Update app.py

app.py

@@ -10,132 +10,125 @@ from docx import Document
 from docx.shared import Pt
 from string import Template
 
-# ---------- Branding ----------
+# =========================
+# Branding & Palette
+# =========================
 APP_NAME = "ScholarLens"
 TAGLINE = "Query your literature, get page-level proof"
 
-# ---------- Palette (guarantees light text on dark, dark text on light) ----------
 PALETTE = {
     "navy": "#083D77",         # dark background
     "gold": "#F2B400",         # primary buttons / accents
     "ice": "#FBF8F9",          # off-white
     "maroon": "#8B1E1E",       # emphasis chips / separators
     "amber": "#F5C26B",        # secondary accent
-    "text_on_dark": "#EAF2FF", # light text for dark surfaces
-    "text_on_light": "#0B1220" # dark text for light (gold/amber/ice)
+    "text_on_dark": "#EAF2FF", # light text on dark
+    "text_on_light": "#0B1220" # dark text on light (gold/amber)
 }
 
 def build_custom_css():
-    """Safe CSS builder using string.Template so { } braces don't break Python."""
+    """Dark UI with light text and gold primary button — safe for Python via Template."""
     tmpl = Template(r"""
-:root {
-  --navy: $navy;
-  --gold: $gold;
-  --ice: $ice;
-  --maroon: $maroon;
-  --amber: $amber;
-  --text-dark: $text_dark;
-  --text-light: $text_light;
+:root{
+  --navy: $navy; --gold: $gold; --ice: $ice; --maroon: $maroon; --amber: $amber;
+  --text-dark: $text_dark; --text-light: $text_light;
+
+  /* Gradio tokens (force our palette) */
+  --body-background-fill: var(--navy);
+  --body-text-color: var(--text-light);
+  --block-background-fill: rgba(8,61,119,0.82);
+  --block-title-text-color: var(--text-light);
+  --border-color-primary: rgba(255,255,255,0.12);
+
+  --button-primary-background-fill: var(--gold);
+  --button-primary-text-color: var(--text-dark);
+  --button-primary-border-color: #c89200;
+
+  --button-secondary-background-fill: var(--amber);
+  --button-secondary-text-color: var(--text-dark);
+  --button-secondary-border-color: #caa157;
+
+  --link-text-color: var(--amber);
 }
 
-/* Global surfaces */
-body, .gradio-container {
-  background: var(--navy) !important;
-  color: var(--text-light) !important;
+/* Global surfaces & text */
+body, .gradio-container{
+  background: var(--body-background-fill) !important;
+  color: var(--body-text-color) !important;
 }
 
-/* Blocks and tabs */
-.gradio-container .block,
-.gradio-container .tabs,
-.gradio-container .tabs > .tabitem {
-  background: color-mix(in srgb, var(--navy) 80%, black 20%) !important;
+/* Blocks / tabs */
+.gradio-container .block, .gradio-container .tabs, .gradio-container .tabs>.tabitem{
+  background: var(--block-background-fill) !important;
   color: var(--text-light) !important;
   border-radius: 12px;
-  border: 1px solid color-mix(in srgb, var(--navy) 70%, white 10%);
+  border: 1px solid var(--border-color-primary);
 }
 
-/* Hero stripe styled with your palette */
-#hero {
-  background:
-    linear-gradient(90deg, var(--navy) 0%, var(--gold) 25%, var(--ice) 45%, var(--maroon) 65%, var(--amber) 85%, transparent 100%);
-  border: 1px solid color-mix(in srgb, var(--gold) 40%, white 20%);
-  border-radius: 14px;
-  padding: 14px 16px;
-  color: var(--text-light);
+/* Hero stripe using your palette */
+#hero{
+  background: linear-gradient(90deg, var(--navy) 0%, var(--gold) 25%, var(--ice) 45%, var(--maroon) 65%, var(--amber) 85%, transparent 100%);
+  border: 1px solid rgba(242,180,0,0.6);
+  border-radius: 14px; padding: 14px 16px; color: var(--text-light);
 }
 
-/* Links */
-a, .prose a { color: var(--amber) !important; text-decoration: none; }
-a:hover { text-decoration: underline; }
-
 /* KPI chips */
-.kpi {
-  text-align:center; padding:12px; border-radius:10px;
-  border:1px solid color-mix(in srgb, var(--ice) 35%, var(--navy) 65%);
-  background: color-mix(in srgb, var(--navy) 75%, black 25%);
-  color: var(--text-light);
-}
+.kpi{ text-align:center; padding:12px; border-radius:10px;
+  border:1px solid rgba(255,255,255,.14); background: rgba(8,61,119,0.65); color: var(--text-light);}
 
 /* Buttons */
-button, .gr-button { border-radius: 10px !important; font-weight: 600 !important; }
-.gr-button, button { color: var(--text-light) !important; }
-.gr-button-primary {
-  background: var(--gold) !important;
-  color: var(--text-dark) !important; /* dark text on light gold */
-  border: 1px solid color-mix(in srgb, var(--gold) 70%, black 10%) !important;
+.gradio-container .gr-button, .gradio-container button{ border-radius:10px !important; font-weight:600 !important; }
+.gradio-container .gr-button-primary, .gradio-container button.primary{
+  background: var(--button-primary-background-fill) !important;
+  color: var(--button-primary-text-color) !important;
+  border: 1px solid var(--button-primary-border-color) !important;
 }
-.gr-button-secondary {
-  background: color-mix(in srgb, var(--amber) 60%, var(--gold) 40%) !important;
-  color: var(--text-dark) !important;
+.gradio-container .gr-button-secondary, .gradio-container button.secondary{
+  background: var(--button-secondary-background-fill) !important;
+  color: var(--button-secondary-text-color) !important;
+  border: 1px solid var(--button-secondary-border-color) !important;
 }
 
 /* Inputs */
-input, textarea, .gr-textbox, .gr-text-area, .gr-slider, .gr-dropdown, .gr-file {
-  background: color-mix(in srgb, var(--navy) 65%, black 35%) !important;
+input, textarea, select, .gr-textbox, .gr-text-area, .gr-dropdown, .gr-file, .gr-slider{
+  background: rgba(8,61,119,0.55) !important;
   color: var(--text-light) !important;
-  border: 1px solid color-mix(in srgb, var(--ice) 25%, var(--navy) 75%) !important;
+  border: 1px solid rgba(255,255,255,0.18) !important;
   border-radius: 10px !important;
 }
-input::placeholder, textarea::placeholder {
-  color: color-mix(in srgb, var(--text-light) 60%, transparent) !important;
-}
+input::placeholder, textarea::placeholder{ color: rgba(234,242,255,0.65) !important; }
 
-/* Text blocks */
-label, .label, .prose h1, .prose h2, .prose h3, .prose p, .markdown-body {
-  color: var(--text-light) !important;
-}
+/* Links & text */
+a, .prose a{ color: var(--amber) !important; text-decoration:none; } a:hover{ text-decoration:underline; }
+label, .label, .prose h1, .prose h2, .prose h3, .prose p, .markdown-body{ color: var(--text-light) !important; }
 
-/* Dataframe / table */
-.dataframe, table, .table, .gr-dataframe * { color: var(--text-light) !important; background: transparent !important; }
-.dataframe th {
-  background: color-mix(in srgb, var(--navy) 70%, black 30%) !important;
-  border-bottom: 1px solid color-mix(in srgb, var(--ice) 20%, var(--navy) 80%) !important;
-}
-.dataframe td {
-  border-bottom: 1px solid color-mix(in srgb, var(--ice) 15%, var(--navy) 85%) !important;
+/* Dataframe */
+.dataframe, table, .table, .gr-dataframe *{ color: var(--text-light) !important; background: transparent !important; }
+.dataframe th{ background: rgba(8,61,119,0.72) !important; border-bottom: 1px solid rgba(255,255,255,0.18) !important; }
+.dataframe td{ border-bottom: 1px solid rgba(255,255,255,0.12) !important; }
+
+/* Accordion */
+.accordion, .gr-accordion{
+  background: rgba(8,61,119,0.65) !important; border: 1px solid rgba(255,255,255,0.14) !important; border-radius: 10px !important;
 }
 
-/* Accordions */
-.accordion, .gr-accordion {
-  background: color-mix(in srgb, var(--navy) 70%, black 30%) !important;
-  border: 1px solid color-mix(in srgb, var(--ice) 20%, var(--navy) 80%) !important;
-  border-radius: 10px !important;
+/* Tabs active underline color */
+.gradio-container .tabs .tab-nav button.selected{
+  box-shadow: inset 0 -3px 0 0 var(--gold) !important; color: var(--text-light) !important;
 }
 
-/* Footer */
-footer, .footer { opacity:.75; color: var(--text-light); }
+/* Optional: center content */
+.gradio-container{ max-width: 1120px; margin: 0 auto; }
 """)
     return tmpl.substitute(
-        navy=PALETTE["navy"],
-        gold=PALETTE["gold"],
-        ice=PALETTE["ice"],
-        maroon=PALETTE["maroon"],
-        amber=PALETTE["amber"],
-        text_dark=PALETTE["text_on_light"],
-        text_light=PALETTE["text_on_dark"],
+        navy=PALETTE["navy"], gold=PALETTE["gold"], ice=PALETTE["ice"],
+        maroon=PALETTE["maroon"], amber=PALETTE["amber"],
+        text_dark=PALETTE["text_on_light"], text_light=PALETTE["text_on_dark"]
     )
 
-# ---------- Config ----------
+# =========================
+# Engine config
+# =========================
 EMBED_MODEL_NAME = "intfloat/multilingual-e5-small"
 CHUNK_SIZE = 1200
 CHUNK_OVERLAP = 200
@@ -156,7 +149,9 @@ embedder = None
 faiss_index = None
 docstore: List[Dict[str, Any]] = []
 
-# ---------- PDF utils ----------
+# =========================
+# PDF utils
+# =========================
 def extract_text_from_pdf(pdf_path: str) -> List[Tuple[int, str]]:
     pages = []
     with fitz.open(pdf_path) as doc:
@@ -165,10 +160,7 @@ def extract_text_from_pdf(pdf_path: str) -> List[Tuple[int, str]]:
             if not txt.strip():
                 blocks = page.get_text("blocks")
                 if isinstance(blocks, list):
-                    txt = "\n".join(
-                        b[4] for b in blocks
-                        if isinstance(b, (list, tuple)) and len(b) > 4
-                    )
+                    txt = "\n".join(b[4] for b in blocks if isinstance(b, (list, tuple)) and len(b) > 4)
             pages.append((i, txt or ""))
     return pages
 
@@ -183,7 +175,9 @@ def chunk_text(text: str, chunk_size=CHUNK_SIZE, overlap=CHUNK_OVERLAP) -> List[
         start = max(end - overlap, start + 1)
     return out
 
-# ---------- Embeddings / FAISS ----------
+# =========================
+# Embeddings / FAISS
+# =========================
 def load_embedder():
     global embedder
     if embedder is None:
@@ -226,7 +220,9 @@ def load_index() -> bool:
         return True
     return False
 
-# ---------- Ingest ----------
+# =========================
+# Ingest
+# =========================
 def _collect_pdf_paths(upload_paths: List[str]) -> List[str]:
     """Accept PDFs and ZIPs of PDFs."""
     if not upload_paths:
@@ -259,10 +255,8 @@ def ingest_pdfs(paths: List[str]) -> Tuple[Any, List[Dict[str, Any]]]:
                     continue
                 for ci, ch in enumerate(chunk_text(ptxt)):
                     entries.append({
-                        "text": ch,
-                        "source": base,
-                        "page_start": pno,
-                        "page_end": pno,
+                        "text": ch, "source": base,
+                        "page_start": pno, "page_end": pno,
                         "chunk_id": f"{base}::p{pno}::c{ci}",
                     })
         except Exception as e:
@@ -274,7 +268,9 @@ def ingest_pdfs(paths: List[str]) -> Tuple[Any, List[Dict[str, Any]]]:
     index = build_faiss(embs)
     return index, entries
 
-# ---------- Retrieval with optional keyword filter ----------
+# =========================
+# Retrieval
+# =========================
 def retrieve(query: str, top_k=5, must_contain: str = ""):
     global faiss_index, docstore
     if faiss_index is None or not docstore:
@@ -304,7 +300,9 @@ def retrieve(query: str, top_k=5, must_contain: str = ""):
         hits.append(item)
     return hits
 
-# ---------- Groq LLM ----------
+# =========================
+# Groq LLM
+# =========================
 def groq_answer(query: str, contexts, model_name="llama-3.3-70b-versatile", temperature=0.2, max_tokens=1000):
     try:
         if not os.environ.get("GROQ_API_KEY"):
@@ -331,9 +329,7 @@ def groq_answer(query: str, contexts, model_name="llama-3.3-70b-versatile", temp
         )
 
         resp = client.chat.completions.create(
-            model=model_name,
-            temperature=float(temperature),
-            max_tokens=int(max_tokens),
+            model=model_name, temperature=float(temperature), max_tokens=int(max_tokens),
             messages=[{"role":"system","content":system_prompt},{"role":"user","content":user_prompt}],
         )
         return resp.choices[0].message.content.strip()
@@ -341,12 +337,13 @@ def groq_answer(query: str, contexts, model_name="llama-3.3-70b-versatile", temp
         import traceback
         return f"Groq API error: {e}\n```\n{traceback.format_exc()}\n```"
 
-# ---------- Export helpers ----------
+# =========================
+# Export helpers
+# =========================
 def export_answer_to_docx(question: str, answer_md: str, rows: List[List[str]]) -> str:
-    """Save Q&A with sources table to a .docx and return path (rows = [Source, Page, Score, Snippet])."""
     doc = Document()
-    styles = doc.styles
     try:
+        styles = doc.styles
         styles['Normal'].font.name = 'Calibri'
         styles['Normal'].font.size = Pt(11)
     except Exception:
@@ -362,10 +359,7 @@ def export_answer_to_docx(question: str, answer_md: str, rows: List[List[str]])
     doc.add_heading("References (Top Passages)", level=2)
     table = doc.add_table(rows=1, cols=4)
     hdr = table.rows[0].cells
-    hdr[0].text = "Source"
-    hdr[1].text = "Page"
-    hdr[2].text = "Score"
-    hdr[3].text = "Snippet"
+    hdr[0].text = "Source"; hdr[1].text = "Page"; hdr[2].text = "Score"; hdr[3].text = "Snippet"
     for r in rows:
         row = table.add_row().cells
         for i, val in enumerate(r):
@@ -375,7 +369,9 @@ def export_answer_to_docx(question: str, answer_md: str, rows: List[List[str]])
     doc.save(path)
     return path
 
-# ---------- UI helpers ----------
+# =========================
+# UI helpers
+# =========================
 def build_index_from_uploads(paths: List[str]) -> str:
     global faiss_index, docstore
     pdfs = _collect_pdf_paths(paths)
@@ -433,40 +429,34 @@ def do_export_docx(question, answer_md, sources_rows):
     if not answer_md or not sources_rows:
         return None
     try:
-        path = export_answer_to_docx(question, answer_md, sources_rows)
-        return path
+        return export_answer_to_docx(question, answer_md, sources_rows)
     except Exception:
         return None
 
-# ---------- Theme (simple; main styling via CSS for compatibility) ----------
+# =========================
+# UI
+# =========================
 theme = gr.themes.Soft(primary_hue="indigo", secondary_hue="blue", neutral_hue="slate")
 
-# ---------- Gradio UI ----------
-with gr.Blocks(
-    title=f"{APP_NAME} | RAG over PDFs",
-    theme=theme,
-    css=build_custom_css()
-) as demo:
-    # --- Header / Hero ---
+with gr.Blocks(title=f"{APP_NAME} | RAG over PDFs", theme=theme, css=build_custom_css()) as demo:
+    # Header / Hero
     with gr.Group(elem_id="hero"):
-        gr.Markdown(
-            f"""
+        gr.Markdown(f"""
 <div style="display:flex;align-items:center;gap:16px;">
   <div style="font-size:36px">📚🔎 <b>{APP_NAME}</b></div>
   <div style="opacity:.9;">{TAGLINE}</div>
 </div>
 <p style="opacity:.85;margin-top:6px;">
 Upload your papers, build an index, and ask research questions with verifiable, page-level citations.
-</p>
-""")
+</p>""")
 
-    # --- KPI row ---
+    # KPI Row
     with gr.Row():
         gr.Markdown("**Meaning-aware retrieval**<br><span class='kpi'>E5 + FAISS</span>", elem_classes=["kpi"])
         gr.Markdown("**Cited answers**<br><span class='kpi'>Page-level proof</span>", elem_classes=["kpi"])
         gr.Markdown("**Runs anywhere**<br><span class='kpi'>HF Spaces or Colab</span>", elem_classes=["kpi"])
 
-    # --- Key / Settings ---
+    # API Key
     with gr.Row():
         api_box = gr.Textbox(label="(Optional) Set GROQ_API_KEY", type="password", placeholder="sk_...")
         set_btn = gr.Button("Set Key")
@@ -474,7 +464,7 @@ Upload your papers, build an index, and ask research questions with verifiable,
     set_btn.click(set_api_key, inputs=[api_box], outputs=[set_out])
 
     with gr.Tabs():
-        # ---------------- Tab 1: Build / Load ----------------
+        # Tab 1: Build / Load
         with gr.Tab("1) Build or Load Index"):
             gr.Markdown("Upload PDFs or a ZIP of PDFs, then click **Build Index**.")
             file_u = gr.Files(label="Upload PDFs or ZIP", file_types=[".pdf", ".zip"], type="filepath")
@@ -496,7 +486,7 @@ Upload your papers, build an index, and ask research questions with verifiable,
             zpath = gr.File(label="Index bundle", interactive=False)
             download_btn.click(fn=download_index_zip, outputs=[zpath])
 
-        # ---------------- Tab 2: Ask ----------------
+        # Tab 2: Ask
         with gr.Tab("2) Ask Questions"):
             with gr.Row():
                 with gr.Column(scale=1):
@@ -515,7 +505,7 @@ Upload your papers, build an index, and ask research questions with verifiable,
                             ["List camera model, sensor type, resolution, and FPS across studies. Cite pages.", "camera, fps, resolution"],
                             ["Extract limitations and future work across the corpus, with page references.", ""],
                             ["Compare GTAW setups: current range, travel speed, torch standoff, sensors.", "GTAW, current, speed, torch"],
-                            ["Summarize the main results tables with metrics and page citations.", "table, accuracy, mAP, F1"]
+                            ["Summarize results tables with metrics and page citations.", "table, accuracy, mAP, F1"]
                         ],
                         inputs=[q, must],
                         label="Quick examples",
@@ -529,15 +519,13 @@ Upload your papers, build an index, and ask research questions with verifiable,
                         export_btn = gr.Button("Export Answer to DOCX", visible=False)
                         exported = gr.File(label="Download answer", visible=True)
 
-            # wire buttons
             ask_btn.click(fn=ask_rag, inputs=[q, topk, model_dd, temp, must], outputs=[ans, src, snippets_md, export_btn])
             export_btn.click(fn=do_export_docx, inputs=[q, ans, src], outputs=[exported])
             clear_btn.click(lambda: ("", [], "", gr.update(visible=False)), outputs=[ans, src, snippets_md, export_btn])
 
-        # ---------------- Tab 3: About ----------------
+        # Tab 3: About
         with gr.Tab("About"):
-            gr.Markdown(
-                """
+            gr.Markdown("""
 **ScholarLens** helps researchers move from reading to results with answers grounded in the papers you upload.
 
 - Meaning-aware retrieval (E5 + FAISS)
@@ -547,10 +535,9 @@ Upload your papers, build an index, and ask research questions with verifiable,
 - Powered by Groq models
 
 *Privacy note:* your files stay on this Space. Only the Groq call is external.
-                """
-            )
+""")
 
-# Broad compatibility for Spaces
+# Run
 demo.queue()
 if __name__ == "__main__":
     demo.launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 7860)))