Upload folder using huggingface_hub

src/eurus/retrieval.py

@@ -31,6 +31,45 @@ from eurus.memory import get_memory
 logger = logging.getLogger(__name__)
 
 
+def _arraylake_snippet(
+    variable: str,
+    query_type: str,
+    start_date: str,
+    end_date: str,
+    min_lat: float,
+    max_lat: float,
+    min_lon: float,
+    max_lon: float,
+) -> str:
+    """Generate a ready-to-paste Python snippet for direct Arraylake access."""
+    # Convert negative lons to 0-360 for ERA5
+    era5_min = min_lon % 360 if min_lon < 0 else min_lon
+    era5_max = max_lon % 360 if max_lon < 0 else max_lon
+    return (
+        f"\n📦 Reproduce this download yourself (copy-paste into Jupyter):\n"
+        f"```python\n"
+        f"import os, xarray as xr\n"
+        f"from arraylake import Client\n"
+        f"\n"
+        f"client  = Client(token=os.environ['ARRAYLAKE_API_KEY'])\n"
+        f"repo    = client.get_repo('{CONFIG.data_source}')\n"
+        f"session = repo.readonly_session('main')\n"
+        f"\n"
+        f"ds = xr.open_dataset(session.store, engine='zarr',\n"
+        f"                     consolidated=False, zarr_format=3,\n"
+        f"                     chunks=None, group='{query_type}')\n"
+        f"\n"
+        f"subset = ds['{variable}'].sel(\n"
+        f"    time=slice('{start_date}', '{end_date}'),\n"
+        f"    latitude=slice({max_lat}, {min_lat}),   # ERA5: descending\n"
+        f"    longitude=slice({era5_min}, {era5_max}),\n"
+        f")\n"
+        f"\n"
+        f"subset.load().to_dataset(name='{variable}').to_zarr('my_data.zarr', mode='w')\n"
+        f"```"
+    )
+
+
 def _format_coord(value: float) -> str:
     """Format coordinates for stable, filename-safe identifiers."""
     if abs(value) < 0.005:
@@ -448,10 +487,18 @@ def retrieve_era5_data(
             # Size guard — prevent downloading datasets larger than the configured limit
             estimated_gb = ds_out.nbytes / (1024 ** 3)
             if estimated_gb > CONFIG.max_download_size_gb:
+                snippet = _arraylake_snippet(
+                    short_var, query_type, start_date, end_date,
+                    min_latitude, max_latitude,
+                    min_longitude if min_longitude >= 0 else min_longitude % 360,
+                    max_longitude if max_longitude >= 0 else max_longitude % 360,
+                )
                 return (
                     f"Error: Estimated download size ({estimated_gb:.1f} GB) exceeds the "
                     f"{CONFIG.max_download_size_gb} GB limit.\n"
-                    f"Try narrowing the time range or spatial area."
+                    f"Try narrowing the time range or spatial area.\n\n"
+                    f"Alternatively, fetch it yourself with this snippet:\n\n"
+                    f"{snippet}"
                 )
             if estimated_gb > 1.0:
                 logger.info(
@@ -527,6 +574,12 @@ def retrieve_era5_data(
                 logger.info(f"Retrying in {wait_time:.1f}s...")
                 time.sleep(wait_time)
             else:
+                snippet = _arraylake_snippet(
+                    short_var, query_type, start_date, end_date,
+                    min_latitude, max_latitude,
+                    min_longitude if min_longitude >= 0 else min_longitude % 360,
+                    max_longitude if max_longitude >= 0 else max_longitude % 360,
+                )
                 return (
                     f"Error: Failed after {CONFIG.max_retries} attempts.\n"
                     f"Last error: {error_msg}\n\n"
@@ -534,7 +587,9 @@ def retrieve_era5_data(
                     f"1. Check your ARRAYLAKE_API_KEY\n"
                     f"2. Verify internet connection\n"
                     f"3. Try a smaller date range or region\n"
-                    f"4. Check if variable '{short_var}' is available"
+                    f"4. Check if variable '{short_var}' is available\n\n"
+                    f"Manual retrieval snippet:\n\n"
+                    f"{snippet}"
                 )
 
     return "Error: Unexpected failure in retrieval logic."

web/agent_wrapper.py

@@ -25,6 +25,7 @@ from langchain.agents import create_agent
 
 # IMPORT FROM EURUS PACKAGE - SINGLE SOURCE OF TRUTH
 from eurus.config import CONFIG, AGENT_SYSTEM_PROMPT
+from eurus.retrieval import _arraylake_snippet
 from eurus.memory import get_memory, SmartConversationMemory  # Singleton for datasets, per-session for chat
 from eurus.tools import get_all_tools
 from eurus.tools.repl import PythonREPLTool
@@ -181,12 +182,14 @@ class AgentSession:
                 lambda: self._agent.invoke({"messages": self._messages}, config=config)
             )
 
-            # Update messages
+            # Only scan NEW messages from this turn
+            prev_count = len(self._messages)
             self._messages = result["messages"]
+            new_messages = self._messages[prev_count:]
             
-            # Parse messages to show tool calls made
+            # Parse NEW messages to show tool calls made
             tool_calls_made = []
-            for msg in self._messages:
+            for msg in new_messages:
                 if hasattr(msg, 'tool_calls') and msg.tool_calls:
                     for tc in msg.tool_calls:
                         tool_name = tc.get('name', 'unknown')
@@ -198,6 +201,24 @@ class AgentSession:
                 await stream_callback("status", f"🛠️ Used tools: {tools_str}")
                 await asyncio.sleep(0.5)
 
+            # Collect Arraylake snippet from NEW messages only
+            arraylake_snippets = []
+            for msg in new_messages:
+                if hasattr(msg, 'tool_calls') and msg.tool_calls:
+                    for tc in msg.tool_calls:
+                        if tc.get('name') == 'retrieve_era5_data':
+                            args = tc.get('args', {})
+                            arraylake_snippets.append(_arraylake_snippet(
+                                variable=args.get('variable_id', 'sst'),
+                                query_type=args.get('query_type', 'spatial'),
+                                start_date=args.get('start_date', ''),
+                                end_date=args.get('end_date', ''),
+                                min_lat=args.get('min_latitude', -90),
+                                max_lat=args.get('max_latitude', 90),
+                                min_lon=args.get('min_longitude', 0),
+                                max_lon=args.get('max_longitude', 360),
+                            ))
+
             # Extract response
             last_message = self._messages[-1]
 
@@ -244,6 +265,10 @@ class AgentSession:
                     # Default to plot (png, jpg, etc.)
                     await stream_callback("plot", "", data=base64_data, path=filepath, code=code)
 
+            # Send Arraylake snippets AFTER response + plots exist in DOM
+            for snippet in arraylake_snippets:
+                await stream_callback("arraylake_snippet", snippet)
+
             # Save to memory
             self._conversation.add_message("assistant", response_text)
 

web/static/css/style.css

@@ -880,4 +880,72 @@ dialog .close-modal:hover {
 .save-keys-btn:disabled {
     opacity: 0.5;
     cursor: not-allowed;
+}
+
+/* ===== ARRAYLAKE SNIPPET (in-message) ===== */
+.arraylake-snippet-section {
+    margin-top: 0.75rem;
+}
+
+.arraylake-actions {
+    display: flex;
+    gap: 0.5rem;
+}
+
+.arraylake-btn {
+    padding: 0.5rem 0.875rem;
+    font-size: 0.75rem;
+    font-weight: 500;
+    border: 1px solid var(--glass-border);
+    border-radius: 0.375rem;
+    background: var(--bg-tertiary);
+    color: var(--text-secondary);
+    cursor: pointer;
+    transition: all 0.15s ease;
+}
+
+.arraylake-btn:hover {
+    border-color: var(--accent-primary);
+    color: var(--accent-primary);
+}
+
+.arraylake-code {
+    margin-top: 0.5rem;
+    border-radius: 0.5rem;
+    overflow: hidden;
+    border: 1px solid var(--glass-border);
+}
+
+.arraylake-code pre {
+    margin: 0;
+    padding: 0.875rem;
+    background: var(--code-bg);
+    overflow-x: auto;
+    font-size: 0.8125rem;
+    line-height: 1.5;
+}
+
+.arraylake-code code {
+    font-family: 'SF Mono', Monaco, Consolas, 'Liberation Mono', monospace;
+}
+
+.copy-snippet-btn {
+    display: block;
+    margin: 0;
+    padding: 0.5rem 0.875rem;
+    width: 100%;
+    font-size: 0.75rem;
+    font-weight: 500;
+    border: none;
+    border-top: 1px solid var(--glass-border);
+    background: var(--bg-tertiary);
+    color: var(--text-secondary);
+    cursor: pointer;
+    transition: all 0.15s ease;
+    text-align: center;
+}
+
+.copy-snippet-btn:hover {
+    color: var(--accent-primary);
+    background: var(--bg-secondary);
 }

web/static/js/chat.js

@@ -430,6 +430,10 @@ class EurusChat {
                 this.sendBtn.disabled = false;
                 break;
 
+            case 'arraylake_snippet':
+                this.addArraylakeSnippet(data.content);
+                break;
+
             case 'error':
                 this.showError(data.content);
                 this.sendBtn.disabled = false;
@@ -677,6 +681,105 @@ class EurusChat {
         this.currentAssistantMessage = null;
     }
 
+    addArraylakeSnippet(snippetText) {
+        // Find the latest assistant message
+        const messages = this.messagesContainer.querySelectorAll('.assistant-message');
+        const targetMsg = messages.length > 0 ? messages[messages.length - 1] : null;
+        if (!targetMsg) return;
+
+        // Strip markdown fences for raw code display
+        let cleanCode = snippetText
+            .replace(/^\n?📦[^\n]*\n/, '')
+            .replace(/^```python\n?/, '')
+            .replace(/\n?```$/, '')
+            .trim();
+
+        // Find the last plot figure's action bar to add button inline
+        const figures = targetMsg.querySelectorAll('.plot-figure');
+        const lastFigure = figures.length > 0 ? figures[figures.length - 1] : null;
+        const actionsDiv = lastFigure ? lastFigure.querySelector('.plot-actions') : null;
+
+        if (actionsDiv) {
+            // Add button inline with Enlarge/Download/Show Code
+            const btn = document.createElement('button');
+            btn.className = 'code-btn';
+            btn.title = 'Arraylake Code';
+            btn.textContent = '📦 Arraylake Code';
+            actionsDiv.appendChild(btn);
+
+            // Add code block to figure (same pattern as Show Code)
+            const codeDiv = document.createElement('div');
+            codeDiv.className = 'plot-code';
+            codeDiv.style.display = 'none';
+
+            const pre = document.createElement('pre');
+            const codeEl = document.createElement('code');
+            codeEl.className = 'language-python hljs';
+            try {
+                codeEl.innerHTML = hljs.highlight(cleanCode, { language: 'python' }).value;
+            } catch (e) {
+                codeEl.textContent = cleanCode;
+            }
+            pre.appendChild(codeEl);
+            codeDiv.appendChild(pre);
+
+            // Copy button inside code block
+            const copyBtn = document.createElement('button');
+            copyBtn.className = 'copy-snippet-btn';
+            copyBtn.textContent = 'Copy Code';
+            copyBtn.addEventListener('click', () => {
+                navigator.clipboard.writeText(cleanCode).then(() => {
+                    copyBtn.textContent = '✓ Copied!';
+                    setTimeout(() => copyBtn.textContent = 'Copy Code', 2000);
+                });
+            });
+            codeDiv.appendChild(copyBtn);
+            lastFigure.appendChild(codeDiv);
+
+            // Toggle
+            btn.addEventListener('click', () => {
+                if (codeDiv.style.display === 'none') {
+                    codeDiv.style.display = 'block';
+                    btn.textContent = '📦 Hide Arraylake';
+                } else {
+                    codeDiv.style.display = 'none';
+                    btn.textContent = '📦 Arraylake Code';
+                }
+            });
+        } else {
+            // No plot figure — add as standalone section under the message
+            const wrapper = document.createElement('div');
+            wrapper.className = 'arraylake-snippet-section';
+            wrapper.innerHTML = `
+                <div class="plot-actions">
+                    <button class="code-btn" title="Arraylake Code">📦 Arraylake Code</button>
+                </div>
+                <div class="plot-code" style="display: none;">
+                    <pre><code class="language-python hljs"></code></pre>
+                </div>
+            `;
+            const codeEl = wrapper.querySelector('code');
+            try {
+                codeEl.innerHTML = hljs.highlight(cleanCode, { language: 'python' }).value;
+            } catch (e) {
+                codeEl.textContent = cleanCode;
+            }
+            const btn = wrapper.querySelector('.code-btn');
+            const codeDiv = wrapper.querySelector('.plot-code');
+            btn.addEventListener('click', () => {
+                if (codeDiv.style.display === 'none') {
+                    codeDiv.style.display = 'block';
+                    btn.textContent = '📦 Hide Arraylake';
+                } else {
+                    codeDiv.style.display = 'none';
+                    btn.textContent = '📦 Arraylake Code';
+                }
+            });
+            targetMsg.appendChild(wrapper);
+        }
+        this.scrollToBottom();
+    }
+
     showError(message) {
         this.removeThinkingIndicator();
 

web/templates/index.html

@@ -60,5 +60,5 @@
 {% endblock %}
 
 {% block scripts %}
-<script src="/static/js/chat.js?v=20260218c"></script>
+<script src="/static/js/chat.js?v=20260219b"></script>
 {% endblock %}