Data preview, execution trace first, styled answer card, Excel upload

agent.py

@@ -121,23 +121,40 @@ def execute_python(code: str, working_dir: str, timeout: int = 30) -> dict:
             os.unlink(tmp_path)
 
 
-def inspect_csv(path: Path) -> dict[str, str]:
+def _read_tabular(path: Path, nrows: int = 200):
     import pandas as pd
 
-    df = pd.read_csv(path, nrows=200)
+    suffix = path.suffix.lower()
+    if suffix in (".xlsx", ".xls"):
+        return pd.read_excel(path, nrows=nrows)
+    return pd.read_csv(path, nrows=nrows)
+
+
+def inspect_data(path: Path) -> dict[str, str]:
+    df = _read_tabular(path, nrows=200)
     schema = "\n".join(f"  {c}: {df[c].dtype}" for c in df.columns)
     sample = df.head(5).to_string(index=False)
+    kind = "excel" if path.suffix.lower() in (".xlsx", ".xls") else "csv"
     return {
-        "type": "csv",
+        "type": kind,
         "schema": schema,
         "sample": sample,
         "row_counts": f"preview_rows={len(df)} (file may be larger)",
     }
 
 
+def inspect_csv(path: Path) -> dict[str, str]:
+    return inspect_data(path)
+
+
 def build_user_message(data_path: Path, task: str) -> str:
-    info = inspect_csv(data_path)
+    info = inspect_data(data_path)
     filename = data_path.name
+    read_hint = (
+        f"pd.read_excel('{filename}')"
+        if info["type"] == "excel"
+        else f"pd.read_csv('{filename}')"
+    )
     lines = [
         f"Data source: {filename}",
         f"Working directory contains: {filename}",
@@ -153,7 +170,7 @@ def build_user_message(data_path: Path, task: str) -> str:
         "",
         f"Task: {task}",
         "",
-        f"Read the file with pandas: pd.read_csv('{filename}')",
+        f"Read the file with pandas: {read_hint}",
     ]
     return "\n".join(lines)
 

app.py

@@ -1,15 +1,17 @@
 """
 DataSense E2B — Hugging Face Space demo
-Execution-grounded data agent (SFT v1) with bundled or uploaded CSVs.
+Execution-grounded data agent (SFT v1) with bundled or uploaded CSVs/Excel.
 """
 
 from __future__ import annotations
 
+import html
 import spaces  # must be first — before any torch/CUDA import
 
 from pathlib import Path
 
 import gradio as gr
+import pandas as pd
 
 from agent import run_agent
 from config import ADAPTER_MODEL, AGENT_MAX_STEPS, DATA_DIR
@@ -32,7 +34,7 @@ CUSTOM_CSS = """
   --ds-accent-dim: #1f6f5c;
   --ds-text: #e8eef5;
   --ds-muted: #8b9cb3;
-  max-width: 1180px !important;
+  max-width: 1280px !important;
   font-family: 'Newsreader', Georgia, serif !important;
 }
 #ds-header {
@@ -61,18 +63,68 @@ CUSTOM_CSS = """
   padding: 0.2rem 0.65rem;
   margin-bottom: 0.75rem;
 }
-#ds-panel {
+#ds-panel, #ds-results {
   background: var(--ds-surface);
   border: 1px solid var(--ds-border);
   border-radius: 14px;
   padding: 1.1rem;
+  min-height: 520px;
 }
+#ds-preview-box {
+  margin-top: 0.5rem;
+  border: 1px solid var(--ds-border);
+  border-radius: 10px;
+  overflow: hidden;
+}
+#ds-preview-box .label-wrap { padding: 0.5rem 0.75rem !important; }
 #run-btn {
   background: linear-gradient(90deg, #1f6f5c, #3ecfae) !important;
   border: none !important;
   font-weight: 600 !important;
   letter-spacing: 0.02em;
 }
+#ds-results .tabs { margin-top: 0 !important; }
+#ds-results .tabitem { padding-top: 0.75rem !important; }
+.ds-answer-card {
+  background: linear-gradient(145deg, #122a2a 0%, #151c26 100%);
+  border: 1px solid #2a3544;
+  border-left: 4px solid #3ecfae;
+  border-radius: 12px;
+  padding: 1.5rem 1.75rem;
+  margin: 0;
+  min-height: 200px;
+}
+.ds-answer-label {
+  font-family: 'IBM Plex Mono', monospace;
+  font-size: 0.72rem;
+  letter-spacing: 0.1em;
+  text-transform: uppercase;
+  color: #3ecfae;
+  margin-bottom: 0.75rem;
+}
+.ds-answer-value {
+  font-family: 'Newsreader', Georgia, serif;
+  font-size: 2rem;
+  font-weight: 600;
+  color: #e8eef5;
+  line-height: 1.3;
+  word-break: break-word;
+}
+.ds-summary {
+  margin-top: 1.25rem;
+  padding-top: 1rem;
+  border-top: 1px solid #2a3544;
+  font-size: 1.05rem;
+  color: #8b9cb3;
+  line-height: 1.55;
+}
+.ds-answer-empty {
+  color: #8b9cb3;
+  font-style: italic;
+  padding: 1rem 0;
+}
+.ds-trace-wrap { margin: 0; padding: 0; }
+.ds-trace-wrap h3 { margin-top: 1rem; color: #3ecfae; font-size: 1rem; }
 footer { visibility: hidden; }
 """
 
@@ -112,7 +164,7 @@ def _load_model():
 
 
 def _resolve_data_path(data_mode: str, dataset_name: str, upload_file) -> Path | None:
-    if data_mode == "Upload your CSV":
+    if data_mode == "Upload your file":
         if upload_file is None:
             return None
         path_str = upload_file[0] if isinstance(upload_file, list) else upload_file
@@ -123,8 +175,41 @@ def _resolve_data_path(data_mode: str, dataset_name: str, upload_file) -> Path |
     return DEMO_DATASETS.get(dataset_name)
 
 
+def _load_preview(data_mode: str, dataset_name: str, upload_file):
+    path = _resolve_data_path(data_mode, dataset_name, upload_file)
+    if path is None:
+        return pd.DataFrame(), "_Select or upload a file to preview._"
+    try:
+        suffix = path.suffix.lower()
+        if suffix in (".xlsx", ".xls"):
+            df = pd.read_excel(path, nrows=100)
+        else:
+            df = pd.read_csv(path, nrows=100)
+        meta = f"**{path.name}** · {len(df)} rows · {len(df.columns)} columns"
+        return df, meta
+    except Exception as exc:
+        return pd.DataFrame(), f"_Could not preview file: {exc}_"
+
+
+def _format_answer_html(answer: str, summary: str = "") -> str:
+    if not answer:
+        return '<div class="ds-answer-empty">Could not parse an answer — check the execution trace tab.</div>'
+    safe_answer = html.escape(answer)
+    summary_block = ""
+    if summary:
+        safe_summary = html.escape(summary)
+        summary_block = f'<p class="ds-summary">{safe_summary}</p>'
+    return (
+        f'<div class="ds-answer-card">'
+        f'<div class="ds-answer-label">Verified answer</div>'
+        f'<div class="ds-answer-value">{safe_answer}</div>'
+        f"{summary_block}"
+        f"</div>"
+    )
+
+
 def _toggle_data_inputs(data_mode: str):
-    is_upload = data_mode == "Upload your CSV"
+    is_upload = data_mode == "Upload your file"
     return (
         gr.update(visible=not is_upload),
         gr.update(visible=is_upload),
@@ -141,13 +226,13 @@ def run_task(
     progress=gr.Progress(),
 ):
     if not task.strip():
-        return "⚠️ Enter a task question.", ""
+        return "", "⚠️ Enter a task question."
 
     progress(0.05, desc="Resolving dataset…")
     data_path = _resolve_data_path(data_mode, dataset_name, upload_file)
     if data_path is None:
-        msg = "⚠️ Upload a `.csv` file first." if data_mode == "Upload your CSV" else f"⚠️ Dataset not found: {dataset_name}"
-        return msg, ""
+        msg = "⚠️ Upload a `.csv` or `.xlsx` file first." if data_mode == "Upload your file" else f"⚠️ Dataset not found: {dataset_name}"
+        return "", msg
 
     try:
         progress(0.15, desc="Loading Gemma-4 + SFT LoRA…")
@@ -162,13 +247,11 @@ def run_task(
             progress=progress,
         )
     except Exception as exc:
-        return f"**Error:** {exc}", ""
+        return "", f"**Error:** {exc}"
 
-    answer_block = f"## {result['answer']}" if result["answer"] else "_Could not parse an answer — check the execution trace._"
-    if result.get("summary"):
-        answer_block += f"\n\n{result['summary']}"
-
-    return answer_block, result["steps_markdown"]
+    answer_html = _format_answer_html(result.get("answer", ""), result.get("summary", ""))
+    trace = f'<div class="ds-trace-wrap">\n\n{result["steps_markdown"]}\n\n</div>'
+    return trace, answer_html
 
 
 @spaces.GPU(duration=300)
@@ -178,6 +261,7 @@ def preload_model():
 
 def build_ui() -> gr.Blocks:
     dataset_choices = list(DEMO_DATASETS.keys())
+    default_df, default_meta = _load_preview("Bundled examples", dataset_choices[0], None)
 
     with gr.Blocks(title="DataSense E2B") as demo:
         with gr.Column(elem_id="ds-header"):
@@ -185,7 +269,7 @@ def build_ui() -> gr.Blocks:
             gr.Markdown(
                 """
 # DataSense E2B
-**Live inference** — Gemma-4 2B + SFT v1 writes Python, runs it on your CSV, reads real stdout/errors.
+**Live inference** — Gemma-4 2B + SFT v1 writes Python, runs it on your data, reads real stdout/errors.
                 """
             )
             gr.Markdown(
@@ -195,10 +279,10 @@ def build_ui() -> gr.Blocks:
             )
 
         with gr.Row(equal_height=False):
-            with gr.Column(scale=4, elem_id="ds-panel"):
+            with gr.Column(scale=5, elem_id="ds-panel"):
                 gr.Markdown("### Configure")
                 data_mode = gr.Radio(
-                    choices=["Bundled examples", "Upload your CSV"],
+                    choices=["Bundled examples", "Upload your file"],
                     value="Bundled examples",
                     label="Data source",
                 )
@@ -208,11 +292,22 @@ def build_ui() -> gr.Blocks:
                     label="Demo dataset",
                 )
                 upload = gr.File(
-                    label="Your CSV file",
-                    file_types=[".csv"],
+                    label="Your CSV or Excel file",
+                    file_types=[".csv", ".xlsx", ".xls"],
                     type="filepath",
                     visible=False,
                 )
+
+                gr.Markdown("### Data preview")
+                preview_meta = gr.Markdown(default_meta)
+                with gr.Group(elem_id="ds-preview-box"):
+                    preview_df = gr.Dataframe(
+                        value=default_df,
+                        interactive=False,
+                        wrap=True,
+                        max_height=280,
+                    )
+
                 task = gr.Textbox(
                     label="Question / task",
                     placeholder="e.g. Which product had the highest total revenue?",
@@ -233,29 +328,30 @@ def build_ui() -> gr.Blocks:
                     label="Quick examples (bundled data)",
                 )
 
-                gr.Markdown(
-                    f"""
----
-<small>Base `unsloth/gemma-4-E2B-it` · bundled CSVs in `{DATA_DIR.name}/`</small>
-                    """
-                )
-
-            with gr.Column(scale=6):
+            with gr.Column(scale=7, elem_id="ds-results"):
+                gr.Markdown("### Results")
                 with gr.Tabs():
-                    with gr.Tab("✅ Answer"):
-                        answer_out = gr.Markdown()
-                    with gr.Tab("🔍 Execution trace"):
+                    with gr.Tab("🔍 Execution trace", id="trace_tab"):
                         steps_out = gr.Markdown()
+                    with gr.Tab("✅ Answer", id="answer_tab"):
+                        answer_out = gr.HTML()
 
-        data_mode.change(_toggle_data_inputs, data_mode, [dataset, upload])
+        preview_inputs = [data_mode, dataset, upload]
+        data_mode.change(_toggle_data_inputs, data_mode, [dataset, upload]).then(
+            _load_preview, preview_inputs, [preview_df, preview_meta]
+        )
+        dataset.change(_load_preview, preview_inputs, [preview_df, preview_meta])
+        upload.change(_load_preview, preview_inputs, [preview_df, preview_meta])
 
         run_btn.click(
             fn=run_task,
             inputs=[data_mode, dataset, upload, task, max_steps],
-            outputs=[answer_out, steps_out],
+            outputs=[steps_out, answer_out],
             show_progress="full",
         )
 
+        demo.load(_load_preview, preview_inputs, [preview_df, preview_meta])
+
     return demo
 
 

requirements.txt

@@ -7,3 +7,4 @@ accelerate>=0.30.0
 bitsandbytes>=0.43.0
 matplotlib>=3.7.0
 sentencepiece>=0.2.0
+openpyxl>=3.1.0