Update app.py

app.py

@@ -485,82 +485,104 @@ def compute_gsd_metrics(diams: List[float], passing: List[float]) -> Dict[str, f
 if "sites" not in st.session_state:
     # initialize with a default site
     st.session_state["sites"] = [{
-            "Site Name": None,
-            "Project Name": "Project",
-            "Site ID": None,
-            "Soil Class": None,
-            "Soil Recognizer Confidence": None,
-            "Coordinates": "",
-            "lat": None,
-            "lon": None,
-            "Project Description": "",
-            # ---------------------------
-            # Site Characterization
-            # ---------------------------
-            "Topography": None,              # manual topo entry
-            "Drainage": None,                # manual drainage notes
-            "Current Land Use": None,        # can be linked to Environmental Data
-            "Regional Geology": None,        # manual geology notes
-            # ---------------------------
-            # Investigations & Lab
-            # ---------------------------
-            "Field Investigation": [],
-            "Laboratory Results": [],
-            "GSD": None,
-            "USCS": None,
-            "AASHTO": None,
-            "GI": None,
-            # ---------------------------
-            # Geotechnical Parameters
-            # ---------------------------
-            "Load Bearing Capacity": None,
-            "Skin Shear Strength": None,
-            "Relative Compaction": None,
-            "Rate of Consolidation": None,
-            "Nature of Construction": None,
-            # ---------------------------
-            # Earth Engine Data
-            # ---------------------------
-            "Soil Profile": {                # SoilGrids (multi-parameter)
-            "Clay": None,                # e.g. % clay at 200 cm
-            "Sand": None,                # % sand
-            "Silt": None,                # % silt
-            "OrganicCarbon": None,       # % organic carbon
-            "pH": None                   # soil pH if available
-            },
-            "Topo Data": None,               # Avg elevation (SRTM DEM)
-            "Seismic Data": None,            # PGA/g (GEM hazard)
-            "Flood Data": None,              # JRC Surface Water occurrence
-            "Environmental Data": {          # Landcover, vegetation, urban, etc.
-            "Landcover Stats": None,     # histogram by class
-            "Forest Loss": None,         # future add: Hansen dataset
-            "Urban Fraction": None       # optional calc from landcover
-            },
-            "Weather Data": {                # daily/monthly climate summaries
+        # ---------------------------
+        # Basic Project Info
+        # ---------------------------
+        "Site Name": None,
+        "Project Name": None,
+        "Site ID": None,
+        "Coordinates": "",
+        "lat": None,
+        "lon": None,
+        "Project Description": "",
+    
+        # ---------------------------
+        # Soil Recognition / Classifier
+        # ---------------------------
+        "Soil Class": None,
+        "Soil Recognizer Confidence": None,
+        "USCS": None,
+        "AASHTO": None,
+        "GI": None,
+        "GSD": None,                        # Grain size distribution metrics
+        "classifier_inputs": {},            # User/lab/OCR extracted inputs
+        "classifier_decision": None,        # Full text decision from classifier
+    
+        # ---------------------------
+        # Site Characterization
+        # ---------------------------
+        "Topography": None,                 # manual topo entry
+        "Drainage": None,                   # manual drainage notes
+        "Current Land Use": None,           # linked to environmental data
+        "Regional Geology": None,           # manual geology notes
+    
+        # ---------------------------
+        # Investigations & Lab
+        # ---------------------------
+        "Field Investigation": [],          # e.g. borehole logs
+        "Laboratory Results": [],           # lab test results
+    
+        # ---------------------------
+        # Geotechnical Parameters (incl. REPORT_FIELDS)
+        # ---------------------------
+        "Load Bearing Capacity": None,      # kPa or psf
+        "Skin Shear Strength": None,        # kPa
+        "Relative Compaction": None,        # %
+        "Rate of Consolidation": None,      # mm/yr or days
+        "Nature of Construction": None,     # text
+        "Borehole Count": None,             # number
+        "Max Depth (m)": None,              # m
+        "SPT N (avg)": None,                # blows/ft
+        "CBR (%)": None,                    # %
+        "Allowable Bearing (kPa)": None,    # kPa
+    
+        # ---------------------------
+        # Earth Engine Data
+        # ---------------------------
+        "Soil Profile": {                   # SoilGrids / OpenLandMap
+            "Clay": None,                   # % clay
+            "Sand": None,                   # % sand
+            "Silt": None,                   # % silt
+            "OrganicCarbon": None,          # % organic carbon
+            "pH": None                      # soil pH
+        },
+        "Topo Data": None,                  # Avg elevation
+        "Seismic Data": None,               # PGA/g
+        "Flood Data": None,                 # % flood occurrence
+        "Environmental Data": {
+            "Landcover Stats": None,        # histogram
+            "Forest Loss": None,            # optional (Hansen)
+            "Urban Fraction": None          # optional calc
+        },
+        "Weather Data": {                   # climate summaries
             "Rainfall": None,
             "Temperature": None,
             "Humidity": None
-            },
-            "Atmospheric Data": {            # optional: pollution, aerosols
+        },
+        "Atmospheric Data": {               # pollution, aerosols
             "AerosolOpticalDepth": None,
             "NO2": None,
-            "CO": None
-            },
-            # ---------------------------
-            # Map & Visualization
-            # ---------------------------
-            "map_snapshot": None,
-            # ---------------------------
-            # AI / Reporting
-            # ---------------------------
-            "chat_history": [],
-            "classifier_inputs": {},
-            "classifier_decision": None,
-            "report_convo_state": 0,
-            "report_missing_fields": [],
-            "report_answers": {}
-             }]
-        
+            "CO": None,
+            "PM2.5": None                   # from Sentinel-5P aerosol index
+        },
+    
+        # ---------------------------
+        # Map & Visualization
+        # ---------------------------
+        "ROI": None,                        # GeoJSON of ROI
+        "roi_coords": None,                 # Flattened coords (lat,lon)
+        "map_snapshot": None,               # PNG snapshot
+    
+        # ---------------------------
+        # AI / Reporting
+        # ---------------------------
+        "chat_history": [],
+        "LLM_Report_Text": None,
+        "report_convo_state": 0,
+        "report_missing_fields": [],
+        "report_answers": {}
+    }]
+
 
 if "active_site" not in st.session_state:
     st.session_state["active_site"] = 0
@@ -568,23 +590,59 @@ if "active_site" not in st.session_state:
 if "llm_model" not in st.session_state:
     st.session_state["llm_model"] = "groq/compound"
 
-# Groq client (simple wrapper)
+# -------------------
+# API Keys
+# -------------------
 GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
-def groq_generate(prompt: str, model: str = None, max_tokens: int = 512) -> str:
-    """Call Groq. If call fails, return an explanatory text."""
+GEM_API_KEY = os.environ.get("GEM_API")
+if "GEM_API" in st.secrets:
+    GEM_API_KEY = st.secrets["GEM_API"]
+
+# -------------------
+# Universal LLM Call
+# -------------------
+def llm_generate(prompt: str, model: str = None, max_tokens: int = 512) -> str:
+    """Universal LLM call for Groq, Gemini, DeepSeek."""
+    model_name = model or st.session_state["llm_model"]
+
     try:
-        client = Groq(api_key=GROQ_API_KEY)
-        model_name = model or st.session_state["llm_model"]
-        completion = client.chat.completions.create(
-            model=model_name,
-            messages=[{"role":"user","content":prompt}],
-            temperature=0.2,
-            max_tokens=max_tokens
-        )
-        text = completion.choices[0].message.content
-        return text
+        # ------------------- GEMINI -------------------
+        if model_name.lower().startswith("gemini"):
+            import requests
+            if not GEM_API_KEY:
+                return "[LLM error: No GEM_API key found in secrets]"
+
+            url = f"https://generativelanguage.googleapis.com/v1beta/models/{model_name}:generateContent?key={GEM_API_KEY}"
+            headers = {"Content-Type": "application/json"}
+            payload = {
+                "contents": [{"role": "user", "parts": [{"text": prompt}]}],
+                "generationConfig": {"temperature": 0.2, "maxOutputTokens": max_tokens}
+            }
+            resp = requests.post(url, headers=headers, json=payload, timeout=60)
+            resp.raise_for_status()
+            data = resp.json()
+            return data["candidates"][0]["content"]["parts"][0]["text"].strip()
+
+        # ------------------- GROQ / DEEPSEEK -------------------
+        elif "groq" in model_name or "llama" in model_name or "deepseek" in model_name:
+            from groq import Groq
+            if not GROQ_API_KEY:
+                return "[LLM error: No GROQ_API key found in secrets/env]"
+
+            client = Groq(api_key=GROQ_API_KEY)
+            completion = client.chat.completions.create(
+                model=model_name,
+                messages=[{"role": "user", "content": prompt}],
+                temperature=0.2,
+                max_tokens=max_tokens
+            )
+            return completion.choices[0].message.content
+
+        else:
+            return f"[LLM error: Unknown model {model_name}]"
+
     except Exception as e:
-        return f"[LLM error or offline: {e}]"
+        return f"[LLM error: {e}]"
 
 # 5) UI helper: nice CSS for chat bubbles & page styling
 st.markdown("""
@@ -592,11 +650,9 @@ st.markdown("""
 /* Background and card styling */
 body { background: #0b0b0b; color: #e9eef6; }
 .stApp > .main > .block-container { padding-top: 18px; }
-
 /* Landing and cards */
 .gm-card { background: linear-gradient(180deg, rgba(255,122,0,0.04), rgba(255,122,0,0.02)); border-radius:12px; padding:14px; border:1px solid rgba(255,122,0,0.06);}
 .gm-cta { background: linear-gradient(90deg,#ff7a00,#ff3a3a); color:white; padding:10px 14px; border-radius:10px; font-weight:700; }
-
 /* Chat bubbles */
 .chat-bot { background: #0f1720; border-left:4px solid #FF7A00; padding:10px 12px; border-radius:12px; margin:6px 0; color:#e9eef6; }
 .chat-user { background: #1a1f27; padding:10px 12px; border-radius:12px; margin:6px 0; color:#cfe6ff; text-align:right;}
@@ -615,7 +671,10 @@ with st.sidebar:
         "llama-3.1-8b-instant",
         "meta-llama/llama-guard-4-12b",
         "llama-3.3-70b-versatile",
-        "groq/compound"
+        "groq/compound",
+        "deepseek-r1-distill-llama-70b",   # ✅ Added DeepSeek
+        "gemini-1.5-pro",                  # ✅ Added Gemini
+        "gemini-1.5-flash"
     ], index=0)
     st.markdown("---")
 
@@ -636,78 +695,97 @@ with st.sidebar:
             "Site Name": new_site_name.strip(),
             "Project Name": "Project - " + new_site_name.strip(),
             "Site ID": idx,
-            "Soil Class": None,
-            "Soil Recognizer Confidence": None,
             "Coordinates": "",
             "lat": None,
             "lon": None,
             "Project Description": "",
+        
+            # ---------------------------
+            # Soil Recognition / Classifier
+            # ---------------------------
+            "Soil Class": None,
+            "Soil Recognizer Confidence": None,
+            "USCS": None,
+            "AASHTO": None,
+            "GI": None,
+            "GSD": None,                        # Grain size distribution metrics
+            "classifier_inputs": {},            # User/lab/OCR extracted inputs
+            "classifier_decision": None,        # Full text decision from classifier
+        
             # ---------------------------
             # Site Characterization
             # ---------------------------
-            "Topography": None,              # manual topo entry
-            "Drainage": None,                # manual drainage notes
-            "Current Land Use": None,        # can be linked to Environmental Data
-            "Regional Geology": None,        # manual geology notes
+            "Topography": None,                 # manual topo entry
+            "Drainage": None,                   # manual drainage notes
+            "Current Land Use": None,           # linked to environmental data
+            "Regional Geology": None,           # manual geology notes
+        
             # ---------------------------
             # Investigations & Lab
             # ---------------------------
-            "Field Investigation": [],
-            "Laboratory Results": [],
-            "GSD": None,
-            "USCS": None,
-            "AASHTO": None,
-            "GI": None,
+            "Field Investigation": [],          # e.g. borehole logs
+            "Laboratory Results": [],           # lab test results
+        
             # ---------------------------
-            # Geotechnical Parameters
+            # Geotechnical Parameters (incl. REPORT_FIELDS)
             # ---------------------------
-            "Load Bearing Capacity": None,
-            "Skin Shear Strength": None,
-            "Relative Compaction": None,
-            "Rate of Consolidation": None,
-            "Nature of Construction": None,
+            "Load Bearing Capacity": None,      # kPa or psf
+            "Skin Shear Strength": None,        # kPa
+            "Relative Compaction": None,        # %
+            "Rate of Consolidation": None,      # mm/yr or days
+            "Nature of Construction": None,     # text
+            "Borehole Count": None,             # number
+            "Max Depth (m)": None,              # m
+            "SPT N (avg)": None,                # blows/ft
+            "CBR (%)": None,                    # %
+            "Allowable Bearing (kPa)": None,    # kPa
+        
             # ---------------------------
             # Earth Engine Data
             # ---------------------------
-            "Soil Profile": {                # SoilGrids (multi-parameter)
-            "Clay": None,                # e.g. % clay at 200 cm
-            "Sand": None,                # % sand
-            "Silt": None,                # % silt
-            "OrganicCarbon": None,       # % organic carbon
-            "pH": None                   # soil pH if available
+            "Soil Profile": {                   # SoilGrids / OpenLandMap
+                "Clay": None,                   # % clay
+                "Sand": None,                   # % sand
+                "Silt": None,                   # % silt
+                "OrganicCarbon": None,          # % organic carbon
+                "pH": None                      # soil pH
             },
-            "Topo Data": None,               # Avg elevation (SRTM DEM)
-            "Seismic Data": None,            # PGA/g (GEM hazard)
-            "Flood Data": None,              # JRC Surface Water occurrence
-            "Environmental Data": {          # Landcover, vegetation, urban, etc.
-            "Landcover Stats": None,     # histogram by class
-            "Forest Loss": None,         # future add: Hansen dataset
-            "Urban Fraction": None       # optional calc from landcover
+            "Topo Data": None,                  # Avg elevation
+            "Seismic Data": None,               # PGA/g
+            "Flood Data": None,                 # % flood occurrence
+            "Environmental Data": {
+                "Landcover Stats": None,        # histogram
+                "Forest Loss": None,            # optional (Hansen)
+                "Urban Fraction": None          # optional calc
             },
-            "Weather Data": {                # daily/monthly climate summaries
-            "Rainfall": None,
-            "Temperature": None,
-            "Humidity": None
+            "Weather Data": {                   # climate summaries
+                "Rainfall": None,
+                "Temperature": None,
+                "Humidity": None
             },
-            "Atmospheric Data": {            # optional: pollution, aerosols
-            "AerosolOpticalDepth": None,
-            "NO2": None,
-            "CO": None
+            "Atmospheric Data": {               # pollution, aerosols
+                "AerosolOpticalDepth": None,
+                "NO2": None,
+                "CO": None,
+                "PM2.5": None                   # from Sentinel-5P aerosol index
             },
+        
             # ---------------------------
             # Map & Visualization
             # ---------------------------
-            "map_snapshot": None,
+            "ROI": None,                        # GeoJSON of ROI
+            "roi_coords": None,                 # Flattened coords (lat,lon)
+            "map_snapshot": None,               # PNG snapshot
+        
             # ---------------------------
             # AI / Reporting
             # ---------------------------
             "chat_history": [],
-            "classifier_inputs": {},
-            "classifier_decision": None,
+            "LLM_Report_Text": None,
             "report_convo_state": 0,
             "report_missing_fields": [],
             "report_answers": {}
-             })
+        }]
 
 
             st.success(f"Site '{new_site_name.strip()}' created.")
@@ -1169,6 +1247,12 @@ def locator_page():
 
     # --- Map setup
     m = geemap.Map(center=[28.0, 72.0], zoom=5, plugin_Draw=True, draw_export=True, locate_control=True)
+    
+    # ✅ Add a basemap explicitly
+    m.add_basemap("HYBRID")      # Google Satellite Hybrid
+    m.add_basemap("ROADMAP")     # Google Roads
+    m.add_basemap("Esri.WorldImagery")
+    m.add_basemap("OpenStreetMap")
 
     # Restore ROI (if available)
     if "roi_geojson" in st.session_state:
@@ -1340,31 +1424,65 @@ def locator_page():
                     st.session_state["sites"][active]["map_snapshot"] = map_bytes
                 st.image(map_bytes, caption="Map Snapshot", use_column_width=True)
 
-            # Display
+            import matplotlib.pyplot as plt
+
+            # Custom theme colors (orange/black/gray)
+            COLOR_SCHEME = ["#FF6600", "#333333", "#FF9933", "#666666", "#FFCC99"]
+            
+            # --- Display results
             st.subheader("📊 Summary")
+            
             st.write(f"**Soil:** {soil_val}")
-            st.write(f"**Elevation:** {elev_val}")
+            st.write(f"**Elevation:** {elev_val} m")
             st.write(f"**Seismic:** {seismic_val}")
             st.write(f"**Flood:** {flood_val}")
             st.write(f"**PM2.5 Index:** {pm25_val}")
-            st.json(lc_stats)
-
+            
+            # --- Landcover pie chart
+            if lc_stats:
+                st.markdown("#### 🌱 Landcover Distribution")
+                labels = list(lc_stats.keys())
+                values = list(lc_stats.values())
+                fig, ax = plt.subplots()
+                ax.pie(values, labels=labels, autopct="%1.1f%%", startangle=90,
+                       colors=COLOR_SCHEME, wedgeprops={'edgecolor': 'white'})
+                ax.set_aspect("equal")
+                st.pyplot(fig)
+            
+            # --- NDVI time series
             if ndvi_ts:
+                st.markdown("#### 🌿 NDVI Trend (2 years)")
                 d, v = zip(*ndvi_ts)
                 fig, ax = plt.subplots()
-                ax.plot(d, v, marker="o"); ax.set_title("NDVI"); ax.set_xlabel("Date")
+                ax.plot(d, v, marker="o", color=COLOR_SCHEME[0])
+                ax.set_title("NDVI", fontsize=12, color=COLOR_SCHEME[1])
+                ax.set_xlabel("Date")
+                ax.set_ylabel("NDVI")
+                ax.grid(True, linestyle="--", alpha=0.6)
                 st.pyplot(fig)
-
+            
+            # --- Precipitation
             if precip_ts:
+                st.markdown("#### 🌧️ Precipitation Trend (1 year)")
                 d, v = zip(*precip_ts)
                 fig, ax = plt.subplots()
-                ax.plot(d, v, marker="o", color="blue"); ax.set_title("Precipitation (mm)"); ax.set_xlabel("Date")
+                ax.plot(d, v, marker="o", color=COLOR_SCHEME[2])
+                ax.set_title("Precipitation (mm)", fontsize=12, color=COLOR_SCHEME[1])
+                ax.set_xlabel("Date")
+                ax.set_ylabel("Rainfall (mm)")
+                ax.grid(True, linestyle="--", alpha=0.6)
                 st.pyplot(fig)
-
+            
+            # --- Temperature
             if temp_ts:
+                st.markdown("#### 🌡️ Land Surface Temp (1 year)")
                 d, v = zip(*temp_ts)
                 fig, ax = plt.subplots()
-                ax.plot(d, v, marker="o", color="red"); ax.set_title("LST Temperature (K)"); ax.set_xlabel("Date")
+                ax.plot(d, v, marker="o", color=COLOR_SCHEME[3])
+                ax.set_title("LST Temperature (K)", fontsize=12, color=COLOR_SCHEME[1])
+                ax.set_xlabel("Date")
+                ax.set_ylabel("Kelvin (K)")
+                ax.grid(True, linestyle="--", alpha=0.6)
                 st.pyplot(fig)
 
 
@@ -1391,12 +1509,11 @@ def load_faiss():
 
 vectorstore = load_faiss()
 
-
 # -------------------
 # RAG Chat Page
 # -------------------
 def rag_page():
-    st.header("🤖 GeoMate Ask (RAG + Groq)")
+    st.header("🤖 GeoMate Ask (RAG + LLM)")
     site = st.session_state["sites"][st.session_state["active_site"]]
 
     # --- Ensure Site ID exists ---
@@ -1466,8 +1583,8 @@ def rag_page():
                 f"If user provides numeric engineering values, return them in the format: [[FIELD: value unit]]."
             )
 
-            # Call your Groq model (replace placeholder with real call)
-            resp = groq_generate(prompt, model=st.session_state["llm_model"], max_tokens=500)
+            # Call the unified LLM function
+            resp = llm_generate(prompt, model=st.session_state["llm_model"], max_tokens=500)
             
             # Save bot reply
             st.session_state["rag_history"][site_id].append({"who": "bot", "text": resp})
@@ -1544,45 +1661,69 @@ from reportlab.lib.units import mm
 
 
 # ----------------------------
-# LLM Helper (Groq API)
+# LLM Helper (Reports Analysis with Orchestration)
 # ----------------------------
-def groq_llm_analyze(prompt: str, section_title: str,
-                     model_name: str = "deepseek-r1-distill-llama-70b",
-                     max_tokens: int = 500) -> str:
+def groq_llm_analyze(prompt: str, section_title: str, max_tokens: int = 500) -> str:
     """
-    Query Groq API for a humanized explanation/analysis of one report section.
-    Returns plain text.
+    Query the selected model (Groq / Gemini / DeepSeek).
+    If token limit or error occurs, automatically switch to backup models
+    and continue the analysis seamlessly.
     """
-    import requests
 
-    key = None
-    if "GROQ_API_KEY" in st.secrets:
-        key = st.secrets["GROQ_API_KEY"]
-    else:
-        key = st.session_state.get("GROQ_API_KEY")
-
-    if not key:
-        return f"[LLM unavailable] {section_title}: {prompt[:200]}..."
-
-    # ✅ FIX: use /openai/v1 instead of /v1
-    url = "https://api.groq.com/openai/v1/chat/completions"
-    headers = {"Authorization": f"Bearer {key}", "Content-Type": "application/json"}
-    payload = {
-        "model": model_name,
-        "messages": [
-            {"role": "system", "content": "You are GeoMate, a geotechnical engineering assistant. Respond professionally with concise analysis and insights."},
-            {"role": "user", "content": f"Section: {section_title}\n\nInput: {prompt}\n\nWrite a professional engineering analysis for this section."}
-        ],
-        "temperature": 0.2,
-        "max_tokens": max_tokens,
-    }
-    try:
-        resp = requests.post(url, headers=headers, json=payload, timeout=60)
-        resp.raise_for_status()
-        data = resp.json()
-        return data["choices"][0]["message"]["content"].strip()
-    except Exception as e:
-        return f"[LLM error in {section_title}] {e}"
+    # Primary model (user choice from sidebar)
+    model_chain = [st.session_state.get("llm_model", "groq/compound")]
+
+    # Add fallback chain (priority order: DeepSeek → Gemini → Groq)
+    if "deepseek" not in model_chain:
+        model_chain.append("deepseek-r1-distill-llama-70b")
+    if "gemini" not in model_chain:
+        model_chain.append("gemini-1.5-pro")
+    if "groq/compound" not in model_chain:
+        model_chain.append("groq/compound")
+
+    system_message = (
+        "You are GeoMate, a geotechnical engineering assistant. "
+        "Respond professionally with concise analysis and insights."
+    )
+
+    full_prompt = (
+        f"{system_message}\n\n"
+        f"Section: {section_title}\n\n"
+        f"Input: {prompt}\n\n"
+        f"Write a professional engineering analysis for this section."
+    )
+
+    final_response = ""
+    remaining_prompt = full_prompt
+
+    # Try each model in the chain until completion
+    for model_name in model_chain:
+        try:
+            response = llm_generate(remaining_prompt, model=model_name, max_tokens=max_tokens)
+
+            if not response or "[LLM error" in response:
+                # If failed, continue to next model
+                continue
+
+            final_response += response.strip()
+
+            # If response length is close to max_tokens, assume it cut off → continue with next model
+            if len(response.split()) >= (max_tokens - 20):
+                # Add continuation instruction for the next model
+                remaining_prompt = (
+                    f"Continue the analysis from where the last model stopped. "
+                    f"So far the draft is:\n\n{final_response}\n\n"
+                    f"Continue writing professionally without repeating."
+                )
+                continue
+            else:
+                break  # Finished properly, exit loop
+
+        except Exception as e:
+            final_response += f"\n[LLM orchestration error @ {model_name}: {e}]\n"
+            continue
+
+    return final_response if final_response else "[LLM error: All models failed]"
 
 
 # =============================
@@ -1726,10 +1867,6 @@ def build_classification_pdf(
     doc.build(elems)
     return filename
 
-
-# -------------------------------
-# Reports Page
-# -------------------------------
 # -------------------------------
 # Reports Page
 # -------------------------------