Code_Analyzer

Sleeping

App Files Files Community

Sayandip commited on Jun 7, 2025

Commit

a254664

verified ·

1 Parent(s): 5c52538

Update app.py

Browse files

Files changed (1) hide show

app.py +26 -118

app.py CHANGED Viewed

@@ -12,7 +12,6 @@ from reportlab.lib.styles import getSampleStyleSheet
 from bs4 import BeautifulSoup
 import re
 import base64
-from streamlit.components.v1 import html
 # Set Gemini API Key
 client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY"))
@@ -100,87 +99,15 @@ def export_conversation_to_pdf(conversation_history):
     doc.build(elements)
     return pdf_path
-# --------- New Media Capture Features ---------
-def capture_photo():
-    # Allow the user to capture a photo
-    photo = st.camera_input("Capture Photo")
-    if photo is not None:
-        # Display the captured photo and process it
-        st.image(photo, caption="Captured Photo", use_container_width=True)
-        # Process the captured photo for Gemini analysis (image processing)
-        st.session_state.image_data = process_image(photo)
-        st.session_state.documents_text += "Captured Image added to context.\n"
-        # Notify the user and process it with Gemini
-        st.success("Photo Captured and Ready for Analysis!")
-def capture_video():
-    # HTML and JS for recording video using the browser
-    video_recorder_html = """
-    <html>
-        <body>
-            <h2>Click to Record a Video</h2>
-            <video id="video" width="320" height="240" autoplay muted></video>
-            <br/>
-            <button id="startStopBtn">Start Recording</button>
-            <br/><br/>
-            <a id="downloadLink" style="display:none" download="captured_video.mp4">
-                <button>Download Recorded Video</button>
-            </a>
-            <script>
-                const videoElement = document.getElementById("video");
-                const startStopBtn = document.getElementById("startStopBtn");
-                const downloadLink = document.getElementById("downloadLink");
-                let mediaRecorder;
-                let recordedChunks = [];
-                // Set up video streaming from the webcam
-                navigator.mediaDevices.getUserMedia({ video: true })
-                    .then(function (stream) {
-                        videoElement.srcObject = stream;
-                        mediaRecorder = new MediaRecorder(stream);
-                        mediaRecorder.ondataavailable = function (e) {
-                            recordedChunks.push(e.data);
-                        };
-                        mediaRecorder.onstop = function () {
-                            const blob = new Blob(recordedChunks, { type: "video/mp4" });
-                            const url = URL.createObjectURL(blob);
-                            downloadLink.href = url;
-                            downloadLink.style.display = 'block';
-                        };
-                    });
-                startStopBtn.onclick = function () {
-                    if (mediaRecorder.state === "inactive") {
-                        mediaRecorder.start();
-                        startStopBtn.innerText = "Stop Recording";
-                    } else {
-                        mediaRecorder.stop();
-                        startStopBtn.innerText = "Start Recording";
-                    }
-                };
-            </script>
-        </body>
-    </html>
-    """
-    html(video_recorder_html, height=400)
-# --------- Main App ---------
 # --------- Main App ---------
 def main():
     st.set_page_config(page_title="Gemini 2.0 Flash Chat Q&A", layout="wide")
     st.title("\U0001F4C4 Team 18 Comprehensive Prototype Using Gemini 2.0 Flash")
-    # Initialize session state variables
     if "conversation" not in st.session_state:
         st.session_state.conversation = []
     if "documents_text" not in st.session_state:
-        st.session_state.documents_text = ""  # Initialize as an empty string
     if "chat_active" not in st.session_state:
         st.session_state.chat_active = True
     if "chat_history" not in st.session_state:
@@ -196,7 +123,6 @@ def main():
         accept_multiple_files=True
     )
-    # Handle the files uploaded
     if uploaded_files:
         all_text_content = ""
         for uploaded_file in uploaded_files:
@@ -231,22 +157,8 @@ def main():
             except Exception as e:
                 st.error(f"Failed to extract/process {uploaded_file.name}: {e}")
-        st.session_state.documents_text += all_text_content  # Append the extracted text
-    # Media Capture Section
-    st.sidebar.header("Capture Media")
-    capture_option = st.sidebar.radio("Select Media Capture Option", ["None", "Capture Photo", "Capture Video"])
-    if capture_option == "Capture Photo":
-        st.session_state.image_data = None
-        st.session_state.video_data = None
-        capture_photo()
-    elif capture_option == "Capture Video":
-        st.session_state.image_data = None
-        st.session_state.video_data = None
-        capture_video()
-    # Display conversation and chat
     if st.session_state.documents_text:
         st.markdown("### \U0001F4AC Conversation")
         for user_q, gemini_a in st.session_state.conversation:
@@ -266,54 +178,52 @@ def main():
                         with st.spinner("Let me think..."):
                             content_blocks = []
                             content_blocks.append({
                                 "text": (
                                     "Your role is to analyze documents, images, and videos to help identify and classify driver states, "
                                     "including: drowsy, alert, yawning, microsleep, eyes/lips open or closed, and other distractions. "
                                     "The user may ask for timestamps of events in video, technical explanations from code/docs, "
                                     "or structured summaries. Prioritize your answers for this use case and act as a helpful assistant "
-                                    "within this domain."
                                 )
                             })
                             if st.session_state.conversation:
                                 history_text = "\n\n".join(
-                                    f"Q: {entry[0]}\nA: {entry[1]}" for entry in st.session_state.conversation
                                 )
                                 content_blocks.append({"text": f"Previous conversation:\n{history_text}"})
                             if st.session_state.documents_text.strip():
                                 content_blocks.append({"text": f"Context:\n{st.session_state.documents_text}"})
-                            # Send image data if available
-                            if "image_data" in st.session_state and st.session_state.image_data:
-                                content_blocks.append({"image": st.session_state.image_data})
-                            # Send video data if available
-                            if "video_data" in st.session_state and st.session_state.video_data:
-                                content_blocks.append({"video": st.session_state.video_data})
                             content_blocks.append({"text": f"Question: {user_input}"})
-                            try:
-                                # Ensure we are sending the data in the correct format
-                                response = client.models.generate_content(
-                                    model="gemini-2.0-flash",
-                                    contents=content_blocks,
-                                    config={"tools": [{"google_search": {}}]}
-                                )
-                                st.session_state.conversation.append((user_input, response.text))
-                                st.success("\U0001F4A1 Answer:")
-                                st.write(response.text)
-                                st.session_state.chat_history.append({
-                                    "question": user_input,
-                                    "answer": response.text
-                                })
-                            except Exception as e:
-                                st.error(f"Error with Gemini API request: {e}")
         if st.session_state.conversation:
             pdf_path = export_conversation_to_pdf(st.session_state.conversation)
@@ -326,6 +236,4 @@ def main():
                 )
 if __name__ == "__main__":
-    main()

 from bs4 import BeautifulSoup
 import re
 import base64
 # Set Gemini API Key
 client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY"))
     doc.build(elements)
     return pdf_path
 # --------- Main App ---------
 def main():
     st.set_page_config(page_title="Gemini 2.0 Flash Chat Q&A", layout="wide")
     st.title("\U0001F4C4 Team 18 Comprehensive Prototype Using Gemini 2.0 Flash")
     if "conversation" not in st.session_state:
         st.session_state.conversation = []
     if "documents_text" not in st.session_state:
+        st.session_state.documents_text = []
     if "chat_active" not in st.session_state:
         st.session_state.chat_active = True
     if "chat_history" not in st.session_state:
         accept_multiple_files=True
     )
     if uploaded_files:
         all_text_content = ""
         for uploaded_file in uploaded_files:
             except Exception as e:
                 st.error(f"Failed to extract/process {uploaded_file.name}: {e}")
+        st.session_state.documents_text = all_text_content
     if st.session_state.documents_text:
         st.markdown("### \U0001F4AC Conversation")
         for user_q, gemini_a in st.session_state.conversation:
                         with st.spinner("Let me think..."):
                             content_blocks = []
+                            # System prompt inserted first here:
                             content_blocks.append({
                                 "text": (
                                     "Your role is to analyze documents, images, and videos to help identify and classify driver states, "
                                     "including: drowsy, alert, yawning, microsleep, eyes/lips open or closed, and other distractions. "
                                     "The user may ask for timestamps of events in video, technical explanations from code/docs, "
                                     "or structured summaries. Prioritize your answers for this use case and act as a helpful assistant "
+                                    "within this domain. "
+                                    "Additionally, your classification may rely on the State Farm Distracted Driver Detection dataset, "
+                                    "which defines 10 distraction categories such as texting, phone use, reaching behind, adjusting the radio, "
+                                    "and safe driving. Use this labeling scheme when interpreting relevant images."
                                 )
                             })
                             if st.session_state.conversation:
                                 history_text = "\n\n".join(
+                                    f"Q: {entry[0]}\nA: {entry[1]}"
+                                    for entry in st.session_state.conversation
                                 )
                                 content_blocks.append({"text": f"Previous conversation:\n{history_text}"})
                             if st.session_state.documents_text.strip():
                                 content_blocks.append({"text": f"Context:\n{st.session_state.documents_text}"})
+                            if "image_data" in st.session_state:
+                                content_blocks.append(st.session_state.image_data)
+                            if "video_data" in st.session_state:
+                                content_blocks.append(st.session_state.video_data)
                             content_blocks.append({"text": f"Question: {user_input}"})
+                            response = client.models.generate_content(
+                                model="gemini-2.0-flash",
+                                contents=content_blocks,
+                                config={"tools": [{"google_search": {}}]}
+                            )
+                            st.session_state.conversation.append((user_input, response.text))
+                            st.success("\U0001F4A1 Answer:")
+                            st.write(response.text)
+                            st.session_state.chat_history.append({
+                                "question": user_input,
+                                "answer": response.text
+                            })
         if st.session_state.conversation:
             pdf_path = export_conversation_to_pdf(st.session_state.conversation)
                 )
 if __name__ == "__main__":
+    main()