Update app.py

app.py

@@ -1,205 +1,660 @@
+# ============================================================
+#                    X-RUDRA CHAT
+#          Hugging Face Gradio Space Edition
+# ============================================================
+
+from __future__ import annotations
+
 import os
-import asyncio
 import json
 import time
+import asyncio
+import traceback
+
 import gradio as gr
 
-from web_search import XrudraWebSearch
 
+# ============================================================
+# CONFIG
+# ============================================================
+
+APP_NAME = "X-RUDRA"
+VERSION = "3.1.0"
+
+M1_REPO = os.getenv(
+    "M1_REPO",
+    "Shrijanagain/M1",
+)
+
+M2_REPO = os.getenv(
+    "M2_REPO",
+    "Shrijanagain/M2",
+)
+
+PORT = int(
+    os.getenv(
+        "PORT",
+        "7860",
+    )
+)
+
+
+# ============================================================
+# LAZY ENGINE
+# ============================================================
+
+_ENGINE = None
+
+
+def get_engine():
+    """
+    IMPORTANT:
+    Do not initialize the web-search engine during
+    Space startup.
+
+    It is created only when the first user request arrives.
+    """
+
+    global _ENGINE
+
+    if _ENGINE is None:
+
+        from web_search import XrudraWebSearch
+
+        _ENGINE = XrudraWebSearch()
+
+    return _ENGINE
+
+
+# ============================================================
+# SAFE VALUE HELPERS
+# ============================================================
+
+def safe_dict(value):
+
+    if isinstance(value, dict):
+        return value
+
+    if hasattr(value, "model_dump"):
+
+        try:
+            return value.model_dump()
+
+        except Exception:
+            pass
+
+    if hasattr(value, "dict"):
+
+        try:
+            return value.dict()
+
+        except Exception:
+            pass
+
+    return {
+        "result": str(value)
+    }
+
+
+def get_value(data, key, default=None):
+
+    if not isinstance(data, dict):
+        return default
+
+    return data.get(
+        key,
+        default,
+    )
+
+
+# ============================================================
+# SOURCE FORMATTER
+# ============================================================
+
+def format_sources(sources):
+
+    if not sources:
+
+        return (
+            "## 📚 Sources\n\n"
+            "No sources were returned."
+        )
+
+    output = [
+        "## 📚 Sources",
+        "",
+    ]
+
+    for index, source in enumerate(
+        sources,
+        start=1,
+    ):
+
+        if not isinstance(source, dict):
+            continue
+
+        title = source.get(
+            "title",
+            "Untitled",
+        )
+
+        url = source.get(
+            "url",
+            "",
+        )
+
+        method = source.get(
+            "fetch_method",
+            "unknown",
+        )
+
+        score = source.get(
+            "source_score",
+            source.get(
+                "score",
+                "N/A",
+            ),
+        )
+
+        if url:
+
+            output.append(
+                f"### {index}. [{title}]({url})"
+            )
+
+        else:
+
+            output.append(
+                f"### {index}. {title}"
+            )
+
+        output.append(
+            f"**Fetcher:** `{method}`"
+        )
+
+        output.append(
+            f"**Source score:** `{score}`"
+        )
+
+        output.append("")
+
+    return "\n".join(output)
+
+
+# ============================================================
+# EVIDENCE FORMATTER
+# ============================================================
+
+def format_evidence(claims):
+
+    if not claims:
+
+        return (
+            "## 🧠 Evidence\n\n"
+            "No structured evidence was returned."
+        )
+
+    output = [
+        "## 🧠 Evidence",
+        "",
+    ]
+
+    for index, claim in enumerate(
+        claims,
+        start=1,
+    ):
+
+        if not isinstance(claim, dict):
+            continue
+
+        text = claim.get(
+            "claim",
+            claim.get(
+                "text",
+                "",
+            ),
+        )
+
+        score = claim.get(
+            "support_score",
+            claim.get(
+                "score",
+                "N/A",
+            ),
+        )
+
+        source = claim.get(
+            "source_url",
+            claim.get(
+                "url",
+                "",
+            ),
+        )
+
+        output.append(
+            f"### Evidence {index}"
+        )
+
+        output.append(
+            str(text)
+        )
+
+        output.append(
+            f"**Support:** `{score}`"
+        )
+
+        if source:
+
+            output.append(
+                f"**Source:** {source}"
+            )
+
+        output.append("---")
+
+    return "\n\n".join(output)
+
+
+# ============================================================
+# VERIFICATION FORMATTER
+# ============================================================
+
+def format_verification(contradictions):
+
+    if not contradictions:
+
+        return (
+            "## ⚖️ Verification\n\n"
+            "✅ No major automatic contradictions "
+            "were detected."
+        )
+
+    output = [
+        "## ⚖️ Verification",
+        "",
+        "⚠️ Potential contradictions detected:",
+        "",
+    ]
+
+    for index, item in enumerate(
+        contradictions,
+        start=1,
+    ):
+
+        if not isinstance(item, dict):
+            continue
+
+        claim_a = item.get(
+            "claim_a",
+            "",
+        )
+
+        claim_b = item.get(
+            "claim_b",
+            "",
+        )
+
+        source_a = item.get(
+            "source_a",
+            "",
+        )
+
+        source_b = item.get(
+            "source_b",
+            "",
+        )
+
+        output.append(
+            f"### Contradiction {index}"
+        )
+
+        output.append(
+            f"**A:** {claim_a}"
+        )
+
+        if source_a:
+
+            output.append(
+                f"Source A: `{source_a}`"
+            )
+
+        output.append("")
+
+        output.append(
+            f"**B:** {claim_b}"
+        )
+
+        if source_b:
+
+            output.append(
+                f"Source B: `{source_b}`"
+            )
+
+        output.append("---")
+
+    return "\n\n".join(output)
+
+
+# ============================================================
+# ANSWER EXTRACTOR
+# ============================================================
+
+def extract_answer(data):
+
+    # Try common final-answer fields first.
+
+    for key in (
+        "final_answer",
+        "answer",
+        "response",
+        "final",
+        "synthesis",
+        "summary",
+    ):
+
+        value = data.get(
+            key,
+            None,
+        )
 
-ENGINE = XrudraWebSearch()
+        if isinstance(
+            value,
+            str,
+        ) and value.strip():
+
+            return value.strip()
+
+    # Otherwise construct an answer from claims.
+
+    claims = data.get(
+        "claims",
+        [],
+    )
+
+    if claims:
+
+        parts = []
+
+        for claim in claims:
+
+            if not isinstance(
+                claim,
+                dict,
+            ):
+                continue
+
+            text = claim.get(
+                "claim",
+                claim.get(
+                    "text",
+                    "",
+                ),
+            )
+
+            if text:
+
+                parts.append(
+                    str(text).strip()
+                )
+
+        if parts:
+
+            return "\n\n".join(
+                parts[:10]
+            )
+
+    return (
+        "Research completed, but the engine "
+        "did not return a final synthesized answer."
+    )
 
 
 # ============================================================
-# RESEARCH ENGINE
+# ACTIVITY PANEL
 # ============================================================
 
-async def research(
+def build_activity(
+    data,
+    elapsed_ms,
+):
+
+    sources = data.get(
+        "sources",
+        [],
+    )
+
+    claims = data.get(
+        "claims",
+        [],
+    )
+
+    contradictions = data.get(
+        "contradictions",
+        [],
+    )
+
+    rounds = data.get(
+        "rounds",
+        data.get(
+            "research_rounds",
+            "N/A",
+        ),
+    )
+
+    return f"""
+## ⚡ X-RUDRA Research
+
+| Stage | Status |
+|---|---|
+| Task analysis | ✅ Complete |
+| M1 research | {"✅ Enabled" if data.get("m1") is not None else "⚙️ Pipeline"} |
+| M2 research | {"✅ Enabled" if data.get("m2") is not None else "⚙️ Pipeline"} |
+| Web discovery | ✅ Complete |
+| Evidence extraction | ✅ Complete |
+| Source verification | ✅ Complete |
+| Contradiction check | {"⚠️ Found" if contradictions else "✅ Clear"} |
+| Final synthesis | ✅ Complete |
+
+**Sources:** `{len(sources)}`  
+**Claims:** `{len(claims)}`  
+**Rounds:** `{rounds}`  
+**Time:** `{elapsed_ms} ms`
+
+### Engine
+
+`M1` → `{M1_REPO}`
+
+`M2` → `{M2_REPO}`
+
+`Web` → `DuckDuckGo`
+
+`Fetcher` → `Scrapling`
+
+`Browser` → `Playwright`
+"""
+
+
+# ============================================================
+# RESEARCH
+# ============================================================
+
+async def do_research(
     question,
     max_results,
     max_rounds,
     use_models,
     freshness,
 ):
-    if not question or not question.strip():
+
+    if not question or not str(
+        question
+    ).strip():
+
         return (
             [],
-            "⚪ Waiting for a question...",
-            "No research started.",
+            "⚪ Enter a question to start.",
+            "",
             "",
         )
 
-    start = time.perf_counter()
+    question = str(
+        question
+    ).strip()
+
+    started = time.perf_counter()
 
     try:
-        report = await ENGINE.search(
-            question=question.strip(),
-            max_results=int(max_results),
-            max_rounds=int(max_rounds),
-            use_models=use_models,
-            freshness_mode=freshness,
-        )
 
-        data = (
-            report.model_dump()
-            if hasattr(report, "model_dump")
-            else report
-        )
+        # ----------------------------------------------------
+        # LAZY INITIALIZATION
+        # ----------------------------------------------------
 
-        sources = data.get("sources", [])
-        claims = data.get("claims", [])
-        contradictions = data.get("contradictions", [])
+        engine = get_engine()
 
         # ----------------------------------------------------
-        # FINAL ANSWER
+        # RUN WEB / MODEL PIPELINE
         # ----------------------------------------------------
 
-        answer_parts = []
+        report = await engine.search(
 
-        for claim in claims[:8]:
-            text = claim.get("claim", "").strip()
+            question=question,
 
-            if text:
-                answer_parts.append(text)
+            max_results=int(
+                max_results
+            ),
 
-        if answer_parts:
-            answer = "\n\n".join(answer_parts)
-        else:
-            answer = (
-                "Research completed, but no high-confidence "
-                "evidence passages were extracted."
-            )
+            max_rounds=int(
+                max_rounds
+            ),
 
-        # ----------------------------------------------------
-        # ACTIVITY
-        # ----------------------------------------------------
+            use_models=bool(
+                use_models
+            ),
 
-        activity = f"""
-### ⚡ X-RUDRA Research Pipeline
+            freshness_mode=str(
+                freshness
+            ),
+        )
 
-✓ Task analyzed  
-✓ M1 + M2 query planning completed  
-✓ DuckDuckGo web discovery completed  
-✓ {len(sources)} sources fetched  
-✓ Evidence extraction completed  
-✓ Source quality evaluated  
-✓ Cross-source verification completed  
-{"⚠️ Contradictions detected" if contradictions else "✓ No major contradictions detected"}  
-✓ Final synthesis completed  
+        data = safe_dict(
+            report
+        )
 
-**Research time:** {int((time.perf_counter() - start) * 1000)} ms
-"""
+        elapsed_ms = int(
+            (
+                time.perf_counter()
+                - started
+            ) * 1000
+        )
 
         # ----------------------------------------------------
-        # SOURCES
+        # OUTPUT
         # ----------------------------------------------------
 
-        source_md = "### 📚 Sources\n\n"
+        answer = extract_answer(
+            data
+        )
 
-        for i, source in enumerate(sources, 1):
-            title = source.get("title") or "Untitled"
-            url = source.get("url", "")
-            method = source.get("fetch_method", "unknown")
-            score = source.get("source_score", 0)
+        sources = data.get(
+            "sources",
+            [],
+        )
 
-            source_md += (
-                f"**{i}. [{title}]({url})**\n\n"
-                f"`{method}` · source score `{score}`\n\n"
-            )
+        claims = data.get(
+            "claims",
+            [],
+        )
 
-        if not sources:
-            source_md += "No sources returned."
+        contradictions = data.get(
+            "contradictions",
+            [],
+        )
 
-        # ----------------------------------------------------
-        # EVIDENCE
-        # ----------------------------------------------------
+        activity = build_activity(
+            data,
+            elapsed_ms,
+        )
 
-        evidence_md = "### 🧠 Evidence\n\n"
+        sources_md = format_sources(
+            sources
+        )
 
-        for i, claim in enumerate(claims[:20], 1):
-            evidence_md += (
-                f"**Evidence {i}**\n\n"
-                f"{claim.get('claim', '')}\n\n"
-                f"Support score: "
-                f"`{claim.get('support_score', 0)}`\n\n"
-                f"Source: `{claim.get('source_url', '')}`\n\n"
-                "---\n\n"
-            )
+        evidence_md = format_evidence(
+            claims
+        )
 
-        if not claims:
-            evidence_md += "No evidence extracted."
+        verification_md = format_verification(
+            contradictions
+        )
 
         # ----------------------------------------------------
-        # CONTRADICTIONS
+        # OLD-COMPATIBLE GRADIO CHAT FORMAT
+        #
+        # IMPORTANT:
+        # No type="messages".
+        # This works with the Gradio version shown
+        # in the uploaded Space logs.
         # ----------------------------------------------------
 
-        if contradictions:
-            verification_md = "### ⚠️ Contradictions\n\n"
-
-            for i, item in enumerate(
-                contradictions,
-                1,
-            ):
-                verification_md += (
-                    f"**Potential contradiction {i}**\n\n"
-                    f"**Source A:**\n"
-                    f"{item.get('claim_a', '')}\n\n"
-                    f"**Source B:**\n"
-                    f"{item.get('claim_b', '')}\n\n"
-                    "---\n\n"
-                )
-        else:
-            verification_md = (
-                "### ✅ Verification\n\n"
-                "No major automatic contradictions detected."
+        history = [
+            (
+                question,
+                answer,
             )
-
-        # ----------------------------------------------------
-        # RETURN CHAT + PANELS
-        # ----------------------------------------------------
-
-        chat = [
-            {
-                "role": "user",
-                "content": question,
-            },
-            {
-                "role": "assistant",
-                "content": answer,
-            },
         ]
 
         return (
-            chat,
+            history,
             activity,
-            source_md,
-            evidence_md + "\n" + verification_md,
+            sources_md,
+            evidence_md,
+            verification_md,
         )
 
     except Exception as exc:
 
-        chat = [
-            {
-                "role": "user",
-                "content": question,
-            },
-            {
-                "role": "assistant",
-                "content": (
-                    "❌ X-RUDRA encountered an error:\n\n"
-                    f"`{type(exc).__name__}: {exc}`"
-                ),
-            },
+        error = (
+            f"❌ **X-RUDRA Error**\n\n"
+            f"`{type(exc).__name__}: {exc}`"
+        )
+
+        print(
+            "\n"
+            + "=" * 70
+        )
+
+        print(
+            "X-RUDRA ERROR"
+        )
+
+        print(
+            traceback.format_exc()
+        )
+
+        print(
+            "=" * 70
+            + "\n"
+        )
+
+        history = [
+            (
+                question,
+                error,
+            )
         ]
 
         return (
-            chat,
+            history,
             "❌ Research failed.",
             "",
             "",
+            "",
         )
 
 
+# ============================================================
+# GRADIO SYNC WRAPPER
+# ============================================================
+
 def run_research(
     question,
     max_results,
@@ -207,8 +662,9 @@ def run_research(
     use_models,
     freshness,
 ):
+
     return asyncio.run(
-        research(
+        do_research(
             question,
             max_results,
             max_rounds,
@@ -218,75 +674,89 @@ def run_research(
     )
 
 
+# ============================================================
+# HEALTH
+# ============================================================
+
+def health_check():
+
+    return f"""
+## 🟢 X-RUDRA Online
+
+**Version:** `{VERSION}`
+
+**M1:** `{M1_REPO}`
+
+**M2:** `{M2_REPO}`
+
+**Web Search:** `DuckDuckGo`
+
+**HTTP Fetch:** `Scrapling`
+
+**Dynamic Fetch:** `Playwright`
+
+**Startup model loading:** `Disabled`
+
+**Engine:** `Lazy initialized`
+"""
+
+
 # ============================================================
 # CSS
 # ============================================================
 
-CSS = r"""
-:root {
-    --radius: 18px;
-}
+CSS = """
 
 body {
-    background: #f7f7f8 !important;
+    background: #f7f7f8;
 }
 
 .gradio-container {
     max-width: 1500px !important;
-    margin: auto !important;
 }
 
 #header {
     text-align: center;
-    padding: 18px 0 8px 0;
+    padding: 20px 0 10px 0;
 }
 
 #logo {
-    font-size: 34px;
+    font-size: 38px;
     font-weight: 800;
-    letter-spacing: -1px;
 }
 
 #tagline {
-    opacity: .65;
-    font-size: 14px;
+    opacity: 0.65;
+    font-size: 15px;
 }
 
 #chat {
-    border-radius: 18px !important;
+    border-radius: 18px;
 }
 
-#research-panel {
-    border-radius: 18px !important;
+#research {
+    border-radius: 18px;
 }
 
-#search-button {
-    border-radius: 14px !important;
-    min-height: 50px !important;
-    font-weight: 700 !important;
-}
-
-.status-box {
-    border-radius: 14px !important;
+#send {
+    min-height: 52px;
+    font-size: 18px;
+    font-weight: 700;
 }
 
 footer {
     display: none !important;
 }
+
 """
 
 
 # ============================================================
-# UI
+# GRADIO APPLICATION
 # ============================================================
 
 with gr.Blocks(
-    title="X-RUDRA",
-    css=CSS,
-    theme=gr.themes.Soft(
-        primary_hue="indigo",
-        neutral_hue="slate",
-    ),
+    title=APP_NAME,
 ) as demo:
 
     # --------------------------------------------------------
@@ -296,22 +766,25 @@ with gr.Blocks(
     gr.HTML(
         """
         <div id="header">
-            <div id="logo">⚡ X-RUDRA</div>
+            <div id="logo">
+                ⚡ X-RUDRA
+            </div>
+
             <div id="tagline">
-                Dual-Model AI Research · Live Web Intelligence · Evidence
+                Dual-Model AI · Live Web Research · Evidence
             </div>
         </div>
         """
     )
 
     # --------------------------------------------------------
-    # MAIN
+    # MAIN CHAT
     # --------------------------------------------------------
 
     with gr.Row():
 
         # ====================================================
-        # CHAT
+        # CHAT COLUMN
         # ====================================================
 
         with gr.Column(
@@ -320,35 +793,30 @@ with gr.Blocks(
 
             chatbot = gr.Chatbot(
                 label="X-RUDRA",
-                type="messages",
-                height=620,
+                height=600,
                 elem_id="chat",
-                placeholder=(
-                    "Ask X-RUDRA anything that needs "
-                    "real-time research..."
-                ),
             )
 
             with gr.Row():
 
                 question = gr.Textbox(
                     placeholder=(
-                        "Ask a research question..."
+                        "Ask X-RUDRA anything..."
                     ),
                     lines=2,
-                    scale=8,
                     show_label=False,
+                    scale=8,
                 )
 
                 send = gr.Button(
                     "➤",
                     variant="primary",
+                    elem_id="send",
                     scale=1,
-                    elem_id="search-button",
                 )
 
         # ====================================================
-        # LIVE RESEARCH PANEL
+        # RESEARCH STATUS
         # ====================================================
 
         with gr.Column(
@@ -360,7 +828,9 @@ with gr.Blocks(
             )
 
             activity = gr.Markdown(
-                "⚪ Waiting for research..."
+                """
+⚪ Waiting for your question.
+"""
             )
 
             gr.Markdown(
@@ -368,32 +838,30 @@ with gr.Blocks(
             )
 
             gr.Markdown(
-                "### ⚙️ Engine"
-            )
+                f"""
+### Model Spaces
 
-            gr.Markdown(
-                """
-**Models**
+**M1**
 
-`M1` · `Shrijanagain/M1`
+`{M1_REPO}`
 
-`M2` · `Shrijanagain/M2`
+**M2**
 
-**Search**
+`{M2_REPO}`
 
-`DuckDuckGo`
+### Web Stack
 
-**Fetching**
+`DuckDuckGo`
 
 `Scrapling`
 
-**Dynamic pages**
-
 `Playwright`
 
-**Verification**
+### Verification
 
 `Evidence Engine`
+
+`Cross-source checking`
 """
             )
 
@@ -443,32 +911,50 @@ with gr.Blocks(
             )
 
     # --------------------------------------------------------
-    # RESEARCH DETAILS
+    # RESEARCH DATA
     # --------------------------------------------------------
 
     with gr.Tabs():
 
         with gr.Tab(
-            "📚 Sources",
+            "📚 Sources"
         ):
+
             sources = gr.Markdown(
-                "Sources will appear after research."
+                "Sources will appear here."
             )
 
         with gr.Tab(
-            "🧠 Evidence",
+            "🧠 Evidence"
         ):
+
             evidence = gr.Markdown(
-                "Evidence will appear after research."
+                "Evidence will appear here."
             )
 
         with gr.Tab(
-            "⚖️ Verification",
+            "⚖️ Verification"
         ):
+
             verification = gr.Markdown(
-                "Verification will appear after research."
+                "Verification will appear here."
             )
 
+    # --------------------------------------------------------
+    # HEALTH
+    # --------------------------------------------------------
+
+    with gr.Accordion(
+        "🩺 System Health",
+        open=False,
+    ):
+
+        health_button = gr.Button(
+            "Check X-RUDRA",
+        )
+
+        health_output = gr.Markdown()
+
     # --------------------------------------------------------
     # EXAMPLES
     # --------------------------------------------------------
@@ -479,10 +965,18 @@ with gr.Blocks(
 
     gr.Examples(
         examples=[
-            ["What are the latest AI education initiatives by UNESCO?"],
-            ["What happened in AI research this week?"],
-            ["Compare the latest open-source AI models."],
-            ["Find reliable evidence about India's AI ecosystem."],
+            [
+                "What are the latest UNESCO AI education initiatives?"
+            ],
+            [
+                "What are the latest developments in open source AI?"
+            ],
+            [
+                "Compare the latest major AI models."
+            ],
+            [
+                "Research India's current AI ecosystem."
+            ],
         ],
         inputs=question,
     )
@@ -504,6 +998,7 @@ with gr.Blocks(
         activity,
         sources,
         evidence,
+        verification,
     ]
 
     send.click(
@@ -518,22 +1013,42 @@ with gr.Blocks(
         outputs=outputs,
     )
 
+    health_button.click(
+        fn=health_check,
+        inputs=[],
+        outputs=[
+            health_output
+        ],
+    )
+
 
 # ============================================================
-# HF SPACES LAUNCH
+# START
 # ============================================================
 
 if __name__ == "__main__":
 
-    port = int(
-        os.environ.get(
-            "PORT",
-            "7860",
-        )
+    print(
+        f"Starting {APP_NAME} {VERSION}"
+    )
+
+    print(
+        "M1:",
+        M1_REPO,
+    )
+
+    print(
+        "M2:",
+        M2_REPO,
+    )
+
+    print(
+        "Lazy engine initialization: ON"
     )
 
     demo.launch(
         server_name="0.0.0.0",
-        server_port=port,
+        server_port=PORT,
+        css=CSS,
         show_error=True,
     )