Code_Analyzer

Sleeping

App Files Files Community

Sayandip commited on Jun 7, 2025

Commit

8d61217

verified ·

1 Parent(s): 35c7d0d

Update app.py

Browse files

Files changed (1) hide show

app.py +136 -106

app.py CHANGED Viewed

@@ -12,6 +12,7 @@ from reportlab.lib.styles import getSampleStyleSheet
 from bs4 import BeautifulSoup
 import re
 import base64
 # Set Gemini API Key
 client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY"))
@@ -99,6 +100,76 @@ def export_conversation_to_pdf(conversation_history):
     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")
@@ -113,117 +184,76 @@ def main():
     if "chat_history" not in st.session_state:
         st.session_state.chat_history = []
-    uploaded_files = st.file_uploader(
-        "Upload files (.txt, .docx, .csv, .xlsx, .pptx, .pdf, .html, .htm, .tex, .jpg, .jpeg, .png, video formats):",
-        type=[
-            "txt", "docx", "csv", "xlsx", "pptx", "pdf", "html", "htm", "tex",
-            "jpg", "jpeg", "png",
-            "mp4", "webm", "mov", "avi", "mkv", "flv", "wmv", "mpeg"
-        ],
-        accept_multiple_files=True
     )
-    if uploaded_files:
-        all_text_content = ""
-        for uploaded_file in uploaded_files:
-            filetype = uploaded_file.name.split(".")[-1].lower()
-            try:
-                if filetype == "txt":
-                    all_text_content += uploaded_file.read().decode("utf-8") + "\n"
-                elif filetype == "docx":
-                    all_text_content += extract_text_from_docx(uploaded_file) + "\n"
-                elif filetype == "csv":
-                    all_text_content += extract_text_from_csv(uploaded_file) + "\n"
-                elif filetype == "xlsx":
-                    all_text_content += extract_text_from_xlsx(uploaded_file) + "\n"
-                elif filetype == "pptx":
-                    all_text_content += extract_text_from_pptx(uploaded_file) + "\n"
-                elif filetype == "pdf":
-                    all_text_content += extract_text_from_pdf(uploaded_file) + "\n"
-                elif filetype in ["html", "htm"]:
-                    all_text_content += extract_text_from_html(uploaded_file) + "\n"
-                elif filetype == "tex":
-                    all_text_content += extract_text_from_tex(uploaded_file) + "\n"
-                elif filetype in ["jpg", "jpeg", "png"]:
-                    st.session_state.image_data = process_image(uploaded_file)
-                    all_text_content += "Image uploaded and processed.\n"
-                    st.image(uploaded_file, caption=uploaded_file.name, use_container_width=True)
-                elif filetype in ["mp4", "webm", "mov", "avi", "mkv", "flv", "wmv", "mpeg"]:
-                    st.session_state.video_data = process_video(uploaded_file)
-                    all_text_content += f"Video file '{uploaded_file.name}' uploaded and processed.\n"
-                    st.video(uploaded_file)
-                else:
-                    st.error(f"Unsupported file format: {uploaded_file.name}")
-            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:
-            st.markdown(f"**You:** {user_q}")
-            st.markdown(f"**Gemini:**\n\n{gemini_a}")
-        if st.session_state.chat_active:
-            with st.form(key="chat_form", clear_on_submit=True):
-                user_input = st.text_input("Ask a question about the documents (type 'exit' to stop):")
-                submit = st.form_submit_button("Send")
-                if submit and user_input:
-                    if user_input.strip().lower() == "exit":
-                        st.session_state.chat_active = False
-                        st.success("Chat ended. Reload to start again.")
-                    else:
-                        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)

 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"))
     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 ---------
 def main():
     st.set_page_config(page_title="Gemini 2.0 Flash Chat Q&A", layout="wide")
     if "chat_history" not in st.session_state:
         st.session_state.chat_history = []
+    action = st.selectbox(
+        "Choose Action",
+        ["Ask a Question", "Capture Photo", "Capture Video"]
     )
+    if action == "Capture Photo":
+        capture_photo()
+    elif action == "Capture Video":
+        capture_video()
+    elif action == "Ask a Question":
+        if st.session_state.documents_text:
+            st.markdown("### \U0001F4AC Conversation")
+            for user_q, gemini_a in st.session_state.conversation:
+                st.markdown(f"**You:** {user_q}")
+                st.markdown(f"**Gemini:**\n\n{gemini_a}")
+            if st.session_state.chat_active:
+                with st.form(key="chat_form", clear_on_submit=True):
+                    user_input = st.text_input("Ask a question about the documents (type 'exit' to stop):")
+                    submit = st.form_submit_button("Send")
+                    if submit and user_input:
+                        if user_input.strip().lower() == "exit":
+                            st.session_state.chat_active = False
+                            st.success("Chat ended. Reload to start again.")
+                        else:
+                            with st.spinner("Let me think..."):
+                                content_blocks = []
+                                # System prompt for context
+                                content_blocks.append({
+                                    "text": (
+                                        "Your role is to analyze documents, images, and videos to help identify and classify driver states..."
+                                    )
+                                })
+                                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)