data Analyst, write the complete conclusion and actionable insight based on the image. Explain it by points.",
-        img
-    ])
-    response.resolve()
-
-    # Assign the response to response5
-    response5 = format_text(response.text)
+    response5 = format_text(generate_gemini_response(plot5_path))
 
-    def generate_gemini_response(plot_path):
-        
-       
-        genai.configure(api_key=api)
-        img = Image.open(plot_path)
-        model = genai.GenerativeModel('gemini-1.5-flash-latest')
-        response = model.generate_content([
-            " As a marketing consultant, I want to understand consumer insights based on the chart and the market context so I can use the key findings to formulate actionable insights", 
-            img
-        ])
-        response.resolve()
-        return response.text
+    
 
     plot1_path = generate_plot(df, 'static/plot1.png', 'countplot')
     plot2_path = generate_plot(df, 'static/plot2.png', 'histplot')
@@ -456,7 +399,7 @@ async def result(request: Request,
     plt.savefig(plot7_path)
     plt.close(fig)
 
-    img = Image.open(plot7_path)
+    img = PIL.Image.open(plot7_path)
     response7 = format_text((generate_gemini_response(plot7_path)))
 
 
@@ -526,7 +469,7 @@ async def result(request: Request,
             ax.set_title('Top 10 Feature Importance (Decision Tree Regressor)', fontsize=18)
             plot8_path = "static/dtree_regressor.png"
             plt.savefig(plot8_path)
-            img = Image.open(plot8_path)
+            img = PIL.Image.open(plot8_path)
             response8 = format_text((generate_gemini_response(plot8_path)))
             
 
@@ -561,7 +504,7 @@ async def result(request: Request,
             ax.set_title('Top 10 Feature Importance (Decision Tree Classifier)', fontsize=18)
             plot9_path = "static/dtree_classifier.png"
             plt.savefig(plot9_path)
-            img = Image.open(plot9_path)
+            img = PIL.Image.open(plot9_path)
             response9 = format_text((generate_gemini_response(plot9_path)))
 
         
@@ -601,7 +544,7 @@ async def result(request: Request,
             ax.set_title('Top 10 Feature Importance (Random Forest Regressor)', fontsize=18)
             plot10_path = "static/rf_regressor.png"
             plt.savefig(plot10_path)
-            img = Image.open(plot10_path)
+            img = PIL.Image.open(plot10_path)
             response10 = format_text((generate_gemini_response(plot10_path)))
 
         elif method == "Classification":
@@ -633,7 +576,7 @@ async def result(request: Request,
             ax.set_title('Top 10 Feature Importance (Random Forest Classifier)', fontsize=18)
             plot11_path = "static/rf_classifier.png"
             plt.savefig(plot11_path)
-            img = Image.open(plot11_path)
+            img = PIL.Image.open(plot11_path)
             response11 = format_text((generate_gemini_response(plot11_path)))
 
 
@@ -704,7 +647,7 @@ async def ask_question(request: Request, question: str = Form(...)):
         raise HTTPException(status_code=400, detail="No file has been uploaded yet.")
 
     # Initialize the LLM model
-    llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash-latest", google_api_key=api)
+    llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash", google_api_key=api)
 
     # Determine the file extension and select the appropriate loader
     file_path = ''
@@ -850,4 +793,6 @@ def save_to_json(question_responses):
 
 
 
-
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="127.0.0.1", port=8000)