Update app.py

app.py

@@ -1,221 +1,53 @@
-import os
 import gradio as gr
-from pypdf import PdfReader
 
-from langchain_text_splitters import RecursiveCharacterTextSplitter
-from langchain_community.vectorstores import FAISS
-from langchain_openai import OpenAIEmbeddings, ChatOpenAI
-from langchain_core.documents import Document
-
-APP_TITLE = "Election Archive"
-APP_TAGLINE = "What Happened Last Elections."
-
-
-CATEGORIES = ["PRESIDENTIAL VOTE", "POLITICANS", "POLITICAL PARTIES", "WHO VIED", "WOMEN IN POLITICS", "CAMPAIGN SLOGANS"]
-
-# ---------- Core logic ----------
-def read_pdf(file_path: str) -> str:
-    reader = PdfReader(file_path)
-    pages = []
-    for p in reader.pages:
-        pages.append(p.extract_text() or "")
-    return "\n".join(pages)
-
-def build_index(files):
-    if not files:
-        return None, "Upload at least one PDF."
-
-    docs = []
-    for f in files:
-        text = read_pdf(f.name)
-        source_name = os.path.basename(f.name)
-        docs.append(Document(page_content=text, metadata={"source": source_name}))
-
-    splitter = RecursiveCharacterTextSplitter(chunk_size=900, chunk_overlap=120)
-    chunks = splitter.split_documents(docs)
-
-    embeddings = OpenAIEmbeddings()
-    index = FAISS.from_documents(chunks, embeddings)
-    return index, f"✅ Memory built: {len(files)} file(s), {len(chunks)} chunks."
-
-def answer_question(index, question):
-    if index is None:
-        return "Upload documents and click **Build Memory** first.", ""
-
-    retriever = index.as_retriever(search_kwargs={"k": 4})
-    hits = retriever.get_relevant_documents(question)
-
-    context_blocks, citations = [], []
-    for i, d in enumerate(hits, start=1):
-        src = d.metadata.get("source", "unknown")
-        snippet = " ".join(d.page_content.split())[:420]
-        context_blocks.append(f"[Source {i}: {src}] {d.page_content[:900]}")
-        citations.append(f"**{i}. {src}** — {snippet}…")
-
-    llm = ChatOpenAI(model="gpt-4o-mini", temperature=0.2)
-
-    prompt = f"""
-You are an archive assistant for an election newsroom.
-Answer using ONLY the context below. If insufficient, say what is missing.
-Give a crisp, newsroom-ready response (bullets where useful).
-End with "Sources used: 1,2,3..." referencing source numbers.
-
-Question: {question}
-
-Context:
-{chr(10).join(context_blocks)}
-""".strip()
-
-    response = llm.invoke(prompt).content
-    return response, "\n\n".join(citations)
-
-# ---------- UI styling ----------
 CUSTOM_CSS = """
-:root{
-  --bg:#ffffff;
-  --text:#111827;
-  --muted:#6b7280;
-  --accent:#4f46e5;
-  --line:#e5e7eb;
-  --radius:16px;
-}
-
-.gradio-container{
-  background: var(--bg) !important;
-  color: var(--text) !important;
-  font-family: Inter, ui-sans-serif, system-ui, -apple-system, Segoe UI, Roboto, Arial !important;
+.gradio-container {
+    background: white !important;
+    font-family: Inter, system-ui, -apple-system, Segoe UI, Roboto !important;
 }
 
-h1{
-  font-size: 48px !important;
-  font-weight: 800 !important;
-  line-height: 1.1 !important;
-  letter-spacing: -1px;
+h1 {
+    font-size: 52px !important;
+    font-weight: 800 !important;
+    line-height: 1.1 !important;
+    letter-spacing: -1px;
 }
 
-h3{
-  font-size: 18px !important;
-  color: var(--muted) !important;
-  font-weight: 400 !important;
+h3 {
+    color: #6b7280 !important;
+    font-weight: 400 !important;
 }
 
-button{
-  border-radius: 12px !important;
-  padding: 12px 20px !important;
-  font-weight: 600 !important;
+button {
+    border-radius: 12px !important;
+    padding: 12px 24px !important;
+    font-weight: 600 !important;
 }
-
-.primary button{
-  background: var(--accent) !important;
-  color: white !important;
-  border: none !important;
-}
-
-.soft button{
-  background: white !important;
-  border: 1px solid var(--line) !important;
-  color: var(--text) !important;
-}
-
-.card{
-  border: 1px solid var(--line);
-  border-radius: var(--radius);
-  padding: 24px;
-  background: white;
-}
-"""
-ARCHIVE_SVG = """
-<div id="archiveArt">
-  <svg width="360" height="220" viewBox="0 0 360 220" fill="none" xmlns="http://www.w3.org/2000/svg" style="opacity:.95">
-    <path d="M85 78L170 45L300 90L215 125L85 78Z" stroke="rgba(233,238,246,.65)" stroke-width="3"/>
-    <path d="M85 78V170L215 215V125L85 78Z" stroke="rgba(233,238,246,.65)" stroke-width="3"/>
-    <path d="M215 125L300 90V180L215 215V125Z" stroke="rgba(233,238,246,.65)" stroke-width="3"/>
-    <path d="M120 90V150" stroke="rgba(233,238,246,.35)" stroke-width="3"/>
-    <path d="M140 98V160" stroke="rgba(233,238,246,.28)" stroke-width="3"/>
-    <path d="M160 105V168" stroke="rgba(233,238,246,.22)" stroke-width="3"/>
-    <rect x="122" y="126" width="48" height="22" rx="6" stroke="rgba(233,238,246,.55)" stroke-width="3"/>
-    <text x="26" y="205" fill="rgba(157,176,198,.95)" font-size="12" font-family="ui-sans-serif, system-ui">Archive box</text>
-  </svg>
-</div>
-"""
-
-HEADER_HTML = f"""
-<div id="topbar">
-  <div>
-    <div id="brand">{APP_TITLE}</div>
-    <div id="tagline">{APP_TAGLINE}</div>
-  </div>
-  <div class="pill">Shielding the Storyteller • Archive Mode</div>
-</div>
 """
 
-# ---------- App ----------
-with gr.Blocks(css=CUSTOM_CSS, title=APP_TITLE) as demo:
+with gr.Blocks(css=CUSTOM_CSS) as demo:
 
-    index_state = gr.State(None)
-    last_category = gr.State("")
+    with gr.Row():
+        with gr.Column(scale=6):
+            gr.Markdown("""
+# How To Keep  
+# Your Files  
+# Organized
 
-    with gr.Tabs():
-        with gr.Tab("Explore"):
-            with gr.Row(equal_height=True):
-                with gr.Column(scale=6):
-                    with gr.Group(elem_classes=["card"]):
-                        gr.HTML(ARCHIVE_SVG)
-                        gr.Markdown(
-                            "Type a question, or pick a category to start exploring.",
-                            elem_classes=["small"],
-                        )
+### Turn your newsroom archive into a searchable memory engine.
+### Upload documents. Ask questions. Get answers with sources.
+""")
 
-                with gr.Column(scale=8):
-                    with gr.Group(elem_classes=["card"]):
-                        query = gr.Textbox(
-                            label="",
-                            placeholder="Explore the archive… e.g. 'What shaped youth voter turnout in 2022?'",
-                        )
-                        with gr.Row():
-                            ask_btn = gr.Button("Search", elem_classes=["primary"])
-                            clear_btn = gr.Button("Clear", elem_classes=["soft"])
-
-                        answer = gr.Markdown(elem_id="results")
-                        sources = gr.Markdown(label="Sources (snippets)")
-
-                    with gr.Group(elem_classes=["card"]):
-                        gr.Markdown("**Or pick a category**")
-                        cat_buttons = []
-                        for c in CATEGORIES:
-                            b = gr.Button(c, elem_classes=["catBtn"])
-                            cat_buttons.append(b)
-
-            def on_category_click(cat):
-                # This just turns category into a query prompt (simple + effective)
-                return f"Give me the most useful archive context on {cat.lower()} during Kenyan elections. Provide 5 bullet themes."
-
-            for c, b in zip(CATEGORIES, cat_buttons):
-                b.click(fn=on_category_click, inputs=[gr.State(c)], outputs=[query])
-
-            def on_clear():
-                return "", "", ""
-
-            clear_btn.click(on_clear, outputs=[query, answer, sources])
-            ask_btn.click(answer_question, inputs=[index_state, query], outputs=[answer, sources])
-
-        with gr.Tab("Upload & Build Memory"):
             with gr.Row():
-                with gr.Column(scale=7):
-                    with gr.Group(elem_classes=["card"]):
-                        uploader = gr.File(
-                            label="Upload PDFs (archive pack)",
-                            file_count="multiple",
-                            file_types=[".pdf"],
-                        )
-                        build_btn = gr.Button("Build Memory", elem_classes=["primary"])
-                        status = gr.Markdown("Upload documents, then click **Build Memory**.", elem_classes=["small"])
-
-                with gr.Column(scale=7):
-                    with gr.Group(elem_classes=["card"]):
-                        gr.Markdown("**How to use**\n- Upload PDFs\n- Build Memory\n- Go to **Explore** and ask questions\n\nTip: Start with 10–30 key docs for fast demos.")
-                        gr.Markdown("**Example questions**\n- What issues dominated 2022 coverage?\n- Summarize history of petitions in a constituency\n- What patterns repeat across election cycles?")
-
-            build_btn.click(build_index, inputs=[uploader], outputs=[index_state, status])
+                gr.Button("Get Started", variant="primary")
+                gr.Button("Upload Archive")
+
+        with gr.Column(scale=6):
+            gr.HTML(
+                """
+                <img src="https://undraw.co/api/illustrations/organizing_files.svg"
+                     style="width:100%; max-width:500px; display:block; margin:auto;">
+                """
+            )
 
 demo.launch()